Depolarization of MEC LII on consecutive days produces consistent changes in grid subfield rates.

A) Firing rate maps of representative grid cells in hM3 mice (top two rows) and a control mouse (Con, bottom row) before and after CNO injection on consecutive days. Color indicates firing rate. Peak firing rate is noted below each rate map. Peak firing rate within each grid subfield is shown below the rate map for that session. BL1, baseline session on Day 1; CNO1, 30-60 min post-CNO injection on Day 1; BL2, baseline session on Day 2; CNO2, 30-60 min post-CNO injection on Day 2. Dashed black lines indicate that BL2 was initiated 12+ hrs. after CNO injection. B) Panel shows grid subfield rate changes between sessions in hM3 (blue) and Con (red) mice. Left, there was no difference between hM3 and Con mice in grid subfield rate changes between BL sessions (hM3 vs. Con: BL1×BL2, hM3 n = 80, Con n = 42, Z = 1.1, p = 0.27; two-sided Wilcoxon rank sum test). Middle, right, grid subfield firing rates changed significantly between BL and CNO sessions on both recording days in hM3 mice versus controls (hM3 vs. Con: BL1×CNO1, hM3 n = 79, Con n = 42, Z = 5.7, p = 5.2 × 10-9; BL2×CNO2, hM3 n = 67, Con n = 40, Z = 7.0, p = 1.5 × 10-12; one-sided Wilcoxon rank sum tests). Top, points represent individual subfields; gapped lines represent mean ± standard deviation. Change refers to an absolute difference score (see Methods). Bottom, black dot: median; black bars: 95% confidence interval; filled curve: sampling-error distribution. C) Scatterplot showing significant correlation between grid subfield rate changes between BL and CNO sessions on Day 1 and Day 2 in hM3 mice (blue, n = 53, r = 0.84, p = 2.3 × 10-18, linear correlation), but not in a shuffled control group (gray, n = 97, r = 0.17, p = 0.10, linear correlation). Points represent individual subfields. Change refers to a difference score. ***p < 0.001.

Depolarization of MEC LII on consecutive days produces a consistent reorganization of CA1 place cell activity.

A) Firing rate maps of representative place cells in hM3 mice (top five rows) and a control mouse (Con, bottom row) before and after CNO injection on Day 1 and Day 2. Color indicates firing rate. Peak firing rate is noted below each rate map. Spatial correlation between rate maps from CNO sessions on each day is noted on the right for each cell. B) Cumulative distribution function shows the spatial correlation between place cell rate maps from BL and CNO sessions on each day in hM3 (BL1×CNO1, light blue; BL2×CNO2, medium blue) and Con (BL1×CNO1, light red; BL2×CNO2, medium red) mice. There was a significant decrease in the spatial correlation of place cells in hM3 versus Con mice on both days (hM3 vs. Con: BL1×CNO1, hM3 n = 106, Con n = 107, D* = 0.60, p = 1.6 × 10-17; BL2×CNO2, hM3 n = 108, Con n = 105, D* = 0.59, p = 7.5 × 10-17; two-sided Kolmogorov-Smirnov tests). C) Cumulative distribution function shows the spatial correlation between place cell rate maps from CNO sessions on each day in hM3 (dark blue) and Con (dark red) mice. There was no difference between hM3 and Con mice in the spatial correlation of place cells between CNO sessions (hM3 vs. Con: CNO1×CNO2, hM3 n = 97, Con n = 104, D* = 0.16, p = 0.15, two-sided Kolmogorov-Smirnov test), indicating that the reorganization of hippocampal place cell activity was consistent across days.

Predictable reorganization of CA1 place cell activity following depolarization of MEC LII.

Ai) Firing rate maps of representative CA1 place cells in hM3 mice before and after CNO injection (one cell per row). Color indicates firing rate. Peak firing rate is noted below each rate map. For each cell, if the location of the primary place field shifted between BL (left, blue circles) and CNO (right, black circles) by more than 20 cm, we identified up to three secondary peaks outside of the primary field in the BL session (left, pink circles) that served as predictions of place field location. We then calculated the distance between each predicted location and the location of the primary place field in the CNO session, defining the shortest of these distances as the ‘prediction offset’. The spatial correlation between rate maps from the BL and CNO session for each cell is noted between columns. The prediction offset for each cell is shown on the right. Note that even when the spatial correlation between sessions was very low, we were often able to predict the location of the new place field for many place cells. Aii) Firing rate map of a representative CA1 place cell in a control mouse exposed to two distinct environments (A and B). Same convention as in Figure 3Ai. B) Histogram shows prediction offsets for all CA1 place cells recorded in hM3 mice (n = 204, median = 12.1 cm, 95% CI, 10.2 – 14.6 cm). C) Histogram shows prediction offsets for CA1 place cells recorded in a separate cohort of mice exposed to distinct environments (A×B, n = 144, median = 16.3 cm, 95% CI, 14.1 – 18.4 cm). Note that prediction offsets were significantly lower for place cells in hM3 mice than for place cells recorded as mice were moved between distinct rooms (hM3 vs. A×B, Z = 2.7, p = 3.4 × 10-3, one-sided Wilcoxon rank sum test), reflecting the predictable reorganization of place fields that occurs during artificial remapping. Vertical lines represent the median of each distribution. D) Kernel smoothed density estimate of prediction offset for place cells in hM3 mice (blue) and a separate cohort of mice exposed to distinct environments (green).

