DGF project 458898724 final report

FINAL REPORT 1 General Information DFG reference number: PR 1803/4-1 Project number: 458898724 Project title: Revealing the Neural Population Code of Auditory Space by Large-Volume Ca2+ Imaging in the Mouse. Name(s) of the applicant(s): Robert Prevedel Official address(es): EMBL, Meyerhofstrasse 1, 69117, Heidelberg, Germany. Name(s) of the co-applicants: Name(s) of the cooperation partners: Hiroki Asari, Brice Bathellier. Reporting period (entire funding period): 03/2022 - 06/2025. 2 Summary Englisch: Sound location coding has been extensively studied at the central nucleus of the mammalian inferior colliculus (CNIC), supporting a population code. However, this population code has not been extensively characterized on the single-trial level with simultaneous recordings or at other anatomical regions like the dorsal cortex of inferior colliculus (DCIC), which is relevant for learninginduced experience dependent plasticity. To address these knowledge gaps, here we made in two complementary ways, large-scale recordings of DCIC populations from awake mice in response to sounds delivered from 13 different frontal horizontal locations (azimuths): volumetric two-photon calcium imaging with 700 cells simultaneously recorded at a relatively low temporal resolution, ∼ and high-density single-unit extracellular recordings with 20 cells simultaneously recorded at a ∼ high temporal resolution. Independent of the method, the recorded DCIC population responses revealed substantial trial-to-trial variation (neuronal noise) which was significantly correlated across pairs of neurons (noise correlations) in the passively listening condition. Nevertheless, decoding analysis supported that these noisy response patterns encode sound location on the single-trial basis, reaching errors that match the discrimination ability of mice. The detected noise correlations contributed to minimize the error of the DCIC population code of sound azimuth. Altogether these findings point out that DCIC can encode sound location in a similar format to what has been proposed for CNIC, opening exciting questions about how noise correlations could shape this code in the context of cortico-collicular input and experience dependent plasticity. Deutsch: Die Ortung von Geräuschen im zentralen Kern des unteren Colliculus (CNIC) von Säugetieren wurde umfassend untersucht, was einen Populationscode unterstützt. Dieser Populationscode wurde jedoch nicht umfassend auf der Ebene einzelner Versuche mit simultanen Aufzeichnungen oder in anderen anatomischen Regionen wie dem dorsalen Kortex des unteren Colliculus (DCIC) charakterisiert, der für lerninduzierte erfahrungsabhängige Plastizität relevant ist. Um diese Wissenslücken zu schließen, haben wir hier auf zwei sich ergänzende Arten groß angelegte Aufzeichnungen von DCIC-Populationen aus wachen Mäusen als Reaktion auf Geräusche aus 13 verschiedenen frontalen horizontalen Positionen (Azimute) vorgenommen: volumetrische ZweiPhotonen-Calcium-Imaging mit 700 Nervenzellen, die gleichzeitig mit einer relativ geringen ∼ zeitlichen Auflösung aufgezeichnet wurden, und hochauflösende EinzelzellExtrazelluläre Aufzeichnungen mit 20 Zellen, die gleichzeitig mit einer hohen zeitlichen Auflösung aufgezeichnet∼ wurden. Unabhängig von der Methode zeigten die aufgezeichneten DCIC-Populationsreaktionen erhebliche Schwankungen von Versuch zu Versuch (neuronales Rauschen), die in der passiven Hörsituation signifikant zwischen Neuronenpaaren korrelierten (Rauschkorrelationen). Dennoch bestätigte unsere Dekodierungsanalyse, dass diese verrauschten Reaktionsmuster die Schallposition auf Einzelversuchsbasis kodieren und Fehler erreichen, die der Unterscheidungsfähigkeit von Mäusen in-vivo entsprechen. Die erkannten Geräuschkorrelationen trugen dazu bei, den Fehler des DCIC-Populationscodes des Schallazimuts zu minimieren. Insgesamt weisen diese Ergebnisse darauf hin, dass DCIC die Schallortung in einem ähnlichen Format kodieren kann wie es für CNIC vorgeschlagen wurde, was spannende Fragen darüber aufwirft, wie Geräuschkorrelationen diesen Code im Zusammenhang mit kortiko-kollikulären Eingaben und erfahrungsabhängiger Plastizität beeinflussen könnten. 3 Progress Report • Background and objectives of the project Background: Locating a sound source facilitates essential behaviors such as foraging, mating and predator avoidance, thanks to the omni-directionality and long reach of the auditory system (King et al., 2001). In vertebrates, sound localization relies on both binaural cues such as interaural level and time differences (ILD, ITD; Knudsen and Konishi, 1979) and monaural cues including spectral notches (SN; Kulkarni and Colburn, 1998). These sound localization cues are processed independently at brainstem nuclei in the ascending auditory pathway and integrated altogether for the first time at the inferior colliculus (IC; Adams, 1979; Brunso-Bechtold et al., 1981; Gourévitch and Portfors, 2018; Grothe et al., 2010). This makes the IC a crucial early relay of the ascending auditory pathway to study how a primary neural representation of auditory space is formed (Gourévitch and Portfors, 2018; Grothe et al., 2010). Furthermore, the IC is also targeted by cortico-fugal interactions involved in higher order functions