NPAS4 expression in CA1 pyramidal neurons in adult mice results in reorganization of inhibition along the somatodendritic axis.

(A) NPAS4 expression in CA1 from mice housed in a standard environment (SE), an enriched environment (EE) for 90 minutes, or following kainic acid (KA) injection. Magenta: NeuN; teal: NPAS4. Scale bar = 50 µm. (Data are mean ± SEM; SE: N = 3 animals, 3 sections each; EE: N = 4 animals, 3 sections each; KA: N = 3 animals, 3 sections each; unpaired t test.) (B) Schematic of the experimental strategy and timeline for whole-cell patch-clamp recordings. (C) eIPSCs evoked in either stratum pyramidale (SP) or stratum radiatum (SR) recorded with simultaneous voltage-clamp from neighboring WT (black) and KO (green) pyramidal neurons. Geometric mean traces shown as percent of WT. SP scale bar = 20 ms, 25% of WT; SR scale bar = 20 ms, 50% of WT. (SP: N = 14 slices from 6 mice; SR: N = 8 slices from 5 mice; ratio paired t test.) (D) Schematic of the experimental strategy and timeline for in vivo electrophysiology recordings. (E) Example image of sparse KO used for in vivo recordings. Scale bar = 50 µm. (F) Schematic of the experimental timeline for each session. Star: reward location. (G) Example rasters (top) and peristimulus time histograms (PSTH; middle) from putative WT (left) and KO (right) neurons. (Bottom) Same PSTHs as above, shown at higher temporal resolution to highlight the 1-second window surrounding light stimulation. Blue arrow and bars: light stimulation (WT example 1: Animal 6 WT Cell 2; WT example 2: Animal 2 WT Cell 3; KO example 1: Animal 1 KO Cell 1; KO example 2: Animal 5 KO Cell 1). (H) Opto-response during low-power light stimulation (∼0.3 mW) in cells categorized as WT (black) or KO (green). Opto-response is defined as the peak PSTH value during light-on minus the peak PSTH value during light-off periods of the optostim protocol. (I) Opto-response during high-power light stimulation (>0.3 mW) in cells categorized as WT (black) or KO (green). As KO neurons were recruited into the pop-spike during high-power light stimulation, their opto-responses decrease substantially. *p < 0.05; **p < 0.01.

NPAS4 knockout neurons have fewer spikes in bursts and are more likely to have high firing rates, but high firing knockout and wild-type neurons exhibit comparable spatial firing.

(A) Schematic of the track recordings. Mice ran in one direction on a rectangular track for 10 trials before being required to switch directions. This behavior was repeated for up to 80 trials or 30 minutes, whichever occurred first. The track was linearized as indicated. (B) Example traces showing x-position (top), y-position (middle), and velocity (bottom) as an animal ran along the track. Time periods during which velocity fell below 2 cm/sec (red bar on velocity plot) were excluded from analysis. (C) Cumulative probability distribution of firing rates across the session. Firing rate was defined as the total number of spikes divided by session duration. Gray shaded region: ± SEM for WT, centered at the mean; green shaded region: ± SEM for KO, centered at the mean. (WT: N = 112 cells; KO: N = 47 cells; Kolmogorov–Smirnov test.) (D) Histogram of interspike intervals (ISIs). Arrow indicates ISIs < 10 ms. (WT: N = 112; KO: N = 47; Kolmogorov–Smirnov test.) (E) Cumulative probability distribution of the proportion of spikes in bursts, defined as ISIs < 10 ms. Shaded regions as in (C). (WT: N = 112; KO: N = 47; Kolmogorov–Smirnov test.) (F) Example linearized rate maps from a ‘low firing’ WT (top) and KO (bottom) neuron. Left: trajectory of the session (black), with spikes (red dots). Right: trial-by-trial linearized rate map for one direction, with the trial-averaged rate map shown below. (G) As in (F), but for ‘high firing’ WT and KO neurons. (H) Percentage of cells classified as ‘low firing’ or ‘high firing’ in one or both directions (p < 0.05; chi-square goodness-of-fit test) (I) Cumulative probability distributions of mean and maximum spatial firing rates for ‘low firing’ cells. Shaded regions as in (C) (WT: N = 86; KO: N = 26; Kolmogorov–Smirnov test). (J) As in (I), but for ‘high firing’ cells. (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test.) *p < 0.05; **p < 0.01.

