Article Figures & Data
Figures
- Figure 1.
Model matches h-current-mediated FDG peaks observed experimentally in human layer 5 cortical pyramidal neurons. A, B, Experimentally calculated averaged non-normalized (A) and normalized FDG (B) of n = 3 L5 human pyramidal neurons with and without blockade of the h-channel with ZD-7288 (see above, Materials and Methods). Under control conditions, there is a clear low-frequency peak at ∼5 Hz. After treatment with ZD-7288, the peak FDG values occur at >10 Hz. Normalized curves exhibit significant differences (p = 0.0017) between 4.6 and 6 Hz, the location of this low-frequency peak. The mean SD of the non-normalized FDG curve is 26.849 before treatment with ZD-7288 and 63.094 after treatment with ZD-7288, and for the normalized FDG curve it is 0.172 before treatment with ZD-7288 and 0.145 after treatment with ZD-7288. C, D, Averaged non-normalized (C) and normalized FDG (D) derived from the model human L5 cortical pyramidal neuron under normal conditions and without h-current activity (see above, Materials and Methods). Normalized plots emphasize the qualitative correspondence between the model and experimental settings; the model exhibits a low-frequency peak (here at ∼3 Hz) under normal conditions, whereas this peak dissipates (yielding a flat FDG profile for >3 Hz) without h-current activity. Non-normalized plots are significantly different for 1–30 Hz with p = 0.0001; normalized plots are significantly different with p = 0.0076 for 1–1.6 Hz and with p = 0.0265 from 10–12.6 Hz. The mean SD of the non-normalized FDG curve is 76.170 under normal conditions and 71.779 with zero h-current activity, and for the normalized FDG curve it is 0.249 under normal conditions and 0.257 with zero h-current activity. Note that the normalized plots (B, D) are not simply a rescaling of the absolute FDGs (A, C), but are processed as defined (see above, Materials and Methods). All significance values are derived from the two-way ANOVA with Bonferroni’s multiple comparisons test. E, F, Example input current (bottom) and output voltage (top) traces in the experimental and model settings. E is replicated with permission from Moradi Chameh et al. (2021) while the default model setting is used in panel F.
- Figure 2.
Currentscape visualization tool facilitates the identification of relative current contributions in spiking simulations. A, Raw data obtained via NEURON of a simulation of a noisy input current injected into the L5 human pyramidal neuron model. Top to bottom: Noisy input current, voltage trace, and ionic current contributions. Of note is that the varying magnitudes of the ionic currents obscure one’s ability to jointly visualize their contribution to the dynamics of the neuron. B, Using the Currentscape tool, the ionic current contributions are recontextualized as the outward and inward percentages of contribution plots, allowing for a more intuitive visualization on a single scale despite the spiking activity of the neuron. Note that Im is not discernible in this visualization, denoting its minimal contribution to the dynamics of the model in this scenario.
- Figure 3.
Currentscape visualization highlights differences in spiking modulated by h-current activity. A, Left, The example noisy input current injected into the L5 human pyramidal neuron model with differing Ih conductance levels. Right, Example individual non-normalized FDGs [plot is Gain (nA) vs log-scaled frequency (Hz)] generated from this specific noisy input in each of the following scenarios: with the Ih maximum conductance doubled (top left, purple), normal (top right, blue), halved (bottom left, teal), or zero (bottom right, black). B–E, Currentscape visualizations corresponding with each of the scenarios outlined above. The h-channel contribution is indicated in the inward current contribution of the visualization in yellow (fourth row, “In %”), alongside the contributions of all the ionic currents and passive currents in the model. Spiking activity differs in each scenario (see regime highlighted by the gold box) driven by the varying Ih conductance and the differing contributions of Ih to the activity of the neuron highlighted by this Currentscape visualization.
- Figure 4.
Currentscape data processing merged with spike-triggered average analysis of human cortical layer 5 neuron dynamics. A–C, Representations of the STA for the percentage contribution to outward (top) and inward (bottom) currents for 30 ms (A), 100 ms (B), and 200 ms (C) before a spike. A unique dynamic is observed in the h-current activity in C; over these 200 ms, the contribution of the h-channel is distinctively nonmonotonic. The shaded portion of each plot represents ± 1 SD over the 30 repetitions with distinct noisy inputs. Im is not included in these plots given its minimal contribution to model dynamics (Fig. 2).
- Figure 5.
The nonmonotonic nature of h-current activity before spiking is exaggerated with an increased h-channel conductance and diminished with a decreased h-channel conductance. A–C, Visualizations of the h-channel contributions preceding spiking with the Ih maximum conductance doubled in A, normal in B, and halved in C, with the shaded regions representing ± 1 SD. D, E, Non-normalized (D) and normalized (E) average FDGs in each of the above scenarios (see above, Materials and Methods). The normalized plots highlight the flatter decay of the FDG at high frequencies when the Ih maximum conductance is halved and the more precipitous drop when the Ih maximum conductance is doubled. The former diminishes the influence of any low-frequency peaks, whereas the latter accentuates it. The mean SD of the non-normalized FDG curve is 76.170 under normal conditions, 74.763 when the Ih maximum conductance is doubled, and 72.422 when it is halved; for the normalized FDG curve the mean SD is 0.249 under normal conditions, 0.203 when the Ih maximum conductance is doubled, and 0.235 when it is halved.
- Figure 6.
FDG properties are distinct in rodent L5 cortical pyramidal neurons, both in vitro and in silico, corresponding with a distinct h-channel contribution. A, B, Non-normalized (A) and normalized (B) FDGs averaged from n = 6 rodent L5 cortical pyramidal neurons firing approximately in the theta frequency range. The mean SD of the non-normalized FDG curve is 55.769 before treatment with ZD-7288 and 66.734 after treatment with ZD-7288, and for the normalized FDG curve it is 0.224 before treatment with ZD-7288 and 0.265 after treatment with ZD-7288. C, D, Non-normalized (C) and normalized (D) FDG analysis of the rodent L5 model of Hay et al. (2011) captures key qualitative properties displayed experimentally, including the lack of a peak in the 2–6 Hz range and minimal change when the h-current is blocked (here the FDG curves are identical), both stark contrasts from the human setting in vitro and in silico. The mean SD of the non-normalized FDG curve is 28.343 under normal conditions and 28.343 with zero h-current activity, and for the normalized FDG curve it is 0.261 under normal conditions and 0.261 with zero h-current activity. E, Comparison between the normalized FDGs of human and rodent L5 cortical pyramidal neurons highlights their differences; these are statistically significant before treatment with ZD-7288 with p = 0.0017 between 5 and 5.4 Hz (2-way ANOVA, Bonferroni’s multiple comparisons test). F, STA analysis of the inward currents for the default model of Hay et al. (2011) (magnified to emphasize the contribution of the h-current). In comparison to the human model, the contribution is much smaller in magnitude and notably more monotonic.
In this issue