concerning sound localization information like experience-dependent plasticity (Bajo et al., 2019; Bajo et al., 2010; Bajo and King, 2012), supporting that IC plays a complex part in the sound localization processing network, contributing from primary representation to shaping behavior. Objective: We aimed to implement Large-Volume Ca2+ Imaging of neuronal population activity at a subdivision of the IC that is optically accessible, the dorsal cortex of inferior colliculus (DCIC) in the Mouse, to interrogate the neuronal population code of sound location, as this is understudied at DCIC. As DCIC is a higher order relay that receives profuse descending corticofugal inputs influencing auditory information processing, including sound location, and generating learning induced plasticity (Bajo et al., 2019; Bajo et al., 2010; Bajo and King, 2012; Lesicko et al., 2022; Winer et al., 2002), elucidating this can uncover new mechanisms relevant for understanding auditory learning in future studies. • Description of the project-specific results and findings. Please refer to the contributions of all applicants as well as other persons involved, cooperation partners, etc. Results that are already generally accessible in published form can be briefly summarised with reference to the publication. Unpublished results are to be described in more detail. The project-specific results and findings are already generally accessible in published form in: Boffi JC et al., (eLife, 2024), https://doi.org/10.7554/eLife.97598.4. Briefly, we produced simultaneous recordings of DCIC population responses to sound location in the horizontal plane (azimuth) through scanned temporal focusing 2-Photon (sTeFo-2P Ca2+) imaging in vivo and validated these data with neuropixels probe recordings. The DCIC population responses collected from passively listening awake mice displayed spontaneous activity that was not time-locked to the stimulation (on-going activity) and sound-evoked response patterns that varied markedly across trials, pointing out that DCIC population responses to sound azimuth are noisy. By monitoring face movements videographically, we observed that the population activity variation correlated to facial movements, specifically of the snout region, pointing out that DCIC populations encode multimodal information. To disentangle sound location responses from this multimodal activity we focused on the fraction of the DCIC populations recorded that carried more azimuth information in their responses, determined as statistical dependency of their responses to the stimulus azimuth. Decoding analysis revealed that small subsets of these neurons carrying higher amounts of azimuth information in their noisy responses would encode sound azimuth redundantly altogether in a population code. Interestingly, correlations in the trial-to-trial response variations from these neurons (noise correlations) positively contributed to the accuracy of the population code, pointing out that in populations with noisy responses like in DCIC, response tuning is not the only feature carrying sound location information, but noise correlations can contribute to this too. Altogether these results lay foundations for studies relying on sound location population coding interrogation at the mammalian DCIC. • Deviations from the original concept; findings that contradict the initial hypothesis The noisy responses from DCIC neurons was quite unexpected and different from previous original studies recording from the central nucleus of IC (CNIC), which constituted an unexpected challenge, as we could not rely on previous decoding models tested at CNIC. This represented a deviation from the original concept of a population code from neurons with average response tuning to the stimulus azimuth, due to the high level of trial-to-trial variation in the responses recorded. To face this challenge, we abandoned average response tuning, as it did not represent the single trial condition, and instead adopted statistical dependency to the stimulus azimuth as a metric for detecting neurons carrying information about stimulus azimuth. Interestingly, this challenge revealed a novel finding in that the correlated trial-to-trial variation in the responses could positively contribute to the accuracy of the DCIC population code for sound azimuth we interrogated. • Activities and approaches to quality-enhancing measures through which the validity or verifiability of your research findings was ensured To enhance the quality of our Large-Volume Ca2+ Imaging datasets, as they were produced with a relatively new imaging methodology that is not yet too widespread (sTeFo 2P) we produced Neuropixels recordings in the same experimental conditions (in the same physical setup as the sTeFo scope) to validate the findings derived from the volumetric imaging datasets. With some minor variation, the same biological conclusions could be drawn from both the imaging and electrophysiological datasets. This was the best approach to validate the findings observed with two independent methodologies to record population activity, with different temporal and spatial resolutions, pointing out that the biological conclusions are independent of the recording modality. Moreover, this controls extensively validate sTeFo 2P as