NPAS4 knockout neurons are less spatially tuned than simultaneously recorded wild-type counterparts.

(A) Example trial-averaged rate maps from a WT (top) and KO (bottom) neuron. Dotted lines mark 10% and 50% of the peak rate and were used to identify place fields (WT example: Animal 1, WT Cell 35; KO example: Animal 6, KO Cell 2). (B) Histogram of the number of place fields per neuron (WT: N = 138; KO: N = 68; Mann–Whitney test). (C) Trial-averaged rate maps from WT (left) and KO (right) neurons. Top: ordered by the peak location in the clockwise (CW) direction, with the same order applied to activity in the counterclockwise (CCW) direction. Bottom: ordered by peak location in the CCW direction (WT: N = 138; KO: N = 68). (D) Cumulative probability distribution of average place field size. Place fields were defined as sets of contiguous bins above 10% of the peak that also included at least one bin above 50%. Gray shaded region: ± SEM for WT, centered at the mean; green shaded region: ± SEM for KO, centered at the mean (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). (E) Trial-averaged in-field rates for WT and KO place fields, centered on the peak of each field. Blue shaded regions indicate bins with p < 0.05 (WT: N = 183 fields over 138 neurons; KO: N = 95 fields over 68 neurons; Kolmogorov–Smirnov test). (F) Cumulative probability distribution of average in-field firing rates. Shaded regions as in (D) (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). (G) As in (E), but firing rates normalized to each neuron’s peak. (H) As in (G), but showing only out-of-field bins. (I) Cumulative probability distribution of average out-of-field firing rates. Shaded regions as in (D) (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). (J) Cumulative probability distribution of signal-to-noise ratio, defined as the average in-field firing rate divided by the average out-of-field firing rate per neuron. Shaded regions as in (D) (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). (K) Cumulative probability distribution of spatial information. Shaded regions as in (D) (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). (L) Cumulative probability distribution of sparsity. Shaded regions as in (D) (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). *p < 0.05; ***p < 0.001.

NPAS4 knockout neurons are less stable and shift their place fields towards the field entrance more rapidly than wild types.

(A) Example trial-averaged rate maps from each of the four epochs (sets of 10 trials) for a WT (left) and KO (right) neuron. For each pairwise comparison (epoch 1 to 2, epoch 2 to 3, and epoch 3 to 4), the Pearson’s Correlation Coefficient (PCC) was calculated. (B) PCC across sequential epoch comparisons for WT (solid black) and KO (solid green) neurons, alongside shuffled controls (WT shuffle: dotted gray; KO shuffle: dotted green). Shuffled distributions were generated by spatially shifting trials randomly (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test for WT vs. KO; Wilcoxon signed-rank test for comparisons to shuffled distributions). (C) Trial-averaged rate maps for WT and KO place cells, separated by epoch and centered on the peak in Epoch 1. Blue shaded regions indicate bins where p < 0.05 (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). (D) Example showing how difference maps are generated: trial-averaged rate maps from each epoch are centered on the peak of Epoch 1, and the difference between sequential epochs is computed. (E) Difference maps for WT (top) and KO (bottom) place fields across all epoch comparisons. Mean difference maps across neurons are shown below (WT: N = 176 fields from 138 neurons; KO: N = 91 fields from 68 neurons). *p < 0.05; **p < 0.01; ****p < 0.0001.

NPAS4 knockout cells are less theta-coupled than wild-type counterparts.

(A) Example of spiking relative to the local field potential (LFP) for WT (top) and KO (bottom) neurons. Spikes are shown as vertical lines above the raw LFP trace with the theta-filtered LFP overlaid. (B) Rose plots show the theta phase of all spikes from the same example neurons with the mean vector overlaid. (C) Rose plot scatter for all neurons. Angular position indicates preferred theta phase; radial position indicates mean vector length (WT: N = 138; KO: N = 68). (D) Preferred phase for WT and KO neurons. Solid line: mean; shaded region: SEM. Concentric circles shown for visualization only (WT: N = 138; KO: N = 68). (E) Cumulative probability distribution of mean vector lengths. Gray shaded region: ± SEM for WT, centered at the mean; green shaded region: ± SEM for KO, centered at the mean (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). (F) Mean vector length as a function of normalized field position. Blue shaded regions indicate bins where p < 0.05 (WT: N = 183 fields from 138 neurons; KO: N = 95 fields from 68 neurons; Kolmogorov–Smirnov test). *p < 0.05.