Grid subfield rate changes are sufficient to produce a predictable reorganization of hippocampal place fields.

A) Firing rate maps of one simulated grid cell before and after modification of grid subfield rates. Color indicates firing rate. We first generated a library of 10,000 grid cells with variable subfield rates (left, BL). We then modified the rate of each subfield by an amount drawn randomly from a distribution of subfield rate changes in hM3 mice (right, CNO; see Figure S6A). The peak firing rate of each simulated grid cell was held constant across sessions to isolate the effect of subfield rate changes on place field location. B) Firing rate maps of four simulated place cells before (left, BL) and after (right, CNO) modification of grid subfield rates. Color indicates firing rate. Same convention for place field prediction as in Figure 3A. The spatial correlation between rate maps from the BL and CNO session for each cell is noted between columns. The prediction offset for each cell is shown on the right. C-D) Histograms show prediction offsets for simulated place cells (C, purple) and a shuffled control dataset (D, gray). Note that prediction offsets were significantly lower for simulated place cells than for the shuffled dataset (simulation n = 875, median = 23.0, 95% CI, 20.2 – 26.0 cm; shuffle n = 1,204, median = 33.0, 95% CI, 31.6 – 34.0 cm; simulation vs. shuffle, Z = 7.9, p = 1.3 × 10-15, one-sided Wilcoxon rank sum test). Vertical lines represent the median of each distribution. E) Kernel smoothed density estimate of prediction offset for simulated place cells (purple) and a shuffled control dataset (gray).

Grid subfield rate changes are sufficient to drive a reproducible reorganization of downstream place fields

A) Firing rate maps of one simulated grid cell before and after modification of grid subfield rates. Color indicates firing rate. We first generated a library of 10,000 grid cells with variable subfield rates (left, BL). We then modified the rate of each subfield by an amount drawn randomly from the distribution of subfield rate changes in hM3 mice (middle, CNO1). Subfield rates were then adjusted a second time by an amount drawn randomly from the distribution of rate changes observed across CNO sessions in hM3 mice (right, CNO2). B) Firing rate maps of three simulated place cells before and after modification of grid subfield rates (left, BL; middle, CNO1; right, CNO2). Color indicates firing rate. The spatial correlation between rate maps from CNO sessions is shown on the right for each cell. Note that all three place cells remapped in response to grid subfield rate changes (BL×CNO1 and BL×CNO2) and that the remapping was consistent between CNO1 and CNO2 (CNO1×CNO2). C) Cumulative distribution function shows the spatial correlation between rate maps of simulated place cells between sessions (BL×CNO1, light blue; BL×CNO2, medium blue; CNO1×CNO2, dark blue). Similar modifications to grid subfield rates produced artificial remapping on each day (spatial correlation, BL×CNO1 vs. BL×CNO2: BL×CNO1 n = 1,975, BL×CNO2 n = 1,988, D* = 0.03, p = 0.18, two-sided Kolmogorov-Smirnov test), but the reorganization of the place cell code was similar across days (spatial correlation, BL×CNO1 vs. CNO1×CNO2: BL×CNO1 n = 1,975, CNO1×CNO2 n = 2,150, D* = 0.53, p = 7.5 × 10-253; BL×CNO2 vs. CNO1×CNO2: BL×CNO2 n = 1988, CNO1×CNO2 n = 2150, D* = 0.51, p = 4.0 × 10-233; two-sided Kolmogorov-Smirnov tests).

Transgenic expression of hM3Dq DREADD receptor in MEC LII

A) Representative image of fluorescent immunohistochemistry of a sagittal brain slice targeting the transgene hM3Dq (magenta). Expression is largely restricted to entorhinal cortex layer II. Counterstain NeuN (cyan). Scale bar: 500 μm. B) Same as in (A) without the NeuN counterstain. C-E) Representative high magnification images of dorsal MEC showing NeuN (cyan) and hM3 (magenta) staining. Cropped images from the same brain-wide section shown in (A) and (B). Scale bar: 100 μm.

