Decreased excitability of dentate mossy cells in Fmr1 KO mice

(A) Schematic illustration of dentate circuit organization and recordings of APs from mossy cells (MC). Note, for simplicity and clarity, we only show one type of interneurons (IN) with axons terminating onto distal dendrites of a granule cell (GC). The arrow indicates axons of granule cells projecting to CA3. ML, molecular layer; GCL, granule cell layer.

(B) Sample traces of spontaneous APs recorded at different membrane potentials from mossy cells in WT (black) and Fmr1 KO (red) mice.

(C) Summary data for experiments exemplified in (A) showing decreased number of APs at membrane potentials of -61 through -55 mV in KO mossy cells. Scatter circles indicate individual data points for this and all subsequent bar graphs in the present study.

(D) Determination of AP threshold and rheobase by a ramp current injection (lower trace, ramp rate 0.15 pA/ms). Only the first APs (arrows, which were expanded and aligned by the time of threshold in insert) were used to estimate AP parameters. The horizontal lines (insert) indicate threshold of the 1st APs. In the lower panel, RbWT and RbKO denote rheobase current intensity at threshold time-point, and the area (integrating time and input current) enclosed by dotted line, current ramp and RbWT (or RbKO) are rheobase charge transfer.

(D) Phase plots of ramp current-evoked AP traces from (C). Arrows showing the threshold of 1st APs.

(E-H) Summary data showing decreased number of APs during 2-s ramp (E), increased voltage threshold (F), rheobase (G) and rheobase charge transfer (H) in KO mossy cells.

(I-N) AP upstroke maximum rise rate (I), rise time (J), fall time (K), duration (L), peak potential (M) and amplitude (N).

*p < 0.05; **p < 0.01; ns, not significant. The statistical data are listed in Supplementary file 1– Supplementary Table 1. Data are mean ± SEM.

Decreased input resistance around threshold potential in Fmr1 KO mossy cells

(A and B) Resting membrane potential (RMP, A) and membrane capacitance (B) of mossy cells.

(C) Input resistance measured at RMP level.

(D) Input resistance measured at -45 mV. Sample traces of the depolarization current step (lowermost panel) induced voltage responses before (Basal, upper panel) and during XE991 (+ XE991, middle panel).

(E) Summary data of input resistance before (basal) and during XE991.

(F) Effects of XE991 on increasing of input resistance. Note XE991 have stronger effect on KO mossy cells.

*p < 0.05; ns, not significant. The statistical data are listed in Supplementary file 1– Supplementary Table 1. Data are mean ± SEM.

Enhanced Kv7 function causes hypo-excitability of Fmr1 KO mossy cells

(A) Changes in holding current at -45 mV in response to XE991. Left, sample traces; Right, summary data.

(B) Changes in membrane potential in response to XE991 when the initial potential being set at -45 mV. Left, sample traces; Right, summary data.

(C) Kv7 current was induced by a ramp protocol (from -95 to +5 mV with a rate of 0.02 mV/ms) and determined by XE991 sensitivity. Inserts, the enlargements of boxed areas in main traces.

(D) The I-V curves were constructed from ramp-evoked Kv7 currents every 5 mV (quasi-steady-state current, averages over 0.01 mV intervals) and normalized to respective cell capacitances. Insert, Kv7 current at -45 mV.

(E-H) Increased threshold (E), rheobase (F) and rheobase charge transfer (H), as well as decreased number of APs (H) in pharmacologically isolated KO MCs. XE991 abolished these differences between genotypes.

*p < 0.05; **p < 0.01; ns, not significant. The statistical data are listed in Supplementary file 1– Supplementary Table 1. Data are mean ± SEM.

Increased E/I ratio of inputs onto Fmr1 KO mossy cells

(A) Distribution of sEPSC instantaneous frequency in MCs. A bin size of 2 Hz was used to calculate sEPSC frequency distribution from a 30-s-long trace per cell. The number of sEPSCs within each bin was normalized to the total number of the respective cells for pooling the data from all cells. Note that sEPSC events in KO mice had a shift toward high frequency. Insert, sample traces of sEPSCs for WT (black) and KO (red) mice.

