Abstract
Navigating a complex world requires integration of multiple spatial reference frames, including information about one's orientation in both allocentric and egocentric coordinates. Combining these two information sources can provide additional information about one's spatial location. Previous studies have demonstrated that both egocentric and allocentric spatial signals are reflected by the firing of neurons in the rat postrhinal cortex (POR), an area that may serve as a hub for integrating allocentric head direction (HD) cell information with egocentric information from center-bearing and center-distance cells. However, we have also demonstrated that POR HD cells are uniquely influenced by the visual properties and locations of visual landmarks, bringing into question whether the POR HD signal is truly allocentric as opposed to simply being a response to visual stimuli. To investigate this issue, we recorded HD cells from the POR of female rats while bilaterally inactivating the anterior thalamus (ATN), a region critical for expression of the “classic” HD signal in cortical areas. We found that ATN inactivation led to a significant decrease in both firing rate and tuning strength for POR HD cells, as well as a disruption in the encoding of allocentric location by conjunctive HD/egocentric cells. In contrast, POR egocentric cells without HD tuning were largely unaffected in a consistent manner by ATN inactivation. These results indicate that the POR HD signal originates at least partially from projections from the ATN and supports the view that the POR acts as a hub for the integration of egocentric and allocentric spatial representations.
Introduction
Efficient navigation depends on the ability to perceive one's orientation in multiple reference frames simultaneously. For example, to solve a spatial task, an organism may be required to calculate its orientation relative to the world at large (e.g., am I facing north or south?) as well as its orientation relative to cues in the local environment (e.g., is landmark A to my left or right?). The former reference frame is considered “allocentric” as it is defined relative to the world, while the latter is considered “egocentric” as it is defined relative to the organism's first-person perspective (Klatzky, 1998). Much theoretical work has focused on exactly how these seemingly disparate signals may interact and inform each other in the brain (McNaughton et al., 1989, 1995; Gallistel, 1990; O’Keefe, 1991; Touretzky et al., 1993; Touretzky and Redish, 1996; Byrne et al., 2007; Bicanski and Burgess, 2018).
One of the most well-known types of neuron with allocentric orientation correlates is the head direction (HD) cell, which fires preferentially when an animal's head points in a certain direction in allocentric space (e.g., north or south; Taube et al., 1990). The HD signal is computed primarily from vestibular information regarding head turns (Stackman and Taube, 1997; Muir et al., 2009), which is projected through a brainstem circuit that culminates in a strongly tuned population of HD cells at the level of the anterior thalamus (ATN; Taube, 1995, 2007; Cullen and Taube, 2017). While HD cells have been identified in multiple additional brain areas throughout the cortex and limbic system, lesion and inactivation studies suggest that much of this tuning relies on intact functioning of the ATN, as manipulating ATN function degrades HD tuning in the postsubiculum (Goodridge and Taube, 1997) and medial entorhinal cortex (MEC; Winter et al., 2015).
Neurons correlated with egocentric orientation have been identified across multiple cortical regions. These neurons are tuned to the egocentric bearing of either the geometric centroid or boundaries of the local environment (Wang et al., 2018; Gofman et al., 2019; Hinman et al., 2019; LaChance et al., 2019; Alexander et al., 2020), as well as objects (Wang et al., 2018) or task-relevant visual cues (Wilber et al., 2014). Some of these studies have also identified cells that conjunctively encode allocentric and egocentric orientation preferences, offering insights into how egocentric and allocentric signals are integrated within single neurons.
One brain area that contains a striking confluence of HD and egocentric orientation tuning is the postrhinal cortex (POR), which is homologous with primate areas TH/TF and the human parahippocampal cortex (Burwell et al., 1995). However, HD-tuned neurons in POR cells differ from more “classic” HD cells in the following ways: First, approximately half of POR HD cells show conjunctive tuning to the egocentric bearing of the geometric centroid or boundaries of the local environment (“center-bearing” tuning; Gofman et al., 2019; LaChance et al., 2019, 2022). Second, they generally have broad tuning profiles, such that their tuning curves often appear sinusoidal (Gofman et al., 2019, LaChance et al., 2019, 2022) compared to more Gaussian or triangular shapes observed in classic HD cells. Third, they can become bidirectionally tuned when a familiar landmark appears in two different locations at the same time (LaChance et al., 2022). For these reasons, we refer to the HD-responsive neurons in POR as the landmark-modulated-HD cells (LM-HD cells).
Whereas POR receives primarily visual and visuospatial inputs from areas such as the retrosplenial cortex, posterior parietal cortex, and lateral posterior thalamus (Burwell and Amaral, 1998a; Pereira et al., 2016; Beltramo and Scanziani, 2019), it also receives a moderate projection from the ATN, including the anterodorsal, anteroventral, and lateral dorsal portions (Pereira et al., 2016), as well as inputs from other brain regions that contain classic HD cells such as the postsubiculum (Taube et al., 1990; Agster and Burwell, 2013), the caudal parasubiculum (PaS; Cacucci et al., 2004; Agster and Burwell, 2013), and the retrosplenial cortex (RSC; Burwell and Amaral, 1998a; Cho and Sharp, 2001). Therefore, while the POR LM-HD signal differs in many ways from classic HD cells, the classic HD signal may still play a role in influencing the POR LM-HD representation, possibly for the purpose of binding egocentric visual information to an allocentric reference frame. Conversely, the purely egocentric signals conveyed by POR cells (e.g., center bearing) are more likely to be sensory in nature and are less likely to be driven significantly by allocentric orientation signals (though see Peyrache et al., 2017).
To investigate the impact of the allocentric HD signal on orientation tuning in POR, we recorded from single neurons in the rat POR while inactivating the ATN [defined here as including the anterodorsal and anteroventral portions, both of which contain HD cells (Taube, 1995; Tsanov et al., 2011), but not the anteromedial portion] (Fig. 1A,B), with the intention of disrupting expression of the allocentric HD signal. We found that, while pure center-bearing cells were largely unaffected by the inactivation, LM-HD cells showed a consistent and significant reduction in firing rate and tuning strength during ATN inactivation. In addition, cells conjunctively tuned to both allocentric and egocentric orientation (i.e., LM-HD × center-bearing cells) showed a preferential disruption in HD tuning compared to egocentric tuning, and their encoding was disrupted at specific allocentric locations in the environment. These results indicate that the POR contains a true interaction between allocentric and egocentric spatial reference frames, with the allocentric portion depending on the integrity of the ATN.
Histology and behavioral effects of ATN inactivation. A, left, a Nissl stained coronal section from one experimental rat (PL98) showing tracks from infusion cannulae aimed at the ATN; right, an expanded view of the left section showing the borders of the anterodorsal (AD) and anteroventral (AV) thalamus. Black arrows indicate where the infusion cannula broke through the membrane along the dorsal border of the ATN. B, a Nissl stained sagittal section from the same rat as (A) showing electrode track and lesion marks, as well as the borders between POR, parasubiculum (PaS), and medial entorhinal cortex (MEC). C, Foraging paths (top) and spatial occupancy histograms (bottom) for two example series of the PRE-LIDO-POST experiment. D, Behavior comparisons between PRE, LIDO, and POST sessions. From left to right: spatial coverage of the environment, with the 80% cutoff for analysis of neural data shown in red; mean running speed over the course of the session; and thigmotaxis, or percent of time spent within 15 cm of a wall. E, Correlative behavior analyses between pairs of sessions. Left, correlations between spatial occupancy histograms (examples given in (C)) for each pair of sessions; right, correlations between HD occupancy histograms for each pair of sessions. Individual data points in D and E represent individual recording sessions from 26 full trials of the inactivation experiment for all five rats. * denotes statistical significance.
