Research at ISTA
Since joining ISTA in 2010, Peter Jonas and his research group planned, executed, and published three significant projects, each funded by substantial grants from the European Research Council and/or the Austrian Science Fund (FWF).
During the initial phase at ISTA, our research focused on the function of GABAergic interneurons, particularly parvalbumin-expressing interneuron subtypes. We specifically investigated the subcellular properties of this important class of interneurons. Our findings revealed that the dendrites of those cells show a low-density of voltage-gated Na+ channels, but a high density of voltage-gated K+ channels. This shapes the somatodendritic integration of the neuron, conveying distributed coincidence detector properties to this cell during activation of synaptic inputs (Hu et al., 2010, Science). We also examined, for the first time, the axons of these cells and discovered that they are enriched in voltage-gated Na+ channels. It is generally thought that high density of voltage-gated Na+ channels is important to convey reliability to action potential propagation along the axon. However, we found that voltage-gated Na+ channels were expressed at “supercritical” density, higher than the absolute minimal level needed to convey reliability of propagation (Hu and Jonas, 2014, Nature Neuroscience). Combining modeling and experimental analysis, we discovered that supercritical Na+ channel density in the axon of these neurons accelerates propagation of the action potential along the axon arbor. Furthermore, we examined the energy efficiency of action potential initiation and propagation in parvalbumin+ interneurons. Our results indicated that the energy efficiency of the action potential was higher than expected based on the short action potential duration, which is crucial for the cell’s energy budget, given its ability to generate high-frequency action potential trains under both in vitro and in vivo conditions (Hu et al., 2018, Neuron). Taken together, these results indicate that both dendritic and axonal properties of parvalbumin-expressing interneurons are specialized, designed to combine fast signaling with high energy efficiency. The project was funded by an ERC Advanced Grant, entitled “Nanophysiology of fast-spiking, parvalbumin-expressing GABAergic interneurons” (2011–2016). We have summarized these results in a review published in Science, and this publication has attracted more than thousand citations (Hu et al., 2014). Through the work carried out by the Jonas group at ISTA, the parvalbumin+ GABAergic interneuron has become one of the most thoroughly characterized neurons in the scientific record.
Subsequently, our research concentrated on the biophysical mechanisms and circuit function of the hippocampal mossy fiber synapse, a key glutamatergic synapse in the trisynaptic circuit of the hippocampus. In addition to the subcellular patch-clamp recording techniques for direct recording from presynaptic terminals that we had already developed, we also innovated techniques to noninvasively stimulate single presynaptic terminals in the so-called bouton-attached configuration (Vandael et al., 2021, Nature Protocols). This advancement enabled us, for the first time, to examine the properties of hippocampal mossy fiber synaptic transmission and plasticity at the single-synapse level. Using this approach, we discovered several novel features of transmission and plasticity at the hippocampal mossy fiber synapses. First, we identified unexpected mechanisms of presynaptic plasticity. We found that a loose coupling configuration between presynaptic Ca2+ channels and release sensors seems to be a unique feature of transmission at this synapse (Vyleta and Jonas, 2014, Science). Intriguingly, we discovered that the unique loose coupling configuration enhances the effects of Ca2+ buffer saturation, which contributes to the large facilitation highly characteristic for this synapse. As a next step, we examined the mechanisms of more long-lasting forms of presynaptic plasticity. At the hippocampal mossy fiber synapse, post-tetanic potentiation (PTP) is very prominent, enhancing the strength of the synapse ~5-fold over a time interval of up to 10 minutes. Through biophysical analysis of synaptic vesicle pool dynamics, we discovered that PTP not only increases release probability, as previously thought, but also enhances the size of the readily releasable pool of synaptic vesicles (Vandael et al., 2020, Neuron). Finally, we tried to address the question of whether unitary excitatory postsynaptic potentials (EPSPs) were sufficient to evoke spikes in the postsynaptic cell (Vyleta et al., 2016, eLife; Vandael et al., 2020, Neuron). We found that although unitary EPSPs had large amplitude, they were below the threshold for action potential initiation. However, both synaptic facilitation and PTP enhanced the probability of initiation of postsynaptic spikes. As facilitation and PTP are mechanistically distinct, these results suggest that orthogonal presynaptic plasticity mechanisms enhance the repertoire of synaptic computations at this important synapse. In parallel with the electrophysiological – biophysical analysis of synaptic transmission and plasticity, we also investigated the possibility of structural changes. To examine this, we implemented a technique called “flash and freeze”, which we applied for the first time to acute slice preparations (Borges-Merjane et al., 2020, Neuron). This approach provides a momentary picture of the synapse through instantaneous freezing, which preserves the ultrastructure by forming vitreous rather than crystalline ice. Our findings revealed that stimulation paradigms designed to induce PTP resulted in a significant increase in the number of docked vesicles (Vandael