NPAS4 KO neurons exhibit shallower phase precession slopes compared to WT counterparts.

(A) Example WT (top) and KO (bottom) neurons showing theta-related spiking during a single field pass. For each example, the top panel shows the trial-averaged rate map across all trials; the panel immediately below shows the rate map for one example trial. The raw local field potential (LFP) during that trial is shown with the theta-filtered LFP overlaid. Spikes are marked as vertical lines below the LFP, and the theta phase of each spike is plotted over time. (B) Three example trials from the neurons shown in (A). Scatter plots show theta phase of each spike over position with linear fits overlaid. (C) Histogram of phase precession slopes across all trials for the neurons in (A). Red line: median slope. (D) Cumulative probability distribution of mean phase precession slopes. Only the field with the highest firing rate from each neuron is included. Gray shaded region: ± SEM for WT, centered at the mean; green shaded region: ± SEM for KO, centered at the mean (WT: N = 70; KO: N = 36; Kolmogorov–Smirnov test). (E) Cumulative probability distribution of mean field size calculated on a trial-by-trial basis. Only cells included in (D) are included. Gray shaded region: ± SEM for WT, centered at the mean; green shaded region: ± SEM for KO, centered at the mean (WT: N = 70; KO: N = 36; Kolmogorov–Smirnov test). (F) Mean field size of each neuron plotted against its mean phase precession slope. Black: WT; Green: KO (WT: N = 70; KO: N = 36; Kolmogorov–Smirnov test). (G) Schematic summarizing results. *p < 0.05.

Representative histology for all sparse-infection animals.

(A) Left: Stitched 10X images showing the anterior extent of the infection. Middle: Site of optetrode implant in CA1, asterisks represent the end of the tetrode tracts. Not all tetrode tracts were able to be identified. Right: Stitched 10X images showing the posterior extent of the infection. (B) Representative 60X images near the site of implant showing sparse infection. Scale bar = 50 µm.

Optical identification and functional characterization of NPAS4 KO neurons in vivo.

(A) Example of unfiltered LFP recording during light stimulation. Blue triangles = light delivery. (B) Example of unfiltered LFP recording during high-power light stimulation. (C) Example of cluster cutting for two WT neurons and a KO neuron during low-level light stimulation. Black dots: spikes from two WT neurons; green dots: spikes from one KO neuron. (D) Same neurons and cluster views as in (C) but for high-power stimulation. Dotted circle: putative population spikes. (E) WT and KO cell counts for each animal during track recordings. (F) Example of cluster cutting for a track recording showing a WT cell (black) and KO cell (green). Insets are average spike waveforms for each of the four tetrode channels for each neuron. (G) L-ratio for WT and KO cells for pre-track home cage (HC1), track, post-track home cage (HC2), and during optostimulation (HC1 WT N=70, KO N=34; track WT N=112, KO N=36; HC2 WT N=78, KO N=15; opto WT N=108, KO N=33; Kolmogorov–Smirnov test). (H) Isolation distance for WT and KO cells for HC1, track, HC2, and opto (HC1 WT N=71, KO N=36; track WT N=111, KO N=37; HC2 WT N=78, KO N=16; opto WT N=108 KO N=34; Kolmogorov–Smirnov test). (I) Example of how the percent infection is obtained from histology. 60X images taken anterior, posterior, medial, and lateral of the implant site. The percent infection per animal shown in (J) is an average of all four sites. Scale bar = 50 µm. (J) The percent of infected cells identified in histology plotted against the percent of optotagged cells identified in vivo. Line of best fit is shown. (K) Distribution of velocity for each session the sparse KO animals ran. (L) Mean velocity for all sparse KO animals. (M) Max velocity for all sparse KO animals. ns = not significant.

Additional information on spatial firing rate properties for NPAS4 WT and KO neurons.