(B) Cumulative probability of sEPSC instantaneous frequency in MCs. Bar graph, number of sEPSCs per minutes. Note that both cumulative probability and number of sEPSCs reveal increased excitatory drive onto MCs.

(C and D) sIPSCs recorded from KO and WT MCs, aligned in the same way as in (A and B), respectively. The IPSC signals were down-ward here and also in Figures 5, Figure 4—figure supplement 1 and Figure 5—figure supplement 1, due to a high chloride electrode solution was used in these experiments. Note the increased sIPSC frequency and number in KO MCs.

(E and F) Summary data for sEPSC amplitude (E) and sIPSC amplitude (F) recorded from MCs. Inserts, sample sEPSC (E) and sIPSC (F) events for WT (black) and KO (red) MCs. Scale: 5 ms (horizontal) and 25 pA (vertical).

(G and H) Mean frequency of sEPSCs (G) and sIPSCs (H) recorded from MCs. Note that loss of FMRP increased mean frequency of both sEPSC and sIPSC.

(I) E/I ratio evaluated by sEPSC and sIPSC frequencies (mean values from G and H, respectively). Note the increased E/I ratio in Fmr1 KO mice.

*p < 0.05; **p < 0.01; ns, not significant. The statistical data are listed in Supplementary file 1– Supplementary Table 1. Data are mean ± SEM.

Decreased E/I ratio of inputs onto Fmr1 KO hilar interneurons

(A) Distribution of sEPSC instantaneous frequency in hilar interneurons. Note that KO sEPSC events had a shift toward low frequency. Insert, sample traces of sEPSCs for WT (black) and KO (red) interneurons.

(B) Cumulative probability of sEPSC instantaneous frequency in interneurons. Bar graph shows number of sEPSCs per minutes. Note that both cumulative probability and number of sEPSCs reveal decreased excitatory drive onto interneurons.

(C and D) sIPSCs recorded from interneurons, aligned in the same way as in (A and B), respectively. Note the decreased sIPSC frequency and number in KO interneurons.

(E and F) Summary data for sEPSC amplitude (E) and sIPSC amplitude (F) recorded from interneurons. Inserts, sample sEPSC (E) and sIPSC (F) events for WT (black) and KO (red) MCs. Scale: 5 ms (horizontal) and 25 pA (vertical).

(G and H) Mean frequency of sEPSCs (G) and sIPSCs (H) recorded from interneurons. Note loss of FMRP decreased mean frequency of both sEPSC and sIPSC.

I) E/I ratio evaluated by sEPSC and sIPSC frequencies in interneurons (mean values from G and H, respectively). Note the decreased E/I ratio in Fmr1 KO mice.

*p < 0.05; **p < 0.01; ns, not significant. The statistical data are listed in Supplementary file 1– Supplementary Table 1. Data are mean ± SEM.

Mossy cells provide the main excitatory drive onto hilar interneurons

(A) Sample traces of sEPSCs recorded from an interneuron of WT mouse before and during DCG-IV.

(B) The same as in (A), but for WIN55212-2 in an interneuron of WT mouse.

(C) Effect of DCG-IV or WIN55212-2 on the sEPSC normalized frequency recorded from interneurons of WT mice.

(D-F) The same as in (A-C) but for interneurons of KO mice. Note that WIN55212-2 had comparable effects on normalized frequency of sEPSCs in KO and WT interneurons, but DCG-IV did not have measurable effects on both genotypes.

(G) Control experiment showing effectiveness of DCG-IV. Sample traces of sEPSCs recorded from MCs before and during DCG-IV.

(H) Sample traces of sEPSCs recorded from MCs before and during WIN 55212-2.

(I) Summary data of changes in the normalized frequency of MC sEPSCs in response to DCG-IV (47.6 ± 7.2% of basal) and WIN55212-2 (114.5 ± 6.7% of basal). Note that, compared to interneurons (A-F), MCs exhibited opposite response to these two agonists, indicating the effectiveness of both agonists.

**p < 0.01; ns, not significant. The statistical data are listed in Supplementary file 1– Supplementary Table 1. Data are mean ± SEM.

