Diverse calcium dynamics underlie place field formation in hippocampal CA1 pyramidal cells
- Laboratory of Cellular Neurophysiology, HUN-REN Institute of Experimental Medicine, Hungary
- Laboratory of Neuronal Signaling, HUN-REN Institute of Experimental Medicine, Hungary
- Roska Tamás Doctoral School of Sciences and Technology, Faculty of Information Technology and Bionics, Pázmány Péter Catholic University, Hungary
Figures
Figure 1
In vivo two-photon [Ca2+] imaging from head-restrained mice during spatial navigation in a virtual corridor.
(A) Double transgenic mice (GCaMP6s and Cre-dependent td-Tomato) were injected with a diluted AAV expressing Cre-recombinase for sparse td-Tomato labeling. Following craniotomy and cannula implantation above the left dorsal CA1 area, animals were trained and imaged with a two-photon (2P) microscope during navigation in an ~8-m-long virtual corridor. The timeline shows the minimum, (median), and maximum number of days between the procedures. (B) Mean 2P image of an imaging field of view. Example cells shown in panels (E), (G), and (I) are indicated by colored circles. (C) Wall pattern of the virtual corridor consisting of low contrast random checkerboard background with six high contrast visual landmarks. (D) Running speed (mean: orange; ± SD: gray; single session) decreased and lick propensity (blue) increased before the animal reached the reward zone (green area). (E) ROI masks (left; scale bar: 5 µm) and [Ca2+] signals (right) of the cells marked in panel (B) during a single session. Gray segments correspond to odd laps in the virtual environment. (F) Position of the animal along the corridor aligned with the fluorescence activity shown in panel (E). (G, H) Same as panels (E) and (F) on an extended time scale showing a single lap (indicated by * in (E)). The time period when the animals’ running speed was below 5 cm/s is shown in red and was excluded from the analysis. Black vertical bars are inferred activity. (I) Raster plots showing the inferred neuronal activity of three cells (color coded cells in panels (B), (E), and (G)) as a function of the lap number and spatial location of the animal. (J) Coverage of the virtual corridor by place fields (PFs). Tuning curves from cells with at least one significant PF are included. The tuning curves are ordered by the location of their largest peak activity. Left panel: Data from a single session. Right panel: Data from all experiments (22,325 place cells recorded across 163 sessions from 45 mice) was randomly down-sampled to match the sample size of the single session. (K) Distributions of the number of PFs of place cells, PF widths, and PF coverage (proportion of the length) of the virtual corridor. Data are presented for all sessions and for a single session (insets).
Figure 2 with 2 supplements
Comparison of [Ca2+] transients underlying different place field formation (PFF) events.
(A) Behavioral time scale synaptic plasticity (BTSP)-like PFFs. Activity raster plots (left) show three cells with BTSP-like PFFs (red arrows). The middle panels zoom in on the activity of newly formed PFs marked by red arrows. Fluorescence traces (right) are shown during PF traversal before, during (red traces), and after PFF (±5 laps). (B) Same as (A), but for non-BTSP-like PFFs (blue arrows and traces). (C) Comparison of [Ca2+] transients before, during, and after PFF. Fluorescence traces for BTSP-like (left; mean trace in red) and non-BTSP-like (middle, mean trace in blue) PFFs are shown for each lap, from the lap before the formation lap (FL−1) to the fourth lap post-formation (FL+4). Traces were aligned to the midpoint of the steepest rising slope for detectable [Ca2+] transients, or to the time point of PF center traversal for traces lacking transients. Insets: Peak-normalized and peak-aligned mean traces are superimposed. Cumulative distributions of the peak [Ca2+] transient amplitudes for BTSP (red) and non-BTSP (blue) events across each lap are shown on the right. (D) Progression of peak [Ca2+] transient amplitudes (ΔF/F) during individual BTSP (red) and non-BTSP (blue) PFF events. Thin lines connect data points in consecutive laps for each PFF. Black circles with whiskers are lap-wise population means ± SD (BTSP: n = 311 events, non-BTSP: n = 58 events from 29 mice). Statistical analysis revealed that the amplitude of the [Ca2+] transients in the formation lap is significantly higher during BTSP-like compared to non-BTSP-like PFFs and the amplitudes in the other laps are not significantly different (mixed ANOVA: main effect for group (BTSP/non-BTSP): p = 0.038; lap: p < 0.001, interaction: p < 0.001; Tukey post hoc test: formation lap: p < 0.001, all other laps: p > 0.711). *** p < 0.001.