Materials and Methods
Subjects
Subjects were 5 female Long-Evans rats aged 5–8 months and weighing 250–340 g prior to surgery. Rats were individually housed in Plexiglas cages and maintained on a 12 h light/dark cycle. Prior to surgery, food and water were provided ad libitum. All experimental procedures involving the rats were performed in compliance with institutional standards as set forth by the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the Dartmouth Institutional Animal Care and Use Committee.
Electrode construction
Animals were implanted with a movable microdrive consisting of a bundle of four tetrodes targeting the POR. The tetrodes were constructed by twisting together four strands of 17-µm nichrome wire. These twisted strands were subsequently threaded through a single 26-gauge stainless steel cannula, and the end of each wire was connected to a single pin of a Mill-Max connector. The two center pins of the connector were attached to the cannula, which acted as an animal ground. Three drive screws (modified 2–56 machine screw) were secured around the connector using dental acrylic, making the electrode driveable in the dorsal-ventral plane.
Electrode and cannula implantation
Animals were anesthetized with isoflurane. They were subsequently placed in a stereotaxic frame, and an incision was made in the scalp to expose the skull. A single craniotomy was drilled above the target structure. Implant coordinates were as follows: 0.45 mm anterior to the transverse sinus, 4.6 mm lateral to lambda, and 0.5–1 mm ventral to the cortical surface. The tetrode tips were also angled 10° forward in the sagittal plane, such that the tetrode tips were pointing anteriorly. Animals were also implanted with two 26-gauge guide cannulae (Plastics One) implanted bilaterally above the ATN. Implant coordinates for the guide cannulae were as follows: AP: −1.5 mm from bregma, ML: ± 1.4 mm from bregma, DV: −3.5 mm ventral to the cortical surface. Guide cannulae were plugged with a piece of 32-gauge stainless steel wire (“dummy cannula”) that could be easily removed for infusions to take place. Drives and guide cannulae were secured to the skull using dental acrylic. In an effort to avoid damaging the ATN permanently, the end of the guide cannulae was purposely positioned at the dorsal border to the ATN (Fig. 1A). When inserted, the drug infusion cannula extended 1 mm beyond the tip of the guide cannula.
Recovery and behavioral training
Rats were allowed 7 d to recover from surgery, after which they were placed on food restriction such that their body weight reached 85–90% of their pre-surgical level. During this time, the rats were also trained to forage for randomly scattered sucrose pellets (20 mg) within a 1.2 × 1.2 m gray square box (walls 50 cm in height) surrounded by a uniform black curtain that formed a circle around the square box. The box itself was featureless but had a single white cardboard sheet (cue card) placed along the south wall. The cue card was 50 cm in height (such that it covered the full vertical extent of the wall) and had a width of 72 cm, such that it covered 60% of the horizontal extent of the wall. The floor was composed of gray photographic backdrop paper which was changed between sessions. Recording began when the animals’ walking paths showed uniform coverage (>80%) of the entire arena during 20 min sessions in the 1.2 m enclosure.
Baseline recording sessions
Over the course of weeks to months, tetrodes were “screened” for cells that contained well-isolated waveforms as the animals foraged for sucrose pellets in the open arena. Electrical signals were pre-amplified using unity-gain operational amplifiers on an HS-18-MM headstage. Signals from each tetrode wire were then differentially referenced against a quiet (low noise) channel from a separate tetrode and bandpass filtered (600 Hz–6 kHz) using a Cheetah Data Acquisition System. If signals on a given tetrode crossed a predefined amplitude threshold (30–50 µV), they were time-stamped and digitized at 32 kHz for 1 ms. The headstage was also equipped with red and green light-emitting diodes (LEDs) spaced ∼6 cm apart over the head and back of the animal, respectively. A color video camera positioned over the arena captured video frames with a sampling rate of 30 Hz, and an automated video tracker extracted the x- and y-positions of the LEDs, as well as their angle in an allocentric frame. The tracking frames were timestamped so they could be matched up to the neural data. If clearly well-isolated waveforms were visually apparent, a 20 min baseline recording session in the 1.2 m square box took place. If no clear waveforms were detected, electrodes were advanced ∼50–100 µm and screened again at least 2 h later or the next day.
Spike sorting
Spike sorting was conducted offline. Spikes collected from a recording session were first automatically sorted into clusters using the automated clustering program Kilosort (Pachitariu et al., 2016), after which manual adjustments were performed using the manual clustering program SpikeSort3D (Neuralynx). If cells were recorded across multiple sessions in a day (i.e., with multiple cue configurations), automatic sorting was performed on a merged dataset to ensure cluster continuity, and then results were separated into individual sessions for manual cleanup and analysis. For the manual step, waveform features including peak, valley, height, width, and principal components were used to visualize the characteristics of individual spikes across multiple wires of a tetrode simultaneously as a 3D scatter plot. Adjustment of automatically sorted clusters, which was not always required, was performed by drawing a polygon around the visually apparent boundaries of each cluster. Single-unit isolation was assessed using metrics such as L-ratio and isolation distance, as well as assessment of temporal autocorrelograms for the presence of a refractory period. L-ratio and isolation distance are complementary methods of measuring the separation between spikes in a cluster and the remaining spikes recorded on the same tetrode. Despite significant advancement of the electrodes between recording sessions, we sometimes found that the same cells were recorded multiple times on the same tetrodes across recording sessions (based on analyzing waveform shape and location in cluster space); in these cases we only used the first recording session of the cell. For each well-isolated cluster, we saved the timestamps for each spike and then analyzed and matched them to the tracking data.
Histology
Once recordings were complete, animals were deeply anesthetized with sodium pentobarbital and small marking lesions were made at the electrode tips by passing a small anodal current (15 µA, 15–20 s) through two active wires from separate tetrodes. Animals were then intracardially perfused with saline followed by 10% formalin solution, after which the brains were removed from the skull and postfixed in 10% formalin solution with 2% potassium ferrocyanide for at least 24 h. The brains were then transferred to 20% sucrose solution for at least 24 h, after which they were blocked such that the posterior half of the brain could be sliced sagittally (to determine electrode placement in POR) and the anterior half could be sliced coronally (to determine cannula placement above the ATN). The brains were frozen and sliced into 30 µm sections using a cryostat. Sections were mounted on glass microscope slides and stained with thionin, after which electrode tracks were examined using a light microscope. Locations of recorded cells were determined by measuring backward from the most ventral location of the marking lesions or, if marking lesions were not visible, the electrode tracks. Delineations of parahippocampal regions were drawn mainly from Boccara et al. (2015) and Burwell (2001), while delineations of the ATN were drawn from Paxinos and Watson (1998).
