(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



Subicular neurons represent multiple variables of a hippocampal-dependent task by using theta rhythm

['Su-Min Lee', 'Department Of Brain', 'Cognitive Sciences', 'Seoul National University', 'Seoul', 'Jae-Min Seol', 'Inah Lee']

Date: 2022-02 

( A ) VSM task. As a trial begins, the rat runs out onto the track of a T-maze from the start box (S), and one of 4 visual scene stimuli (Zebra, Z; Bamboo, B; Pebbles, P; Mountain, M) is presented on LCD monitors. Each scene stimulus is associated with either the left or right arm of the T-maze. ( B ) Behavioral performance during recording sessions (21 sessions from 5 rats). Each dot corresponds to the percent correct for each scene stimulus of a session and is color coded for individual rats. Box plot indicates interquartile range and median value. The median values exceeded the performance criterion (dashed line, 75%) for all scenes. ( C ) Photomicrographs of Nissl-stained coronal brain sections with verified electrode tips (black arrows). Numbers above the arrows indicate normalized positions of marked recording sites along the proximodistal axis. Dashed lines represent the anatomical boundaries of the CA1 and subiculum. Upper and lower rows show recording sites from the subiculum and CA1, respectively. ( D ) Proportional distribution of cells recorded in the CA1 (blue) and SUB (red) along the proximodistal axis (CA1, n = 270; SUB, n = 151). The positions are normalized to account for differences in relative length between 2 regions. The dashed line at 0.36 indicates the boundary between 2 regions. Data associated with this figure can be found in S1 Data file. SUB, subiculum; VSM, visual scene memory.

In the VSM task, rats (n = 5) learned to associate each scene stimulus with either a left or right turn response on the T-maze ( Fig 1A ). During recording sessions, rats performed the VSM task well above performance criterion (75%) for all stimuli (p-values < 0.0004 for all scenes, one-sample Wilcoxon signed-rank test; Fig 1B ). Tetrodes located at the boundaries of either the CA1 or subiculum (including the border between them) were excluded from the analysis ( Fig 1C ). To quantify the anatomical distributions of recording locations for the CA1 and subiculum along the proximodistal axis, we measured the relative positions from which individual cells were recorded and normalized them across rats ( Fig 1D ). Only complex-spiking cells satisfying our unit-filtering criteria (CA1, n = 270; subiculum, n = 151; see Methods ) were used for analysis. Subicular cells were found along the entire proximodistal axis, whereas CA1 cells were mainly recorded from the intermediate to proximal portions of the CA1. More details can be found in our previous study [ 11 ].

( A , B ) Firing rate maps of single cells (left) and cell populations (right) in the CA1 (A) and subiculum (B), plotted as a function of the linearized position on the T-maze from the st box to the fw in both arms. Red arrowheads indicate the choice point after which rats’ positions diverged between the left and right choice trials. On the firing rate maps of single cells, legitimate place fields are overlaid with thick black lines, and non-place fields that did not pass the place field criteria are marked by gray lines. Serial numbers on the upper left corner are cell IDs. Population firing rate maps are sorted according to peak firing rate of each cell on the T-maze. White dashed lines and red arrowheads indicate the choice points. ( C ) Proportional differences of place cells between the subiculum and CA1, defined by the firing rate–based method. Cells are classified into 3 groups according to the number of place fields per cell: “SF” for one field, “MF” for more than one field, and “NF” for zero field. ***p < 0.0001. ( D–F ) Differences in mean firing rate (D), SI score (E), and place field width (F) of recorded cells between the CA1 and subiculum. ***p < 0.0001. Data associated with this figure can be found in S1 Data file. fw, food well; MF, multi-field; NF, non-place field; SF, single-field; SI, spatial information; st box, start box; SUB, subiculum.