Recording sites in CA1 and MEC.

A) Representative coronal section used to identify tetrode tracks in CA1. B) Representative sagittal section used to identify tetrode tracks in superficial MEC. C) Tetrode locations in CA1 identified in three coronal sections in hM3 mice (left, blue), control mice (middle, red), and a separate group of mice exposed to two distinct environments (right, green). Numbers indicate distance from bregma. D) Tetrode locations in superficial layers (II/III) of MEC identified in three sagittal sections in hM3 (left, blue) and control (right, red) mice. Numbers indicate distance from midline. SUB, subiculum; PrS, presubiculum; dsc, lamina desiccans.

Putative excitatory neurons in MEC exhibit similar changes in firing rate and field size across days.

A-B) Panels show significant changes in firing rate (A, left) and field size (B, right) of putative excitatory neurons in MEC between BL and CNO sessions on Day 1 and Day 2 in hM3 (blue) versus Con (red) mice (rate change, hM3 vs. Con: BL1×CNO1, hM3 n = 61, Con n = 46, Z = 4.8, p = 8.8 × 10-7; BL2×CNO2, hM3 n = 58, Con n = 39, Z = 5.0, p = 3.3 x10-7; size change, hM3 vs. Con: BL1×CNO1, hM3 n = 46, Con n = 42, Z = 1.7, p = 0.04; BL2×CNO2, hM3 n = 41, Con n = 33, Z = 2.5, p = 6.6 × 10-3; one-sided Wilcoxon rank sum tests). Points represent individual MEC neurons; gapped lines represent mean ± standard deviation. Change refers to an absolute difference score (see Methods). C-D) Scatterplots show significant correlation between firing rate (C, left) and field size (D, right) changes of putative excitatory neurons in MEC between BL and CNO sessions on Day 1 and Day 2 in hM3 mice (rate change, BL1×CNO1 vs. BL2×CNO2: n = 49, r = 0.71, p = 1.0 × 10-8; size change, BL1×CNO1 vs. BL2×CNO2: n = 30, r = 0.76, p = 9.8 × 10-7; linear correlations). Points represent individual MEC neurons. ***p < 0.001. E) Panel shows that there was no difference in the spatial correlation of MEC neurons (including grid cells) following CNO injection between hM3 and control mice (spatial correlation, hM3 vs. Con: BL1×CNO1, hM3 n = 48, Con n = 42, Z = 1.6; p = 0.10; BL2×CNO2, hM3 n = 43, Con n = 34, Z = 1.8; p = 0.07; two-sided Wilcoxon rank sum tests). Points represent individual MEC neurons; gapped lines represent mean ± standard deviation. F) Scatterplot showing significant correlation between grid subfield relationships during CNO session on Day 1 and Day 2 in hM3 (blue) and Con mice (red) (CNO1 vs. CNO2: hM3 n = 66, r = 0.60, p = 2.0 × 10-7; Con n = 64, r = 0.57, p = 1.4 × 10-5; linear correlations). For each grid cell, subfield rates were ordered from highest to lowest. We then calculated the normalized difference in peak firing rate between all subfield pairs. Points represent the normalized difference between each subfield pair in CNO1 versus CNO2. ***p < 0.001. For lower panels in (A), (B) and (E), black dot: median; black bars: 95% confidence interval; filled curv.e: sampling-error distribution.

CA1 place cells exhibit consistent changes in firing rate and field size across days.

A-B) Panels show significant changes in firing rate (A, left) and field size (B, right) of CA1 place cells between BL and CNO sessions on Day 1 and Day 2 in hM3 (blue) versus Con (red) mice (rate change, hM3 vs. Con: BL1×CNO1, hM3 n = 113, Con n = 109, Z = 4.1, p = 1.8 × 10-5; BL2×CNO2, hM3 n = 109, Con n = 111, Z = 6.3, p = 1.6 × 10-10; size change, hM3 vs. Con: BL1×CNO1, hM3 n = 86, Con n = 101, Z = 6.1; p = 4.6 × 10-10; BL2×CNO2, hM3 n = 80, Con n = 102, Z =5.5, p = 2.0 × 10-8; one-sided Wilcoxon rank sum tests). Points represent CA1 place cells; gapped lines represent mean ± standard deviation. Change refers to an absolute difference score (see Methods). C-D) Scatterplots show significant correlation between firing rate (C, left) and field size (D, right) changes of CA1 place cells between BL and CNO sessions on Day 1 and Day 2 in hM3 mice (rate change, BL1×CNO1 vs. BL2×CNO2: n = 104, r = 0.60, p = 2.1 × 10-11; size change, BL×CNO1 vs. BL×CNO2: n = 64, r = 0.45, p = 1.8 × 10-4; linear correlations). Points represent CA1 place cells. E) Cumulative distribution function shows spatial correlation of rate maps from BL sessions on each recording day for place cells in hM3 (light blue) and Con (light orange) mice. Note that there was no difference between groups (spatial correlation, BL1×BL2: hM3 n = 126, median = 0.79, 95% CI, 0.69 – 0.82; Con n = 114, median = 0.79, 95% CI, 0.74 – 0.83; hM3 vs. Con, D* = 0.10, p = 0.57, two-sided Kolmogorov-Smirnov test), indicating that place cells in hM3 mice returned to their BL representations 12+ hrs. after CNO injection. For lower panels in (A) and (B), black dot: median; black bars: 95% confidence interval; filled curve: sampling-error distribution.