Recording protocol
On each day of recording, subjects were initially allowed to forage for sugar pellets during a 20 min baseline recording session in the 1.2 × 1.2 m enclosure (“PRE” session). Recorded cells were subsequently classified as encoding up to four behavioral variables (see classification criteria below). If at least one LM-HD cell or center-bearing cell was identified, the lidocaine infusion process took place. First, the animal was wrapped snugly in a towel, and the dummy cannulae were removed. Next, two 32 gauge infusion cannulae were inserted bilaterally into the guide cannulae, with the two infusion cannulae connected to an infusion pump. Over the course of the next 8 min, 1 µl of lidocaine solution (1 g lidocaine per 1 ml 0.9% saline) was slowly infused into the ATN bilaterally through the infusion cannulae. Following infusion, the infusion cannulae were allowed to stay in place for 4 min to allow diffusion of the drug. After this, the infusion cannulae were removed, and the dummy cannulae were replaced to prevent fluid from running back up the guide cannulae. The animal was subsequently placed into the recording arena for a 20 min session (“LIDO” session). The animal was then returned to its home cage, and later a final 20 min recording session was run approximately 90 min following the completion of drug infusion (“POST” session).
In some cases, in order to assess the impact of the infusion process itself on cell firing, saline alone was infused instead of lidocaine. The infusion session for these cases is referred to as a “SALINE” session instead of “LIDO”.
Behavioral analyses
To assess the behavioral effects of ATN inactivation, we considered the following three main factors with regard to the rats' movement, spatial location, and directional heading: (i) spatial coverage of the environment, (ii) average movement speed over the course of the recording session, and (iii) location and direction preferences. Changes in the first two factors may indicate behavioral impairments that need to be addressed when assessing cellular tuning strengths. The third factor may indicate changes in movement patterns that may relate to a loss of allocentric directional processing.
To assess spatial coverage of the environment, we partitioned the environment into 2 × 2 cm bins (60 bins along each axis or 3,600 total bins) and calculated the number of bins sampled by the animal over the course of the recording session. This number was divided by the total number of bins to compute a coverage percentage. We examined this coverage percentage across PRE, LIDO, and POST sessions to determine if spatial coverage differed during the LIDO session. For analysis of cellular tuning strengths, sessions with <80% coverage of the environment were removed from the dataset.
To assess the animals' average movement speed across sessions of the experiment, we first calculated the animals' instantaneous linear speeds for each video frame over the course of each recording session. First, the tracking data was separated into x- and y-components, and a line of best fit was computed for a series of five video frames centered on each time point. The slope of each line was then extracted to indicate the x- or y-component of the velocity vector at that time point. These components were used to compute the animals' instantaneous linear speed at each time point. The average speed for the entire recording session was then computed by taking the average of all instantaneous speeds over the course of the session.
We determined the animals' sampling of space and direction during each recording session. One measure we used was thigmotaxis, which is the tendency for rodents to maintain close proximity to environmental boundaries. We assessed the degree of thigmotaxis for each session by computing the percent of the session that the animal spent within 15 cm of a wall. We also looked at the animals' overall location biases by computing a 2D spatial occupancy histogram, which indicates the amount of time the animal spent in each 2 × 2 cm bin of the environment over the course of a recording session. These histograms often showed biases toward specific locations in the environment, such as corners, which the animals tended to revisit over the course of the session (Fig. 1). We assessed the consistency of these occupancy histograms across sessions by computing the Pearson correlation between them for each pair of sessions (PRE-POST, PRE-LIDO, and POST-LIDO). If ATN inactivation impaired the animals' sense of allocentric orientation, we might expect them to visit different locations during the LIDO session than in the PRE or POST sessions, and therefore we would expect to see high PRE-POST correlations compared to PRE-LIDO and PRE-POST correlations. We also computed an HD occupancy histogram for each session using 12° bins, and compared HD occupancy between sessions using a Pearson correlation. Visually, HD occupancy histograms tended to be much less biased than the spatial occupancy histograms, suggesting that the animals overall sampled the full range of HDs in each session. However, a bias toward certain HDs in the LIDO session could bias the firing of POR LM-HD cells, and therefore it was important to check for differences in HD preferences across sessions.
Cell classifications with a generalized linear model
Cells were initially classified as encoding up to four behavioral variables using 10-fold cross-validation with a Poisson generalized linear model (GLM; Hardcastle et al., 2017; LaChance et al., 2019). The behavioral variables were: allocentric HD, egocentric bearing of the environment center, egocentric distance of the environment center, and linear speed. For each model, the firing rate vector r for a single cell over all time points was modeled as follows:
For cross-validation, data for a session was split into training (9/10 of the session) and test (1/10 of the session) data (k = 10 folds). Parameter vectors were estimated by minimizing the objective function on the training data using the full model with all four variables. Drawing parameter estimates from the full model helps to reduce correlation artifacts between variables (Burgess et al., 2005) and makes models with different variable combinations more comparable. Log-likelihoods for models with all possible variable combinations were computed. This procedure was repeated until all portions of the data had been used as test data (10 folds).
To select the best model, the log-likelihood values from the best two-variable model were compared to those from the best one-variable model. If the two-variable model showed significant improvement from the one-variable model (using a one-sided Wilcoxon signed-rank test), then the best three-variable model was compared to the two-variable model, and so on. If the more complex model was not significantly better, the simpler model was chosen. If the chosen model performed significantly better than an intercept-only model, the chosen model was used as the cell's classification. Otherwise, the cell was marked “unclassified” (Hardcastle et al., 2017).
HD tuning curves and LM-HD cell classification
HD tuning curves were created using 12° bins, typical for analysis of POR cells (LaChance et al., 2019). For each cell, the amount of time that each bin was sampled and the number of spikes fired per bin over the course of a session were calculated, and the tuning curve was computed by dividing the number of spikes per bin by the amount of sampling time per bin. The mean vector length and preferred firing direction (PFD) of the HD tuning curve were extracted to indicate tuning strength and PFD, respectively. A neuron was considered an LM-HD cell if it (i) passed the classification procedure for HD modulation, (ii) had mean vector length >99th percentile of a within-cell shuffle distribution (discussed below), and (iii) had maximum firing rate >1 Hz in its HD tuning curve.
Computation of egocentric bearing
To compute the egocentric bearing of some reference location, we first computed the allocentric bearing of that location from the animal (defined as the angle between the positive x axis with origin centered on the animal and a line drawn from the animal to the reference location) for each time point in the recording session, using the following equation:
Center-bearing cell classification
Center-bearing tuning curves were created using 12° bins. For each cell, the amount of time that each bin was sampled and the number of spikes fired per bin over the course of a session were calculated, and the tuning curve was computed by dividing the number of spikes per bin by the amount of sampling time per bin. The mean vector length and mean angle were then extracted to indicate tuning strength and preferred firing direction, respectively. A neuron was considered a center-bearing cell if it (i) passed the classification procedure for center-bearing modulation, (ii) had mean vector length >99th percentile of a within-cell shuffle distribution (discussed below), and (iii) had maximum firing rate >1 Hz in its center-bearing tuning curve.
Egocentric bearing reference point analysis
Egocentric bearing tuning curves were constructed relative to a 20 × 20 array of potential reference locations (spaced 6 cm apart in x- and y- dimensions). Mean vector lengths were calculated for each tuning curve, and the location with the highest mean vector length was considered the cell's preferred egocentric bearing reference location (“MVLmax location”). The distance of the MVLmax location from the environment center was considered the cell's allocentric MVLmax bias.
