Unbalanced excitability underlies offline reactivation of behaviorally activated neurons

Journal name:
Nature Neuroscience
Volume:
17,
Pages:
503–505
Year published:
DOI:
doi:10.1038/nn.3674
Received
Accepted
Published online

Hippocampal sharp waves (SWs)/ripples represent the reactivation of neurons involved in recently acquired memory and are crucial for memory consolidation. By labeling active cells with fluorescent protein under the control of an immediate-early gene promoter, we found that neurons that had been activated while mice explored a novel environment were preferentially reactivated during spontaneous SWs in hippocampal slices in vitro. During SWs, the reactivated neurons received strong excitatory synaptic inputs as opposed to a globally tuned network balance between excitation and inhibition.

At a glance

Figures

  1. Reactivation of behaviorally activated neurons during in vitro SWs.
    Figure 1: Reactivation of behaviorally activated neurons during in vitro SWs.

    (a) Calcium imaging from dVenus and dVenus+ CA1 neurons loaded with CaSiR-1. (b) dVenus+ neurons participated more frequently in SWs than dVenus neurons. **P = 3.6 × 10−7, Fisher's exact test, n = 197 dVenus and 90 dVenus+ cells. (c) Spontaneous activity rates in slices treated with ZIP and scrambled ZIP (scrZIP), peptides known to eliminate previously occurring long-term potentiation, did not differ between dVenus and dVenus+ neurons. n = 123 or 115 dVenus and 75 or 51 dVenus+ cells. Error bars are s.d. of 6 slices from 3 mice. (d) dVenus (top) and dVenus+ (bottom) neurons were loaded with Alexa Fluor 568 and current-clamped at I = 0. (e) dVenus+ neurons participated more frequently in SWs than dVenus neurons. **P = 0.0057, Mann-Whitney U test, n = 20 dVenus and 19 dVenus+ cells in 21 slices from 21 mice.

  2. E/I imbalance during SWs.
    Figure 2: E/I imbalance during SWs.

    (a) EPSCs (top) and IPSCs (bottom) in a CA1 pyramidal cell during SWs. (b) Mean ± s.d. of 122 EPSCs (top) and 86 IPSCs (bottom) relative to the SW peak timings in the same neuron as in a. All SW-locked EPSCs and IPSCs recorded from this neuron were pooled. (c) Time evolution of the mean EPSC and IPSC in b. (d) The E/I time evolution averaged across all SWs recorded from 39 cells (from 21 slices in 21 mice). (eh) The mean peak amplitude of SW-relevant EPSCs (e), the mean peak amplitude of SW-relevant IPSCs (f), the EPSC-to-IPSC ratio of the mean charge (g), and the mean delay between EPSC and IPSC peak timings (h), compared between 28 SW nonparticipants and 11 SW participants. Error bars, s.d. The y axes are logarithmic in eg. **P = 0.0008, U11,28 = 47, Mann-Whitney-Wilcoxon test.

  3. Linearly tuned inhibitory activity.
    Figure 3: Linearly tuned inhibitory activity.

    (a) SW-relevant EPSC sizes recorded in CA1 pyramidal cells (top) were less correlated than the IPSC sizes (bottom) with the corresponding SW sizes: r2 = 0.15, n = 1,542 EPSCs; r2 = 0.46, n = 1,139 IPSCs; from 39 cells (in 21 slices from 21 mice). (b) A PV+ interneuron, which exhibited nonadaptive spiking (bottom), was targeted for whole-cell recording (bottom) using Alexa Fluor 568 in a PV-GFP transgenic mouse slice (top). Scale bar, 50 μm. (c) A putative PV+ basket cell, reconstructed from a biocytin–Alexa Fluor 594 photograph (inset). Scale bar, 100 μm. (d) The number of spikes emitted by PV+ cells during SWs was linear with the SW size. Gray lines indicate individual cells, and the black line represents the means ± s.d. of 2,901 SWs from 10 cells (in 6 slices from 4 mice). (e) The sizes of SW-relevant EPSCs recorded in PV+ cells correlated positively with the SW size: r2 = 0.47, n = 3,032 SWs from 10 cells (in 6 slices from 4 mice).