To quantitatively compare differential firing patterns between the 2 regions, we first classified cells according to the number of place fields: no place field, a single field, or multiple fields. A spatial firing distribution was considered a place field if its peak firing rate exceeded 1 Hz and its spatial information content (bits/spike) exceeded 0.5. Field boundaries were set at the spatial bins in which the associated firing rates dropped below 33% of the peak firing rate (see Methods ). Of cells that were active during the rat’s outbound journey on the T-maze, approximately 90% were single-field cells in the CA1, while only about half of cells exhibited either single- or multi-fields in the subiculum ( = 122.96, p < 0.0001; chi-squared test; Fig 2C ). With respect to basic firing properties, cells in the CA1 showed lower firing rates (Z = 5.14, p < 0.0001; Fig 2D ) with higher spatial information (Z = 14.2317, p < 0.0001; Wilcoxon rank-sum test; Fig 2E ), compared with those in the subiculum. Overall, as we reported previously (Lee and colleagues, 2018), subicular cells exhibited less spatial tuning than CA1 cells (spatial selectivity: 4.04 ± 0.08 in CA1, 2.24 ± 0.07 in subiculum; sparsity: 0.41 ± 0.01 in CA1, 0.74 ± 0.17 in subiculum; mean ± SEM; p-values < 0.0001; Wilcoxon rank-sum test). In addition, field width was larger in subicular place cells than those in CA1 (Z = 5.96, p < 0.0001; Wilcoxon rank-sum test) for both single- and multi-field cells ( Fig 2F ). Taken together with the comparisons made between the 2 regions based on other traditional measures in our previous study [ 11 ], these spatial firing patterns made it difficult to define place fields for individual neurons in the subiculum compared with those in the CA1.

Prior studies [ 9 – 12 ] reported that cells in the subiculum fire at higher rates with lower spatial selectivity than those in the CA1, a finding also confirmed in our study. That is, cells recorded from the CA1 fired at focal and restricted locations along the T-maze ( Fig 2A ), whereas cells recorded from the subiculum tended to show broad and continuous firing fields ( Fig 2B ), making it challenging to identify a place field using the conventional field detection method based on spatial firing rates. Specifically, although some subicular cells exhibited spatially tuned place fields similar to CA1 place fields (cells 234–4–1–5 and 232–5–4–8 in Fig 2B ), some background spiking activity continued to occur outside their place fields. Furthermore, other subicular cells fired continuously across the entire track (cells 232–4–17–1 and 232–5–20–1 in Fig 2B ), complicating efforts to define the field boundaries for these cells. These differences in field characteristics between the CA1 and subiculum can be more clearly observed in population rate maps constructed by stacking the rate maps of individual cells ( Fig 2A and 2B ).

Identification of latent place fields based on theta phase precession of spiking

Our findings show that the fundamental differences in spatial firing characteristics between the CA1 and subiculum make it difficult to use conventional approaches commonly employed for analyzing place fields in both regions because these approaches have mostly been developed for place fields of cells in the hippocampus and not for those in the subiculum. In fact, a large number of subicular cells that would have been defined as no-field cells by conventional field detection methods did fire more vigorously at particular locations of the track (Fig 2B, cells 232–4–17–1 and 232–5–20–1). However, the conventional field detection algorithm was unable to detect such spatial firing patterns because of the higher spontaneous firing activities throughout the track in subicular cells compared with CA1 neurons. Our previous study tried to locate field boundaries in these cells by adjusting the threshold for detecting field boundaries or by finding local minima through statistical comparisons of trial-by-trial firing rates between neighboring bins. However, such methods still defined some subicular cells as having no field. Furthermore, the conventional field detection algorithm tended to ignore a small subfield if there was one dominant field with a very high firing peak.

To overcome such limitations, we explored the possibility of defining place fields using theta phase precession, a well-known phenomenon in which theta-related phases of spikes of a neuron gradually shift to earlier phases as the rat repeatedly passes through the cell’s place field [25,26]. In particular, we examined whether the broad firing field of a subicular neuron could be divided into multiple subfields if it were defined by theta phases of spikes. As shown in Fig 3, theta phase precession occurred robustly within the identified unitary place field in both the CA1 and subiculum as the rat ran along the track (CA1 single-field cells 234–2–12–2 and 561–2–3–1 in Fig 3A; subicular single-field cells 232–5–4–8 and 234–4–1–5 in Fig 3B). Importantly, those cells classified as having no place field exhibited multiple cycles of robust theta phase precessions in the subiculum (subicular non-place field cells 232–7–17–1 and 232–4–17–1 in Fig 3C).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Robust multiple theta phase precessions in the subiculum. (A–C) Representative examples of theta phase precession in the single-field cells in CA1 (A), subiculum (B), and non-place field cells in the subiculum (C). The left column within each cell consists of linearized position (top), a raw trace of theta oscillation (middle), and spiking theta phases (bottom) in the temporal axis in a single trial. Spikes in the raw theta traces are marked by red circular dots. Scale bar, 250 μV. Spiking theta phases are plotted within a range of 360°, and the initial phase is adjusted for clear observation of theta phase precession. Serial numbers in the upper right corner are cell IDs. The right column displays a linearized firing rate map (top) and a position phase plot (bottom) on the spatial axis across a session. Black solid lines overlaid on the firing rate maps indicate verified place fields, whereas black dotted lines are non-place fields. Numbers above firing rate maps denote peak firing rates (Hz) and spatial information scores (bit/spike) of place or non-place fields. Red arrowheads and red dashed lines mark choice points. Note that subicular cells showed multiple cycles of theta phase precession, some of which were as robust as those of CA1 cells. fw, food well; NF, non-place field; SF, single-field; st box, start box; SUB, subiculum. https://doi.org/10.1371/journal.pbio.3001546.g003