Allocentric location rate maps
The animal's location in the environment was binned into 4 cm bins. For each cell, the amount of time that each bin was sampled and the number of spikes fired per bin over the course of a session were calculated, and the rate map was computed by dividing the number of spikes per bin by the amount of sampling time per bin. This rate map was then smoothed with a Gaussian filter. We extracted the peak and mean firing rates, and computed the relative variance of the rate map by dividing its variance by its mean.
Classification of MEC/PaS cells
A small number of cells were recorded from the medial entorhinal cortex or PaS during the LIDO experiment (n = 3 animals). MEC/PaS cells were included in the HD cell analysis if they had a mean vector length >0.2 in both PRE and POST sessions (n = 9 cells), and in the grid cell analysis if they had a grid score >0.4 in both PRE and POST sessions (n = 3 cells). Grid scores were calculated as described previously (Winter et al., 2015).
Directional spike plots
To visualize the directional firing of cells across space, we created directional spike plots that plot the path taken by the animal during foraging (gray trace) overlaid with dots indicating the animal's location when a single cell fired a spike. The dots are colored (circular rainbow color palette) according to the animal's allocentric HD when the spike was fired; red = 0°, green = 90°, blue = 180°, and purple = 270°.
Shuffling procedure
Each cell's spike train was randomly shifted by at least 30 s, with entries beyond the end wrapped to the beginning, to offset the spike data from the behavioral data without interrupting its temporal structure. Relevant tuning scores were then computed based on the shifted spike train (LaChance et al., 2019). This procedure was repeated 400 times for each cell, and a within-cell 99th percentile cutoff was used to determine tuning significance for individual cells.
Statistics
Statistical analyses were performed using Python code. All tests were two sided and used an α level of 0.05 (except for GLM classifier cross-validation comparisons, which were one sided; Hardcastle et al., 2017; LaChance et al., 2019). Comparisons across sessions were made using a one-way repeated-measures ANOVA, with Greenhouse–Geisser correction applied if samples violated sphericity. Post hoc pairwise comparisons were performed using Bonferroni-corrected paired t tests.
Results
We recorded 240 single neurons from the POR of five rats as they completed a baseline 20 min foraging session in a 1.2 × 1.2 m square enclosure (“PRE” session). Of these 240 cells, 118 were classified as encoding the animal's allocentric HD (49%; LM-HD cells), 170 were classified as encoding the egocentric bearing of the environment centroid (71%; center-bearing cells), and 44 were classified as encoding the egocentric distance of the environment centroid (18%; center-distance cells). Many of these cells were conjunctively tuned to multiple variables (Table 1). On days when one or more of these functional cell types was recorded, we subsequently inactivated the ATN by bilaterally infusing lidocaine and performed another recording session with the ATN inactivated (“LIDO” session). This LIDO session was followed by a final baseline session, which began ∼90 min following the completion of lidocaine infusion (“POST” session).
Summary of conjunctive cell types recorded in baseline session
Behavioral effects of ATN inactivation
We first examined whether lidocaine infusion produced any observable behavioral effects by looking for motor deficits, which could result in a reduction in spatial sampling of the environment or a reduction in movement speed. Across 26 inactivation experiments from the five rats, we found no difference in spatial coverage of the environment across the PRE, LIDO, and POST sessions (F(2,50) = 0.56, p = 0.58), as well as no difference in average movement speed across the three sessions (F(2,50) = 2.00, p = 0.15; Fig. 1D). From this analysis we conclude that the animals were able to locomote just as quickly and extensively during the ATN inactivation as they were during the nondrug sessions.
We next determined if ATN inactivation caused a change in the animals' location or heading preferences due to disruption of allocentric spatial processing. To assess if the animals displayed different location preferences across the sessions, we created two-dimensional (2D) occupancy histograms that indicated how much time the animals spent in each 2 × 2 cm bin of the enclosure (Fig. 1C). We then computed the correlation between the occupancy histograms for each pair of sessions (PRE-POST, PRE-LIDO, and POST-LIDO). We found that correlations were significantly lower when comparing LIDO with both PRE and POST sessions than when comparing PRE and POST to each other (F(2,50) = 10.66, p = 1.37 × 10−3; PRE-POST vs PRE-LIDO t(25) = 3.13, p = 0.013; PRE-POST vs POST-LIDO t(25) = 3.68, p = 3.40 × 10−3; PRE-LIDO vs POST-LIDO t(25) = 0.85, p > 0.99; Fig. 1E). Thus, despite covering a similar extent of the environment across sessions, the animals tended to spend time in different locations in the LIDO sessions compared to PRE or POST sessions. This difference was not due to a change in thigmotaxic behavior, as the percent of time spent within 15 cm of a wall did not differ across the sessions (F(2,50) = 3.13, p = 0.06; Fig. 1D). Instead, the change in the pattern of spatial occupancy may be due to a disruption of the animals' processing of allocentric location.
We also assessed if the animals showed any change in their heading preferences across sessions. For this purpose, we created an occupancy histogram for each session that indicated how much time the animals spent sampling each 12° HD bin. Unlike the spatial occupancy histograms, which showed clear biases toward specific locations in the environment, the HD occupancy histograms were largely unbiased across the PRE and POST sessions (mean PRE mean vector length: 0.11, mean LIDO mean vector length: 0.09, mean POST mean vector length: 0.12). Importantly, when we computed the correlation between the HD occupancy histograms for each pair of sessions (PRE-POST, PRE-LIDO, and POST-LIDO), we found no difference in correlations across the different pairs of sessions (F(2,50) = 0.11, p = 0.89; Fig. 1E). This result suggests that any significant changes in the tuning of HD cells due to ATN inactivation is not due to a change in the animals' orientation preferences across sessions.
Effects of ATN inactivation on orientation tuning in POR
We recorded 131 POR cells (n = 5 rats) across the PRE, LIDO, and POST sessions (n = 26 complete experiments). However, after establishing a cutoff for minimum spatial coverage of the environment for every session (>80% coverage), that number decreased slightly to 117 POR cells recorded across 18 complete lidocaine-infusion experiments. Of these 117 cells, 52 cells (44%) were classified as encoding the animal's allocentric HD in both the PRE and POST sessions, and were therefore used for analysis. In agreement with previous experiments (LaChance et al., 2019, 2022), a large proportion of the POR LM-HD cells showed conjunctive tuning to center-bearing (41 of 52 cells; 79%).
We hypothesized that the ATN might provide excitatory drive to LM-HD cells in POR, and therefore we would expect to see a decrease in peak firing rate during the LIDO session compared to the PRE and POST sessions. This hypothesized response is what we observed (F(2,102) = 29.22, p = 9.26 × 10−11; PRE vs LIDO t(51) = −5.88, p = 9.35 × 10−7; POST vs LIDO t(51) = −7.43, p = 3.38 × 10−9; Fig. 2A–D). LM-HD cells showed an average peak firing rate decrease of 25.21 ± 4.10% in the LIDO session compared to the PRE session. Peak firing rates did not differ between the PRE and POST sessions (t(51) = 0.47, p > 0.99). We also saw a reduction in tuning strength following lidocaine infusion, as measured using mean vector length (F(2,102) = 21.03, p = 1.05 × 10−7; PRE vs LIDO t(51) = −5.41, p = 5.03 × 10−6; POST vs LIDO t(51) = −4.88 p = 3.30 × 10−5; Fig. 2C). Mean vector lengths did not differ between PRE and POST sessions (t(51) = −1.94, p = 0.17). Despite these overall effects, it should be noted that POR LM-HD cells displayed heterogeneity in their responses to ATN inactivation, with strongly disrupted cells sometimes being simultaneously recorded with cells that appeared unaffected by the manipulation (Fig. 2A,B). However, it is clear that ATN inactivation overall reduces the firing rates and tuning strengths of POR LM-HD cells (Table 2).
