Figures and data
HCN expression and effect of Cs+ on AP firing in cortical L2/3 PCs
a. HCN gene subunits show uniform expression pattern among cortical PC types. Data is collected from the online available dataset(29). b. HCN1 immunohistochemistry reveals dense somatic expression in layer 2/3 of the visual cortex c. HCN channels control the excitability of L2/3 PCs. Schematic illustration of the experimental arrangement of cell attached recordings during extracellular stimulation of L4, and the shape of the elicited action potential. d. Representative traces showing increased firing upon ZD-7288 application. e. ZD-7288 increases the firing of L2/3 PCs upon electrical stimulation (0.1 ± 0.03 spikes/stimuli vs 0.4 ± 0.12 spikes/stimuli in control vs ZD-7288 application, respectively, p=0.04, t(4)=-3.0, Student’s paired-sample t-test, n=5). f. Increased spiking of L2/3 PC upon blockade of HCN channels with 2 mM CsMeSO4 (right). g. Pharmacological block of Ih significantly increases the firing response of L2/3 PCs (0.22 ± 0.06 spikes/stimuli vs. 0.48 ± 0.09 spikes/stimuli for control vs CsMeSO4 bath application, p=0.002, t(9)=-4.42, Student’s paired t-test, n=10. h. Presynaptic contributions were examined using the paired-pulse ratio (PPR) in whole cell current clamp recordings (1.09 ± 0.12 spikes/stimuli vs. 1.17 ± 0.15 spikes/stimuli for control vs CsMeSO4 bath application, p=0.422, t(6)=-0.8612, Student’s paired t-test, n=7).
Functionally relevant Ih in L2/3 PCs
a. Whole cell current clamp recordings showing the presence and absence of sag response to very negative current commands in control conditions (black) and upon Cs+ application (red). b. Steady-state voltage response rectification upon hyperpolarizing current injections is mediated by Ih. c. Sag voltage is abolished by blocking Ih. d. Sag amplitude is Cs+ sensitive (0.7 ± 0.49 mV vs. 0.05 ± 0.1 mV for control vs CsMeSO4 application, p=2.46-4, t(22)=4.37, Student’s two-sample t-test, n=13 and 11). e. Resting membrane potential is modulated by Ih (-76.75 ± 6.37 mV vs. -80.29 ± 6.48 mV for control vs CsMeSO4 bath application, p=1.98-4, t(9)=6.02, Student’s paired t-test, n=10). f. Input resistance is altered by Ih (74.78 ± 42.34 mV vs. 104.99 ± 52.91 mV for control vs CsMeSO4 bath application, p=0.048, t(6)=-2.48, Student’s paired t-test, n=7). g. The time course of voltage responses to hyperpolarizing current on the magnitude of the current injection is abolished by Cs+. h. Example voltage clamp recording of cesium sensitive currents in a L2/3 PC. i. Voltage dependence of cesium sensitive current activation, j. Voltage dependent activation kinetics of cesium sensitive currents.
Noncanonical effect of Ih block on EPSPs in L2/3 PCs