Circuit-wide inhibition of Kv7 channels boosted inhibitory drive onto granule cells and corrected granule cell excitability in Fmr1 KO mice

(A) Sample traces of simultaneous recording of sEPSC (down-ward events) and sIPSC (up-ward events) from granule cells (also see Figure 7—figure supplement 2A).

(B) Summary data for sEPSC mean frequency in basal (upper) and during XE991 (lower). Horizontal lines (with or without dropdown) denote comparison between genotypes; vertical lines indicate comparison between before and during XE991 within genotypes

(C) The same as in (B), but for sIPSC simultaneously recorded from the same granule cells. Note that XE991 increase sIPSC frequency only in KO mice, but not in WT mice.

(D) E/I ratio evaluated by frequency. XE991 (lower) significantly decreased the E/I ratio in KO mice only.

(E) Evaluation of granule cell excitability by recording APs. Sample traces for multistep-current (Lowermost panel) evoked APs in WT and KO granule cells in the pharmacologically isolated granule cells in the absence (upper) or presence of XE991 (lower).

(F) Summary data for number of APs exemplified in (E) showing increased excitability of the pharmacologically isolated granule cells in the absence (upper) or presence of XE991 (lower). Also, note that XE991 increased number of AP in both WT and KO granule cells. Horizontal lines (with dropdown) denote comparison between genotypes; vertical lines indicate comparison between in the absence and presence of XE991 within genotypes.

(G) The same as in (E), but for granule cells with intact dentate circuit.

(H) The same as in (F), but for granule cells with intact circuit. Note that XE991 did not increase number of AP in KO granule cells; rather it increased number of AP only in WT granule cells.

*p < 0.05; **p < 0.01; ns, not significant. The statistical data are listed in Supplementary file 1– Supplementary Table 1. Data are mean ± SEM.

Circuit-wide inhibition of Kv7 channels restored dentate output during theta-gamma coupling stimulation in Fmr1 KO mice

(A) Sample traces of PP-stimulation evoked compound postsynaptic currents (cPSC) and their respective underlying EPSC and IPSC, in the basal state. For better comparison, the traces were normalized to their own underlying EPSC, which reflects stimulation intensity. Red vertical lines denote the excitation window that was summarized in (H). Stimulation artifacts were removed and baseline before stimulation was shifted to be 0 for presentation purpose.

(B) The same as in (A), but in the presence of XE991.

(C-E) Summary data of normalized excitatory component (C), inhibitory component (D) and underlying IPSC (E), in the absence (basal, upper panels) and in the presence of XE991 (+ XE991, lower panels).

(F) Summary data of E/I ratio evaluated by the peaks of underlying EPSC and IPSC in the absence (basal, upper panel) and in the presence of XE991 (+ XE991, lower panel).

(G) The same as in (F), but evaluated by the charge transfers of underlying EPSC and IPSC.

(H) Summary data of excitation window in the absence (basal, upper panel) and in the presence of XE991 (+ XE991, lower panel).

(I) Sample traces of theta-gamma coupling stimulation-evoked APs in granule cells from WT mice, in the absence of (basal, upper panel) or in the presence of XE991 (+XE991, middle panel). Lower panel showing stimulation protocols: Control, 15 stimuli at 5 Hz; Test, a burst of gamma stimulation (5 stimuli at 50 Hz, arrow) 200 ms before 15 stimuli at 5 Hz. AP probability in test train was calculated in 1-second-bin (ie, binned in 1st, 2nd or 3rd second) and plotted in (K).

(J) The same as in (I), but for KO mice.

(K) Summary data of gamma-suppression of dentate output in response to PP stimulation at theta frequency in the absence of (basal, upper) or in the presence of XE991 (+XE991, lower). Note loss of FMRP compromised gamma-suppression of AP output in granule cells, and Kv7 blocker XE991 restored the gamma burst-induced suppressive effect on dentate output in Fmr1 KO mice.

*p < 0.05; **p < 0.01; ns, not significant. Horizontal lines (with or without dropdown) denote comparison between genotypes; vertical lines indicate comparison between in the absence and presence of XE991 within genotypes. The statistical data are listed in Supplementary file 1– Supplementary Table 1. Data are mean ± SEM.