ATN inactivation disrupts POR LM-HD cells. A, Directional spike plots and HD tuning curves for three simultaneously recorded LM-HD cells across the PRE-LIDO-POST sequence that showed a reduction in firing rate and tuning strength in the LIDO session. B, HD tuning curves for four LM-HD cells recorded simultaneously with the cells in (A) showing heterogeneous responses to ATN inactivation. C, Normalized HD tuning curves for all LM-HD cells recorded in the PRE-LIDO-POST sequence. D, Change in peak firing rates (Δ PFR)(left) and mean vector lengths (Δ MVL)(right) in the LIDO and POST sessions compared to the PRE session. Note that both peak firing rates and mean vector lengths decreased in the LIDO session. E,F, Same as (C,D), but for LM-HD cells recorded in the PRE-SALINE-POST sequence. Note that no significant changes occurred in the SALINE session. Individual data points (in D and F) represent all individual LM-HD cells recorded from all animals. * denotes statistical significance.
Summary of LIDO effects on head direction tuning
Because lidocaine works most effectively immediately following infusion and can lose its effectiveness over the course of a recording session, we examined whether the effects on peak firing rate and mean vector length for POR LM-HD cells decreased over the course of the recording session. We split the LIDO session into four 5 min blocks (Fig. 3A,B) and computed peak firing rate and mean vector length values for each block. We then compared these values to the final 5 min block of the PRE session. We found that peak firing rates were significantly reduced during the first three blocks of the session and did not significantly differ from baseline in the fourth block (F(4,204) = 18.26, p = 3.37 × 10−9; Block 1 t(51) = −6.44, p = 4.15 × 10−7; Block 2 t(51) = −5.06, p = 5.75 × 10−5; Block 3 t(51) = −3.69, p = 5.49 × 10−3; Block 4 t(51) = −2.87, p = 0.060; Fig. 3C), suggesting that the peak firing rates of POR LM-HD cells returned to baseline over the course of the 20 min LIDO session. Curiously (although similar to previous results in MEC; Winter et al., 2015), mean vector lengths among POR LM-HD cells did not show a time-dependent effect and were only significantly reduced during the second and fourth blocks of the session (F(4,204) = 7.88, p = 5.60 × 10−5; Block 1 t(51) = −2.07, p = 0.434; Block 2 t(51) = −5.47, p = 1.37 × 10−5; Block 3 t(51) = −2.64, p = 0.11; Block 4 t(51) = −3.84, p = 3.35 × 10−3; Fig. 3C). This result may be due to stochasticity or noise in the firing of the cells that becomes more apparent when firing rates are overall reduced and tuning curves are constructed using data from short time periods (i.e., 5 min blocks).
Time-dependent effects of ATN inactivation on POR LM-HD cells. A, Directional spike plots and HD tuning curves computed for the fourth quarter of the PRE session (PRE B4) and all four quarters of the LIDO session (B1, B2, B3, B4) showing a general increase in firing rate over the course of the LIDO session. Cells in the second and third row were simultaneously recorded. B, Normalized HD tuning curves for all LM-HD cells computed for the fourth quarter of the PRE session (PRE B4) and all four quarters of the LIDO session (B1, B2, B3, B4). Note that the HD signal was largely lost in LIDO B1, but strengthened significantly by LIDO B4. C, Change in peak firing rates (Δ PFR) (left) and mean vector lengths (Δ MVL) (right) in all four quarters of the LIDO session compared to the fourth quarter of the PRE session. Note that peak firing rates tended to increase over time, while mean vector lengths showed a less clear pattern, although they were generally decreased across all four 5 min blocks. D,E) Same as (B,C), but for LM-HD cells recorded in the PRE-SALINE-POST sequence. Note that no significant changes occurred during any block of the SALINE session. Individual points in C and E represent all individual LM-HD cells recorded from all animals. * denotes statistical significance.
To ensure that the effects of ATN inactivation on POR HD tuning were not due to the infusion process itself, we recorded 26 POR LM-HD cells in a series of sessions where saline was infused instead of lidocaine (session order: PRE-SALINE-POST; Fig. 2D). These cells showed no difference in peak firing rate (F(2,50) = 0.91, p = 0.38) or mean vector length (F(2,50) = 0.81, p = 0.45; Fig. 2E) across any of the sessions, indicating that the infusion process itself was not responsible for the change in POR LM-HD cell firing properties. We also found no change in peak firing rate (F(4,100) = 0.40, p = 0.76) or mean vector length (F(4,100) = 1.07, p = 0.37) across any of the 5 min blocks of the SALINE session (Fig. 3D,E).
Given these results, it is clear that ATN inactivation disrupts the firing of POR LM-HD cells, but does it also impact egocentric tuning in POR? To address this issue, we focused on LM-HD cells that also encoded egocentric center-bearing (i.e., conjunctively tuned). Analyzing conjunctive cells should allow us to assess if ATN inactivation had disparate effects on allocentric (HD) and egocentric (center-bearing) tuning within the same cell (Fig. 4A). As these conjunctive cells (n = 41) made up the majority of the LM-HD cells recorded in the experiment, it was not surprising that they showed a significant decrease in both HD peak firing rate (F(2,80) = 28.08, p = 5.77 × 10−10; PRE vs LIDO t(40) = −6.18, p = 7.96 × 10−7; POST vs LIDO t(40) = −7.02, p = 5.23 × 10−8) and HD mean vector length (F(2,80) = 17.55, p = 2.00 × 10−6; PRE vs LIDO t(40) = −4.93, p = 4.43 × 10−5; POST vs LIDO t(40) = −4.32, p = 3.01 × 10−4) during the LIDO session, despite consistent values between PRE and POST sessions (peak firing rate PRE vs POST t(40) = −0.15, p > 0.99; mean vector length PRE vs POST t(40) = −2.01, p = 0.15; Fig. 4B,C). We expected that this decrease in firing rate would also decrease the peak firing rates of the conjunctive cells' center-bearing tuning curves (Fig. 4D), but if the ATN was not critical for supporting center-bearing tuning, we should not see a reduction in the cells' center-bearing mean vector lengths. This effect is what we found: center-bearing peak firing rates for conjunctive cells were reduced in the LIDO session (F(2,80) = −17.25, p = 5.90 × 10−7; PRE vs LIDO t(40) = −5.00, p = 3.51 × 10−5; POST vs LIDO t(40) = −5.51, p = 6.87 × 10−6), but their mean vector lengths were not significantly impacted (F(2,80) = 3.32, p = 0.048; all pairwise comparisons p > 0.05; Fig. 4E). Center-bearing peak firing rates did not differ between PRE and POST sessions (PRE vs POST t(40) = −0.54, p > 0.99). Thus, ATN inactivation causes a significant reduction in overall firing rate and directional tuning strength (i.e., HD aspects of firing) for POR conjunctive cells, but it does not impact center-bearing tuning strength (Tables 2, 3).
