Periodic Reporting for period 5 - IN-Fo-trace-DG (Role of GABAergic interneurons in the formation of new memory traces in the Dentate Gyrus ofbehaving mice)
Okres sprawozdawczy: 2025-01-01 do 2025-06-30
A central question in memory neuroscience is how the hippocampus distinguishes between similar but discrete information and encodes them as independent memories. Theories proposed that the dentate gyrus (DG), as the input gate of the hippocampus has been thought to be the primary area achieving this function, but experimental evidences were lacking. The DG receives a rich input stream from the entorhinal cortex via the performant path, which carries information of various modalities of external cues (e.g. shape, colour, of the car). The DG has been thought to segregate this input into non-overlapping sparse memories; a process called pattern separation and allows thereby a high resolution of information. Consistent with the sparse coding theory, granule cells (GCs), the glutamatergic principal cells in the DG, discharge at low mean frequency. However, how sparse coding is realized on the level of the DG network has been controversially discussed. We proposed that particularly GABAergic inhibitory interneurons have a large impact on population activity in the DG network by means of their inhibitory output synapses. However, how cell associations representing the external world emerge in space and time and how interneurons may contribute to this process was largely unknown. IN-Fo-Trace-DG tackled this problem by recording population activity in the intact DG at high temporal and spatial resolution using 2-Photon imaging of head-fixed mice performing pattern separation tasks in virtual-realities.
We demonstrated cell assemblies in the DG, representing virtual environments, emerge slowly across ~3 subsequent days of spatial learning and once formed, are highly stable over months. DG interneurons contribute substantially to the formation of cell assemblies during spatial learning. Perisoma-targeting interneurons define spike timing, whereas dendrite-targeting interneurons ensure non-overlapping contextual representations. This kind of ‘division of labor’ among interneuron types enables GC populations to discriminate between contexts. Moreover, interneurons undergo structural and functional plasticity that support their dynamic integration into the circuitry in dependence of experience. Taken together, these findings are of fundamental relevance for both basic and translational neuroscience, as several diseases, including Alzheimer’s disease and schizophrenia, have been linked to interneuron dysfunctions in association with impairments in distinguishing between familiar and novel experiences.
We demonstrated cell assemblies in the DG, representing virtual environments, emerge slowly across ~3 subsequent days of spatial learning and once formed, are highly stable over months. DG interneurons contribute substantially to the formation of cell assemblies during spatial learning. Perisoma-targeting interneurons define spike timing, whereas dendrite-targeting interneurons ensure non-overlapping contextual representations. This kind of ‘division of labor’ among interneuron types enables GC populations to discriminate between contexts. Moreover, interneurons undergo structural and functional plasticity that support their dynamic integration into the circuitry in dependence of experience. Taken together, these findings are of fundamental relevance for both basic and translational neuroscience, as several diseases, including Alzheimer’s disease and schizophrenia, have been linked to interneuron dysfunctions in association with impairments in distinguishing between familiar and novel experiences.
By investigating the complex functions of interneurons in the DG, my team has made a far-reaching contribution to memory research. Our main findings are as follows:
Lifetime memories in the mouse DG. We provide first evidence that GC assemblies, representing virtual environments, emerge slowly across ~3 subsequent days of spatial learning and once formed, are highly stable over months (Hainmueller & Bartos, Nature 558, 2018; Cholvin et al., Neuron 109, 2021; Cholvin & Bartos, Nature Com 13, 2022). In marked contrast, spatial representations are highly dynamic in downstream hippocampal areas CA1-3. We propose that lifetime memories of the external world might act as reference frame, which can be combined with the varying details of the environment (see also Muysers et al., Nature Com 15, 2024; Cell Rep 44, 2025).
Novelty detection by mossy cells. The second major principal cell type of the DG are mossy cells. By applying data-driven machine-learning tools (e.g. decoders, generalized linear models, population geometry analysis), we showed that mossy cells rapidly reconfigure their activity in response to environmental changes indicating their role in novelty detection (Huang et al., Cell Reports 24, 2024).
Division of labor among interneuron types in shaping memories. By dissecting the role of synaptic inhibition in population dynamics of the DG, we provide first evidence that parvalbumin-expressing, perisoma-targeting interneurons (PVIs) and somatostatin-positive, dendrite-targeting interneurons (SOMIs) play key roles in shaping memory traces during learning. PVIs support spike timing and facilitate generalization across environments, whereas SOMIs limit GC activity and enhance context discrimination at the level of GC populations (Hainmueller & Cazala et al., Nature Com 15, 2024).
Interneuron plasticity supports mnemonic functions. We dissected the role of interneuron plasticity in the DG by designing a Cre-dependent shRNA to allow cell type-specific knockdown of the metabotropic glutamate receptor 1- α (mGluR1 α) selectively in SOMIs. We show that loss of mGluR1α prevents long-term plasticity at glutamatergic inputs onto SOMIs and impairs memory on object locations (Grigoryan et al., PNAS USA 120, 2023).