et al., 2020, Neuron). Furthermore, we found that chemical potentiation of synaptic transmission by the adenylylcyclase activator forskolin initiated a comparable increase in the docked vesicle pool size (Kim et al., 2024). In parallel, we investigated the coupling nanotopography using freeze fracture replica labeling, in close collaboration with Ryuichi Shigemoto and Walter Kaufmann at ISTA. Specifically, we determined the distance between presynaptic P/Q-type Ca2+ channels and Munc13-1 immunoreactivity, a putative marker of primed vesicles. Our findings showed that the nearest neighbor distance and the weighted pairwise distance were reasonably consistent with the coupling distance measured in biophysical experiments with exogenous Ca2+ chelators (Kim et al., 2024). Interestingly, we also found a slight reduction in the coupling distance after chemical potentiation. This implies that the coupling configuration is not constant, but modulated in an activity- and plasticity-dependent manner. Our results suggested a model in which the formation of vesicle pool engrams could contribute to both short- and long-term synaptic plasticity. They also revealed the biophysical basis of several forms of presynaptic plasticity, as well as the relation between functional changes and structural alterations at this central glutamatergic synapse. The project was funded by an ERC Advanced Grant, entitled “Biophysics and circuit function of a giant cortical glutamatergic synapse” (2017–2022). We have summarized several of the new results in a review article recently published in Science, which has received a lot of attention in the scientific community (Vandael and Jonas, 2024).
In parallel, we developed several new techniques. These included methods for high resolution whole-cell patch-clamp recordings from neurons in vivo (Zhang et al., 2020, Neuron), as well as methodologies for rabies virus-mediated circuit mapping (Ben-Simon et al., 2022, Nature Communications; Sumser et al., 2022, eLife). We also advanced methodologies for expansion superresolution microscopy, in close collaboration with Johann Danzl at ISTA, and techniques for electron microscopy (Borges-Merjane et al., 2020, Neuron; Chen et al., 2024, Neuron; Kim et al., 2024), including functional electron microscopy, “flash and freeze”, and freeze fracture replica labeling. These innovations were made possible by funding from the Wittgenstein award of the FWF, which allowed maximal flexibility in planning and executing experiments. The broadening of the technological platform over the past few years provides a solid basis for upcoming projects in the next five years, which will crucially depend on the combinatorial application of several cutting-edge electrophysiological and morphological techniques.
Future research
In the next five years, we will concentrate on the connectivity, unitary synaptic transmission, and circuit function of the hippocampal CA3 network. This network is of great importance for several reasons. First, the CA3 region plays a crucial role in the memory function of the hippocampus. For example, bilateral lesions of the CA3 or suppression of synaptic output from CA3 neurons affect episodic-like or contextual memory. Second, CA3 pyramidal neurons are highly connected to each other, forming the largest recurrent network throughout the brain (Guzman et al., 2016, Science; Watson et al., 2025). Thus, recurrent collateral synapses could represent a huge-capacity system for engram storage. Third, CA3–CA3 synapses show “Hebbian” spike-timing dependent plasticity with a broad induction window (Mishra et al., 2016, Nature Communications). These properties appear to be ideal for engram formation. Fourth, network models of the CA3 region suggested that the hippocampal CA3 can not only store information through Hebbian synaptic plasticity, but also recall this information through a process known as pattern completion. Thus, the CA3 circuit forms a content-addressable memory system in which synaptic engrams are rapidly and reliably retrieved. Finally, engram neurons in the dentate gyrus (granule cells) directly innervate CA3 pyramidal neurons via the powerful “detonator” mossy fiber synapses (Vyleta et al., 2016, eLife; Vandael and Jonas, 2024, Science). Thus, activity in engram cells (granule cells) is expected to directly translate into firing of downstream CA3 pyramidal neurons.
To date, there is substantial evidence that the CA3 network plays a critical role in learning, memory, and behavior. However, compared to other brain regions, the function of this specific circuit remains poorly understood. Therefore, over the next 5 years we intend to focus our research on uncovering the functional and structural properties of this important circuit. Our specific goals are the following:
First, we want to characterize the functional and structural connectivity in the CA3 circuit. To address this question, we will combine multicellular patch clamp-based circuit mapping with expansion superresolution microscopy, in close collaboration with Johann Danzl (Velicky et al., 2023, Nat. Methods; Michalska et al., 2023, Nat. Biotechnology; Watson et al., 2025). Through this combined approach, we will examine the functional and structural connectome of CA3 in three dimensions along the radial (superficial-deep), transverse (proximo-distal), and longitudinal (septo-temporal) axes. In particular, we will try to identify signature features of engram formation in the circuit, including correlations of structural properties of synapses formed by the same axon on the same postsynaptic target cell, as previously suggested in Sejnowski’s work. We also intend to label engram cells based on the activity of the immediate early gene c-Fos, and to compare the properties with those of non-engram cells.