(A) Additional examples of rate maps from five WT cells. For each cell, both clockwise (CW) and counterclockwise (CCW) directions are shown; the direction not used for analysis is greyed out. The summary data shown above each rate map (Place Field Number [PF #], Average Size [Ave. Size], and Maximum Firing Rate [Max. FR]) corresponds to the analyzed direction only. (B) As in (A) but for KO cells. (C) Histogram of mean firing rate for all neurons (both low firing and high firing; WT: N = 224; KO: 94). (D) As in (C) but for the maximum firing rate. (E) Pearson’s Correlation Coefficient (PCC) between the trial-averaged CW rate map for each neuron and the trial-averaged CCW rate map (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). (F) The percentage of all spikes for all neurons that occurred in-field and out-of-field for WT and KO neurons (WT: N = 140; KO: N = 68; p = 0.21; chi-square goodness-of-fit test). (G) Average place field size as calculated in the main figure with the exception that the minimum threshold used was 20% of max firing instead of 10% of max firing (WT: N = 138; KO: N = 68; Wilcoxon rank sum test). (H) As in (G) but with a minimum threshold of 30% (WT: N = 138; KO: N = 68; Wilcoxon rank sum test).

Spatial deficits persist across firing rate thresholds, matched firing rates, and independent replication.

(A) Cumulative probability distribution of average place field size when only including neurons with mean firing rates > 0.5 Hz or max firing rates > 5 Hz. Gray shaded region: ± SEM for WT, centered at the mean; green shaded region: ± SEM for KO, centered at the mean (WT: N = 81; KO: N = 36; Kolmogorov–Smirnov test). (B) As in (A) but for spatial information. (C) As in (A) but for sparsity. (D) Histogram of the number of place fields per neuron when the number of spikes is the same across all neurons (WT: N = 138; KO: N = 68; Mann–Whitney test). (E) Cumulative probability distribution of average place field size when the number of spikes is the same across all neurons. Gray shaded region: ± SEM for WT, centered at the mean; green shaded region: ± SEM for KO, centered at the mean (WT: N = 81; KO: N = 36; Kolmogorov–Smirnov test). (F) As in (E) but for spatial information. (G) As in (E) but for sparsity. *p < 0.05; **p < 0.01.

NPAS4 stability deficits are the result of spurious out-of-field firing and shifts in the place field towards field entrance.

(A) Left: Four representative trials for a WT example cell. The number of fields, size, information, and sparsity are calculated independently for each trial. Right: For the same example cell, histograms for (top to bottom) the number of place fields, the average size of the place fields, the information, and the sparsity calculated for each trial independently and shown for all trials. Red line depicts the average. (B) As in (A) but for an example KO cell. (C) Median number of place fields per neuron. The number of fields was computed independently for each trial, then the median value was calculated for each neuron (WT N=140, KO N=68; Mann-Whitney test). (D) As in (C) but for the average place field size (WT N=140, KO N=68; Kolmogorov–Smirnov test). (E) As in (C) but for the spatial information (WT N=140, KO N=68; Kolmogorov–Smirnov test). (F) As in (C) but for the sparsity (WT N=140, KO N=68; Kolmogorov–Smirnov test). (G) Correlation between sequential sets of epochs (E1 to E2, E2 to E3, and E3 to E4) for all bins across the track. Gray dots: WT neurons; green dots: KO neurons; dotted gray: shuffled WT; dotted green: shuffled KO (WT N=140, KO N=68; Kolmogorov–Smirnov test). (H) As in (G) but for only the in-field bins for each neuron (WT N=140, KO N=68; Kolmogorov–Smirnov test). (I) As in (G) but for only the out-of-field bins for each neuron (WT N=140, KO N=68; Kolmogorov–Smirnov test). (J) Example trial-averaged rate maps for each epoch from one WT (left) and one KO (right) neuron to show shift calculation. Shift (denoted as Δ) is negative when the field shifts towards field entrance and positive when it shifts towards field exit. (K) Shift values for WT (top) and KO (bottom) across sequential sets of epochs (E2-E1, E3-E2, E4-E3). Dark shaded regions: fields with shift less than -1; unshaded regions: fields with shift between -1 and 1; light shaded regions: fields with shift greater than 1. Significance indicates comparisons between WT and KO (WT: N = 176 fields from 138 neurons; KO: N = 91 fields from 68 neurons). *p < 0.05; **p < 0.01; ****p < 0.0001.