Figure 2—figure supplement 1
Characteristics of place field formation (PFF) and stability.
(A) Histogram of the probability that non-behavioral time scale synaptic plasticity (BTSP) PFFs are misclassified due to random [Ca2+] activity occurring before a BTSP-like PFF event (see Methods), calculated for each non-BTSP PFF. The red dotted line indicates the significance threshold. (B) Fraction of active laps following BTSP (red) and non-BTSP (blue) PFFs (average of 15 laps, p = 0.871, unpaired t-test). (C) Number of new BTSP-like (top row) and non-BTSP-like (bottom row) PFF for individual animals (colored circles) plotted against time elapsed from the start of training (left) and the number of training or imaging sessions (right) in the regular maze.
Figure 2—figure supplement 2
Place field formations (PFFs) in a VR corridor with a complex wall pattern.
(A) A complex visual pattern composed of overlapping shapes, creating a rich visual environment within the VR corridor. (B) Running speed (orange: mean; ± SD: gray, single session) decreased and lick propensity (blue) increased before the animal entered the reward zone (green area). (C) Raster plots showing the inferred neuronal activity of three cells as a function of lap number and spatial location of the animal. (D) Coverage of the virtual corridor by place fields (PFs). Tuning curves from cells with at least one significant PF are included. The tuning curves are ordered by the location of their largest peak activity. Left panel: Data from a single session. Right panel: Data from all experiments (2578 place cells from 12 sessions of 4 mice) was randomly down-sampled to match the sample size of the single session. (E) Activity raster plots (left) show two cells with behavioral time scale synaptic plasticity (BTSP)-like PFFs (red arrows). The middle panels display the isolated activity of newly formed PFs. Fluorescence traces (right) are shown before, during (red traces), and after PFFs (±5 laps). (F) Same as (E), but for non-BTSP-like PFFs (blue arrows and traces). (G) The rates of BTSP- and non-BTSP-like PFFs are comparable between the landmark-dominated and visually complex virtual environments (Chi-square test: p = 0.507).
Figure 3 with 2 supplements
Newly formed place fields (PFs) cover the entire virtual corridor and have non-uniform birth rates.
(A) The tuning curves of behavioral time scale synaptic plasticity (BTSP)- (left) and non-BTSP-like (right) PFs were ordered by the center position of the newly born PFs (white space bins). (B) The width of BTSP-like newly formed PFs is correlated with the animal’s running speed during PFF. Scatter plot shows individual BTSP- (red) and non-BTSP-like (blue) PF formation (PFF) events (Spearman correlations: BTSP (n = 261): R = 0.29, p = 2.7 × 10–6; non-BTSP (n = 47): R = 0.015, p = 0.92). (C) Correlation between the width of PFs formed by BTSP- (red) and non-BTSP-like (blue) mechanisms and the distance from the nearest visual landmark (Spearman correlations: BTSP (n = 261): R = 0.59, p = 1.4 × 10–25; non-BTSP (n = 47): R = 0.67, p = 2.4 × 10–7). (D) Explained variance of PF width by the running speed in the formation lap, distance to nearest visual landmark at formation, or both, for BTSP- (red) non-BTSP- like (blue) or combined (black) PFF events. (E) Histogram of the number of newly formed PFs per single session in the virtual corridor. A few sessions have unusually high (>20) new PFF events. Note that the analysis excludes PFFs in the first 17 laps of the session (Methods). (F) Number of newly formed PFs by laps. Example sessions showing high number of BTSP- (red) and non-BTSP-like (blue) PFFs (upper panels, two of the three sessions with the largest numbers of PFFs in the histogram in panel (E)) and sessions with moderate number of PFFs (lower panels). The histograms are aligned with lap-by-lap population vector (PV) correlograms, which reveal sudden change in population activity for the upper panels and lack of such representational switch for the lower panels. Color scale applies to all panels. (G) [Ca2+] transients of PFF events during representational switches (switch, isolated from sessions with the three highest number of PFF events in the histogram in panel (E)) are compared to PFF events outside representational switches (no switch) for