  4. SWs in vitro.
    Supplementary Fig. 1: SWs in vitro.

    (a) Representative LFP trace (top) and 120–250 Hz filtered trace (bottom) recordings from CA1 stratum pyramidale. SW timings are shown with blue lines. Three SWs are magnified in the left insets. The peak ripple oscillation frequency was 191 ± 22 Hz (mean ± SD). (b) The brains were sliced horizontally (top) or at an angle of 12.7° to the fronto-occipital axis (bottom). (c) CA3 stratum pyramidale was stimulated at various intensities, and neurons that fired in response to the stimulation were monitored in CA1 stratum pyramidale using functional multineuron calcium imaging with Oregon Green 488 BAPTA-1. The mean percentage of spiking neurons to the total neurons recorded in horizontal and oblique slices are plotted as a function of the stimulation intensity. These results imply that CA3-to-CA1 axonal projections were more preserved in oblique slices than in horizontal slices (**P = 6.6×10-5, F1,28 = 21.9, two-way ANOVA). Error bars are SDs of 4 or 5 slices from 3 mice. (d) Paired LFP recordings from CA3 and CA1. SWs in CA1 occurred together with CA3 network activity in an intact hippocampal slice. In slices that received surgical lesions between CA3 and CA1, CA3 network activity still remained, but SWs in CA1 disappeared. (n = 3 slices from 2 mice each).

  5. Calcium imaging of SW-relevant activities.
    Supplementary Fig. 2: Calcium imaging of SW-relevant activities.

    (a) Functional multineuron calcium imaging from the CA1 stratum pyramidale of an Oregon Green 488 BAPTA-1-loaded slice and recordings of local field potentials (LFP) (top). Simultaneous loose cell-attached recording and calcium imaging (bottom). (b) Representative raster plot of the spiking of CA1 neurons (bottom). SW-relevant activity is shown in blue. SW times (middle) were determined from simultaneously recorded LFPs (top). (c) Peri-SW time histogram of calcium transients (n = 1,625 cells in 15 slices from 15 mice). (d) Neurons activated during each SW event.

  6. dVenus expression after behavioral exploration of Arc-dVenus mice.
    Supplementary Fig. 3: dVenus expression after behavioral exploration of Arc-dVenus mice.

    (a) Schematic illustration of experimental procedures. SWs were monitored in hippocampal slices prepared from Arc-dVenus transgenic mice that had been allowed to explore a novel enriched environment for 30 min. (b) Representative control images of dVenus fluorescence (green) and Hoechst counterstain (blue) in control naïve mice (Home cage), animals that explored the novel enriched environment A for 30 min (30 min in A), animals that explored the novel but less-enriched environment B (30 min in B), and animals that sequentially explored environments A and B for 15 min each (15 min in A/15 min in B). Scale bar = 20 μm. (c) Percentages of dVenus(+) neurons to the total neurons were compared in home-cage mice (n = 3 mice), animals that explored environment A for 30 min (n = 4 mice), animals that explored environment B (30 min in B, n = 4 mice), and animals that explored sequentially explored A and B for 15 min each (15 min in A/15 min in B, n = 3 mice). Three 200-μm-thick sections were prepared per mouse from a middle part (2,600 μm ventral from bregma) of the hippocampus 5 h after the exploration, and 21 confocal sections were taken at a Z interval of 1 μm. Cells were considered dVenus(+) when the mean fluorescence intensity in the soma (ROI Φ = 5 μm) was higher than the background fluorescence intensity by twice its SD. The background intensities were measured from 50 ROIs (Φ = 5 μm) selected randomly from cell body-absent regions in the stratum radiatum in the same histological sections, and their mean and SD were used to detect dVenus(+) neurons. Neurons were identified by their nuclear morphology. Error bars represent the SEM. *P = 0.011, Q = 4.04 versus Home cage; **P = 0.0052, Q = 4.50 versus Home cage; Tukey's multiple comparison test after one-way ANOVA.

  7. Preferential SW participation of dVenus+ neurons in control mice.
    Supplementary Fig. 4: Preferential SW participation of dVenus+ neurons in control mice.

    dVenus(+) neurons in naïve mice that were not exposed to novel environments were more frequently activated during SWs when compared to dVenus(–) neurons in those mice. **P = 6.7×10-3, Fisher's exact probability test, n = 82 dVenus(–) and 31 dVenus(+) cells in 4 slices in 2 mice

  8. No difference in SW-irrelevant, basal synaptic activity between SW participants and nonparticipants.
    Supplementary Fig. 5: No difference in SW-irrelevant, basal synaptic activity between SW participants and nonparticipants.