a. Experimental arrangement of extracellular targeted stimulation of a L2/3 PC dendrite. b. Application of ZD-7288 revealed a proximal bias for EPSP modulation by Ih (Exponential fit τ=104.7 µm, R2=0.33, n=18). Inset shows a stimulating location 28 µm away from the soma in control conditions (black trace) and upon Z-7288 application (green trace). c. EPSP halfwidth is significantly modulated along the dendritic axis by Ih, (proximal dendritic locations: 17.42 ± 1.68 ms vs. 40.95 ± 24.78 ms for control vs 50µm ZD-7288 bath application, p=0.03, t(6)=2.84, Student’s paired t-test, n=7, distal dendritic locations: 22.68 ± 1.78 ms vs. 28.65 ± 2.17 ms for control vs 50µm ZD-7288 bath application, p=2.1- 4, t(8)=-6.39, Student’s paired t-test, n=9). d. Schematic illustration of the experimental arrangement. L2/3 PCs were recorded in whole-cell current clamp mode, and a stimulating electrode was placed either in L1 or L4. e. Ih modulates EPSP halfwidth for L4 stimulation, but not for L1 (13.87 ± 1.37 ms vs. 15.8 ± 1.67 ms for control vs CsMeSO4 bath application in L1, p=0.38, t(18)=-0.89, Student’s two- sample t-test, n=10 each, 10.65 ± 0.64 ms vs. 20.17 ± 2.88 ms for control vs CsMeSO4 bath application for L4-Type I inputs, p=0.005, t(18)=-3.22, Student’s two-sample t-test, n=10 each, 11.01 ± 0.85 ms vs. 17.05 ± 2.41 ms for control vs CsMeSO4 bath application for L4-Type II, p=0.03, t(18)=-2.36, Student’s two-sample t-test, n=10 each). f. Pathway-specific Ih effect on unitary EPSP voltage integral (9.35 ± 2.54 mV*ms vs. 12.9 ± 2.04 mV*ms for control vs CsMeSO4 bath application in L1, p=0.3, t(19)=-1.07, Student’s two-sample t-test, n=11 and n=10 respectively, 7.85 ± 1.24 mV*ms vs. 17.39 ± 2.5 mV*ms for control vs CsMeSO4 bath application for L4-Type I inputs, p=0.002, t(20)=-3.6, Student’s two-sample t-test, n=12 and n=10 respectively, 28.01 ± 3.45 mV*ms vs. 44.89 ± 6.42 mV*ms for control vs CsMeSO4 bath application for L4-Type II inputs, p=0.03, t(18)=-2.31, Student’s two-sample t-test, n=10 each). g. Representative recordings of 50 Hz repeated extracellular stimuli (L1 – grey, L4-Type I – blue, putative L4-Type II -green). h. Temporal summation of L4-Type I input stimulation in control conditions (black) and during CsMeSO4 application (red). i. Pathway-specific Ih modulation of synaptic summation (81.05 ± 14.5 mV*ms vs. 71.07 ± 11.19 mV*ms for control vs CsMeSO4 bath application in L1, p=0.6, t(18)=0.54, Student’s two-sample t-test, n=10 each, 60.06 ± 9.03 mV*ms vs. 95.15 ± 11.29 mV*ms for control vs CsMeSO4 bath application for L4-Type I, p=0.03, t(18)=-2.42, Student’s two-sample t-test, n=10 each, 182.08 ± 23.36 mV*ms vs. 182.05 ± 28.59 mV*ms for control vs CsMeSO4 bath application for L4-Type II, p=0.99, t(18)=8.74*10-4, Student’s two-sample t-test, n=10 each). j. Schematic illustration of the experimental setup. Mice were injected with AAV.CAMKII.C1V1/eYFP either in the latermodial visual area (LM, to label axons arriving to L1) or in the LGN (to label proximal targeting axons). k. Example trace showing optogenetic activation of LM (top) and LGN (bottom) axons in control conditions (black) and upon Cs+ application (red). Traces are normalized. Blue line marks the timing of the optogenetic stimulus. l. Cs+ effect on optogenetically elicited EPSPs from cortico-cortical (LM, left; 35.06 ± 3.06 vs. 31.04 ± 3.82 for control vs Cs+ respectively, p=0.99, t(7)=0.001, Student’s paired-sample t-test, n=8) and thalamic (LGN, right; 29.31 ± 3.68 vs. 40.11 ± 3.95 for control vs Cs+ respectively, p=0.01, t(10)=- 2.95, Student’s paired-sample t-test, n=11) sources.
