Higher hyperpolarization activated current (I_h) in a subpopulation of interneurons in stratum oriens of area CA1 in the hippocampus of Fragile X mice

Abstract

Fragile X syndrome is the most common inherited form of intellectual disability and the leading monogenetic cause of autism. Studies in mouse models of autism spectrum disorders, including the Fmr1 knockout (FX) mouse, suggest that abnormal inhibition in hippocampal circuits contributes to behavioral phenotypes. In FX mice, changes in multiple voltage-gated ion channels occur in excitatory pyramidal neurons of the hippocampus. Whether there are also changes in the intrinsic properties of hippocampal inhibitory interneurons, however, remains largely unknown. We made whole-cell current clamp recordings from both fast-spiking (FS) and low threshold spiking (LTS) interneurons in the stratum oriens region of the hippocampus. We found that LTS, but not FS, interneurons in FX mice had lower input resistance and action potential firing compared to wild type. When we subdivided LTS interneurons into low-threshold high I_h (LTH) and putative oreins-lacunosum moleculare (OLM) cells (Hewitt et al., 2021), we found that it was the LTH subgroup that had significantly lower input resistance in FX mice. The difference in input resistance between wild type and FX LTH interneurons was absent in the presence of the h-channel blocker ZD7288, suggesting a greater contribution of I_h in FX LTH interneurons. Voltage clamp recordings found that indeed, I_h was significantly higher in FX LTH interneurons compared to wild type. Our results suggest that altered inhibition in the hippocampus of FX mice may be due in part to changes in the intrinsic excitability of LTH inhibitory interneurons.

Graphical Abstract

New and Noteworthy

In this paper we use physiological and biochemical approaches to investigate the intrinsic excitability of inhibitory interneurons in hippocampal area CA1 of the Fragile X mouse. We found that higher I_h lowers the intrinsic excitability of one specific type of interneuron. This study highlights how changes to voltage-gated ion channels in specific neuronal populations may contribute to the altered excitatory/inhibitory balance in Fragile X syndrome.

Introduction

Fragile X Syndrome (FXS) is the primary monogenetic cause of autism and the most common form of inherited intellectual disability. Patients with FXS exhibit epilepsy, sensory hypersensitivity, and anxiety among other cognitive impairments (1–4). One working hypothesis is that excitatory/inhibitory (E/I) balance, which is essential for normal brain circuit function, is disrupted in FXS. The hippocampus contains a diverse population of inhibitory interneurons that form distinct circuits that control the precise timing of excitatory activity in the hippocampus (5–8). Given this diversity of interneuron functions, changes in even a single population of interneurons, could skew the balance between excitation and inhibition (9–12).

FXS is characterized by the transcriptional silencing of the Fmr1 gene (13) leading to a loss of Fragile X Messenger Ribonucleoprotein 1 (FMRP) (14) (fmr1 gene ID: 2332, NCBI, 2022). Studies demonstrated altered excitability in the hippocampus of Fmr1 knockout (FX) mice (15–17). FMRP regulates the function and expression of numerous voltage-gated ion channels across both cell types and brain regions (18–20). There are known changes in voltage-gated ion channels in both CA3 and CA1 pyramidal neurons in the hippocampus of FX mice (21–25). It is unknown however, if there are changes in the intrinsic properties of hippocampal inhibitory interneurons in FX mice.

There are many types of inhibitory interneurons in the hippocampus that gate activity flow, influence synaptic plasticity, and exhibit control over rhythmic activity (5, 26–29). Fast spiking (FS) interneurons make up roughly 14% of the CA1 interneuron population and are located near the pyramidal cell layer or in the stratum oriens where many predominantly exhibit strong inhibition on the cell body of pyramidal neurons (30). Fast spiking interneurons in area CA1 typically provide strong feedforward inhibition as they receive excitatory input from many CA3 pyramidal neurons (31, 32). Low threshold spiking interneurons account for approximately 4.5% of total CA1 interneurons and predominantly target the distal dendrites of pyramidal neurons (30). Low threshold spiking interneurons in area CA1 form a canonical feedback inhibition circuit and receive most of their excitatory input from CA1 pyramidal neurons (33, 34). While these neurons make up a small percentage of the interneuron population, proper activity of LTS neurons shapes critical hippocampal circuitry and behavior (7, 28, 35). We recently showed that stratum oriens LTS interneurons in wild type mice can be further divided into two distinct classes based on electrophysiological properties (36). We categorized these as putative oriens-lacunosum moleculare (OLM) interneurons and low threshold, high I_h (LTH) interneurons, nomenclature we adopted for this paper.