Predictable reorganization of place code despite robust artificial remapping.

A) First eight rows show spike path plots and firing rate maps of BL and CNO sessions for eight CA1 place cells in hM3 mice (one cell per row). In spike path plots (first two columns), the mouse’s trajectory is shown in gray and action potentials are represented in black. For firing rate maps (last two columns), color indicates firing rate. Peak firing rate is noted below each rate map. Same convention for place field prediction as in Figure 3A (blue circles, primary place field in BL session; pink circles, predictions of place field location; black circles, primary place field in CNO session). The prediction offset and the spatial correlation between rate maps from the BL and CNO sessions are shown on the right for each cell. Bottom row shows spike path plots and firing rate maps for one CA1 place cell from a control mouse exposed to two distinct environments (A and B). Same convention as above. B) Histogram shows prediction offsets for a shuffled control dataset (see Methods). Note that prediction offsets were significantly lower in hM3 mice (see Figure 3C) than in the shuffled dataset (hM3 n = 204, median = 12.1 cm, 95% CI, 10.2 – 14.6 cm; shuffle n = 261, median = 16.1 cm, 95% CI, 14.4 – 17.9 cm; hM3 vs. shuffle, Z = 3.9, p = 5.6 × 10-5, one-sided Wilcoxon rank sum test). Vertical lines represent the median of the distribution. C) Panel shows the firing rate change within the primary place field from the BL session for place cells in hM3 (blue) and Con (red) mice. Between sessions, there was a significant decrease in firing rate within the BL primary field in hM3 mice relative to controls (Con n = 320, median = −0.10; hM3 n = 204, median = −0.42; Con vs. hM3, Z = 8.4, p = 5.3 × 10-17, two-sided Wilcoxon rank sum test). Points represent CA1 place cells; gapped lines represent mean ± standard deviation. Change refers to a difference score (see Methods). D) Panel shows the firing rate change within the primary place field from the CNO session for place cells in hM3 (blue) and Con (red) mice. Between sessions, there was a significant increase in firing rate within the CNO primary field in hM3 mice relative to controls (Con n = 320, median = 0.03; hM3 n = 204, median = 0.39; Con vs. hM3, Z = 8.4, p = 4.25 × 10-17, two-sided Wilcoxon rank sum test). Points represent CA1 place cells; gapped lines represent mean ± standard deviation. Change refers to a difference score (see Methods). E) Scatterplot shows no relationship between the peak firing rate in the predicted location in the BL session and the prediction offset for place cells in hM3 mice (n = 204, r = −0.11, p = 0.12, linear correlation). Points represent CA1 place cells. F) Scatterplot shows weak relationship between the degree of remapping following CNO injection and the prediction offset for place cells in hM3 mice (n = 204, r = −0.26, p = 1.9 × 10-4, linear correlation). Points represent CA1 place cells. ***p < 0.001. For lower panels in (C) and (D), black dot: median; black bars: 95% confidence interval; filled curve: sampling-error distribution.

Grid subfield rate changes redistribute activity among spatially stable grid inputs.

