Grid cells require excitatory drive from the hippocampus

Journal name:
Nature Neuroscience
Volume:
16,
Pages:
309–317
Year published:
DOI:
doi:10.1038/nn.3311
Received
Accepted
Published online

Abstract

To determine how hippocampal backprojections influence spatially periodic firing in grid cells, we recorded neural activity in the medial entorhinal cortex (MEC) of rats after temporary inactivation of the hippocampus. We report two major changes in entorhinal grid cells. First, hippocampal inactivation gradually and selectively extinguished the grid pattern. Second, the same grid cells that lost their grid fields acquired substantial tuning to the direction of the rat's head. This transition in firing properties was contingent on a drop in the average firing rate of the grid cells and could be replicated by the removal of an external excitatory drive in an attractor network model in which grid structure emerges by velocity-dependent translation of activity across a network with inhibitory connections. These results point to excitatory drive from the hippocampus, and possibly other regions, as one prerequisite for the formation and translocation of grid patterns in the MEC.

At a glance

Figures

  1. Muscimol-induced inactivation of the dorsal hippocampus.
    Figure 1: Muscimol-induced inactivation of the dorsal hippocampus.

    (a) Sagittal brain sections showing positions of tetrodes in the MEC (left) and cannulae and tetrodes in the hippocampus (middle and right, respectively). The locations of the tetrode tips are indicated by arrows. The anteromedial-posterolateral distance between the cannulae and tetrodes in the hippocampus is 2.2 mm. Scale bars, 1 mm. (b) Trajectory (black) with superimposed spike positions (red) for a representative place cell before and after infusion of the GABAergic agonist muscimol (gray shading indicates presence of muscimol) into the hippocampus. The cell is completely silenced within 20 min. (c) Mean firing rate (± s.e.m.) before and after muscimol infusion (all cells in this analysis were hippocampal pyramidal cells). Time t = 0 indicates the start of infusion. (d) Percentage of the entire cell sample that fired less than 1% of the baseline average in trial blocks before and after muscimol infusion.

  2. Disruption of entorhinal grid structure after inactivation of the hippocampus.
    Figure 2: Disruption of entorhinal grid structure after inactivation of the hippocampus.

    (a) Trajectory with spike positions (top), rate map (middle) and autocorrelation map (bottom) for successive trial blocks in a grid cell after muscimol infusion in the hippocampus. Rate maps and autocorrelation maps are color coded with cold (blue) and warm (red) colors indicating low and high rates and correlation values, respectively. The correlation colors are scaled to the observed range. Peak rates and grid scores are indicated over the rate maps and the autocorrelation maps, respectively. Baseline and 6-h maps were recorded over a 20-min period. Pixels not covered are white. The scale of the autocorrelation diagrams is twice the scale of the rate maps (side lengths of 200 cm compared to 100 cm, respectively). For additional examples, see Supplementary Figure 6. (b) Percentage of cell samples that passed the 95th percentile criterion for grid cells in 10-min trial blocks before and after hippocampal muscimol infusion. Percentages refer to the total number of cells that passed the criterion in an analysis of the entire baseline period (20 min). The red line indicates the chance value (5%). (c) Mean grid scores (± s.e.m.) for trial blocks of 10 min starting 20 min before infusion. Grid scores measure the degree of sixfold rotational symmetry in the spatial autocorrelation map of the cell. The red line indicates the 95th percentile of a distribution of shuffled spike times for the same cells (400 permutations); the blue line indicates the 50th percentile.