We used patch clamp electrophysiology to investigate differences in CA1 stratum oriens interneurons between wild type and FX mice. There were no significant differences in input resistance and action potential firing frequency in FS and OLM interneurons between wild type and FX mice. In contrast, input resistance and action potential firing frequency were significantly lower in LTH interneurons from FX mice compared to wild type. Previous studies showed that changes in the functional expression of HCN channels (I_h) altered the intrinsic excitability of pyramidal neurons in a cell-type specific manner in FX mice (21, 25, 37, 38). Given the relatively higher contribution of I_h in LTH interneurons, we tested the hypothesis that the lower intrinsic excitability of FX LTH interneurons was due to a greater contribution of I_h. In support of this hypothesis, we found that differences in input resistance between wild type and FX LTH interneurons were absent in the presence of the I_h blocker ZD7288. Furthermore, direct measurement of I_h using voltage-clamp recordings found that the amplitude of I_h in FX LTH interneurons was significantly higher in compared to wild type.

Methods

Slice preparation:

All animal procedures were performed in accordance with the Univeristy of Texas at Austin and University of Nevada at Las Vegas animal care committee’s regulations. Mice had free access to food and water and were housed in a reverse light-dark cycle of 12 hr on/ 12 hr off. Experiments used male 2–4-month-old wild type and fmr1 knockout (FX) mice on a C57/Bl6 background (JAX: strain #000664). Mice were anesthetized using a ketamine/xylazine cocktail (100/10 mg/kg) and then underwent cardiac perfusions with ice-cold saline consisting of (in mM): 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 0.5 CaCl₂, 7 MgCl₂, 7 dextrose, 205 sucrose, 1.3 ascorbic acid, and 3 sodium pyruvate (bubbled constantly with 95% O₂/5% CO₂ to maintain pH at ~7.4). The brain was removed and sliced into 300 µM parasagittal sections from the middle hippocampus using a vibrating tissue slicer (Vibratome 300, Vibratome Inc). The dorsal-ventral position of slices was estimated as 0.17±0.07 along the dorsal-ventral axis, where the range is −3 (most ventral) to 3 (most dorsal) using the method described in (39). The slices were placed in a chamber filled with artificial cerebral spinal fluid (aCSF) consisting of (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 2 CaCl₂, 2 MgCl₂, 10 dextrose, 1.3 ascorbic acid and 3 sodium pyruvate (bubbled constantly with 95% O₂/5% CO₂) for 30 minutes at 35°C and then held at room temperature until time of recording.

Electrophysiology

Slices were placed in a submerged, heated (33–34 C°) recording chamber and continually perfused with aCSF (in mM): 125 NaCl, 3 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 2 CaCl₂, 1 MgCl₂, 10 dextrose, and 3 sodium pyruvate (bubbled constantly with 95% O₂/5% CO₂). Ionotropic glutamatergic and GABAergic synaptic transmission were blocked with 20 µM DNQX, 25 µM D-AP5, and 2 µM gabazine. Interneurons within the stratum oriens region of CA1 in slices were visualized with a Zeiss AxioScope or AxioExaminer under 60x magnification.

Current clamp recordings were made using a Dagan BVC-700 amplifier and custom written acquisition software using Igor Pro (WaveMetrics) or Axograph X (Axograph). Data were sampled at 20–50 kHz, filtered at 3 kHz, and then digitized by an InstruTECH ITC-18 interface (HEKA). The internal recording solution consisted of (in mM): 135 K-gluconate, 10 HEPES, 7 NaCl, 7 K₂-phosphocreatine, 0.3 Na-GTP, 4 Mg-ATP (pH corrected to 7.3 with KOH). In some cases, 0.3% neurobiotin was added to the recording solution for post-hoc morphological reconstruction (see below). Recording electrodes were pulled from borosilicate glass and had an open tip resistance of 4–6 MΩ. Series resistance was compensated using the bridge balance circuit and was monitored throughout the experiment. The median series resistance of recordings was 26 MΩ and recordings in which the series resistance exceeded 35 MΩ or increased by more than 30% were excluded. Resting membrane potential was noted immediately after establishing the whole cell recording configuration.