Interneurons predict experience-dependent goal locations. We provide key evidence that SOMI activity predicts expected reward locations in expert mice, characterized by goal-anticipatory behavior, but not in non-experts. Moreover, predictive goal coding is rapidly lost once rewards are no longer available indicating that SOMIs encode current and past experiences to bias behavioral outcomes (Yuan et al., Nature Com 16:2025).
PVIs show structural plasticity in relation to novel experiences. By applying a broad set of techniques, including electron microscopy, labeling of pre- and postsynaptic partners with eGRASP, in vitro whole-cell recordings, and behavioral analysis, we demonstrated that DG PVIs possess dendritic spines that undergo novelty-related structural dynamics. Spine growth boosts functional integration of PVIs into the DG circuitry upon novel experiences (Kaufhold et al., Cell Rep 43, 2024).
Our work was presented at several national and international conferences in the form of posters, lectures, and keynote talks. It was published in high-impact journals and communicated to the public through newspapers and participation in science outreach events. In this way, we disseminated our findings to both the scientific community and the public.
Lifetime memories in the mouse DG. We provide first evidence that GC assemblies, representing virtual environments, emerge slowly across ~3 subsequent days of spatial learning and once formed, are highly stable over months (Hainmueller & Bartos, Nature 558, 2018; Cholvin et al., Neuron 109, 2021; Cholvin & Bartos, Nature Com 13, 2022). In marked contrast, spatial representations are highly dynamic in downstream hippocampal areas CA1-3. We propose that lifetime memories of the external world might act as reference frame, which can be combined with the varying details of the environment (see also Muysers et al., Nature Com 15, 2024; Cell Rep 44, 2025).
Novelty detection by mossy cells. The second major principal cell type of the DG are mossy cells. By applying data-driven machine-learning tools (e.g. decoders, generalized linear models, population geometry analysis), we showed that mossy cells rapidly reconfigure their activity in response to environmental changes indicating their role in novelty detection (Huang et al., Cell Reports 24, 2024).
Division of labor among interneuron types in shaping memories. By dissecting the role of synaptic inhibition in population dynamics of the DG, we provide first evidence that parvalbumin-expressing, perisoma-targeting interneurons (PVIs) and somatostatin-positive, dendrite-targeting interneurons (SOMIs) play key roles in shaping memory traces during learning. PVIs support spike timing and facilitate generalization across environments, whereas SOMIs limit GC activity and enhance context discrimination at the level of GC populations (Hainmueller & Cazala et al., Nature Com 15, 2024).
Interneuron plasticity supports mnemonic functions. We dissected the role of interneuron plasticity in the DG by designing a Cre-dependent shRNA to allow cell type-specific knockdown of the metabotropic glutamate receptor 1- α (mGluR1 α) selectively in SOMIs. We show that loss of mGluR1α prevents long-term plasticity at glutamatergic inputs onto SOMIs and impairs memory on object locations (Grigoryan et al., PNAS USA 120, 2023).
Interneurons predict experience-dependent goal locations. We provide key evidence that SOMI activity predicts expected reward locations in expert mice, characterized by goal-anticipatory behavior, but not in non-experts. Moreover, predictive goal coding is rapidly lost once rewards are no longer available indicating that SOMIs encode current and past experiences to bias behavioral outcomes (Yuan et al., Nature Com 16:2025).
PVIs show structural plasticity in relation to novel experiences. By applying a broad set of techniques, including electron microscopy, labeling of pre- and postsynaptic partners with eGRASP, in vitro whole-cell recordings, and behavioral analysis, we demonstrated that DG PVIs possess dendritic spines that undergo novelty-related structural dynamics. Spine growth boosts functional integration of PVIs into the DG circuitry upon novel experiences (Kaufhold et al., Cell Rep 43, 2024).
Our work was presented at several national and international conferences in the form of posters, lectures, and keynote talks. It was published in high-impact journals and communicated to the public through newspapers and participation in science outreach events. In this way, we disseminated our findings to both the scientific community and the public.
In summary, IN-Fo-Trace-DG substantially changed our understanding of DG circuit function. In particular, it shed new light on the emergence, differentiation, and reliability of spatial representations formed by GC assemblies compared to neighboring hippocampal areas CA1-3. Our data demonstrate that the DG provides a highly stable spatial code that serves as a blueprint for the formation of specific engrams of temporally varying contents and contexts in downstream hippocampal areas. Such an encoding scheme enables the association of memories acquired within the same global environment while still allowing discrimination between slightly different or temporally separated instances of those memories. We further highlight the roles of different interneuron types in the DG, demonstrating that they dynamically shape the activity of both individual cells and GC populations during learning. Thus, interneurons do not only counterbalance excitation to prevent runaway activity but also encode information and play an active role in the emergence of new memory traces within the DG. By contributing critically to pattern separation processes in the DG, they are essential for the formation and discrimination of lifetime memories.