  3. Disappearance of the grid pattern in dynamic maps.
    Figure 3: Disappearance of the grid pattern in dynamic maps.

    (a) Spike-triggered (dynamic) spatial autocorrelation maps showing the grid structure for 10-s time windows referenced to the location of each spike. One map is shown for each time block before and after muscimol infusion for a representative cell. The width of each map is 200 cm. The correlations are color coded and range from 0 (dark blue) to the maximum correlation value indicated at the bottom right of each map (dark red). Grid scores are indicated for each map. Baseline and 6-h maps were recorded over a 20-min period. (b) Grid scores (mean ± s.e.m.) of dynamic autocorrelation maps. Time is measured from the end of the hippocampal infusion. The 95th percentile criterion is indicated by the red line; the blue line indicates the 50th percentile. Note the disappearance of the grid structure after hippocampal inactivation despite the use of a moving reference point. (c) Spike-triggered (dynamic) spatial crosscorrelogram showing the effect of hippocampal inactivation on the spatial-phase relationship in successive trial blocks in a representative pair of simultaneously recorded grid cells (regular rate maps and autocorrelation maps for the same cells are shown in Supplementary Fig. 6). Each map shows the distribution of correlation between the first cell's firing rate at the trigger location and the second cell's firing rate at all relative locations visited during the subsequent 1-s interval. The correlations are color coded as in a. The width of each map is 60 cm (30 cm away from the origin). Cross-correlation values are indicated for each map (baseline and 6-h maps are 20 min). Note that the spatial-phase relationship of the two grid cells is disrupted by inactivation of the hippocampus, suggesting that the original spatial structure of the grid cells is abolished even at time intervals as short as 1 s. (d) Correlation between dynamic cross-correlation maps in the first 10-min block of baseline recording and in subsequent 10-min blocks (second block of baseline, muscimol infusion and recovery; means ± s.e.m.).

  4. Loss of grid structure leads to directional tuning.
    Figure 4: Loss of grid structure leads to directional tuning.

    (a) Development of directional tuning after the disruption of grid structure in two example cells (with the top three rows showing one cell and the bottom three rows showing the other cell). The top two rows for each cell show the trajectory with spike locations and the spatial autocorrelation maps, respectively. The grid scores are indicated above the autocorrelation maps. The bottom row for each cell shows polar plots of firing rate as a function of head direction. The peak firing rate is indicated over each plot. For additional examples, see Supplementary Figure 6. (b,c) Development of mean vector length after hippocampal muscimol infusion (means ± s.e.m.), with results shown for all grid cells that were nondirectional in the baseline trial (b) and all cells with direction modulation in the baseline trial (c). The gray area indicates the presence of muscimol. The red line indicates the 95th percentile value for mean vector length in the corresponding shuffled distributions; the blue line indicates the 50th percentile. (d) Percentage of initially nondirectional grid cells that passed the 95th percentile criterion for head-direction cells in trial blocks after hippocampal infusion of muscimol. Percentages refer to the total number of cells that passed the criterion in an analysis of the entire baseline period (20 min). The gray box indicates the presence of muscimol. The red line indicates the chance value (5%).

  5. Loss of the grid pattern depends on the decrease in firing rates of grid cells.
    Figure 5: Loss of the grid pattern depends on the decrease in firing rates of grid cells.

    (a) Mean firing rates (± s.e.m.) before and after muscimol infusion for the entire sample of entorhinal cells (left, grid cells with no directional modulation; middle, conjunctive grid and head-direction cells; right, head-direction and border cells combined). (b) Relationship between the normalized average firing rate (relative to the baseline rate) and either grid score (left) or mean vector length (right) after hippocampal inactivation (mean values for all simultaneously recorded grid cells are shown). (c,d) Grid scores and mean vector lengths for the trial before infusion as a function of downsampling of spike numbers. (c) Examples of trajectory maps, rate maps and directional plots before and after downsampling (10-min trials). The grid score is indicated above each rate map, and the mean vector length is indicated over each directional plot. (d) Line diagram showing the decrease of grid scores and the increase of mean vector length (black and red line, respectively; means ± s.e.m.) as a function of downsampling of data from 10-min blocks of baseline recording. Note that both measures tolerate considerable downsampling well beyond the reduction of firing rates observed after muscimol infusions in the experimental data (dashed line).

  6. Remaining nonperiodic spatial firing after hippocampal inactivation.
    Figure 6: Remaining nonperiodic spatial firing after hippocampal inactivation.

    (a) Localized nonperiodic firing in two example cells after hippocampal muscimol infusions that removed all visible grid structure (from top to bottom, trajectory with spike locations, rate map and autocorrelation map; peak rates and grid scores are indicated). (b,c) Spatial coherence (b) and spatial stability (c) of all grid cells from all rats (means ± s.e.m.). Spatial coherence is the average correlation between the firing rate in neighboring bins of the recording box (first-order spatial autocorrelation); spatial stability is the bin-by-bin correlation of firing rates between the first and second half of the trial. The red line indicates the 95th percentile value of the shuffled distribution, and the blue line indicates the chance value (50th percentile). The chance value for spatial stability is 0.

