Noisy neuronal populations effectively encode sound localization in the dorsal inferior colliculus of awake mice
- Cell Biology and Biophysics Unit, European Molecular Biology Laboratory, Germany
- Epigenetics and Neurobiology Unit, European Molecular Biology Laboratory, Italy
- Université Paris Cité, Institut Pasteur, AP-HP, Inserm, Fondation Pour l'Audition, Institut de l’Audition, IHU reConnect, France
- Developmental Biology Unit, European Molecular Biology Laboratory, Germany
- Molecular Medicine Partnership Unit, European Molecular Biology Laboratory, Germany
- Interdisciplinary Center for Neurosciences, Heidelberg University, Germany
Figures
Figure 1 with 2 supplements
Simultaneous recording of DCIC population responses to sound azimuth through sTeFo-2P Ca2+ imaging and neuropixels probes.
(A) Representative histological section showing AAV transduced jRGECO1a expression across IC. Middle panel inset: Contrast enhanced commissure region of the slice to visualize commissural projections from jRGECO1a expressing IC neurons. Bottom panel: Dotted lines delimit anatomical IC regions according to Paxinos and Franklin, 2001; dashed lines delimit approximate area targeted for imaging. Scale bar: 200 μm. DCIC: dorsal cortex from inferior colliculus. CNIC: Central nucleus from inferior colliculus. ECIC: External cortex from inferior colliculus. Com.: Commissure from inferior colliculus. (B) Schematic representation of the experimental design, incorporating sTeFo 2 P for Ca2+ imaging. (C) Neuropil-corrected and denoised jRGECO1a signals extracted from a representative full dataset. Extracted signals are arranged from dorsal (top) to ventral (bottom) ROI position across the DCIC volume imaged. (D) Representative neuropil corrected and denoised jRGECO1a traces (blue) with their corresponding spike probability traces (gray) and stimulation epochs (color-coded based on stimulus azimuth angle according to (B)) super-imposed. (E) Representative simultaneous recording of a DCIC population from an awake, passively listening mouse, displaying spontaneous, on-going activity (not synchronized to stimulation, arrowheads) and variable sound-evoked response patterns (during sound stimuli). Top trace is the population average response. Sound stimulation epochs are color-coded based on azimuth. (F) Representative histological section showing DiI labeled neuropixels electrode tract across IC. Dotted lines delimit anatomical IC regions according to Paxinos and Franklin, 2001. Scale bar: 500 μm. (G) Same as (B) but representing integration with electrophysiological recording of DCIC population activity with a neuropixels probe. (H) Representative high pass filtered (>300 Hz) voltage traces simultaneously recorded from 100 channels spanning across DCIC in a neuropixels probe shank during an experiment displaying clear unit waveforms captured across neighbouring channels. (I) Same as (E) but showing a representative raster plot of the spike sorted DCIC single-unit activity simultaneously recorded during an experiment.
Figure 1—figure supplement 1
Examples of extracted signals for imaging and electrophysiology, together with sound frequency tuning at DCIC.
(A) Representative median intensity projection from one sTeFo-2P imaging plane timelapse, extracted from one of our volumetric imaging datasets (left panel, scale bar 100 μm) with superimposed footprints from CaImAn segmented and manually curated ROIs corresponding to recorded units (right panel). (B) Representative raw dF/F0 traces extracted from CaImAn segmented and manually curated ROIs (units, blue) with CaImAn neuropil corrected and denoised output superimposed (red). (C) Schematic representation of a neuropixels probe shank location across DCIC (approximated histologically) highlighting in different colors the channels across which curated single-units were detected after spike sorting with kilosort 2.5 and manual curation with Phy (see methods). (D) 100 superimposed traces from aligned spikes recorded from the color coded channels highlighted in (C), displaying spike amplitude and somatic waveform consistency and spatial attenuation. (E) Correlogram plots from the curated single-units shown in (C,D), displaying very few refractory period violations (vertical dotted lines: 2ms refractory period time window considered). (F) Representative responses to sound frequency (pure tone stimuli) of a frequency-sensitive single-unit. Top panels show the peri-stimulus time histograms and the bottom panels show the corresponding spike raster plots across trials. (G) Representative sound frequency tuning curves from two DCIC sound frequency dependent single-units recorded with neuropixels. Mean and standard deviation are plotted. (H) Relationship between the best frequency of DCIC sound frequency dependent single-units and recording depth. Data from n=4 mice, point color corresponds to the same mouse. X axis is in logarithmic scale. (I) Neuronal signal to noise level (nS/N) histograms from neuropixels recorded DCIC single-units in the absence of sound stimuli (on going activity, gray) and in response to pure tone stimuli trials at the best frequency for all collected single-units (blue) and the sound frequency sensitive subset (red).