Pathway-specific NMDA receptor boosting of synaptic information
a. Schematic illustration of intracellular pipette solution perfusion during whole-cell patch clamp recordings. Briefly, the pipette was filled with control recording solution (black trace), which was exchanged to an intracellular solution containing 50 µM MK-801 (purple trace) during the recording. b. Mean peak amplitude of sEPSPs measured at two different membrane potentials and two different recording conditions (ctrl -80 mV vs. ctrl prediction: p=4.72*10-6, t(10) = 8.86, ctrl -80 mV vs ctrl -55 mV: p=0.02, t(10) = -2.81, ctrl prediction vs ctrl -55 mV: p=1.63*10-5, t(10) = -7.71, MK-801 -80 mV vs. MK-801 prediction: p=1.15*10-6, t(10) = 10.36, MK-801 -80 mV vs MK-801 -55 mV: p=0.58, t(10) = 0.57, MK-801 prediction vs MK-801 -55 mV: p=2.69*10-3, t(10) = -3.95, Student’s paired-sample t-test, n = 11 for each). c. Synaptic amplification estimated by dividing the predicted peak amplitude from the hyperpolarized recordings and the measured peak at depolarized membrane potentials (1.69 ± 0.09 and 1.43 ± 0.1 for ctrl and MK-801 amplification respectively, p=0.01, t(10) = 3.04, Student’s paired-sample t-test, n=11). d. Schematic illustration of the multiparameter NEURON fitting procedure for dendritic NMDA. Conductance localization, dispersions, and conductance densities of NMDA were varied along a somato-dendritic axis as set by a Gaussian curve, which was shifted in space, in peak, and in width (top panel). Simulations were carried out using fully reconstructed L2/3 PC from the Allen Institute open-source database (bottom panel). e. Inserting NMDA receptors into proximally located simulated synapses recreated the experimentally determined distance-halfwidth relationship. Notice the best fit of the control conditions (where proximal/distal EPSP halfwidth is similar), is characterized by a Gaussian curve with proximal peak center with a moderate width (dispersion). f. Best fit of the distance-dependent EPSP halfwidth with no HCN in the system. Red asterisk denotes the best fit parameter-space on panel. Proximally (gray) and distally (black) originating EPSPs from the model are shown in the inset. e. g. Amplitude-normalized EPSPs recorded from L2/3 PCs by minimal stimulation in either in L1 (dark color tones) or L4 (light color tones) in different recording conditions (ctrl: black, extracellular cesium: green, extracellular cesium + intracellular MK-801: red, intracellular MK-801: purple). Cells were recorded for each pharmacological condition independently with L4 and L1 stimulation occurring in each recording. h. As predicted (panel f) proximal inputs from L4-originating EPSPs were faster than those from distal L1 synapses in control conditions. The broadening effect of cesium on L4 (proximal) originating EPSPs is occluded by NMDA receptor block (14.25 ± 1.17 ms and 10.11 ± 0.56 ms for L1 vs L4 stimuli in control conditions, p=6.73*10-3, t(9)=3.5, Student’s paired-sample t-test, n=10 each, 16.22 ± 1.91 ms and 23.61 ± 4.04 ms for L1 vs L4 stimuli during Cs bath application, p=0.03, t(9)=-2.67, Student’s paired-sample t-test, n=10 each, 19.92 ± 2.65 ms and 11.8 ± 1.55 ms for L1 vs L4 stimuli during Cs bath application and MK-801 intracellular application, p=0.02, t(7)=3.01, Student’s paired-sample t-test, n=8 each, 13.62 ± 1.43 ms and 8.97 ± 0.72 ms for L1 vs L4 stimuli during MK-801 intracellular application, p=0.025, t(7)=2.84, Student’s paired-sample t-test, n=8 each).
Proximal bias in HCN channel expression dampens proximal NMDA receptor-dependent synaptic boosting in a NEURON model simulation.
a. Schematic illustration of the multiparameter NEURON fitting procedure for dendritic HCN. Simulations here included the NMDA dendritic parameters established in Fig. 4. HCN conductance localization, dispersions, and conductance densities were varied along the somato-dendritic axis as set by a Gaussian curve, which was shifted in space, in peak, and in width as well (top panel). Simulations were carried out using fully reconstructed L2/3 PC from the Allen Institute open-source database (bottom panel). b. A proximal abundance bias of HCN channels recapitulates experimental findings in control conditions. Best fit in the parameter space tested is denoted with red asterisk, where a high density of HCN channels were proximally localized. c. Model-derived data showing the effect of HCN blockade in the best-fit scheme when NMDA is also included as shown in Fig. 4. Note that the effect of HCN removal in this model closely recapitulates our experimental findings using ZD-7288 earlier.