BTSP- (left) and non-BTSP-like (right) events. Traces are shown from the lap preceding PFF (FL−1) to the lap following PFF (FL+1). Traces were aligned to the midpoint of the steepest rising slope, or to the time point at which the PF center was reached (for traces without detectable transients). The mean traces are shown in orange, red, cyan, and blue. Insets: peak-normalized and peak-aligned mean traces. The cumulative distributions show the peak [Ca2+] transient amplitudes for BTSP (with switch: yellow, without switch: red) and non-BTSP (with switch: cyan, without switch: blue) PFF events across each lap. (H) Progression of peak amplitudes (ΔF/F) during individual switch and no switch BTSP (yellow, red) and non-BTSP (cyan, blue) formation events. Individual data points are connected to show the change in peak amplitude across laps for each event. Black circles with whiskers are lap-wise population mean ± SD (BTSP no switch: n = 271, BTSP switch: n = 40, non-BTSP no switch: n = 36, non-BTSP switch: n = 22 events from 29 mice). The peak amplitudes of [Ca2+] transients are not significantly different for the BTSP-like events (mixed ANOVA: main effect for group (switch/no switch): p = 0.457; lap: p < 0.001, interaction: p = 0.214) and for the formation lap of the non-BTSP-like events but are significantly higher for non-BTSP-like events during switches than outside switches in the lap after the formation lap (mixed ANOVA: main effect for group (switch/no switch): p < 0.001; lap: p < 0.001, interaction: p < 0.001; Tukey’s post hoc test: formation lap –1: P=0.999, formation lap: P=0.774, formation lap +1: P<0.001). *** p < 0.001.
Figure 3—figure supplement 1
Factors contributing to place field width and correlations between fluorescence and inferred spike rates.
(A) Activity raster plots of two cells demonstrate large variability in place field (PF) width. (B) Place field width and running speed across virtual corridor locations. (Top) Mean PF width plotted against PF center location. (Bottom) Mean running speed of the animals at corresponding locations (n = 25,852 PFs from 80 sessions of 29 animals). (C) Comparison of inferred spike activity (top) and GCaMP6s fluorescence (bottom). Vertical bars indicate PF boundaries determined from inferred spiking. (D) Scatter plots of peak [Ca2+] transient amplitudes vs. maximum inferred spike rates within PFs (Pearson correlation coefficients: top: 0.95, bottom: 0.88) for the two cells shown in C. Data points are color-coded to correspond to the boxed regions in C. (E) Cumulative distribution of Pearson correlation coefficients between lap-wise maximum [Ca2+] transient and inferred spike activity within PFs (black line; n = 353 cells with PFFs from n = 80 sessions, n = 29 animals). Gray line: null distribution generated by shuffling lap identity labels inside PFs (see Methods).
Figure 3—figure supplement 2
Spontaneous representation switches within a virtual environment.
(A) Lap-by-lap population vector correlogram reveals a significant representation rearrangement around lap 16 (corresponding laps are color coded through A–D). (B) Three place cells showing altered firing locations before and after the representation switch highlighted in (A). (C) Stability of the behavior before and after the representation switch. Running speed (orange line: mean; gray shading: ± SD) and lick propensity (blue line) are shown across the whole session (top), before the switch (middle), and after the switch (bottom). (D) Multiday stability of switching representations. A lap-by-lap population vector correlogram from multiday recordings demonstrates that both the original and switched representations remain stable across multiple days (scale bar: same as in panel A). Note that the switch on Day 4 occurred at lap 11, that is earlier than the period where we analyzed the density of place field formation (PFF) events (≥Lap 18); therefore, this session did not appear as a PF surge in the analysis in Figure 3E. (E) Reversible switches within a single session. Left: Lap-by-lap population vector correlogram revealing multiple reversible switches during one session (scale bar: same as in panel A). Right: The firing location of a place cell is altered parallel with the population representation switch (corresponding laps are color coded).