During voltage-clamp recordings, slices were perfused with aCSF containing (in mM): 0.001 tetrodotoxin, 10 tetraethylammonium chloride, 5 4-aminopyridine, 0.1 BaCl₂, 0.1 CdCl₂, and 0.05 NiCl₂ to block voltage-gated sodium, potassium, and calcium channels. Voltage clamp data were acquired using an Axopatch 200B amplifier (Molecular Devices) with custom written acquisition software in Igor Pro (WaveMetrics). Data were digitized at 20 kHz and filtered at 3 kHz. Inward currents were recorded in response to a series of 1000 ms hyperpolarizing voltage commands (−60 mV to −130 mV in −10 mV increments) from a holding potential of −30 mV.

Drugs

All drugs were obtained from Tocris, Abcam pharmaceutical, or Sigma. Drugs were prepared from a 1000x stock solution in water.

Data analysis and statistics

Custom written software for either Igor Pro 7 or Axograph was used to analyze all physiology data. Input resistance was calculated from the linear portion of voltage-current relationship in response to a family of 1.5 s current injections of −40 pA to 40 pA in steps of 10 pA. Voltage sag responses were calculated from the ratio of maximum voltage deflection and steady-state voltage from the −40 pA current step. Rebound slope was acquired as a function of the steady-state membrane potential. Membrane time constant was calculated using the −10 pA current step and was determined as the slow component of a double exponential fit to the voltage decay. Action potential firing frequency was calculated from the number of action potentials during a family of 1.5 s depolarizing current steps from 0 pA to 250 pA in steps of 25 pA. Action potential threshold was determined as the voltage where the first derivative exceeded 20 mV/ms. The interspike interval (ISI) was defined as the time between threshold of the first action potential and the subsequent action potential. The ISI ratio for interneuron classification was calculated by dividing ISI for last two action potentials and the first two action potentials in a train of 8–10 action potentials. All the statistical tests (unpaired t-tests, ANOVA, repeated measures of ANOVA, and Pearson’s R) were performed using Prism (GraphPad). All data are shown as the mean ± standard error of the mean (SEM). Effect size was calculated using G*Power and is listed in the figure legend using the canonical definitions of large (0.8), medium (0.5), and small (0.2).

Immunohistochemistry

Mice were deeply anesthetized using a ketamine/xylazine cocktail (100/10 mg/kg) and perfusion-fixed with ~20 mL of cutting saline (see above) or phosphate buffered saline (PBS) followed by ~25 mL 4% paraformaldehyde (PFA) in cutting saline or phosphate buffer at 4°C. The brain was removed, postfixed in PFA for 2 hours at RT, and placed in 30% sucrose in PBS overnight at 4°C. Horizontal thin sections (50 µm) containing the middle hippocampus (see above) were made on a cryostat (Leica) and placed in PBS. FMRP staining protocol was adapted from (40). First, slices underwent antigen retrieval and were heated to between 85 °C and 95 °C for 30 minutes in 0.01M sodium citrate (pH adjusted to 6.0 with HCl) (Sigma). Slices were moved to a new plate and washed in 0.02% PBS-T 3 × 5 minutes. Slices were then incubated in a blocking buffer solution (5% normal goat serum, 0.2% triton+PBS) at RT for 1 hr, followed by primary antibodies in blocking solution for 48 hr at 4°C. Primary antibodies used were: rabbit anti-parvalbumin (1:500, Swant), rabbit anti-somatostatin (1:500, ImmunoStar) and mouse anti-FMRP (1:3 dilution, final concentration of 4–5µg/mL) (developed by Tartakoff, A.M. / Fallon, J.R.; Case Western Reserve University), obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology, Iowa City, IA 52242.) Sections were then washed with 1x PBS 3 × 10 min and placed in blocking solution containing secondary antibodies (1:500 488 anti-mouse, 1:500 594 anti-rabbit) and Hoerst (1:2000) for 24 hr at 4°C. The slices were washed 4 × 10 min in 1x PBS and mounted on microscope slides and cover slipped with Fluormount-g (Invitrogen).