Developmental regulation of Ih expression
a. Representative firing patterns recorded from one week old (left, blue), two weeks old (middle, grey) and six weeks old L2/3 PCs. b. Developmental regulation of sag ratio (0.0 ± 0.0, 0.12 ± 0.01 and 0.07 ± 0.01 for one- (n=8), two- (n=5) and six week old L2/3 PCs (n=6), respectively, p=0 for one week old vs two weeks old cells, p=0.01, t(9)=3.04 for two vs six weeks old cells and p=0 for one vs six weeks old cells, Student’s two-sample t-test), input resistance (450.6 ± 57.91 MΩ, 175.66 ± 13.66 MΩ and 98.34 ± 18.86 MΩ for one- (n=8), two- (n=5) and six week old L2/3 PCs (n=7), respectively, p=0.004, t(11)=3.65 for one week old vs two weeks old cells, p=0.01, t(10)=3.06 for two vs six weeks old cells and p=1.11*10-4, t(13)=5.54 for one vs six weeks old cells, Student’s two-sample t-test) and resting membrane potential (-56.05 ± 2.33 mV, -73.13 ± 2.55 mV and -79.47 ± 1.33 MΩ for one- (n=8), two- (n=5) and six week old L2/3 PCs (n=7), respectively, p=5.9*10-4, t(11)=4.76 for one week old vs two weeks old cells, p=0.04, t(10)=2.39 for two vs six weeks old cells and p=1.36*10-6, t(13)=8.37 for one vs six weeks old cells, Student’s two-sample t-test). c. Representative voltage clamp recordings of hyperpolarization activated currents in one week old (left, blue), two weeks old (middle, gray) and six weeks old L2/3 PCs (right, black). d. Voltage dependence of hyperpolarization activated currents in one week old (blue), two weeks old (gray) and six weeks old L2/3 PCs (black). e. Negative correlation between measured hyperpolarization activated current and resting membrane potential during development (Linear fit (red) slope: -2.84 mV/nS, intercept: -57.67 mV, R2=0.71, n=20). f. Two- photon microscopy image depicting a proximal and a distal extracellular stimulating location positioned closely to an Alexa-594 filled L2/3 PC dendrite (left), and the resulting EPSPs (proximal – black, distal – green). g. EPSP halfwidth decreases with dendritic distance (Linear fit (red) slope: -0.42 ms/µm, intercept: 161.01 ms, R2=0.4, n=35). h. EPSPs were significantly slower in proximal dendritic locations (147.07 ± 12.87 ms and 85.57 ± 6.83 ms for proximal and distal events, respectively, p=5.31*10-5, t(33)=4.64, Student’s two-sample t-test, n=13 and 22, respectively).
Neuromodulation shifts L2/3 PC excitability via HCN channels
a. 5HT7 receptor mRNA is abundantly expressed in L2/3 PCs. Data is collected from the online available dataset(29). b. Example cell attached recording of a L2/3 PC (left) upon 50 Hz extracellular stimulation in L4 in control conditions (black) and 5-CT bath application (orange). 5-CT significantly reduced L2/3 PC spiking (0.2 ± 0.04 spike/stimuli vs. 0.1 ± 0.04 spike/stimuli for control vs 5-CT application, p=3.57*10-6, t(11)=7.92, Student’s paired-sample t-test, n=12 each) c. Example current clamp recording of voltage responses to hyperpolarizing current injections (left; control – black, 5-CT – orange). Steady-state voltage response rectification is amplified by 5HT7R blockade (right). d. 5-CT shifts the voltage dependence of hyperpolarization activated currents (-109.06 ± 0.47 mV vs -102.35 ± 1.09 mV half activation for control vs 5-CT bath application, R2 = 0.99 for both). e. 5-CT did not alter resting membrane potential (-82.41 ± 1.43 mV vs -80.14 ± 1.38 mV in control vs 5-CT bath application, respectively, p=0.89, t(5)=-1.41, Student’s paired-sample t-test, n=6 each), sag ratio at -2 pA/pF current injection (0.02 ± 2.55*10-3 vs 0.03 ± 0.01 in control vs 5-CT bath application, respectively, p=0.82, t(5)=-1.01, Student’s paired-sample t-test, n=6 each) and action potential threshold (-41.13 ± 2.6 mV, -40.08 ± 3.05 mV in control vs 5-CT bath application, respectively, p=0.64, t(4)=-0.4, Student’s paired-sample t-test, n=5 each), but significantly reduced input resistance (108.21± 10.3 mΩ, 92.29 ± 9.9 MΩ in control vs 5-CT bath application, respectively, p=0.02, t(5)=2.75, Student’s paired-sample t-test, n=6 each). f. Example recording showing no effect of 5-CT bath application when HCN channels were blocked (left: example recordings, right: 0.3 ± 0.11 spike/stimuli vs 0.34 ± 0.12 spike/stimuli in control vs 5-CT bath application, respectively, p=0.13, t(5)=-1.81, Student’s paired-sample t-test, n=6 each). g. NMDA independent 5-CT effect on L2/3 PC excitability (0.27 ± 0.03 spike/stimuli vs 0.14 ± 0.04 spike/stimuli in control vs 5-CT bath application, respectively, p=0.01, t(7)=3.66, Student’s paired-sample t-test, n=7 each).