    (a-g) The mean sizes (a,c) and frequencies (b,d) of spontaneous EPSCs (a,b) or IPSCs (c,d) and the EPSC-to-IPSC ratio of the mean size (e), the mean frequency (f), and the mean charge (g) during periods without SWs were compared between SW nonparticipants (open circles) and participants (closed circles). Error bars represent the SDs of 28 SW nonparticipants and 11 SW participants. All parameters scored P > 0.1, Mann-Whitney-Wilcoxon test.

  9. No difference in intrinsic properties between SW participants and nonparticipants.
    Supplementary Fig. 6: No difference in intrinsic properties between SW participants and nonparticipants.

    (a) Baseline membrane potential (Vmrest) of SW nonparticipants (open circles) and participants (closed circles) is shown. (b) Input resistance (Rinput) of SW nonparticipants and participants is shown. Error bars represent the SDs of 28 SW nonparticipants and 11 SW participants.

  10. Dynamic-clamp stimulation of a CA1 pyramidal cell.
    Supplementary Fig. 7: Dynamic-clamp stimulation of a CA1 pyramidal cell.

    (a) Joint conductances of SW-relevant EPSC-like and IPSC-like waveforms were injected into a CA1 pyramidal cell under the pharmacological blockade of fast synaptic transmission. (b) Raw traces of the spike responses to the conductance injection in a representative CA1 neuron are shown with truncated spikes. The conductance peak amplitude was 0.53–4 nS for Ge(t) and 4 nS for Gi(t), whereas the time difference between Ge(t) and Gi(t) ranged from 0 to 6 ms. We tested 49 (= 7 × 7) combinations of EPSC and IPSC waveforms. (c) Pseudocolored map of the probability that CA1 pyramidal neurons fire spikes in response to dynamic-clamp conductance injection with various combinations of the EPSC-to-IPSC ratios and timings. Data were averaged from 9 neurons (20 trials each, from 5 slices in 5mice). (d) The same data shown in (c) are plotted as the means ± SD.

Videos

  1. Time-lapse confocal imaging of calcium activity from neurons.
    Video 1: Time-lapse confocal imaging of calcium activity from neurons.
    The CA1 stratum pyramidale of a hippocampal slice was loaded with Oregon Green 488 BAPTA-1, and neuronal spiking activity was monitored at 50 Hz with simultaneous LFP recording. The timings of individual SWs are indicated by flashes of a blue box.

References

  1. Lee, A.K. & Wilson, M.A. Neuron 36, 11831194 (2002).
  2. Wilson, M.A. & McNaughton, B.L. Science 265, 676679 (1994).
  3. Girardeau, G., Benchenane, K., Wiener, S.I., Buzsaki, G. & Zugaro, M.B. Nat. Neurosci. 12, 12221223 (2009).
  4. Eguchi, M. & Yamaguchi, S. Neuroimage 44, 12741283 (2009).
  5. Egawa, T. et al. J. Am. Chem. Soc. 133, 1415714159 (2011).
  6. Volk, L.J., Bachman, J.L., Johnson, R., Yu, Y. & Huganir, R.L. Nature 493, 420423 (2013).
  7. Kameda, H. et al. Eur. J. Neurosci. 35, 838854 (2012).
  8. Hájos, N. et al. J. Neurosci. 33, 1167711691 (2013).
  9. Norimoto, H., Mizunuma, M., Ishikawa, D., Matsuki, N. & Ikegaya, Y. Brain Res. 1461, 19 (2012).
  10. Woodruff, A., Xu, Q., Anderson, S.A. & Yuste, R. Front. Neural Circuits 3, 15 (2009).
  11. Takahashi, N. et al. Curr. Protoc. Neurosci. 2, 14 (2011).
  12. Sasaki, T., Takahashi, N., Matsuki, N. & Ikegaya, Y. J. Neurophysiol. 100, 16681676 (2008).
  13. Takahashi, N., Sasaki, T., Matsumoto, W., Matsuki, N. & Ikegaya, Y. Proc. Natl. Acad. Sci. USA 107, 1024410249 (2010).
  14. Sun, Y., Norimoto, H., Pu, X.P., Matsuki, N. & Ikegaya, Y. J. Pharmacol. Sci. 118, 288294 (2012).
  15. Maier, N. et al. Neuron 72, 137152 (2011).