A) Panel shows distribution of grid subfield rate changes in hM3 (blue) and Con (red) mice between BL and CNO sessions. In our simulation of the grid-to-place cell transformation, grid subfield rates were modified by values drawn randomly from the distribution of grid field rate changes in hM3 mice (median = 0.55, 95% CI, 0.48 −0.65). Note that there was significantly more change in grid subfield rates between BL and CNO sessions in hM3 mice relative to controls (hM3 vs. Con: BL×CNO, hM3 n = 117, Con = 80, p = 2.7 × 10-19, two-sided Wilcoxon rank sum test). Top, points represent grid subfields; gapped lines represent mean ± standard deviation. Change refers to an absolute difference score (see Methods). Bottom, test statistic is the median difference, shown on the y axis as a bootstrap sampling distribution. Black dot: median; black bars: 95% confidence interval; filled curve: sampling-error distribution. p value is from Wilcoxon rank sum test; ***p < 0.001. B)Cumulative distribution function shows spatial correlation between rate maps from BL and CNO sessions for place cells in hM3 mice (blue) and simulated place cells. When grid subfield rates were modified by values drawn randomly from the distribution of grid field rate changes in hM3 mice (Sim Δ = 0.6, purple), the extent of remapping was similar for simulated place cells and place cells in hM3 mice (spatial correlation, BL×CNO: hM3 n = 394, median = 0.14, 95% CI, 0.09 – 0.17; Sim Δ = 0.6 n = 1,975, median = 0.13, 95% CI, 0.11 – 0.16; hM3 vs. Sim Δ = 0.6, Z = 1.7, p = 0.09, two-sided Wilcoxon rank sum test). Decreasing (Sim Δ = 0.3, green) or increasing (Sim Δ = 0.9, orange) the extent of grid subfield rate changes (by adjusting the median of the distribution from which subfield rates were sampled) modulated the degree of remapping among simulated place cells (spatial correlation, BL×CNO: Sim Δ = 0.9 n = 2,033, median = 0.01, 95% CI, 0.00 – 0.03; Sim Δ = 0.3 n = 1,897, median = 0.31, 95% CI, 0.28 – 0.33). C) Panels show excitation maps representing the summed grid cell input to a single simulated place cell before and after grid subfield rate change (top) or independent realignment of grid modules (bottom). Excitation maps (top row) depict strength of summed grid input (from blue to red). Color in corresponding place cell rate maps (bottom row) indicates firing rate. To predict changes in the location of summed grid inputs (rather than hippocampal place fields), we used the same method as in Figure 3A (blue circles, previous peak; pink circles, predictions; black circles, new peak). The prediction offset and the spatial correlation between excitation maps from each session are shown on the right. Note that grid subfield rate changes typically caused the location of the primary field to shift to an alternate peak in the input pattern rather than a random location, resulting in low prediction offsets and high spatial correlations between sessions (median prediction offset = 16.0 cm, 95% CI, 13.0 – 19.0 cm). The location of the primary field typically shifted to an unpredicted location following independent realignment of grid modules, resulting in high prediction offsets and low spatial correlations between sessions (median prediction offset = 40.2 cm, 95% CI, 39.0 – 41.2 cm). The prediction offset was significantly lower after grid subfield rate changes than independent realignment (subfield rate change n = 3,730, independent realignment n = 4,616, Z = 21.8, p = 9.3 × 10-106, two-sided Wilcoxon rank sum test). D) Cumulative distribution functions show spatial correlation between excitation maps before and after grid subfield rate change (gray) or independent realignment of grid modules (black). Grid subfield rate changes resulted in a predictable reorganization of grid cell input, resulting in significantly higher spatial correlation between excitation maps from each session than following independent realignment (spatial correlation: subfield rate change n = 5,000, median = 0.575, 95% CI, 0.571 – 0.579; independent realignment n = 5,000, median = 0.054, 95% CI, 0.047 – 0.061; subfield rate change vs. independent realignment, D* = 0.91, p < 2.2 × 10-16, two-sided Kolmogorov-Smirnov test).

Grid subfield rate changes elicit predictable and reproducible reorganization of hippocampal place fields.

A) Histogram shows prediction offsets between BL and CNO2 for simulated place cells. Note that we were able to predict place field locations equally well during both runs of the simulation (prediction offset: BL×CNO2 n = 895, median = 22.2 cm, 95% CI, 19.4 – 24.1 cm; BL×CNO1 vs. BL×CNO2, Z = 1.0, p = 0.31, two-sided Wilcoxon rank sum test). Vertical line represents the median of the distribution. B) Histograms show prediction offsets between BL and CNO2 for a shuffled control dataset (see Methods; BL×CNO2 n = 1,218, median = 31.7 cm, 95% CI, 30.2 – 33.1 cm). Note that prediction offsets for simulated place cells between BL and CNO2 were significantly lower for the shuffled control dataset (simulation vs. shuffle, Z = 8.7, p = 4.4 × 10-18, two-sided Wilcoxon rank sum test). Vertical line represents the median of the distribution. C) Kernel smoothed density estimate of prediction offset for simulated place cells (purple) and a shuffled control dataset (gray) between the BL session and CNO2 (BL×CNO2).