ZD-7288 application results in similar active and passive electrical alterations.
a. Representative effect of ZD-7288 on voltage responses to hyperpolarizing current injections. b. ZD-7288 abolished sag voltage (0.42 ± 0.07 mV vs 0.1 ± 0.04 mV in control vs ZD-7288 application, respectively, p=4.5*10-4, t(15)=4.47, Student’s paired-sample t-test, n=16), decreased the resting membrane potential (-75.86 ± 0.84 mV vs -79.73 ± 0.86 mV in control vs ZD-7288 application, respectively, p=4.41*10-4, t(15)=4.48, Student’s paired-sample t-test, n=16), and increased the input resistance (92.74 ± 6.59 MΩ vs 140.8 ± 13.07 MΩ in control vs ZD-7288 application, respectively, p=1.79*10-4, t(15)=-4.93, Student’s paired-sample t-test, n=16).
Similar Ih current and effect in different cortical areas.
a. Steady-state rectification of L2/3 PCs in the primary somatosensory cortex (S1), primary motor cortex (M1), primary auditory cortex (A1) and primary visual cortex (V1). b. Similar resting membrane potential (left), sag ratio (middle) and input resistance (right) of L2/3 PCs in different cortical areas. c. Example hyperpolarization activated current recordings from four cortical areas (left), with similar voltage dependent activation (right).
L4-Type I and L4-Type II inputs are well separated by k-means clustering
k-means clustering results plotted along the first and second principal component. Ellipses denote the confidence interval; edges are connected to the cluster center. Hollow circles represent L4 Type II inputs and filled circles represent L4-Type I inputs determined by the experimenter.
The presence of dendritic sodium and potassium channels cannot explain wider proximal synaptic inputs
a. Sodium channels cannot explain the distance-EPSP halfwidth relationship observed when Ih was blocked. Color coding denotes the ratio between proximal and distal EPSP halfwidths. Notice the miniscule range of the color coding. C. Same as panel b. but for voltage dependent potassium channels.
Pipette perfusion does not alter spontaneous circuit activity measurements
a. sEPSP frequency is unaltered by pipette perfusion (6.63 ± 0.3 and 6.4 ± 0.44 for ctrl -80 mV and perfused ctrl -80 mV, p=0.57, t(6) = 0.59, 6.1 ± 0.25 and 6.03 ± 0.33 for ctrl -55 mV and perfused ctrl -55 mV, p=0.83, t(6) = 0.23, n=7 for each, Student’s paired-sample t-test). b. sEPSP peak amplitudes are stable during pipette perfusion experiments (0.23 ± 0.01 and 0.25 ± 0.01 for ctrl -80 mV and perfused ctrl -80 mV, p=0.43, t(6) = -0.83, 0.16 ± 0.01 and 0.17 ± 0.01 for ctrl prediction and perfused prediction, p=0.43, t(6) =-0.83, 0.31 ± 0.03 and 0.34 ± 0.02 for ctrl -55 mV and perfused ctrl -55 mV, p=0.24, t(6) = -1.31, n=7 each, Student’s paired-sample t-test). c. Pipette perfusion does not alter measured synaptic amplification (1.92 ± 0.07 and 2.01 ± 0.24 for ctrl and perfused ctrl, p=0.65, t(6) = -0.48, n=7 each, Student’s paired-sample t-test).
Elicited EPSP halfwidth in not dependent on event amplitude
a. EPSPs elicited by electrical stimulation show no correlation between event amplitude and halfwidth (R2= 0.021, n=53). b. EPSP peak is not dependent on dendritic distance from the soma (R2= 0.021, n=53).