ATN inactivation preferentially disrupts the HD tuning of conjunctive cells. A, Directional spike plots, HD tuning curves, and center-bearing tuning curves for four example cells recorded across the PRE-LIDO-POST sequence that were tuned conjunctively to HD and center-bearing. Cells in the top two rows were simultaneously recorded. B, Normalized HD tuning curves for conjunctive cells recorded across the PRE-LIDO-POST sequence (C) Change in HD peak firing rates (Δ PFR) (left) and HD mean vector lengths (Δ MVL) (right) in the LIDO and POST sessions compared to the PRE session. Note the reduction in both peak firing rate and mean vector length in the LIDO session. D,E, Same as (B,C), but for center-bearing tuning curves, center-bearing peak firing rates, and center-bearing mean vector lengths. Note that there was a decrease in peak firing rate, but not mean vector length in the LIDO session. Individual points (in C and E) represent all individual conjunctive cells recorded from all animals. * denotes statistical significance.
To further assess the impact of ATN inactivation on egocentric processing in POR, we focused on cells that were tuned to center-bearing without any allocentric HD sensitivity (“pure” center-bearing cells; n = 36 cells; Fig. 5A,B). In contrast to LM-HD cells, pure center-bearing cells did not show any overall change in peak firing rate across the sessions (F(2,70) = 2.37, p = 0.10; Fig. 5C). However, they did show a small but significant decrease in mean vector lengths between the PRE and LIDO sessions (F(2,70) = 5.30, p = 7.18 × 10−3; PRE vs LIDO t(35) = −2.94, p = 0.017), although no difference was observed between POST and LIDO or between PRE and POST sessions (POST vs LIDO t(35) = −1.42 p = 0.49; PRE vs POST t(35) = −2.11, p = 0.12; Fig. 5C; Table 3). This absence of a strong and coherent overall response of POR egocentric cells to ATN inactivation supports a stronger role for the ATN in driving HD representations in POR.
ATN inactivation largely spares center-bearing tuning. A, Directional spike plots and center-bearing tuning curves for three example “pure” center-bearing cells recorded across the PRE-LIDO-POST sequence that did not show a change in PFR or MVL across the sessions. B, Normalized center-bearing tuning curves for “pure” center-bearing cells recorded across the PRE-LIDO-POST sequence. C, Change in center-bearing peak firing rates (Δ CB PFR) (left) and center-bearing mean vector lengths (Δ CB MVL) (right) in the LIDO and POST sessions compared to the PRE session. Note that there was no change in peak firing rate during the LIDO session, but there was a significant decrease in mean vector length in the LIDO session compared to the PRE session, but not compared to the POST session. Individual points in C represent all individual “pure” center-bearing cells recorded from all animals. * denotes statistical significance, n.s. = not significant.
Summary of LIDO effects on center-bearing tuning
Effects of ATN inactivation on allocentric spatial coding in POR
It has been proposed that POR neurons may combine egocentric bearing and allocentric HD signals to represent specific allocentric locations throughout the environment (LaChance et al., 2019; LaChance and Taube, 2023). One way this process might occur within conjunctive cells is by using an HD representation to shift the reference point of the center-bearing signal from the environment center to some other location. Figure 6A shows how summing an HD signal with a center-bearing signal effectively shifts the reference point of the representation from the environment center to some other location, according to the intensity and direction of the HD signal (LaChance and Taube, 2023). The resulting conjunctive representation indicates the egocentric bearing of a specific allocentric location, and could be used by downstream brain regions to guide behavior toward that location or to construct a “place” representation (O’Keefe, 1991).
ATN inactivation impairs egocentric coding of allocentric locations in POR. A, Top-bottom schematic illustrating how center-bearing and HD signals can combine to produce a representation specifying the egocentric bearing of an allocentric location (denoted by a red X). Arrow directions and lengths represent the preferred firing directions and firing rates, respectively, for idealized cells at different allocentric locations. B, Top-bottom schematic illustrating the measurement of egocentric bearing relative to a 20 × 20 array of potential egocentric bearing reference points spaced throughout the environment. C, Top, directional spike plot for a “pure” center-bearing cell; bottom, MVL map showing the mean vector lengths of egocentric bearing tuning curves for the same cell constructed relative to the 20 × 20 array of locations shown in (B). Note that this particular cell has its highest MVL (MVLmax location) in the center of the environment. D, Histogram of MVLmax locations for all “pure” center-bearing cells recorded across the three sessions of the inactivation experiment. E, Scatter plots comparing each cell’s MVLmax bias (distance of MVLmax location from environment center) between PRE and both LIDO (left) and POST (right) sessions. F, Change in MVLmax bias for all “pure” bearing cells in the LIDO and POST sessions compared to the PRE session. G, Same as (C), but for a conjunctive cell. Note that this cell’s MVLmax location is offset toward the east wall of the environment. H–J) same as (D–F), but for conjunctive cells. Note that MVLmax biases are significantly reduced in the LIDO session, suggesting that HD tuning was disrupted relative to center-bearing tuning. Individual points in (E,F) represent all individual “pure” center-bearing cells recorded from all animals, while those in (I,J) represent all individual conjunctive cells. * denotes statistical significance.
To assess the egocentric encoding of specific allocentric locations by POR cells, we created egocentric bearing tuning curves relative to a 20 × 20 array of potential reference locations spread uniformly across the environment, and selected the location with the highest mean vector length as each cell’s preferred egocentric bearing reference point (“MVLmax location”; Fig. 6B,C). For “pure” center-bearing cells, which lack an HD component, MVLmax locations tended to be concentrated in the center of the environment across all three sessions of the experiment (Fig. 6C,D). To quantify this effect, we calculated the distance of the MVLmax location from the environment center, which we refer to as the MVLmax bias. For “pure” center-bearing cells, there was no change in MVLmax bias across the three sessions (F(2,70) = 0.19, p = 0.83; Fig. 6E,F), suggesting that they were consistently tuned to the egocentric bearing of the environment center.
In contrast, conjunctive cells (LM-HD x center-bearing) tended to encode MVLmax locations that were scattered across the environment (Fig. 6G,H). Critically, the MVLmax biases of the conjunctive cells were significantly reduced in the LIDO session compared to the PRE and POST sessions (F(2,80) = 10.00, p = 1.33 × 10−4; PRE vs LIDO t(40) = −4.05, p = 6.83 × 10−4; POST vs LIDO t(40) = −3.18, p = 8.55 × 10−3; PRE vs POST t(40) = −1.47, p = 0.45; Fig. 6I,J), such that their MVLmax locations tended to contract toward the center of the environment in the LIDO session. Thus, ATN inactivation appeared to cause conjunctive cells to behave more like “pure” center-bearing cells in terms of their egocentric bearing preferences, while their allocentric coding was reduced.
Another way that allocentric HD and egocentric bearing can combine to produce allocentric spatial coding is by biasing cells to fire more strongly in specific allocentric locations. While a “pure” center-bearing cell or a “pure” HD cell would be expected to fire relatively uniformly over the surface of a given environment (assuming uniform sampling of center-bearing and HD in those locations), a conjunctive cell should show a varied firing rate over the surface, with higher firing rates in locations where the cell’s preferred HD and center-bearing are most closely aligned. For example, a cell that fires when the animal is facing north and when the animal is facing the environment center will have its highest firing rate when the animal is in the southern portion of the environment. Removing the HD inputs to a conjunctive cell by silencing the ATN would be expected to reduce this firing variance, causing the cell’s firing to be more uniformly distributed throughout allocentric space.
