Figure 1.
Quasi-synchrony and topography of spontaneous and evoked KCs in EEG and ECOG.
A, An individual spontaneous KC shows quasi-synchrony over frontal, central, and occipital scalp EEG channels. Overlaid channels participating in the KC are color-coded by scalp position: frontal channels in green, central in pink, and occipital in purple. There is no evidence of a systematic propagation of green frontal to purple posterior channels. The vertical black line is plotted at the time point where there is a 0 ms delay in the timing of the most negative peak of the KC in channels across the anterior to posterior axis: Fp1, Fpz, F1, Fc2, C4, Po3, and Po8. B, Almost all ECOG channels show a quasi-synchronous surface negative potential at the time of a single spontaneous KC detected at the scalp. The left plot depicts the color coded local field potentials while the right plot depicts the waveforms of this KC. Regions listed are: temporal pole (TempPole), lateral occipital (LatOc), anterior orbitofrontal (antOrbF), anterior prefrontal (antPreF), and ventral occipital (VenOc). C, Both spontaneous (black) and evoked (red) KCs in ECOG are characterized by a quasi-synchronous and widespread surface negativity. KCs were averaged over trials on the most negative peak of the KC as detected at the pictured scalp EEG electrode, C4. Waveforms are the average of 86 spontaneous (black) and 42 evoked (red) KCs. Patient is the same as in B. D, Traces of KC averages from all channels pictured in C are superimposed to demonstrate the regularity of the waveforms across sites. E, On both an average and single trial basis, spontaneous and evoked KCs show similar topographical profiles that are quasi-synchronous across the cortex. The scalp EEG electrode and ECOG channels from across the cortex (selected to be maximally far apart) are pictured for the averaged spontaneous or evoked KC, as depicted in C & D, as well as a single trial. The vertical black line indicates time zero of the most negative peak of the scalp KC on the average (left) and the single trial (right). Channel location regions listed are the same as in B.
Figure 2.
Bipolar SEEG recordings illustrate widespread spontaneous KC quasi-synchrony across lobes and hemispheres.
Single spontaneous (dashed lines) and averaged (solid lines) KCs detected on Fz for each patient and computed for each selected pair of bipolar contacts. The color of each waveform corresponds to the colored arrow indicating its location on the reconstructed hemispheres and coronal MRI sections for each patient. A, A single trial and the average of 70 KCs for Patient 2 with contacts across the frontal, temporal, and parietal lobes. On average, there is a 32 ms delay between the anterior (inferior frontal sulcus, LbiFg) and posterior (angular gyrus, LAng) channels shown by the yellow arrows. B, A single trial and the average of 129 KCs for Patient 3 with contacts in both hemispheres. The yellow arrow marks mirrored locations in the left and right superior frontal gyrus (RsFg and LsFg), which have an average KC peak latency delay of 2 ms. C, A single trial and the average of 172 KCs for Patient 4 with contacts in both hemispheres for subfrontal and frontal regions. In mirrored sites in the left and right precentral sulcus (LpCs and RpCs), highlighted by the yellow arrow, there is an average 18 ms peak latency difference.
Figure 3.
Network geometry and altered projections of thalamocortical model.
A, The network contained 100 PY, 25 IN, 50 TC, and 50 RE neurons. The number of projections between PY and RE neurons was altered to test two ways of evoking KCs. B, In the original network geometry, each PY neuron projected to 5 RE neurons. To evoke KCs, all RE neurons were depolarized. C, In the altered connectivity, a subset of PY neurons projected to all RE neurons, while the remaining PY neurons projected to 5 RE neurons. To evoke KCs, only the PY neurons projecting to all RE neurons were stimulated.
Figure 4.
Characteristics of a single spontaneous KC.
A KC generated when 15 PY neurons projected to all RE neurons. For each of the neuronal populations, the membrane potential of the individual neurons (Vm), the average membrane potential (Avg Vm), the spiking of a single neuron (Cell), and the average spiking rate (Avg Spikes/Sec) are pictured. In addition, high gamma power is plotted for PY neurons and spindle power is plotted for RE and TC neurons. The KC is characterized by a cessation of firing by all cell types, with a drop in membrane potential (blue asterisks), and high gamma power (black asterisk) in PY neurons. The model exhibited spindling (green asterisks), which dropped in the RE and TC neurons during the cortical KC (red asterisks). The red arrow indicates the 15 PY neurons that are connected to all RE neurons, and the red vertical lines mark the start of RE spindle disruption. The membrane potential color scale displayed in the middle of the figure is the same for all cell populations.