  7. Preserved theta activity during hippocampal inactivation.
    Figure 7: Preserved theta activity during hippocampal inactivation.

    (a) Local entorhinal EEG readings showing 2.0 s of activity during running in the open field before (top) and 30 min after (bottom) hippocampal muscimol infusion. (b) Power spectrum showing the persistence of theta activity in entorhinal EEG readings after hippocampal muscimol infusion during a representative trial (T30T160, 1–125 Hz filter). FFT, fast Fourier transform. (c) Spike time autocorrelation diagram showing the persistence of theta modulation (spiking at ~125-ms intervals) in a grid cell after muscimol infusion (T30T160). (d) Relative power of theta activity in entorhinal EEG readings as a function of time after hippocampal muscimol infusion (mean ± s.e.m.). Theta power was normalized by the power in the entire 1–125 Hz band. (e) Peak theta frequency of spike-time autocorrelation functions of individual grid cells (autocorrelogram, gray) and theta peak frequency in simultaneously recorded local EEG (black) as a function of time after muscimol infusion (mean ± s.e.m.). Note that individual cells are modulated at a higher theta frequency than the field EEG readings both before and after muscimol infusion, suggesting that phase precession is preserved25.

  8. The effect of hippocampus inactivation in an attractor model of grid cells.
    Figure 8: The effect of hippocampus inactivation in an attractor model of grid cells.

    (a) Cartoon showing how the movement of activity across the neuronal sheet (bottom three images) results in the spatially selective firing of a single cell (top image). At time t = 1, the pointer is recording from a cell that is currently inactive. From t = 1 to t = 2, the activity pattern is translated with the movement of the animal such that the indicated cell has an elevated firing rate and is therefore in its spatial field. The cell then leaves its spatial field as the population activity shifts at t = 3. Supplementary Video 1 shows an example of this translation of activity and how it is reflected in single-cell firing. (b) Grid scores and mean vector length (mean ± s.e.m.) compared to the strength of external input (l). For large external inputs, high grid scores are achieved through the mechanism described in a; see also Supplementary Video 1. As the external input is decreased below a critical amount, the activity on the neuronal sheet, even if hexagonal, is not stable in the sense that it is easily distorted (Supplementary Video 2). At the single-cell level, therefore, no grid-like activity is present, and the grid scores drop. At the same time, the head-directional input becomes the dominant source of input to the cells, and the neurons show high directional tuning. For each value of external input, we simulated the network when a virtual rat navigated nine experimentally recorded tracks, and each time we took the response of 50 random neurons from the network. The averages and error bars are over these 450 neurons. (c) Spike distribution plots and directional tuning curves generated from example cells for the extreme cases of weak hippocampal input (bottom) and strong hippocampal input (top).