To assess the possibility of ATN inactivation disrupting allocentric location coding by POR conjunctive cells, we began by computing Pearson correlations between the allocentric rate maps of the conjunctive cells across the PRE, LIDO, and POST sessions to determine if firing patterns differed across the sessions. The LIDO rate maps were significantly less correlated with PRE and POST sessions than the PRE and POST rate maps were with each other (F(2,80) = 6.91, p = 1.70 × 10−3; PRE-POST vs PRE-LIDO t(40) = −2.89, p = 0.019; PRE-POST vs POST-LIDO t(40) = −3.18, p = 8.50 × 10−3; PRE-LIDO vs POST-LIDO t(40) = −0.29, p > 0.99; Fig. 7A,B), suggesting that the conjunctive cells’ allocentric location preferences changed somehow in the LIDO session. Much like with the HD tuning curves, we also observed a decrease in rate map peak firing rates in the LIDO session (F(2,80) = 36.72, p = 4.86 × 10−12; PRE vs LIDO t(40) = −7.13, p = 3.72 × 10−8; POST vs LIDO t(40) = −7.40, p = 1.59 × 10−8; PRE vs POST t(40) = 0.50, p > 0.99; Fig. 7C), as well as the mean firing rate of the rate maps (F(2,80) = 12.73, p = 1.59 × 10−5; PRE vs LIDO t(40) = −4.50, p = 1.74 × 10−4; POST vs LIDO t(40) = −4.18, p = 4.67 × 10−4; PRE vs POST t(40) = 0.85, p > 0.99; Fig. 7D). To determine if the allocentric rate maps became more uniform in the LIDO session while controlling for these changes in firing rate, we computed the relative variance of each rate map (variance normalized by mean firing rate). Relative variance was significantly decreased in the LIDO session (F(2,80) = 27.08, p = 1.05 × 10−9; PRE vs LIDO t(40) = −6.25, p = 6.35 × 10−7; POST vs LIDO t(40) = −6.30, p = 5.44 × 10−7; PRE vs POST t(40) = −0.73, p > 0.99; Fig. 7E). Taken together, these results suggest that ATN inactivation had a disruptive effect on allocentric location coding by POR conjunctive cells, reducing their allocentric location firing preferences.
ATN inactivation disrupts the allocentric location firing biases of POR cells. A, Directional spike plots (top) and allocentric location rate maps (bottom) for four example conjunctive cells that displayed a visual reduction in their allocentric location firing biases in the LIDO session. B, Correlations between pairs of rate maps for all conjunctive cells recorded across all three sessions of the inactivation experiment. C, Change in rate map peak firing rate in the LIDO and POST sessions compared to the PRE session. D, Same as (C), but for the mean of the rate map. E, Same as (C), but for the relative variance of the rate map. F, Directional spike plots (top) and allocentric location rate maps (bottom) for two example “pure” center-bearing cells. G–J, Same as (B–E), but for all “pure” center-bearing cells. Individual data points in (B–E) represent all individual conjunctive cells recorded from all animals, while those in (G–J) represent all individual “pure” center-bearing cells. * denotes statistical significance.
While “pure” center-bearing cells also showed overall lower correlations of their location rate maps between PRE-LIDO and POST-LIDO compared to PRE-POST (F(2,70) = 8.66, p = 9.32 × 10−4; PRE-POST vs PRE-LIDO t(35) = −3.51, p = 3.77 × 10−3; PRE-POST vs POST-LIDO t(35) = −2.75, p = 0.028; PRE-LIDO vs POST-LIDO t(35) = 1.60, p = 0.36; Fig. 7F,G), there was no overall difference in rate map peak firing rate (F(2,70) = 1.99, p = 0.14; Fig. 7H), mean firing rate (F(2,70) = 2.14, p = 0.14; Fig. 7I), or relative variance (F(2,70) = 4.89, p = 0.017; all pairwise tests p > 0.05; Fig. 7J). The lack of a clear overall effect on POR egocentric cells further highlights the ATN’s coherent control over the firing rates and tuning strengths of POR LM-HD cells.
Effects of ATN inactivation on allocentric coding in MEC/PaS
In some animals (n = 3), tetrodes were advanced beyond the ventral border of POR into the medial entorhinal cortex or parasubiculum (MEC/PaS), where a small number of HD cells (n = 9) and grid cells (n = 3) were recorded during the inactivation experiment. In agreement with a previous study (Winter et al., 2015), HD cells showed a significant decrease in peak firing rate during the LIDO session (F(2,16) = 9.23, p = 2.16 × 10−3; PRE vs LIDO t(8) = −3.94, p = 0.013; POST vs LIDO t(8) = −3.99, p = 0.012; PRE vs POST t(8) = −0.68, p > 0.99), although the decrease in mean vector length was only significant relative to the POST session (F(2,16) = 9.55, p = 1.87 × 10−3; PRE vs LIDO t(8) = −2.85, p = 0.064; POST vs LIDO t(8) = −6.06, p = 9.03 × 10−4; PRE vs POST t(8) = 0.34, p > 0.99; Fig. 8A,B). Relative to their PRE peak firing rates, the MEC/PaS HD cells showed a decrease in peak firing rate that was similar to the most affected POR LM-HD cells (Fig. 8C). The three grid cells all showed decreased grid scores in the LIDO session compared to PRE or POST (Fig. 8D), which is similar to the findings reported in the earlier study.
ATN inactivation disrupts allocentric spatial tuning in MEC/PaS. A, HD tuning curves for two example MEC/PaS HD cells that were recorded across the three sessions of the inactivation experiment. B, Change in HD peak firing rates (Δ PFR) (left) and HD mean vector lengths (Δ MVL) (right) in the LIDO and POST sessions compared to the PRE session for all recorded MEC HD cells. C, Comparison with POR LM-HD cells. Peak firing rates (left) and mean vector lengths (right) have been normalized relative to their PRE values in order to control for potential firing rate differences between the brain areas. Individual data points represent all POR LM-HD cells (left of each plot) or all MEC/PaS HD cells (right of each plot). D, Path and spike plots (top) and allocentric location rate maps (bottom) for an example grid cell recorded across the three sessions of the inactivation experiment. Note the disruption of the grid pattern in the LIDO session. E, Change in grid score in the LIDO and POST sessions compared to the PRE session for all three grid cells. * denotes statistical significance.
Discussion
Our results demonstrate that POR LM-HD tuning is driven at least partially by either direct or indirect inputs from the ATN, with ATN inactivation disrupting allocentric orientation and spatial coding by these cells. It is therefore likely that the directional tuning properties of POR LM-HD cells are derived in part from vestibular-based allocentric HD cells in the ATN. In contrast, egocentric center-bearing tuning in POR was not coherently or consistently affected by ATN inactivation, suggesting that allocentric tuning properties of POR cells are preferentially impacted by the HD circuit.
