Research ArticleNEUROSCIENCE

Optogenetic pacing of medial septum parvalbumin-positive cells disrupts temporal but not spatial firing in grid cells

See allHide authors and affiliations

Science Advances  05 May 2021:
Vol. 7, no. 19, eabd5684
DOI: 10.1126/sciadv.abd5684
  • Fig. 1 PV+ cells in MSA selectively express channelrhodopsin (ChR2) after injection of virus in PvalbCre rats.

    (A) Illustration of a rat brain seen slightly from above on the left side highlights the MSA in purple and MEC in blue/cyan, with long-range projections from MSA to MEC (green). These projections target all layers of MEC, from the dorsal to the ventral area. A viral construct carrying ChR2 was injected in MSA. (B) Three coronal sections from a representative animal show the extent of the virus expression (green) in MSA at three different anterior-posterior positions. Expression covered large parts of MSA. (C) Virus expression in MSA was restricted to PV+ cells (red). White arrowheads mark overlap between virus (ChR2) and PV+ cells (PV) (top row). Virus-expressing projections were found in MEC, parasubiculum (PaS), and all regions of the hippocampus [CA1, CA3, and dentate gyrus (DG)] (bottom row). The area of MEC chosen for the magnified images is indicated by a small square. ChR2-labeled septal projections target PV+ cells in MEC (small outline). (D) Illustration of experimental setup. PvalbCre rats were implanted with optic fiber in MSA and recording electrodes in MEC (two animals also had optic fibers together with recording electrodes in MEC). Blue laser light (470 nm) was used to activate PV+ cells in MSA at two different frequencies, 11 and 30 Hz.

  • Fig. 2 Optogenetic pacing of PV+ cells in MSA abolish endogenous theta in MEC.

    (A) Illustration of a presumed MSA-to-MEC connection showing that PV+ cells (PV) in MSA terminate on interneurons (IN) in MEC that together with grid cells (GC) form a recurrent circuit. Optogenetic activation of MEC projecting PV+ cells in MSA (blue shading) caused a frequency shift in oscillations in MEC as shown by time-frequency representations of LFP during open field exploration. (B) Average PSD for all baseline and stimulus sessions. (C) Cumulative density plots of relative theta power and relative power in the stimulated band show that the power of the endogenous theta was strongly reduced for both stimulation frequencies. (D) Illustration of optogenetic stimulations in MEC. Optogenetic activation of MSA PV+ cell terminals in MEC did not reduce the endogenous theta frequency. Time-frequency representations of LFP in two simultaneously recorded hemispheres where one is stimulated with 11 Hz. (E) Similar to (B). The endogenous theta remained when stimulating PV+ cell terminals in MEC at 11 Hz. (F) Cumulative density plots of relative theta power show that the power of endogenous theta was reduced when stimulating PV+ cell terminals in MEC. (G) Raster plot of single-unit spiking responses of one grid cell and one narrow spiking inhibited (NSi) cell. Probability density for the whole population of grid cells and NSi cells shows fast inhibition of NSi cells followed by grid cell activation. Thirty-hertz stimulation led to a second activation of both cell types after approximately 20 ms. (H) Raster plots from one grid cell and one NS cell during optogenetic stimulation in MEC. Probability densities of the two units are overlaid in the bottom graph to illustrate difference in response times.

  • Fig. 3 Grid cells stop phase precessing during MSA stimulation.

    (A) Theta rhythmicity estimated by the power spectral density of the neuronal firing rate obtained by kernel estimation with a Gaussian kernel of width 10 ms. (B) Cumulative density of the peak theta rhythmicity (6 to 10 Hz) divided by the average power of 1-Hz wide adjacent bands. (C) Spike LFP coherence in grid cells for baseline sessions compared to 11- and 30-Hz stimulation. Grid cells showed strong spike coherency to the endogenous theta frequency (around 8 Hz) in baseline sessions, which was reduced during stimulation. (D) Cumulative density plots show that grid cells significantly reduced spike coherency at the endogenous theta frequency during both 11- and 30-Hz stimulation. Both stimulated bands caused similar peak coherence. (E) Polar plot displaying theta phase preference and vector length of all grid cells during baseline (endogenous theta) and stimulation sessions (stimulated band). (F) Swarm plot of P values (only neurons with P value below 0.10 are shown) for neurons phase precessing (circular linear correlation r < 0) during Baseline I, 11 Hz, Baseline II, and 30 Hz sessions. We found no significant phase precessing neurons during stimulation as indicated by no points below the black line (P < 0.01). (G) Rows show color-coded rate maps from two grid cells during baseline followed by an 11-Hz stimulation session. Rat and unit numbers are indicated to the left, and peak rate and gridness are indicated above. Raster plots represent spike phase relative to LFP (band-pass filtered at 6 to 10 Hz) versus distance traveled through grid fields. The regression line indicates that the spike phase predicts the position of the animal within the field. P value of phase precession is shown to the right. Both example units ceased phase precessing during stimulation, shown by the loss of significant phase precession (P < 0.01).