Figure 5.
Characteristics of 39 averaged spontaneous KCs.
KCs generated when 15 PY neurons projected to all RE neurons, as in Figure 4. For each of the populations, the membrane potential of the individual neurons (Vm), the average membrane potential (Avg Vm), and the average spiking rate (Avg Spikes/Sec) (averaged over the 39 KCs) are pictured. High gamma power is plotted for all PY neurons. In addition, the membrane potential (Vm) and the averaged spiking (Avg Vm) of the 15 PY neurons projecting to all RE neurons, averaged over the 39 KCs, are plotted separately. The red arrow indicates the 15 PY neurons that are connected to all RE neurons. The color scale of the PY membrane potential is the same for all PY neurons and these 15 PY neurons. The averaged KC exhibited the same characteristics as the individual KC shown in Figure 4: cessation of firing by all cell types, and a drop in all PY neurons in membrane potential (blue asterisks), and high gamma power (black asterisk). Spindling also dramatically decreased in the RE and TC neurons during the cortical KC (red asterisks). The 15 PY neurons showed a marked increase in spontaneous firing before the KC (yellow asterisk).
Figure 6.
Parametric modulation of spontaneous KC frequency by changing the projection from PY to RE.
A, Increasing the number of PY neurons projecting to all RE neurons increased the number of spontaneous KCs in a near linear fashion. B, In contrast, increasing the strength of projections from PY neurons to RE neurons had little effect until a threshold strength was reached and the system began oscillating almost continuously (i.e., generated SOs rather than KCs). In A, the PY to RE configuration schematic on the left illustrates one PY cell projecting to all RE neurons, while each of the remaining PY neurons maintain a projection to 5 RE neurons (as in Figure 3C). The schematic on the right shows how the projection pattern changes with the addition of a second PY neuron projecting to all RE neurons. In B, the schematic on the left illustrates the original model configuration (as in Figure 3B), while the right schematic shows the increasing strength of PY to RE connections. For each value on the x-axis, 10 simulations of 200 s with different random seeds were run. The number of spontaneous KCs was calculated for each run; each value represents the average number of KCs per minute in the 10 runs ± SEM. The black arrow marks 10 PY neurons projecting to all RE neurons while the red arrow marks the original connectivity with a strength of 1 µS between PY and RE neurons.
Figure 7.
Evoking KCs by directly versus indirectly depolarizing RE.
A, KCs are evoked by directly depolarizing all RE neurons for 350 ms at 85.8pA. B, KCs are evoked by stimulating 6 PY neurons, indicated by the red arrow, for 200 ms at 15pA. In A & B, the membrane potential of individual neurons (Vm), the average membrane potential (Avg Vm), the spiking of a single cell, and the average spiking rate (Avg Spikes/Sec) are graphed for each population. The membrane potential color scales are the same for both panels for each cell population. The number of the individual cell graphed is also the same between panels. The length of the RE depolarization is outlined by the orange box and the length of the stimulation of the 6 PY neurons is outlined by the black box. In both cases, the KC was quantified by a drop in the PY membrane potential and the spiking in cell populations (blue asterisks and “KC”). There was a marked increase in the membrane potential of the stimulated 6 PY neurons projecting to all RE neurons (yellow asterisk). In both cases, RE membrane potential was depolarized for the duration of RE current injection or PY stimulation (purple asterisks), but this did not correspond to an increase in RE spiking (orange asterisks). Like the spontaneous KC (Figures 4 & 5), a drop in the TC and RE spindling occurred at the same time as the evoked KC (red asterisks) in both cases.
Figure 8.
Parametric analysis of evoking KCs by directly versus indirectly depolarizing RE.
The effect of increasing levels of RE depolarization or PY stimulation was inspected at individual time points. The schematic to the left outlines the connectivity and the order of effects on: 1) RE neurons, 2) TC neurons, and 3) PY neurons, based on RE depolarization (A) or PY stimulation (B). Five simulations with different random seeds were run for each value plotted, as well as a no stimulation baseline run. Baseline correction was performed for membrane potential (Vm) and high gamma power. Percent change was calculated for spindling power and firing. The average of five runs at a particular stimulation value and time point is plotted ± SEM. A, As the level of applied RE depolarization increases, so does the RE membrane potential (purple asterisk); however, RE firing does not increase until high levels of depolarization are reached (red box and orange asterisk). The green box highlights a level of RE depolarization that is subthreshold for evoking KCs. At 114.4pA (purple box), PY firing and high gamma power drop (blue and black asterisks, respectively), indicating a KC. Spindling also decreases in all three cell populations (red asterisk). The red box outlines a level of stimulation where RE firing increases, leading to the production of a KC. B, When stimulating 6 PY neurons projecting to all RE neurons, the spindling in all three populations drops 100 ms after stimulation is applied at a level of 20pA or higher (red asterisk). At 20pA (purple box), the firing in all cell populations (blue asterisk) and high gamma power in PY neurons (black asterisk) drop 300 ms after stimulation, indicating a KC. The green box highlights a subthreshold level of PY stimulation. At 80pA of PY stimulation (grey box), there is a drop in spindling but no KC, presumably because the direct cortical excitation is sufficient to counteract the removal of thalamic input (green asterisks).