Imaging and Image Analysis

The labelled sections were visualized on a Zeiss AxioImager fluorescent microscope with an ApoTome at 10x for overview images, 20x for analysis images and 40x for representative cell images. The images were captured by CCD camera, acquired by Stereo Investigator software (MBF Bioscience), and analyzed using Image J (NIH). Images were taken of the Stratum Oriens CA1 area of the hippocampus defined as the area between the CA1 pyramidal layer and the alveus and in between the subiculum and CA2/CA3, and counts were performed on the entirety of the CA1 stratum oriens contained in each section. Co-expression analysis was completed using the cell counter software in Image J (NIH). Neurons were first counted and traced in the 594 channel to identify PV+ and SOM+ interneurons, and then overlayed in the 488 channel to identify co-expression of FMRP. Neurons were counted if they co-expressed the interneuron marker (PV or SOM) and FMRP. Slices from 4 wild type mice across 15 sections were included in the analysis. In total, there were 171 PV and 174 SOM positive interneurons that were counted for analysis, we then took the percentage of PV and SOM positive interneurons that also expressed FMRP.

Results section:

Both FS and LTS interneurons co-express FMRP

FMRP is highly expressed in the hippocampus in excitatory pyramidal neurons (41); however, there are currently no studies that report the co-expression of FMRP within specific subtypes of inhibitory interneurons in the CA1 hippocampus. Somatostatin (SOM) and parvalbumin (PV) are established markers of LTS and FS hippocampal interneurons respectively (but see (42, 43)). We performed immunohistochemistry using antibodies against FMRP (Figure 1A-D) and against either PV or SOM to identify putative FS and LTS interneurons in the stratum oriens of area CA1 in wild type mice and measure the co-expression of FMRP (Figure 1). We found that FMRP was present in both PV+ (Figure 1 E-G) and SOM+ cells (Figure 1 H-J) in the stratum oriens of area CA1. When expressed as a population of all PV+ and SOM+ interneurons, the percentage of SOM+ interneurons (putative LTS interneurons) which co-express FMRP is greater than PV+ interneurons (putative FS interneurons) which co-express FMRP (unpaired t-test; df = 6; t=5.119; p=0.002; Figure 1K).

Figure 1: Co-expression of FMRP with parvalbumin and somatostatin in stratum oriens of the hippocampus.

A) Immunohistochemical labeling of FMRP (green) in the hippocampus of a wildtype mouse. DAPI stain is shown in blue. B) 40x magnification of box area in panel (A). C) Immunohistochemical labeling of FMRP in the hippocampus of an FX mouse. D) 40x magnification of box area in panel (C). Note the absence of FMRP labeling in C-D indicative of antibody specificity. E) Immunohistochemical labeling of FMRP (green) and parvalbumin (red) in the CA1 region of the hippocampus. DAPI stain is shown in blue. F) 20X magnification image of box in (E). G) Immunohistochemical labeling of FMRP (green) and somatostatin (red) in the CA1 region of the hippocampus. DAPI stain is shown in blue. H) 20X magnification image of the box indicated in (G). I) Individual PV+ interneuron shown at 40x magnification showing co-expression of FMRP from the indicated region in (F). J) Individual SOM+ interneuron shown at 40x magnification showing co-expression of FMRP from the indicated region in (H). In panels F and I, s.o.: stratum oriens, s.p.: stratum pyramidale. K) Percent of PV+ or SOM+ neurons that also co-express FMRP (FMRP+ neurons / interneuron marker). In stratum oriens of CA1, more SOM+ interneurons show FMRP co-expression compared to PV+ interneurons (PV images n= 4 mice (filled symbols), 14 slices (open symbols), SOM images n= 4 mice (filled symbols), 14 slices (open symbols) unpaired t-test; df = 6; t=5.119; p=0.002).

The intrinsic excitability of LTS, but not FS, interneurons is different in FX mice

We made whole-cell current clamp recordings from interneurons in the stratum oriens of area CA1 in the hippocampus from wild type (WT) and Fmr1 knock-out (FX) mice. After noting the resting membrane potential (RMP), interneurons were held at −70 mV and two families of current injections were used to identify the type of interneuron. First, a series of small subthreshold current injections was used to measure the input resistance (R_N). Second, a series of 1.5-sec long depolarizing current injections was used to measure action potential firing. Fast spiking (FS) interneurons were identified by having a maximum action potential firing rate >100 Hz, no spike frequency adaptation, and low R_N (<200 MΩ). Low threshold spiking (LTS) interneurons by contrast had a maximum action potential firing rate <80Hz, readily apparent spike frequency adaptation, and high R_N (>200 MΩ).