Figure 1—figure supplement 2
Relationship between DCIC on-going activity and face movements.
(A) Schematic representation of the strategy employed to videographically track face movements (ear and snout) during imaging experiments. (B) Representative traces of imaged population activity and ear or snout movement across stimulus azimuth trials. (C) Correlation between the average population activity trace and ear or snout movements across mice. Median value across mice is represented by a horizontal line. (D, E) Histograms of the correlation coefficients between ear or snout movement and the imaged activity of each unit (D) or the top ranked unit subset azimuth dependent, (E) across mice.
Figure 2 with 1 supplement
Decoding imaged single-trial DCIC population responses to sound azimuth.
(A) Schematic representation of the decoding strategy using multi-class classification on the recorded simultaneous population responses. (B) Cumulative distribution plots of the absolute cross-validated single-trial prediction errors obtained using naive Bayes classification (blue) and chance level distribution associated with our stimulation paradigm obtained by considering all possible prediction errors for the 13 azimuths tested (gray). K.S.: Kolmogorov-Smirnov test, n.s.: (p>0.05). (C) Schematic representation of the decoding strategy using the first PCs of the recorded population responses. Inset: % of explained variance obtained using PCA for dimensionality reduction on the complete population responses. Median (blue line) and median absolute deviation (shaded blue area) are plotted for (n=12 mice/imaging sessions). (D) Same as (B) but for decoding different numbers of first PCs from the recorded complete population responses. (E) Significance of classification performance with respect to chance level for different numbers of first PCs, determined via Kolmogorov-Smirnov tests with Sidak correction for multiple comparisons. Arrowhead indicates model loss of performance associated with fitting more parameters for a larger feature space (# PCs) with the same dataset size (# trials collected).
Figure 2—figure supplement 1
Alternative decoding models tested.
(A) Cumulative distribution plots of the absolute cross-validated single-trial prediction errors obtained using different classifiers (blue; KNN: K-nearest neighbors; SVM: support vector machine ensemble) and chance level distribution (gray) on the complete populations of imaged units. Cumulative distribution plots of the absolute cross-validated single-trial prediction errors obtained using a KNN classifier to decode the single-trial response patterns from the 31 top ranked units in the simultaneously imaged datasets across mice (cyan), modeled decorrelated datasets (orange) and the chance level distribution associated with our stimulation paradigm (gray). Vertical dashed lines show the medians of cumulative distributions. K.S. w/Sidak: Kolmogorov-Smirnov with Sidak.
Figure 3
Decoding neuropixels recorded single-trial DCIC population responses to sound azimuth.
(A) Cumulative distribution plots of the absolute cross-validated single-trial prediction errors obtained decoding the complete simultaneously recorded population responses with neuropixels probes across mice (blue) and chance level distribution associated with our stimulation paradigm (gray). K.S.: Kolmogorov-Smirnov test, n.s.: (p>0.05). (B) % of observed variance explained across PC number for the complete population responses recorded with neuropixels. Median (blue line) and median absolute deviation (shaded blue area) are plotted for n=4 mice. (C) Same as (A) but for decoding different numbers of first PCs. (D) Significance of decoding performance shown in (C) with respect to chance level for different numbers of first PCs, determined via Kolmogorov-Smirnov tests with Sidak correction for multiple comparisons. Shaded areas show the corresponding median decoding errors to the points within the area.
Figure 4 with 1 supplement
Sound azimuth information is carried by specific units from the imaged DCIC populations.
(A) Histogram of the nS/N ratios from the recorded units across mice during sound stimulation or during the inter trial periods without sound stimulation (on going). (B) Representative stimulus azimuth tuning curves from units with significant median response tuning detected using non-parametric one-way ANOVA (Kruskal-Wallis test). Median and absolute median deviation are plotted. The imaging depth from the corresponding units is displayed in gray. Azimuth selectivity is color-coded based on Figure 1B. (C) Percentage of the simultaneously recorded units across mice that showed significant median response tuning, compared to false positive detection rate (α=0.05, chance level). (D) Response dependency to stimulus azimuth, determined via χ2 tests (see Materials and methods), for simultaneously recorded units ranked in descending order of significance. Left inset: Representative responses from the top ranked 7 units with significant response dependency to stimulus azimuth. Response amplitudes are displayed with a continuous trace for visualization purposes, the displayed response order was sorted as a function of stimulus azimuth and does not represent the experimental stimulus delivery order (random). Right inset: Same as (A) but for the subset of units displaying response dependency to stimulus azimuth. (E) Percentage of the simultaneously recorded units across mice that showed significant response dependency to stimulus azimuth, compared to false positive detection rate (α=0.05, chance level). (F) Schematic representation of the decoding strategy using the top ranked units from the recorded population responses. (G) Top: Cumulative distribution plot of the absolute cross-validated single-trial prediction errors obtained with a Bayes classifier (N. Bayes, naive approximation for computation efficiency). The number of top ranked units considered for decoding their simultaneously recorded single-trial population response patterns is color coded from cyan (4 top ranked units) to purple (10 top ranked units) and the chance level distribution associated to our stimulation paradigm, obtained by considering all possible prediction errors for the 13 azimuths tested, is displayed in gray. Bottom: Significance of classification performance with respect to chance level for 4–30 decoded top ranked units, determined via Kolmogorov-Smirnov tests with Sidak correction for multiple comparisons. Arrowhead indicates model loss of performance associated with fitting more parameters for a larger feature space (# units) with the same dataset size (# trials collected).