Figure 9.
Currents and conductances underlying KCs.
A, A spontaneous KC (same KC as in Figure 4), B, a KC evoked by depolarizing all RE neurons (same KC as in Figure 7A), or C, a KC evoked by stimulating 6 PY neurons projecting to all RE neurons (same KC as in Figure 7B), all show a drop in PY spiking (“KC”), and in RE and TC spindling (red asterisks). The orange box in B and the black box in C indicate the length of the applied RE depolarization or PY stimulation, respectively. In all three cases, these characteristics of the KC were preceded by RE IT inactivation dropping towards zero (RE IT h, purple asterisks), indicating greater inactivation. A decrease in PY to RE currents and a decrease in TC IT accompanied this RE IT inactivation drop. During the KC, an increase in RE IT h indicates greater deinactivation (green asterisks), and rebound spindling.
Figure 10.
Inducing KCs by abruptly increasing RE IT inactivation (RE IT h).
A, In the original configuration and, B, in the 6 PY neurons projecting to all RE neurons configuration, RE IT h is scaled to 40% of its original value (purple asterisks) at 1 s (vertical black line). In other words, the proportion of deinactivated IT channels is abruptly decreased. In both cases, this drop in RE IT h leads to a decrease in PY to RE currents and TC IT, as well as a drop in RE and TC spindling (red asterisks), which ultimately leads to a KC, as indicated by decreased PY spiking (“KC”).
Figure 11.
Hyperpolarizing TC neurons produces a KC by decreasing the thalamocortical drive to cortex.
All TC neurons were hyperpolarized at 116pA for 300 ms. The length of TC hyperpolarization is outlined by the blue box. For the duration of TC hyperpolarization (blue asterisk), RE and TC spindling dropped (red asterisk), RE IT became deinactivated (green asterisk), and PY spiking dropped to zero (“KC”), indicating a KC. This was followed by a rebound upstate, as marked by TC depolarization, increased RE and TC spindling above baseline levels, and high PY spiking (black arrows).
Figure 12.
Disruption of spindling in the human thalamus precedes the cortical KC.
KCs chosen on a single prefrontal bipolar SEEG channel located in Brodmann's area 10 are displayed as single trials (A, 50 randomly selected individual KCs) and, B, the average of all 229 KCs, band pass filtered to 0.1 to 5 Hz. C, Time frequency analysis (5–120 Hz) in the 1st (most medial) thalamic bipolar SEEG channel using the times of the frontal KCs in B, thresholded at p<0.01 (uncorrected) compared to the −1.5 to −0.5 second baseline. The blue arrow shows that there appears to be a drop in spindle power. D, The average absolute value of the Hilbert transform applied to the frontal KCs in B, band pass filtered for high gamma (60–120 Hz), is plotted in black. The grey box indicates the time period where high gamma drops significantly compared to baseline, outlined with a black box (−1.5 to −1 seconds, p<0.01, FDR corrected). The blue line indicates the average absolute value of the Hilbert transform applied to the 1st thalamic bipolar SEEG channel band pass filtered for spindling (12–16 Hz), using the times of the frontal KCs. The blue box indicates the time period where spindling drops significantly compared to baseline (p<0.01, FDR corrected). The drop in thalamic spindling (blue box) occurs prior to the drop in cortical high gamma (grey box). E& F, The same analysis as outlined in D is applied to the 2nd thalamic bipolar channel (E) and the 3rd thalamic bipolar channel (F). The cortical data are the same in all three subplots; the thalamic spindling amplitude scales are individualized for each thalamic bipolar contact in D–F. For all three thalamic bipolar channels, thalamic spindling drops significantly (blue box) prior to the significant cortical high gamma drop (grey box).
Table 1.
Patient demographics and clinical characteristics.