Voltage responses to current injections in FS interneurons from both wild type and FX mice exhibited no sag or rebound and had comparatively low input resistances (Figure 2A). We found no significant differences in the RMP (WT: −61.4 ± 2.73 mV; FX: −55.9 ± 1.45 mV; unpaired t-test; df = 15; t=1.834; p=0.09) or input resistance (WT: 140 ± 8.6 MΩ; FX: 143 ± 12.7 MΩ; unpaired t-test; df = 13; t=0.215; p=0.83) between wild type and FX FS interneurons (Figure 2B-C). FS interneurons from both wild type and FX mice exhibited the expected high firing frequency and displayed no spike frequency adaptation. The number of action potentials fired in response to a range of depolarizing current amplitudes (two-way ANOVA, main effect of genotype F(1,29)=1.107; p=0.301; Figure 2D-E) and the maximum firing rate (WT: 97 ± 21.2 Hz; FX: 105 ± 13.6 Hz ; unpaired t-test; df = 15; t=0.3396; p=0.7389) were not significantly different between wild type and FX FS interneurons. Analysis of single action potentials found no difference in action potential parameters between WT and FX FS interneurons (Table 1).

Figure 2: Intrinsic excitability of low-threshold spiking, but not fast spiking, stratum oriens interneurons in is lower in FX mice.

A) Voltage responses to a family of small current steps in WT and FX FS interneurons. B) Resting membrane potential is not different between WT and FX FS interneurons (unpaired t-test; df = 15; t=1.834; p=0.08). C) Steady state input resistance is not different between WT and FX FS interneurons (unpaired t-test; df = 14; t=0.065; p=0.52) D) Action potential firing in WT and FX FS interneurons in response to a 250 pA current injection. E) There was no difference in action potential firing output across a range of depolarizing current steps between WT and FX interneurons (repeated measures two-way ANOVA, main effect of genotype F(1,29)=1.107; p=0.301). WT n=8 cells/6 mice; FX n=9 cells/7 mice, effect size small. F) Voltage traces of WT and FX LTS interneurons in response to a family of hyperpolarizing current steps. G) No differences in the resting membrane potential between WT and FX LTS interneurons (unpaired t-test; df = 54; t=0.69; p=0.49). H) FX LTS interneurons have a lower input resistance than WT LTS interneurons (unpaired t-test; df = 53; t=2.622; p=0.01) I) Action potential firing in WT and FX LTS interneurons in response to a 250 pA current step. J) FX LTS interneurons fire significantly fewer action potentials in response to a family of depolarizing current steps (repeated measures ANOVA, main effect of genotype (F (1,50) = 5.69; p = 0.02)).

Table 1: FS neuron action potential properties are not different between wild type and FX mice.

Analysis of single AP properties showed no difference between wild type and FX LTH neurons. AP threshold: (unpaired t-test, df = 15, t=0.1433, p=0.89); AP Amplitude: (unpaired t-test, df = 15, t=1.303, p=0.21); AP half-width: (unpaired t-test, df = 15, t=0.27, p=0.79); AHP: (unpaired t-test, df = 15, t=0.38, p=0.88; Max rate of rise: (unpaired t-test, df = 15, t=0.34, p=0.74). Data are presented as the mean ± SEM.

	wild type (n=11)	FX (n=7)	p-value
AP threshold (mV)	−40.2 ± 1.06	−40.5 ± 1.43	0.89
AP amplitude (mV)	84 ± 4.1	90 ± 2.5	0.21
AP half-width (ms)	0.58 ± 0.04	0.59 ± 0.04	0.79
AHP (mV)	−21.4 ± 1.54	−22.2 ± 1.45	0.71
Max. rate of rise (mV/ms)	256.6 ± 21.8	266.5 ± 20.0	0.74

In contrast to FS interneurons, subthreshold voltage responses from LTS interneurons exhibited a prominent sag and rebound in response to current injections (Figure 2F), and a comparatively high input resistance. Although there was no significant difference in RMP between wild type and FX LTS interneurons (WT: −52.8 ± 1.29 mV; FX: −51.9 ± 0.67 mV; unpaired t-test; df = 54; t=0.69; p=0.49; Figure 2G), FX LTS interneurons had a significantly lower R_N compared to wild type (WT: 413 ± 33.7 MΩ; FX: 305 ± 23.6 MΩ; unpaired t-test; df = 53; t=2.622; p=0.011; Figure 2H). As expected, both wild type and FX hippocampal LTS interneurons displayed spike frequency adaptation. When compared across a range of depolarizing current steps, FX LTS interneurons fired significantly fewer action potentials compared to wild type (two-way ANOVA, main effect of genotype F(1,50)=5.69; p=0.02; Figure 2I-J). These data therefore suggest that the intrinsic excitability of LTS interneurons is lower in FX mice.