No changes in KCNQ2 and KCNQ3 expression in Fmr1 KO mice

a-b Western blot analysis of KCNQ2 (a) and KCNQ3 (b) in the whole brain lysate. Ponceau staining was used as a loading control for the lysate (lower panel).

c-d The same as (a-b), but for dentate gyrus lysate. ns, not significant. The statistical data are listed in Supplementary file 1–Supplementary Table 1. Data are mean ± SEM.

Changes in miniature synaptic inputs onto mossy cells in Fmr1 KO mice

(A) Distribution of mEPSC instantaneous frequency in MCs. A bin size of 2 Hz was used to calculate mEPSC frequency distribution from a 30-s-long trace per cell. The number of mEPSCs within each bin was normalized to the total number of the respective cells for pooling the data from all cells. Note that mEPSC events in KO mice had a shift toward high frequency. Insert, sample traces of mEPSCs for WT (black) and KO (red) mice.

(B) Cumulative probability of mEPSC instantaneous frequency in MCs. bar graph, number of mEPSCs per minutes. Note that both cumulative probability and number of mEPSCs reveal increased excitatory drive onto MCs.

(C) Summary data for mEPSC amplitude. Insert, sample mEPSC events for WT (black) and KO (red) MCs. Scale: 5 ms (horizontal) and 25 pA (vertical).

(D-F) mIPSCs recorded from KO and WT MCs, aligned in the same way as in (A-C), respectively.

*p < 0.05; **p < 0.01. The statistical data are listed in Supplementary file 1–Supplementary Table 1. Data are mean ± SEM.

Changes in miniature synaptic inputs onto hilar interneurons in Fmr1 KO mice

(A) Sample traces (Left) of 3 types of interneurons classified by firing pattern. Upper trace, fast-spiking interneurons with high-frequency non-adapting firing (WT 20/37 and KO 18/29, where numerator denotes the number of cells for a given type and denominator is the total number of tested cells for a given genotype); Middle trace, regular-spiking interneurons with slower and adapting firing (WT 4/37, KO 3/29); Lower trace, stuttering-like interneurons with high-frequency irregular bursting firing (WT 13/37, KO 8/29). Stack bar graph (Right) shows the percentage of 3 interneuron types (color-coded as the traces in the Left panel). No significant differences in the ratios of 3 interneuron types between WT and KO mice.

(B-E) We pooled RMP (B), capacitance (C), input resistance (D) and threshold (E) from 3 types of interneurons due to no significant differences observed among types.

(F) Distribution of mEPSC instantaneous frequency in interneurons. A bin size of 2 Hz was used to calculate mEPSC frequency distribution from a 30-s-long trace per cell. The number of mPSCs within each bin was normalized to the total number of the respective cells for pooling the data from all cells. Note that mEPSC events in KO mice had a shift toward low frequency. Insert, sample traces of mEPSCs for WT (black) and KO (red) mice.

(G) Cumulative probability of mEPSC instantaneous frequency in interneurons. Bar graph shows number of mEPSCs per minutes. Note that both cumulative probability and number of mEPSCs reveal decreased excitatory drive onto interneurons.

(H) Summary data for mEPSC amplitude. Insert, sample mEPSC events for WT (black) and KO (red) MCs. Scale: 5 ms (horizontal) and 25 pA (vertical).

(I-K) mIPSCs recorded from KO and WT interneurons, aligned in the same way as in (F-H), respectively.

*p < 0.05; **p < 0.01; ns, not significant. The statistical data are listed in Supplementary file 1– Supplementary Table 1. Data are mean ± SEM.

Effect of XE991 on spontaneous synaptic inputs onto granule cells

(A) Example traces for simultaneous recording of sEPSC and sIPSC from granule cells (holding at -40 mV). In the absence of blockers against glutamate and GABA ionotropic receptors, both sEPSC (down-ward) and sIPSC (up-ward) were exhibited (upper trace). In the presence of APV and DNQX (blockade of NMDA and AMPA receptors), only up-ward sIPSC was kept (middle trace). In contrast, in the presence of gabazine (blockade of GABAA receptors), only down-ward sEPSC was observed (lower trace).