To identify a spike cluster that belonged to a single theta precession cycle in the phase position plot, we used the DBSCAN (Density-Based Spatial Clustering with Applications of Noise) algorithm (see Methods for details). We compared the results of two different methods for detecting place fields: a firing rate–based method that finds a rate-based field, and the theta phase precession–based DBSCAN method, which finds a phase-based field. Both algorithms produced the same results in some cells in both the CA1 and subiculum (Fig 4A). However, we were also able to find new place fields for other cells based on the phase-based method. Specifically, some cells that were originally classified as single-field cells were converted into multi-field cells by application of the theta phase–based clustering algorithm (Fig 4B–4D). That is, in some cells, existing rate-based fields were subdivided into more than 2 phase-based fields (Fig 4B). In other cells, additional place fields that might not have been detectable by the rate-based method (mostly owing to low firing peaks) were revealed by the phase-based clustering (Fig 4C). In a final group of cells, the phase-based method separated an existing field and added a new field at the same time (Fig 4D).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Identification of place fields based on spiking theta phases in the CA1 and subiculum. (A–D) For each cell example, a linearized firing field based on firing rates (rate-based field indicated by solid black line; top), linearized firing fields based on theta phases (phase-based fields denoted by different colors; middle), and a position phase plot on the spatial axis across a session (bottom) are shown. Gray dotted line indicates the mean firing field. The numbers on the right corner indicate spatial information scores using firing rates (bit/spike; front) and spiking theta phases (bit/cm; back) obtained from the entire firing activities of a cell (gray) or individual phase-based fields (color coded). Red arrowheads denote choice points. Serial numbers above the firing rate maps are cell IDs. Spike clusters in position phase plots are color coded with the same colors used for the firing rate maps. Black straight lines on spike clusters indicate the circular–linear regression line. The numbers and asterisks in the box with colored borders are circular–linear correlation coefficients and their significance for phase-based fields in the same color. *p < 0.05, **p < 0.01, ***p < 0.0001. fw, food well; st box, start box; SUB, subiculum. https://doi.org/10.1371/journal.pbio.3001546.g004

The proportions of cells showing different numbers of place fields changed when using the phase-based clustering method compared with the rate-based method. Specifically, phase-based clustering classified 20% of CA1 cells and 62% of subicular cells as multi-field cells and only 9% of subicular cells as having no field ( = 106.70, p < 0.0001; chi-squared test; Figs 5A and S1). When the categorical changes were examined for each cell group, it turned out that three-quarters of the rate-based non-place cells in the CA1 and subiculum exhibited multiple phase-based place fields (Fig 5B). In addition, some rate-based single-field cells in the CA1 (14%) and subiculum (45%) were converted to multi-field cells by the phase-based clustering. We also found that some rate-based multi-field cells in the subiculum exhibited additional phase-based fields after applying the phase-based protocol (MF added in Fig 5B). Although the widths of phase-based place fields remained still significantly larger in the subiculum than in the CA1 (Z = 4.08, p < 0.0001; Wilcoxon rank-sum test; Fig 5C), other firing properties of individual fields defined by theta phase became comparable between the 2 regions. In particular, when we compared the 2 regions by both rate-based spatial information (bit/spike) and phase-based spatial information (bit/cm) measured for all spiking activities associated with outbound journeys in individual neurons (see Methods for details), the distributions were well separated between the CA1 and subiculum (S2A Fig). However, when the same distributions were obtained by using only the spiking activities within the boundaries of the phase-based fields, the 2 distributions of CA1 and subiculum largely overlapped (S2B Fig). These findings suggest that the place cells in both regions may share more common firing properties than previously thought when the place fields are defined by theta phases of spikes instead of the conventional rate–based method.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Advantages of theta phase–based field detection. (A) Difference between the CA1 and subiculum in the proportion of place cells associated with different numbers of place fields, when defined using the theta phase–based method. Cells are classified into 3 groups: “SF” for one field, “MF” for more than one field, and “NF” for zero field. ***p < 0.0001. (B) Categorical changes of cells within each rate-based cell group (NF, SF, and MF on the x axis) as the field identification method was shifted to the one using theta phase. The bar graph shows what proportion of cells in each cell group (classified by the rate-based method) was recategorized after the phase-based method. (C) Regional differences in place field width after phase-based field detection. ***p < 0.0001. (D–E) Cumulative distributions of TPP slope (D) and correlation coefficient (E) of place cells for each method (rate-based, FR, and phase-based, TP) and each region. Line graphs on the right side of each panel display mean values and standard errors for the same data. (F–G) Regional differences in the onset phase (H) and the phase shift range (G) of the TPP. These measurements were obtained only from the phase-based fields that showed significant TPP. *p < 0.05, ***p < 0.0001. (H) Changes in the proportion of place cells that exhibited significant TPPs when the fields were identified by the conventional rate–based method (FR) versus the theta phase–based method (TP) in the CA1 and subiculum. ***p < 0.0001. Data associated with this figure can be found in S1 Data file. FR, firing rate; MF, multi-field; NF, non-place field; SF, single-field; TP, theta phase; TPP, theta phase precession. https://doi.org/10.1371/journal.pbio.3001546.g005