Impaired theta modulation in NPAS4 KO neurons is carried by single spikes but not bursts and the accompanying phase precession phenotype is related to differences in the size of the place fields.

(A) Velocity (top) and spectrogram (bottom) for a representative session. Shaded red bars are periods of time when the velocity is below 2 cm/sec. (B) Theta power after accounting for the aperiodic offset (data are mean ± SEM; control N=6 animals, sparse N=8; Kolmogorov–Smirnov test). (C) Cumulative probability distribution of mean vector lengths for the in-field spikes (left) or out-of-field (right) spikes. Gray shaded region: ± SEM for WT, centered at the mean; green shaded region: ± SEM for KO, centered at the mean (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). (D) As in (F) but for spikes in a burst that occurred in-field (left) or singles (spikes not in a burst) that occurred in-field (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). (E) As in (F) but for spikes in a burst that occurred out-of-field (left) or singles (spikes not in a burst) that occurred out-of-field (WT: N = 138; KO: N = 68; Kolmogorov–Smirnov test). (F) Percentage of neurons retained after sequential thresholding steps applied during phase precession slope estimation. Criteria included: 1. Max ISI - spikes within each trial must have interspike intervals ≤ 1 s; 2. Theta Cycles - trials must span ≥ 3 theta cycles; 3. Sig - trial-level circular-linear regression must yield a p-value < 0.05; 4. Number of Trials - neurons must have ≥ 3 trials meeting the above criteria to be included in downstream analyses (WT: N = 70; KO: N = 36). (G) Cumulative probability distribution of the standard deviation of phase precession slopes across trials for WT and KO neurons. Gray shaded region: ± SEM for WT, centered at the mean; green shaded region: ± SEM for KO, centered at the mean (WT: N = 70; KO: N = 36; Kolmogorov–Smirnov test). (H-K) Shuffle analyses to assess whether the group difference in phase precession slopes reflects structured relationships between spike phase and position. (H) Histogram of mean slope differences (WT minus KO) across 100 iterations of theta phase shuffling, where theta phases were randomly permuted within each trial to disrupt spike–theta alignment. Red dotted line: the value derived from the actual data (WT: N = 70; KO: N = 36). (I) Histogram of p-values from Kolmogorov-Smirnov (KS) tests comparing WT and KO slopes in each phase-shuffled iteration (shuffling as in [H]). Red dotted line: the value derived from the actual data (WT: N = 70; KO: N = 36). (J) As in (H), but for position shuffling, in which spike positions were randomly permuted within trials (WT: N = 70; KO: N = 36). (K) As in (H), but for position shuffling as in [J] (WT: N = 70; KO: N = 36). (L-M) Bootstrapping analysis to estimate the reliability of the observed group difference. (L) Histogram of mean slope differences (WT minus KO) across 1,000 bootstrap iterations, resampling neurons with replacement while maintaining the original sample size per genotype (WT: N = 70; KO: N = 36). (M) Histogram of p-values from Kolmogorov-Smirnov (KS) tests comparing WT and KO slopes in each bootstrap iteration (WT: N = 70; KO: N = 36). (N) Cumulative probability distribution of the median field size for neurons included in the phase precession analysis. Gray shaded region: ± SEM for WT, centered at the mean; green shaded region: ± SEM for KO, centered at the mean (WT: N = 70; KO: N = 36; Kolmogorov–Smirnov test). (O) Relationship between phase precession slope and log-transformed field size across neurons (WT: gray; KO: green; WT: N = 70; KO: N = 36). (P) Relationship between phase precession slope and theta modulation strength (MVL; WT: gray; KO: green; WT: N = 70; KO: N = 36). *p < 0.05; **p < 0.01, ****p < 0.0001.

Linear regression models examining predictors of phase precession slope.

Models were fit using MATLAB’s fitlm function. Each model included different combinations of predictors: genotype (WT vs. KO), field size (log-transformed), and theta modulation (log-transform of the mean vector length). The purpose of these models was to determine which variables best account for variability in phase precession slope and to assess the unique contribution of each predictor. Adjusted R² values were used to compare model fit while penalizing for model