(B and C) Amplitudes of simultaneously recorded sEPSC (B) and sIPSC (C) before (Basal, upper bars) and during XE991 (+ XE991, lower bars).

(D) E/I ratio evaluated by amplitude before (Basal, upper bars) and during XE991 (+ XE991, lower bars).

(E and F) The same as (B and C), but for charge transfers of simultaneously recorded sEPSC (E) and sIPSC (F).

(G) The same as (D), but evaluated by charge transfer.

*p < 0.05; **p < 0.01; ns, not significant. Horizontal lines denote comparison between genotypes; vertical lines indicate comparison between before and during XE991 within genotypes. The statistical data are listed in Supplementary file 1–Supplementary Table 1. Data are mean ± SEM.

Estimation of direct and circuit effects of XE991 on granule cell excitability

(A) XE991 can directly increase excitability of granule cells, leading to increase AP firing. The direct effect of XE991 on AP firing were defined as the differences between before and during XE991 in the isolated granule cells (Figure 7F), ie, values of (7Flower – 7Fupper), where 7Flowerand 7Fupper are the mean values from Figure 7F lower and upper panels, respectively. Note that XE991 directly increases number of AP largely independent of step current intensity in both genotypes, but with relative larger effect in KO mice. GCs, granule cells.

(B) XE991 can also modulate granule cell excitability to change AP firing via its circuit effect. Values were estimated from the differences before and during XE991, between isolated granule cells (Figure 7F) and granule cells with intact circuit (Figure 7H), ie, values of [(7Hlower – 7Hupper) - (7Flower – 7Fupper)], where 7F and 7H are the mean values from Figures 7F and 7H (lower or upper panel, accordingly). Note that the circuit effect of XE991 on AP firing was also independent of step current intensity (little effect in WT mice as shown by values fluctuating around 0; but dampening ∼6 APs in KO mice).

Isolation of underlying EPSC and IPSC from compound PSC

(A and B) Compound postsynaptic current (cPSC) was recorded from granule cells by holding membrane potential at -45 mV. The cPSC (A) shows an initial excitatory component (down-ward) and followed by an inhibitory component (up-ward). Dotted line denotes the baseline before stimulation. At the end of each recording, the pure EPSC (B) was recorded by keeping the same stimulation intensity and in the presence of gabazine, which was used to create EPSC template (an average of at least 20 uncontaminated EPSCs) for the same cell. Note that the amplitude and decay trajectory of pure EPSC was largely different from excitatory component of cPSC. Stimulation artifacts were removed. Boxed area in (A) was enlarged to show the procedure of approximation of underlying EPSC.

(C) Zoom-in of boxed area in (A) showing approximation of underlying EPSC. The EPSC template was repeatedly scaled to each data point (black circles) of cPSC “approximating segment” (AS) to obtain a set of scaled EPSCs (gray traces). All scaled EPSCs were then averaged to approximate an underlying EPSC (orange trace) for a given cPSC (black trace). The cPSC “approximating segment” was defined as 25–65% height of cPSC excitatory component (but not beyond 2.5 ms after stimulation). Excitation window was defined as the full duration of cPSC excitatory component. Stimulation artifact was removed. “S” denotes the stimulation time point.

(D) The underlying IPSC (blue trace) was isolated by subtracting the underlying EPSC (orange trace as that in C) from corresponding cPSC (black trace). For the purpose of view, the baseline before stimulation was shifted to be 0 in the figure. The two vertical red lines delimit the excitation window.

(E) Normalized traces of cPSC, underlying EPSC and IPSC. For better comparison, we normalized the cPSC and underlying IPSC to their respective underlying EPSC, which reflects the PP stimulation intensity. The underlying EPSC and IPSC were then used to estimate E/I ratio by their peak amplitudes or charge transfers (amplitude-time integration within 100 ms). The letters c, d and e indicate the normalized peak amplitudes of cPSC excitatory, inhibitory components and underlying IPSC, which then summarized in Figures 8C, 8D and 8E, respectively.