Figure 4—figure supplement 1
Top ranked units are scattered throughout the imaged DCIC volumes.
(A) Histogram plots of the distribution of top ranked unit position in the imaged volume across each anatomical axis (20 μm bins) obtained from all imaged mice, either for the complete sample of units (purple) or the subsample of top ranked units (~32% of all the units imaged per mice, blue). K.S.: Kolmogorov-Smirnov test. (B) Scatter plots of the centroid position throughout the anatomical axes in the imaged volume from the detected top ranked units across mice.
Figure 5
Neuropixels recordings support observations drawn from sTeFo 2 P Ca2+ imaging experiments.
(A) Representative responses to stimulus azimuth of a top ranked unit recorded with neuropixels. Top panels show the peri-stimulus time histograms and the bottom panels show the corresponding spike raster plots across trials. (B) Left: Schematic representation of a neuropixels probe shank highlighting in different colors the position from channels where representative top ranked single-units were detected across DCIC (approximated histologically, same units as displayed in Figure 1—figure supplement 1C—E). Right Representative azimuth tuning curves from three DCIC top ranked single-units recorded with neuropixels, plot colors correspond to position in the shank schematic. Mean and standard deviation are plotted. (C) Neuronal signal to noise level (nS/N) histograms from neuropixels recorded DCIC single-units in the absence of sound stimuli (on going activity, gray) and in response to the best azimuth trials, for all collected single-units (blue) and the top ranked single-unit subset (red). (D) Percentages of sound azimuth dependent units (top ranked units), sound frequency dependent units and both azimuth and frequency dependent units across mice. Median value across mice is represented by a horizontal line. (E) Relationship between azimuth sensitivity and best frequency of DCIC sound frequency and azimuth dependent single-units across mice. Data from n=4 mice, point color corresponds to the same mouse. X axis scale is logarithmic. (F) Cumulative distribution plots of the absolute cross-validated single-trial prediction errors obtained by decoding the responses from different numbers of top ranked units simultaneously recorded with neuropixels probes across mice and chance level distribution associated with our stimulation paradigm (gray). (G) Significance of decoding performance shown in (F) with respect to chance level for different numbers of top ranked units decoded, determined via Kolmogorov-Smirnov tests with Sidak correction for multiple comparisons. Shaded areas show the corresponding median decoding errors to the points within the area. Sample sizes (number of mice) is informed at the top of the graph for each point.
Figure 6
Noise correlations in DCIC population activity contribute to encode sound azimuth.
(A) Simplified schematic representation of the possible effects from (positive) noise correlations on the response separability of a theoretical population consisting of 2 units, and within class randomization strategy to model decorrelated datasets lacking noise correlations. (B, C) Left top: Representative correlation matrices of pairwise correlations between the responses of top ranked units detected in simultaneous recordings during sound stimuli for representative azimuths. The simultaneously imaged units are sorted in the correlation matrices based on cross validated hierarchical clustering (see Materials and methods). Left bottom: Distribution histograms for the pairwise correlation coefficients (Kendall tau) from pairs of simultaneously recorded top ranked units across mice (blue) compared to the chance level distribution obtained through randomization of the temporal structure of each unit’s activity to break correlations (purple). Vertical dashed lines show the medians of these distributions. *: p<0.05, ***: p<0.0001, Kolmogorov-Smirnov with Sidak. Right: Cumulative distribution plots of the absolute cross-validated single-trial prediction errors obtained using a Bayes classifier (naive approximation for computation efficiency) to decode the single-trial response patterns from the 6 (neuropixels) or 7 (sTeFo 2 P imaging) top ranked units in the simultaneously acquired datasets across mice (cyan), modeled decorrelated datasets (orange) and the chance level distribution associated with our stimulation paradigm (gray).
Author response image 1
Percentage of the neuropixels recorded DCIC single units across mice that showed significant median response tuning, compared to false positive detection rate (α = 0.05, chance level).
Additional files
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Noisy neuronal populations effectively encode sound localization in the dorsal inferior colliculus of awake mice
eLife 13:RP97598.
https://doi.org/10.7554/eLife.97598.4