  • Fig. 4 Grid cells are spatially stable during optogenetic stimulation of MSA.

    (A) Color-coded rate maps (top) and running path with spikes superimposed (bottom) from three grid cells recorded in all four sessions of 10 min each. Rate maps are adjusted with color coding corresponding to the heat color of the first recording session. Peak rates and gridness are denoted above each rate map. Rat number and unit ID are indicated to the left of the rate maps. (B) Scatter plots showing relative changes in spatial shift of grid cells followed across multiple recording sessions. There were no detectable significant changes compared to Baseline Ia or IIa versus Baseline Ia or IIb (first half versus second half of Baseline I or Baseline II, respectively). (C) Cumulative density plots of average rate, maximum rate, gridness, spatial information, and spatial information specificity. There were no detectable differences in either rate measures; however, there was a nonsignificant change in gridness between baseline and stimulation and a significant change in spatial information rate when comparing Baseline I and 11 Hz. Spatial information specificity showed no significant change. (D) Raster plot from an example grid cell where 0 marks 11-Hz stimulation onset. Spikes in the preresponse (Pre: green/purple; −5 to 5 ms) and response period (Resp: orange/pink; 5 to 11 ms) and the between pulses (Between: blue; 15 to −5 ms) are marked. The rat’s running trajectory of the recording session with spikes from Pre, Resp, and Between periods superimposed. Black circles indicate outline of identified grid fields. (E) Similar to (D) but showing 30-Hz stimulation. Violin plots [minimum to maximum and median (black line)] show increased out-of-field spiking activity during response periods. Width of violin corresponds to number of samples for each value. *P < 0.05 and ***P < 0.001; LMM test. ns, not significant.

  • Fig. 5 Neuronal speed modulation is stable despite disrupted speed modulation in theta.

    (A) Example of speed modulation in one grid cell and one NS cell (NSi) during baseline and stimulation sessions. Speed score is denoted above each speed graph. Vertical axis represents normalized firing rate. (B) Violin plots show speed scores in grid cells (top) and NSi cells (bottom) between baseline and stimulation sessions. Neither 11- nor 30-Hz stimulation caused any significant change in speed scores for either cell type. (C) Violin plots showing paired comparisons of running speed of animals in all recording sessions. Running speed decreased slightly from Baseline I to 11 Hz, from Baseline I to Baseline II, and from Baseline II to 30 Hz. (D) Correlation between running speed and theta peak frequency was strong in both Baseline I and Baseline II but disrupted during 11- and 30-Hz stimulation. (E) Frequency score, as represented by the correlation between running speed and peak frequency. This was significantly reduced from Baseline I to 11 Hz and from Baseline II to 30 Hz. (F) Correlation between running speed and theta power was strong in both Baseline I and Baseline II but disrupted during 11- and 30-Hz stimulation. (G) Power score, as represented by the correlation between running speed and power. This was significantly reduced from Baseline I to 11 Hz and from Baseline II to 30 Hz. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001, Wilcoxon signed-rank test.

Supplementary Materials

  • Supplementary Materials

    Optogenetic pacing of medial septum parvalbumin-positive cells disrupts temporal but not spatial firing in grid cells

    Mikkel Elle Lepperød, Ane Charlotte Christensen, Kristian Kinden Lensjø, Alessio Paolo Buccino, Jai Yu, Marianne Fyhn, Torkel Hafting

    Download Supplement

    This PDF file includes:

    • Figs. S1 to S11
    • Tables S1 to S6
    • Supplementary discussion related to Fig. 3

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article