Second, we want to obtain more information about the relation between structure and function at CA3–CA3 recurrent collateral synapses. This is critical for understanding the physical and chemical changes during storage of information – in other words, synaptic engrams. We plan to combine recordings from synaptically connected neurons, biocytin labeling of pre-and postsynaptic neurons, and expansion superresolution microscopy of cells and synaptic connections. Through this combination of advanced techniques, we can reliably measure the number of contacts per connection, the number of active zones–postsynaptic densities per contact, previously a domain of electron microscopy, and the nanoscopic determinants of synaptic transmission. In particular, we plan to examine functionally relevant proteins and their effects: Cav2.1 P/Q-type Ca2+ channels as a determinant of release probability, Munc13-1 as a proxy for primed vesicles, and AMPA-type glutamate receptor subunits (GluA1 and GluA2) as a factor presumably determining quantal size.
Third, we will examine the mechanisms of synaptic plasticity at glutamatergic CA3–CA3 synapses at the single synapse level. This set of experiments will be greatly facilitated by our previous developments, which enable us to make simultaneous recordings from up to eight neurons in each slice preparation (Guzman et al., 2016; Watson et al., 2025). Moreover, we propose to perform these experiments in mice, which show a higher synaptic connectivity than rats or humans (Watson et al., 2025). Exploiting the relatively high synaptic connectivity in mice, we will examine convergence motifs in which several presynaptic neurons are connected to a given postsynaptic cell. We will then, for the first time, probe the main characteristics of plasticity: specificity, cooperativity, and associativity of plasticity at CA3–CA3 synapses at the single-synapse level. For specificity and associativity experiments, we will compare responses generated by a potentiated test input to an unpotentiated control input. For cooperativity experiments, we will combine intracellular stimulation of a single presynaptic neuron with optogenetic stimulation of multiple adjacent neurons.
Fourth, we plan to explore the mechanisms of recall of information in the CA3 network under in vivo conditions. In particular, we want to focus on “pattern completion”. Until now, mechanisms of iterative recall and pattern completion have been worked out in models at a high level of mathematical precision. However, whether the recall processes in the natural network are the same than those in the model remains to be determined. To address this question, we will examine the properties of iterative recall in vivo by combination of holographic genetics and Ca2+ imaging in the CA3 network. The results will be important, because they will reveal the properties of pattern completion and engram recall in the real circuit for the first time.
Fifth, we want to compare the properties of synaptic transmission and connectivity in different species, up to the level of humans. This is critical, because we finally want to understand the properties of the human brain, including scaling rules and potential aspects of uniqueness. In particular, we want to test the following hypotheses: that connectivity in different species is inversely related to cell number, that specific features of synaptic transmission contribute to the unique computational power, and that the functional properties of human synapses increase the energy efficiency of the human brain. The already established collaboration with the Neurosurgery Department of the Medical University of Vienna (Karl Rössler) will facilitate these experiments. In particular, the mentioned department provides us access to magnetic resonance imaging-negative patients with minimal structural alterations in the hippocampus, which provides a unique opportunity to examine the human CA3 circuit as closely to the natural conditions as possible.
Finally, we want to develop models of the hippocampal CA3 autoassociative recurrent network. To achieve this, we will employ bottom-up and top-down approaches in parallel. For the bottom-up approach, we will use previously acquired data on single synapse properties, synaptic connectivity, passive and active properties of neurons, and dendritic morphology. Starting from our simplified binary models of pattern completion (Guzman et al., 2016, Science; Watson et al., 2025), we will make the model increasingly more realistic. The final goal is to obtain “CA3 digital twins”, which not only capture the average properties of network parameters, but also the interindividual differences. If correlated with behavioral parameters, these results may help to understand the circuit basis of individuality. In parallel, we want to use a normative approach, in which we vary the properties of a recurrent network to maximize its information content and storage capacity (Navas-Olive et al., 2024). If the two approaches converge, this would lead to a unified understanding of the hippocampal CA3 memory system.
The proposed research requires the combined application of several cutting-edge techniques. Many of these techniques have been already established in the Jonas group, in ISTA’s scientific service units, and in collaborating research teams. Thus, ISTA is the ideal place for executing such ambitious research projects.
In summary, the proposed research will not only provide crucial information about cells, synapses, and microcircuits of the CA3 region, but also potentially address a fundamental question in neurobiology: where, how, and why engrams are formed in the CA3 network. We are confident that, more than 100 years after the proposal of the engram concept by Richard Semon, the “search for the engram” (Lashley, 1950) may come to a successful end.