Figure 4
Similar [Ca²+] transients underlie place field formation (PFF) in familiar and novel environments.
(A) PFF in a familiar environment with a novel extension. A familiar corridor was extended with a new 6-m-long segment. New PFs appear already during the first traversal (top) by both behavioral time scale synaptic plasticity (BTSP)- (red arrow) and non-BTSP-like (blue arrow) dynamics. Bottom, [Ca2+] transients during and after the PFFs marked on the raster plot. (B) Histogram of BTSP- and non-BTSP-like PFFs during the first session in the extended maze region (n = 6 mice). (C) [Ca2+] transients of the formation lap and the following lap (FL+1) of BTSP- (left panels, mean trace in orange) and non-BTSP-like (middle panels, mean trace in cyan) PFF events in the new extension. Insets show peak-normalized, peak-aligned mean traces. Cumulative distributions of the peak [Ca2+] transient amplitudes for BTSP and non-BTSP events across each lap are shown on the right. (BTSPNEW, orange, n = 24, non-BTSPNEW: cyan, n = 16 events) of the novel environment (n = 6 animals), and familiar environment (BTSP: red, non-BTSP: blue; data from Figure 2). (D) Progression of peak [Ca2+] transient amplitudes during BTSP (top) and non-BTSP (bottom) PFF events in the extended maze (new). BTSP and non-BTSP data from Figure 2 (fam) are also shown. Thin lines connect data points in consecutive laps for each PFF. The amplitudes of [Ca2+] transients in the formation lap and the lap after formation were not significantly different between the new and familiar environments for both BTSP- and non-BTSP-like events (two-way mixed ANOVA: main effect for PFF type (BTSP/non-BTSP): p < 0.001, corridor (familiar/new): p = 0.701, type vs. corridor: p = 0.193, lap: p = 0.018, lap vs. type: p < 0.001, lap vs. corridor: p = 0.738, lap vs. type vs. corridor: p = 0.025; Tukey post hoc tests: lap 0: BTSP new vs. familiar: p = 0.999, non-BTSP new vs. familiar: p = 0.999; lap 1: BTSP new vs. familiar: p = 0.633, non-BTSP new vs. familiar: p = 0.780). *** p < 0.001.
Figure 5
Large somatic [Ca2+] events are often not sufficient to evoke novel place fields.
(A) Raster plots of inferred activity (top) and [Ca2+] transients (bottom) vs. spatial location in each lap for three cells reveals behavioral time scale synaptic plasticity (BTSP)-like new place field formations (PFFs; red arrows) as well as large amplitude activities (yellow arrows) which did not evoke new PFs (solitary events). (B) Fluorescence traces around the BTSP-like PFF events (red) marked in panel (A) and larger amplitude [Ca2+] transients not associated with PFF (yellow) within the same sessions are shown. (C) Traces from the lap before (EL−1), during (Event Lap), and the lap after (EL+1) of BTSP-like PFF events (left traces) and solitary non-PF-evoking events (right traces) identified within the same session. In EL−1, traces were aligned to the time point of crossing the center of the PF. Pre- and post-solitary event traces were aligned to the crossing time of the first space bin of the corresponding solitary event. Inset: peak-normalized and peak-aligned mean fluorescence traces. Bar graphs depict the peak amplitudes of [Ca2+] transients (mean ± SD; BTSP: n = 59, solitary: n = 59 events, n = 22 mice) in the given laps. The amplitudes of [Ca2+] transients of the solitary events in the event lap were significantly higher, whereas in the EL+1 they were smaller compared to those of BTSP-like events (mixed ANOVA: main effect for group (BTSP/solitary): p = 0.682, lap: p < 0.001, interaction: p < 0.001; Tukey post hoc test: lap –1: p = 1, event lap: p < 0.001, lap +1: p < 0.001). *** p < 0.001.
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)
Diverse calcium dynamics underlie place field formation in hippocampal CA1 pyramidal cells
eLife 13:RP103676.
https://doi.org/10.7554/eLife.103676.3