References

  1. O'Keefe, J. & Dostrovsky, J. The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely-moving rat. Brain Res. 34, 171175 (1971).
  2. Fyhn, M., Molden, S., Witter, M.P., Moser, E.I. & Moser, M.B. Spatial representation in the entorhinal cortex. Science 305, 12581264 (2004).
  3. Hafting, T., Fyhn, M., Molden, S., Moser, M.B. & Moser, E.I. Microstructure of a spatial map in the entorhinal cortex. Nature 436, 801806 (2005).
  4. Moser, E.I., Kropff, E. & Moser, M.-B. Place cells, grid cells, and the brain's spatial representation system. Annu. Rev. Neurosci. 31, 6989 (2008).
  5. Giocomo, L.M., Moser, M.-B. & Moser, E.I. Computational models of grid cells. Neuron 71, 589603 (2011).
  6. McNaughton, B.L., Battaglia, F.P., Jensen, O., Moser, E.I. & Moser, M.-B. Path integration and the neural basis of the “cognitive map.”. Nat. Rev. Neurosci. 7, 663678 (2006).
  7. Taube, J.S., Muller, R.U. & Ranck, J.B. Jr. Head-direction cells recorded from the postsubiculum in freely moving rats. I. Description and quantitative analysis. J. Neurosci. 10, 420435 (1990).
  8. Sargolini, F. et al. Conjunctive representation of position, direction, and velocity in entorhinal cortex. Science 312, 758762 (2006).
  9. Savelli, F., Yoganarasimha, D. & Knierim, J.J. Influence of boundary removal on the spatial representations of the medial entorhinal cortex. Hippocampus 18, 12701282 (2008).
  10. Solstad, T., Boccara, C.N., Kropff, E., Moser, M.B. & Moser, E.I. Representation of geometric borders in the entorhinal cortex. Science 322, 18651868 (2008).
  11. Fyhn, M., Hafting, T., Treves, A., Moser, M.-B. & Moser, E.I. Hippocampal remapping and grid realignment in entorhinal cortex. Nature 446, 190194 (2007).
  12. Fuhs, M.C. & Touretzky, D.S. A spin glass model of path integration in rat medial entorhinal cortex. J. Neurosci. 26, 42664276 (2006).
  13. Solstad, T., Moser, E.I. & Einevoll, G.T. From grid cells to place cells: a mathematical model. Hippocampus 16, 10261031 (2006).
  14. Molter, C. & Yamaguchi, Y. Entorhinal theta phase precession sculpts dentate gyrus place fields. Hippocampus 18, 919930 (2008).
  15. Rolls, E.T., Stringer, S.M. & Elliot, T. Entorhinal cortex grid cells can map to hippocampal place cells by competitive learning. Network 17, 447465 (2006).
  16. de Almeida, L., Idiart, M. & Lisman, J.E. The input-output transformation of the hippocampal granule cells: from grid cells to place fields. J. Neurosci. 29, 75047512 (2009).
  17. Savelli, F. & Knierim, J.J. Hebbian analysis of the transformation of medial entorhinal grid-cell inputs to hippocampal place fields. J. Neurophysiol. 103, 31673183 (2010).
  18. Si, B. & Treves, A. The role of competitive learning in the generation of DG fields from EC inputs. Cogn. Neurodyn. 3, 177187 (2009).
  19. Monaco, J.D. & Abbott, L.F. Modular realignment of entorhinal grid cell activity as a basis for hippocampal remapping. J. Neurosci. 31, 94149425 (2011); erratum 31, 11096 (2011).
  20. O'Keefe, J. & Burgess, N. Dual phase and rate coding in hippocampal place cells: theoretical significance and relationship to entorhinal grid cells. Hippocampus 15, 853866 (2005).
  21. Burgess, N., Barry, C. & O'Keefe, J. An oscillatory interference model of grid cell firing. Hippocampus 17, 801812 (2007).
  22. Kropff, E. & Treves, A. The emergence of grid cells: intelligent design or just adaptation? Hippocampus 18, 12561269 (2008).
  23. Sreenivasan, S. & Fiete, I. Grid cells generate an analog error-correcting code for singularly precise neural computation. Nat. Neurosci. 14, 13301337 (2011).
  24. Samu, D., Erös, P., Ujfalussy, B. & Kiss, T. Robust path integration in the entorhinal grid cell system with hippocampal feed-back. Biol. Cybern. 101, 1934 (2009).
  25. Hafting, T., Fyhn, M., Bonnevie, T., Moser, M.-B. & Moser, E.I. Hippocampus-independent phase precession in entorhinal grid cells. Nature 453, 12481252 (2008).
  26. Allen, T.A. et al. Imaging the spread of reversible brain inactivations using fluorescent muscimol. J. Neurosci. Methods 171, 3038 (2008).
  27. Boccara, C.N. et al. Grid cells in pre- and parasubiculum. Nat. Neurosci. 13, 987994 (2010).
  28. Langston, R.F. et al. Development of the spatial representation system in the rat. Science 328, 15761580 (2010).
  29. O'Keefe, J. & Recce, M.L. Phase relationship between hippocampal place units and the EEG theta rhythm. Hippocampus 3, 317330 (1993).
  30. Burak, Y. & Fiete, I.R. Accurate path integration in continuous attractor network models of grid cells. PLOS Comput. Biol. 5, e1000291 (2009).
  31. Couey, J.J. et al. Recurrent inhibitory circuitry as a mechanism for grid formation. Nat. Neurosci. doi:doi:10.1038/nn.3310 (20 January 2013).
  32. van Haeften, T., Baks-te-Bulte, L., Goede, P.H., Wouterlood, F.G. & Witter, M.P. Morphological and numerical analysis of synaptic interactions between neurons in deep and superficial layers of the entorhinal cortex of the rat. Hippocampus 13, 943952 (2003).
  33. Kloosterman, F., Van Haeften, T., Witter, M.P. & Lopes Da Silva, F.H. Electrophysiological characterization of interlaminar entorhinal connections: an essential link for re-entrance in the hippocampal-entorhinal system. Eur. J. Neurosci. 18, 30373052 (2003).
  34. Brandon, M.P. et al. Reduction of theta rhythm dissociates grid cell spatial periodicity from directional tuning. Science 332, 595599 (2011).
  35. Koenig, J., Linder, A.N., Leutgeb, J.K. & Leutgeb, S. The spatial periodicity of grid cells is not sustained during reduced theta oscillations. Science 332, 592595 (2011).
  36. Alonso, A. & Klink, R. Differential electroresponsiveness of stellate and pyramidal-like cells of medial entorhinal cortex layer II. J. Neurophysiol. 70, 128143 (1993).
  37. Yoshida, M., Giocomo, L.G. & Hasselmo, M.E. Frequency of subthreshold oscillations at different membrane potential voltages in neurons at different anatomical positions on the dorsoventral axis in the rat medial entorhinal cortex. J. Neurosci. 31, 1268312694 (2011).
  38. Zilli, E.A. & Hasselmo, M.E. Coupled noisy spiking neurons as velocity-controlled oscillators in a model of grid cell spatial firing. J. Neurosci. 30, 1385013860 (2010).
  39. Stensola, H. et al. The entorhinal grid map is discretized. Nature 492, 7278 (2012).
  40. Yartsev, M.M., Witter, M.P. & Ulanovsky, N. Grid cells without theta oscillations in the entorhinal cortex of bats. Nature 479, 103107 (2011).
  41. Wills, T.J., Cacucci, F., Burgess, N. & O'Keefe, J. Development of the hippocampal cognitive map in preweanling rats. Science 328, 15731576 (2010).
  42. Burwell, R.D. & Hafeman, D.M. Positional firing properties of postrhinal cortex neurons. Neuroscience 119, 577588 (2003).
  43. Muller, R.U. & Kubie, J.L. The firing of hippocampal place cells predicts the future position of freely moving rat. J. Neurosci. 9, 41014110 (1989).