Download references

Author information

Affiliations

  1. Laboratory of Chemical Pharmacology, Graduate School of Pharmaceutical Sciences, University of Tokyo, Tokyo, Japan.

    • Mika Mizunuma,
    • Hiroaki Norimoto,
    • Kentaro Tao,
    • Tetsuya Sakaguchi,
    • Norio Matsuki &
    • Yuji Ikegaya
  2. Laboratory of Chemistry and Biology, Graduate School of Pharmaceutical Sciences, University of Tokyo, Tokyo, Japan.

    • Takahiro Egawa,
    • Kenjiro Hanaoka &
    • Tetsuo Nagano
  3. Department of Morphological Brain Science, Graduate School of Medicine, Kyoto University, Kyoto, Japan.

    • Hiroyuki Hioki &
    • Takeshi Kaneko
  4. Division of Morphological Neuroscience, Gifu University Graduate School of Medicine, Gifu, Japan.

    • Shun Yamaguchi
  5. PRESTO, Japan Science and Technology Agency, Saitama, Japan.

    • Shun Yamaguchi
  6. Center for Information and Neural Networks, Suita City, Osaka, Japan.

    • Yuji Ikegaya

Contributions

M.M. and Y.I. designed the study. M.M., H.N. and K.T. conducted the experiments. M.M., H.N., K.T., T.S. and Y.I. analyzed the data. T.E., K.H. and T.N. prepared calcium indicator. H.H., T.K. and S.Y. prepared transgenic mice. M.M. and Y.I. wrote the paper. The study was managed by Y.I. and N.M.

Competing financial interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to:

Author details

Supplementary information

Supplementary Figures

  1. Supplementary Figure 1: SWs in vitro. (155 KB)

    (a) Representative LFP trace (top) and 120–250 Hz filtered trace (bottom) recordings from CA1 stratum pyramidale. SW timings are shown with blue lines. Three SWs are magnified in the left insets. The peak ripple oscillation frequency was 191 ± 22 Hz (mean ± SD). (b) The brains were sliced horizontally (top) or at an angle of 12.7° to the fronto-occipital axis (bottom). (c) CA3 stratum pyramidale was stimulated at various intensities, and neurons that fired in response to the stimulation were monitored in CA1 stratum pyramidale using functional multineuron calcium imaging with Oregon Green 488 BAPTA-1. The mean percentage of spiking neurons to the total neurons recorded in horizontal and oblique slices are plotted as a function of the stimulation intensity. These results imply that CA3-to-CA1 axonal projections were more preserved in oblique slices than in horizontal slices (**P = 6.6×10-5, F1,28 = 21.9, two-way ANOVA). Error bars are SDs of 4 or 5 slices from 3 mice. (d) Paired LFP recordings from CA3 and CA1. SWs in CA1 occurred together with CA3 network activity in an intact hippocampal slice. In slices that received surgical lesions between CA3 and CA1, CA3 network activity still remained, but SWs in CA1 disappeared. (n = 3 slices from 2 mice each).

  2. Supplementary Figure 2: Calcium imaging of SW-relevant activities. (239 KB)

    (a) Functional multineuron calcium imaging from the CA1 stratum pyramidale of an Oregon Green 488 BAPTA-1-loaded slice and recordings of local field potentials (LFP) (top). Simultaneous loose cell-attached recording and calcium imaging (bottom). (b) Representative raster plot of the spiking of CA1 neurons (bottom). SW-relevant activity is shown in blue. SW times (middle) were determined from simultaneously recorded LFPs (top). (c) Peri-SW time histogram of calcium transients (n = 1,625 cells in 15 slices from 15 mice). (d) Neurons activated during each SW event.