Circuit-wide inhibition of Kv7 channels enhanced the gamma burst-induced suppression of EPSP integration in Fmr1 KO mice

(A) Sample Traces of EPSP in response to theta-gamma coupling stimulation of PP. Traces were zoomed in vertically (ie, amplitude dimension) and APs were truncated (indicated by double-slash) to emphasize EPSP size. EPSC amplitude was defined by the voltage difference between EPSC peak and -70 mV level (cyan line). Lowermost panel shows the stimulation paradigm. Up-pointing arrow indicates gamma-stimulation.

(B) Summary data for experiments exemplified in (A) showing stable EPSP amplitude in control stimulation (left panel) and gamma-suppression of EPSP amplitude (right panel). Insert, real EPSP amplitude in control stimulation.

(C) The same as in (A), but in the presence of XE991.

(D) The same as in (B), but in the presence of XE991. Note that XE991 significantly dampened EPSP amplitude in KO mice (right panel) and the EPSP were largely comparable between genotypes.

*p < 0.05; **p < 0.01; ns, not significant. The statistical data are listed in Supplementary file 1– Supplementary Table 1. Data are mean ± SEM.

Circuit-wide inhibition of Kv7 channels corrected dentate gyrus output in Fmr1 KO mice

(A) The diagram shows dentate gyrus output in response to theta-gamma coupling stimulation of PP in the WT mice. Here, we focus on the three-synapse indirect feedback inhibition pathway: Granule cells (GCs)➔MCs-+Interneurons (INs)-+GCs (thickness of arrows indicating synaptic drive weight). Because MCs (rather than granule cells) provide the main source of excitatory drive onto interneurons, the three-synapse indirect feedback inhibition likely surpasses the canonical two-synapse feedback inhibition (ie, GCs➔INs➔GCs). This model indicates that, in the normal condition, MCs integrate the incoming excitatory input and then relay and expand it to interneurons, which secure a sufficient and well-timed inhibition feedback onto granule cells to maintain a dynamically balanced E/I inputs onto granule cells and a narrow excitation window, and thus ensures sparse AP firing in granule cells. Red arrows designate excitation and blue arrows inhibition. The arrow entering shading areas from left side represents PP input (theta-gamma coupling stimulation). The downward arrow exiting the shading area indicates dentate output to CA3. The thinly semi-transparent arrows are synaptic connections whose contribution to cellular excitability has not been evaluated in the present study. The thickness of arrows denotes the synaptic drive weight of excitation (or inhibition). It is noteworthy that the actual synaptic drive weight of each synapse varies dynamically to maintain the precisely well-timed circuit E/I balance and then proper information processing. Thus, one may imagine that the thickness of each arrow (ie, synaptic drive weight) changes dynamically and sequentially (due to neurotransmission direction and synaptic delays) to understand dynamic E/I balance.

(B) The same as in (a), but for KO mice. Loss of FMRP caused these changes (for details, see Figures 1, 4, 5, 6 and 8): 1) the granule cells are hyperexcitable (highlighted by yellow explosion marker), and dentate output are increased (thicker arrow compared with that of WT mice in A); 2) MCs are hypo-excitable (shadow font) and MCs output are decreased (thinner arrow); 3) both excitatory and inhibitory inputs onto MCs are increased (thicker arrows); and 4) both excitatory and inhibitory inputs onto INs are reduced (thinner arrows).

(C) The same as in (B), but in the presence of XE991 (circuit-wide inhibition of Kv7 channels). XE991 caused these changes in the KO mice (for details, see Figure 8 and Figure 8—figure supplement 2): 1) Owing to the abnormal enhanced Kv7 function in KO MCs, XE991 boosted up MC excitability (explosion marker) to enhance excitatory output (thicker red arrow) onto interneurons, which increases interneurons excitability (explosion marker); and 2) XE991 increased inhibitory drive onto granule cells (thicker blue arrow onto granule cells compared with that of in (B), which dampens EPSP integration (due to summation of enhanced inhibitory input) in granule cells and suppresses dentate gyrus output.