It is important to note that POR LM-HD cells did not completely become quiescent during ATN inactivation, and while their tuning was disrupted, overall they continued to fire directionally according to their established HD preferences (i.e., maintaining the same preferred firing direction as in the PRE session). This result contrasts somewhat with previous results regarding the impact of ATN inactivation on HD cells in MEC, which appeared to cease firing completely at the start of the inactivation session (Winter et al., 2015). One possibility is that our inactivations were not as complete or effective as those in the previous study. Alternatively, this inconsistency may reflect different inputs targeting the two areas; while POR is primarily targeted by visual and visuospatial (i.e., egocentric) areas along with more moderate inputs from areas associated with allocentric spatial coding, MEC is primarily targeted by inputs from areas that are associated with allocentric spatial coding, such as the postsubiculum (van Groen & Wyss, 1990; Agster and Burwell, 2013). Given this pattern of connections, we might expect the MEC to contain a more “classic” HD representation that depends strongly on the ATN, while the POR LM-HD signal is more strongly defined by visual elements of the world that remain when the ATN is inactivated. The POR LM-HD signal likely represents a confluence between allocentric HD and egocentric visual inputs. One important future experiment will be to record from POR LM-HD cells during ATN inactivation in darkness. If visual inputs are responsible for the residual directional firing among POR LM-HD cells when the ATN is inactivated, then removing the influence of visual cues by recording in darkness may cause the LM-HD cells to lose their directional tuning completely. In either case, it is important to keep in mind that if a cell is firing at a lower rate than previously, then its impact on downstream neurons is going to be lessened; consequently, other inputs (i.e., non-HD ones) will have an increased impact on cell firing.
The results of this study help to contextualize our previous findings regarding the visual landmark-referenced firing properties of POR LM-HD cells. For example, when a familiar visual landmark was removed from the recording environment, many POR LM-HD cells showed a significant decrease in firing rate but maintained their orientation preferences relative to the previous location of the landmark (LaChance et al., 2022). This retained tuning suggested that the cells maintained a sense of the animal’s true allocentric orientation despite changes in the visual scene; while we had removed the visual component of POR LM-HD firing, a directional component remained. This result may be due to inputs from allocentric HD cells. Likewise, when we attempted to remove the directional component of LM-HD firing in the current study using ATN inactivation, we found a similar decrease in firing rate and tuning strength, but the cells' overall tuning preferences (PFDs) remained consistent with the initial session, likely due to remaining tuning to visual cues still present in the environment. To fully disrupt POR HD tuning, both ATN inactivation and visual cue removal might be necessary.
Interestingly, training animals with two identical cues from the beginning in our previous experiment caused the LM-HD signal to be largely unidirectional, suggesting that the LM-HD signal was not tied to each cue individually, but instead estimated a unidirectional HD signal based on the entire visual scene (Yan et al., 2021; LaChance et al., 2022). This unidirectional tuning despite symmetrical cues may be attributed to the influence from the ATN-based HD signal. It should be noted, however, that ATN HD cells were always robustly unidirectional in the face of visual cue manipulations (LaChance et al., 2022). We previously demonstrated that lesioning POR neither disrupted HD cell firing in the ATN nor the control of their PFDs by a prominent visual landmark (Peck and Taube, 2017). Taken together, our current study demonstrates that there is functional connectivity between POR and ATN in terms of HD tuning, but it appears to be unidirectional, such that the ATN contributes its HD representation to POR, but does not depend on inputs from POR to maintain its own HD signal. It remains to be determined whether other brain regions depend on POR input to shape or maintain their orientation preferences.
While ATN inactivation disrupted the LM-HD signal in POR, it remains unknown whether this result is due to monosynaptic projections from ATN, or projections from other brain areas that also receive ATN inputs. Examples of such regions that contain HD cells and project moderately or strongly to POR include the postsubiculum (Taube et al., 1990; Agster and Burwell, 2013), caudal PaS (Cacucci et al., 2004; Agster and Burwell, 2013), retrosplenial cortex (Burwell and Amaral, 1998a; Cho and Sharp, 2001), and posterior parietal cortex (Burwell and Amaral, 1998a; Wilber et al., 2014; Fig. 9). It is also possible that the disrupted HD tuning following lidocaine infusion was partially due to silencing of fibers of passage, and so future experiments should infuse a chemical other than lidocaine that is more selective for cell bodies and for which the area of effect can be more readily visualized (e.g., muscimol; Frost et al., 2021).
Potential routes for egocentric and allocentric spatial signals to reach POR. Schematic diagram showing possible pathways by which HD signals (red arrows) and egocentric bearing signals (blue arrows) might reach POR. Note that HD tuning may be received directly from the ATN or via indirect connections through the subicular complex or retrosplenial cortex.
Egocentric center-bearing cells without HD tuning did not exhibit the coherent decrease in firing rate and tuning strength shown by LM-HD cells, though some cells showed a small but inconsistent decrease in egocentric bearing MVLs during the inactivation (Fig. 5C). This effect could reflect a loss of indirect ATN inputs via brain areas that combine HD and egocentric bearing tuning, such as the postsubiculum (Peyrache et al., 2017) or the PaS (Gofman et al., 2019), which contain HD cells that fire preferentially when the animal is near a particular wall. Otherwise, POR egocentric tuning may derive from projections from other brain areas that contain similar “pure” egocentric bearing-tuned cells, such as the retrosplenial cortex (Alexander et al., 2020), or from more canonical visual areas of the brain, such as the tectal visual pathway involving the lateral posterior thalamic nucleus, which provides the largest subcortical input to POR (more so than ATN; Pereira et al., 2016). This tectal pathway conveys visual information from the superior colliculus to POR (Beltramo and Scanziani, 2019; Fig. 9). Much like the LM-HD signal, the true answer is likely a combination of different inputs.
We have proposed previously that POR may combine egocentric and allocentric spatial variables in order to support the construction of allocentric spatial maps, the disambiguation of spatial contexts, or the guidance of goal-directed navigation (LaChance et al., 2019, 2022; LaChance and Taube, 2022, 2023). The current study provides strong support for this proposal by demonstrating a dissociation between the allocentric HD signal and the egocentric center-bearing signal in POR, even within individual conjunctively tuned neurons. In addition, ATN inactivation disrupted allocentric spatial coding by conjunctive POR cells in two ways: (1) by causing the conjunctive cells' preferred egocentric bearing reference locations to contract toward the center of the environment (Fig. 6) and (2) by reducing the conjunctive cells’ allocentric location firing biases (Fig. 7). Both of these effects would be expected to impair allocentric coding in downstream regions that may depend on POR input, including potentially the entorhinal cortex (Burwell and Amaral, 1998b; Koganezawa et al., 2015; Doan et al., 2019) or subiculum (Naber et al., 2001), both of which show disrupted spatial tuning in response to ATN inactivation (Winter et al., 2015; Frost et al., 2021). Overall, our findings provide insights into how largely unimodal representations such as the vestibular-based ATN HD signal may be integrated into a more abstract polymodal representation such as the POR LM-HD signal.
Footnotes
This work was supported by the National Institute of Neurological Disorders and Stroke (NINDS)(NS053907) and National Institute on Deafness and Other Communication Disorders (NIDCD) (DC009318).
The authors declare no competing financial interests.
- Correspondence should be addressed to Jeffrey S. Taube at jeffrey.s.taube{at}dartmouth.edu.