  3. Supplementary Figure 3: dVenus expression after behavioral exploration of Arc-dVenus mice. (190 KB)

    (a) Schematic illustration of experimental procedures. SWs were monitored in hippocampal slices prepared from Arc-dVenus transgenic mice that had been allowed to explore a novel enriched environment for 30 min. (b) Representative control images of dVenus fluorescence (green) and Hoechst counterstain (blue) in control naïve mice (Home cage), animals that explored the novel enriched environment A for 30 min (30 min in A), animals that explored the novel but less-enriched environment B (30 min in B), and animals that sequentially explored environments A and B for 15 min each (15 min in A/15 min in B). Scale bar = 20 μm. (c) Percentages of dVenus(+) neurons to the total neurons were compared in home-cage mice (n = 3 mice), animals that explored environment A for 30 min (n = 4 mice), animals that explored environment B (30 min in B, n = 4 mice), and animals that explored sequentially explored A and B for 15 min each (15 min in A/15 min in B, n = 3 mice). Three 200-μm-thick sections were prepared per mouse from a middle part (2,600 μm ventral from bregma) of the hippocampus 5 h after the exploration, and 21 confocal sections were taken at a Z interval of 1 μm. Cells were considered dVenus(+) when the mean fluorescence intensity in the soma (ROI Φ = 5 μm) was higher than the background fluorescence intensity by twice its SD. The background intensities were measured from 50 ROIs (Φ = 5 μm) selected randomly from cell body-absent regions in the stratum radiatum in the same histological sections, and their mean and SD were used to detect dVenus(+) neurons. Neurons were identified by their nuclear morphology. Error bars represent the SEM. *P = 0.011, Q = 4.04 versus Home cage; **P = 0.0052, Q = 4.50 versus Home cage; Tukey's multiple comparison test after one-way ANOVA.

  4. Supplementary Figure 4: Preferential SW participation of dVenus+ neurons in control mice. (23 KB)

    dVenus(+) neurons in naïve mice that were not exposed to novel environments were more frequently activated during SWs when compared to dVenus(–) neurons in those mice. **P = 6.7×10-3, Fisher's exact probability test, n = 82 dVenus(–) and 31 dVenus(+) cells in 4 slices in 2 mice

  5. Supplementary Figure 5: No difference in SW-irrelevant, basal synaptic activity between SW participants and nonparticipants. (111 KB)

    (a-g) The mean sizes (a,c) and frequencies (b,d) of spontaneous EPSCs (a,b) or IPSCs (c,d) and the EPSC-to-IPSC ratio of the mean size (e), the mean frequency (f), and the mean charge (g) during periods without SWs were compared between SW nonparticipants (open circles) and participants (closed circles). Error bars represent the SDs of 28 SW nonparticipants and 11 SW participants. All parameters scored P > 0.1, Mann-Whitney-Wilcoxon test.

  6. Supplementary Figure 6: No difference in intrinsic properties between SW participants and nonparticipants. (69 KB)

    (a) Baseline membrane potential (Vmrest) of SW nonparticipants (open circles) and participants (closed circles) is shown. (b) Input resistance (Rinput) of SW nonparticipants and participants is shown. Error bars represent the SDs of 28 SW nonparticipants and 11 SW participants.

  7. Supplementary Figure 7: Dynamic-clamp stimulation of a CA1 pyramidal cell. (139 KB)

    (a) Joint conductances of SW-relevant EPSC-like and IPSC-like waveforms were injected into a CA1 pyramidal cell under the pharmacological blockade of fast synaptic transmission. (b) Raw traces of the spike responses to the conductance injection in a representative CA1 neuron are shown with truncated spikes. The conductance peak amplitude was 0.53–4 nS for Ge(t) and 4 nS for Gi(t), whereas the time difference between Ge(t) and Gi(t) ranged from 0 to 6 ms. We tested 49 (= 7 × 7) combinations of EPSC and IPSC waveforms. (c) Pseudocolored map of the probability that CA1 pyramidal neurons fire spikes in response to dynamic-clamp conductance injection with various combinations of the EPSC-to-IPSC ratios and timings. Data were averaged from 9 neurons (20 trials each, from 5 slices in 5mice). (d) The same data shown in (c) are plotted as the means ± SD.

Video

  1. Video 1: Time-lapse confocal imaging of calcium activity from neurons. (2.07 MB, Download)
    The CA1 stratum pyramidale of a hippocampal slice was loaded with Oregon Green 488 BAPTA-1, and neuronal spiking activity was monitored at 50 Hz with simultaneous LFP recording. The timings of individual SWs are indicated by flashes of a blue box.

PDF files

  1. Supplementary Text and Figures (1,205 KB)

    Supplementary Figures 1–7 and Supplementary Table 1

Additional data