We next examined the robustness of theta phase precession of place cells in the subiculum compared with that in the CA1 using circular statistics (linear regression and linear correlation) for each spike cluster [27]. We found that the slope of theta phase precession was significantly different between the 2 regions (F (1,435) = 4.43, p = 0.036), but it was not affected by the field–identification method (F (1,613) = 0.59, p = 0.44, two-way mixed ANOVA with region as the between-subject factor and the field identification method as the within-subject factor; Fig 5D). There was no interaction between the region and field detection method (F (1,613) = 0.52, p = 0.47), mostly attributable to the reduced regional difference when the phase precession slope was calculated based on the phase-based fields compared to the rate-based fields. The precession slope of rate-based fields tended to be steeper in the CA1 than in the subiculum (t (805) = 2.01, p = 0.045 for Bonferroni-corrected unpaired two-sample t test; corrected α = 0.0125), an outcome that could be expected based on the larger field width of subicular place cells. However, the regional difference in slope diminished when using the phase-based method (t (496) = 1.64, p = 0.102). The slope of theta phase precession was not affected by the field detection method within each region (CA1, t (555) = 1.39, p = 0.16; SUB, t (637) = 0.03, p = 0.98). On the other hand, the strength of theta phase precession of place cells evaluated by circular–linear correlation coefficient was significantly different between the CA1 and subiculum (F (1,445) = 31.57, p < 0.0001) and between the 2 field detection methods (F (1,655) = 50.25, p < 0.0001; two-way mixed ANOVA; Fig 5E). The interaction between the region and method was not significant (F (1,655) = 3.68, p = 0.055). Post hoc tests revealed that the phase precession strength increased in phase-based fields compared with rate-based fields in both regions (CA1, t (565) = 4.77, p < 0.0001; subiculum, t (695) = 5.36, p < 0.0001; unpaired two-sample t test with Bonferroni correction; corrected α = 0.0125). Although precession strength was significantly lower in the subiculum than in the CA1 even based on the phase-based field detection (rate-based, t (896) = 5.08, p < 0.0001; phase-based, t (541) = 4.03, p < 0.0001), the precession strength in the subiculum increased closer to that of the CA1.

To further compare the basic properties of theta phase precession between CA1 and subiculum, we screened for the phase-based fields that showed significant theta phase precession based on the following criteria: (i) the range of phase shift ≥90°; (ii) the slope of circular–linear regression line <0; and (iii) p-value of circular–linear correlation ≤0.05. In both CA1 and subiculum, 70% of phase-based fields exhibited significant theta phase precession by meeting all 3 criteria (CA1, n = 235/337; subiculum, n = 212/300). In the subiculum, the onset phase was slightly, yet significantly, earlier (Z = 2.81, p = 0.005; Fig 5F) and the range of phase shift was significantly smaller (Z = 4.48, p < 0.0001, Wilcoxon rank-sum test; Fig 5G) than in CA1. The proportion of place cells showing significant theta phase precession increased in the subiculum as the phase-based method was applied compared to using the conventional rate–based method ( = 61.33, p < 0.0001; chi-squared test; Fig 5H), but this was not the case in the CA1 ( = 4.95, p = 0.084). These findings indicate that the field detection method based on theta phase precession of spikes effectively identified multiple subfields enveloped in the broad firing activities of the subicular cells.

[END]

[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001546

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/