Download references

Author information

Affiliations

  1. Kavli Institute for Systems Neuroscience and Centre for the Biology of Memory, Norwegian Brain Centre, Norwegian University of Science and Technology (NTNU), Trondheim, Norway.

    • Tora Bonnevie,
    • Benjamin Dunn,
    • Marianne Fyhn,
    • Torkel Hafting,
    • Dori Derdikman,
    • Yasser Roudi,
    • Edvard I Moser &
    • May-Britt Moser
  2. Department of Cell Biology, State University of New York Downstate Medical Center, Brooklyn, New York, USA.

    • John L Kubie
  3. Present addresses: Department of Molecular Biosciences, University of Oslo, Kristine Bonnevies hus, Oslo, Norway (M.F. & T.H.) and Technion, Israel Institute of Technology, Rappaport Faculty of Medicine, Bat Galim, Haifa, Israel (D.D.).

    • Marianne Fyhn,
    • Torkel Hafting &
    • Dori Derdikman

Contributions

T.B. and M.F. performed the majority of the experiments; T.B. did the majority of the analyses; B.D. and Y.R. did the network simulations; E.I.M. and T.B. wrote the manuscript, except for the computational model (B.D. and Y.R.); and M.-B.M. supervised the project. All authors contributed to discussion and interpretation.

Competing financial interests

The authors declare no competing financial interests.

Corresponding authors

Correspondence to:

Author details

Supplementary information

PDF files

  1. Supplementary Text and Figures (9M)

    Supplementary Figures 1–8

Movies

  1. Supplementary Video 1 (24M)

    External input is large enough (100% hippocampal activity). This corresponds to an external input to the right of the transition in Fig. 8b. In this case, the activity on the neuronal sheet is a hexagonal grid. When the animal moves, this activity is translated on the network and follows the movement of the animal without being distorted. The resulting activity at the single cell levels is a hexagonal grid and thus high grid score.

  2. Supplementary Video 2 (24M)

    External input is below the transition. In this case, most of the time the activity on the neuronal sheet is still grid like, but the grid changes size, amplitude and orientation as it tries to follow the animals movement sometime turning into stripe patterns (see e.g. t = 0:16). The peak activity also substantially changes between different time steps. Consequently, no grid firing will be observed at the single cell level and substantially low grid scores are found.

Additional data