a very promising informative methodology for in vivo high throughput neuronal population activity interrogation. • Description of the handling of research data generated in the project and the data infrastructures used, if any (use the following checklist as a guide: www.dfg.de/research_data/checklist) Research data generated in the project was collected and handled for analysis in local storage devices from PCs and the EMBL file server. After publication of the outcomes of the project in preprint form datasets and code supporting the study were made publicly available in BioStudies: https://www.ebi.ac.uk/biostudies/bioimages/studies/S-BIAD1064, and archived in tapes at the EMBL IT department in Heidelberg. • Description of any research data, methods, standards, software or infrastructures generated in the project that are re-usable and openly accessible to others Research data and code generated in this project are re-usable and publicly available from: https://www.ebi.ac.uk/biostudies/bioimages/studies/S-BIAD1064. Methods are described in: https://doi.org/10.7554/eLife.97598.4 Microscopy infrastructure for building an sTeFo 2P was described in detail in: https://doi.org/10.1038/nmeth.4040 The sTeFo 2P microscope built for this study is available at EMBL Heidelberg and openly accessible to collaborators upon agreement with the Prevedel Lab. • Implementation of scientific events, science communication measures The outcomes of the project were communicated at the FENS forum 2022 in Paris, in poster format, at the EMBO workshop on subcortical sensory circuits 2023, in talk format, to an international general neuroscience audience. Furthermore, the project results were presented as a seminar at the UCL Ear Institute in London, The Francis Crick Institute and the Physiology Department of Oxford University in Oct-Nov 2023. In addition, the work has been presented locally at EMBL internal meetings and seminars and at MMPU seminars at Heidelberg University. • Bibliography (list of works you referred to in describing the scientific results generated by the project and putting these in context. This might include your own work and that of other researchers.) Adams JC (1979) Ascending projections to the inferior colliculus. The Journal of Comparative Neurology 183:519–538. Bajo VM, Nodal FR, Moore DR, King AJ (2010) The descending corticocollicular pathway mediates learning-induced auditory plasticity. Nature Neuroscience 13:253–260. Bajo VM, King AJ (2012) Cortical modulation of auditory processing in the midbrain. Frontiers in Neural Circuits 6:114. Bajo VM, Nodal FR, Korn C, Constantinescu AO, Mann EO, King AJ (2019) Silencing cortical activity during sound-localization training impairs auditory perceptual learning. Nature Communications 10:3075. Brunso-Bechtold JK, Thompson GC, Masterton RB (1981) HRP study of the organization of auditory afferents ascending to central nucleus of inferior colliculus in cat. The Journal of Comparative Neurology 197:705–722. Gourévitch, B Portfors CV (2018) Subcortical pathways: towards a better understanding of auditory disorders. Hearing Research 362:48–60. Grothe B, Pecka M, McAlpine D (2010) Mechanisms of sound localization in mammals Physiological Reviews 90:983–1012. King AJ, Schnupp JWH, Doubell TP (2001) The shape of ears to come: dynamic coding of auditory space. Trends in Cognitive Sciences 5:261–270. Knudsen EI, Konishi M (1979) Mechanisms of sound localization in the barn owl (Tyto alba) Journal of Comparative Physiology? A 133:13–21. Kulkarni A, Colburn HS (1998) Role of spectral detail in sound-source localization Nature 396:747–749. 4 Published Project Results 4.1 Category A – Articles in peer-reviewed journals, contributions to peer-reviewed conferences or to anthology volumes, and book publications Open access: Boffi JC, Bathellier B, Asari H, Prevedel R. Noisy neuronal populations effectively encode sound localization in the dorsal inferior colliculus of awake mice. eLife (2024) doi: 10.7554/eLife.97598.1 4.2 Category B – Any other form of published results Open acces dataset associated to Boffi et al., 2024: https://www.ebi.ac.uk/biostudies/bioimages/studies/S-BIAD1064 Open access preprints: P. Aymard, J-C. Boffi, H. Asari, R. Prevedel, D. Holcman Column-Like Subnetwork Reconstruction in Motor Cortex from Graph-Based 3D High-Density TwoPhoton Calcium Imaging BioRxiv (2025) doi: 10.1101/2025.06.17.660119 Kamm GB*, Boffi JC*, Abd El Hay MY*, Rajot D, Cukić A, Havenith MN, Scholvinck M, Renier N, Asari H, Prevedel R Central infusion of prostaglandin E2 reveals a unified representation of sickness in the mouse insular cortex bioRxiv (2025) doi: 10.1101/2025.04.28.651028 *Shared first authorship Conference poster: Juan C. Boffi, Brice Bathellier, Hiroki Asari & Robert Prevedel Elucidating the neuronal population code of sound location at the inferior colliculus of awake mice using a fast volumetric calcium imaging approach. Federation of European Neuroscience Societies (FENS forum), 2022, Paris. Conference talk: Juan C. Boffi Effective population coding of sound location by noisy neurons from the dorsal cortex of inferior colliculus in awake, passively listening mice. EMBO Workshop on subcortical sensory circuits, November 2023, Heidelberg. 4.3 Patents (applied for and granted) None.