LTH and OLM interneurons are present in FX mice

We previously demonstrated that LTS interneurons in stratum oriens of area CA1 can be separated based on a constellation of physiological measurements into at least two distinct groups, putative oriens lacunosum-moleculare interneurons (OLM) and low threshold high I_h interneurons (LTH) (36). We wanted to determine if the differences in intrinsic excitability between wild type and FX LTS interneurons occurred in both groups of interneurons. Our prior K-means clustering revealed that two of the physiological parameters that made the highest contribution to separating LTS interneurons were steady-state input resistance and inter-spike interval (36). Furthermore, there was a negative correlation between steady-state input resistance and ISI ratio. In this study, we saw that steady-state input resistance was negatively correlated with ISI ratio for both wild type (Pearson’s r = −0.3878, p = 0.0102) and FX (Pearson’s r = −0.4868, p = 0.0007) LTS interneurons (Figure 3A-B). Based on our observation that input resistance is significantly different between wild type and FX LTS interneurons, we used ISI ratio to separate LTS interneurons into putative OLM interneurons (ISI ratio ≤2.2) and LTH interneurons (ISI ratio >2.2) (36). Spike frequency adaptation was not different between wild type and FX mice for either LTH or OLM interneurons (two-way ANOVA, main effect of genotype F(1, 65) = 0.0105, p = 0.9189; Figure 3C). We previously reported that are no gross morphological differences between OLM and LTH neuron in wild type mice (36). As a proxy for cell size, we estimated the membrane capacitance from the measurements of the membrane time constant and input resistance. In agreement with our prior results, we found no significant difference in membrane capacitance between OLM and LTH neurons (WT: 85 ± 2.8 pF; FX: 71 ± 12.7 pF; unpaired t-test; t=1.191, df = 17, p = 0.251). There was also no difference in the proportion of OLM and LTH interneurons between wild type and FX mice (Figure 3D).

Figure 3: Both OLM and LTH cell types of low-threshold spiking interneurons are found in the FX hippocampus.

A) Voltage traces from the first depolarizing current step to elicit action potentials in LTS interneurons, separated by ISI ratio, from WT (black) and FX (red) mice. B) Steady state input resistance is correlated with the ISI ratio for both wild type and FX LTS interneurons. Note: the slope is reduced in FX LTS interneurons because of the lower steady-state input resistance. C) ISI ratio is not different between WT and FX mice for OLM or LTH interneurons. D) The proportion of OLM and LTH interneurons is not different between WT and FX mice (total = the number recorded cells). WT recordings: OLM n=10 cells/10 mice and LTH n=8 cells/8 mice; FX recordings: OLM n=22 cells/15 mice and LTH n=16 cells/12 mice.

LTH interneurons have lower intrinsic excitability in FX mice.

Using a series of hyperpolarizing current injections, we found that FX LTH interneurons had a lower R_N compared to wild type LTH interneurons (WT: 305 ± 20.9 MΩ; FX: 197 ± 17.4 MΩ; unpaired t-test; df = 31; t=3.989; p=0.0004; Figure 4A-B). In response to depolarizing current injections, FX LTH interneurons fired significantly fewer action potentials compared to wild type (two-way ANOVA, main effect of genotype F(1, 23)=8.803; p=0.0069; Figure 4C-D). Additionally, the maximum firing rate in response to the largest depolarizing current step was significantly smaller in FX LTH interneurons compared to wild type (WT: 48.6 ± 2.3 Hz; FX: 35.8 ± 2.9 Hz; unpaired t-test; df = 22; t=2.667; p=0.0141). Surprisingly, resting membrane potential, sag, and rebound were not significantly different between wild type and FX LTH interneurons (Table 2). In contrast to LTH interneurons, we found no significant difference in R_N (WT: 466 ± 45 MΩ; FX: 389 ± 29.3 MΩ; unpaired t-test; df = 33; t = 1.457; p=0.1546; Figure 4E-F) or resting membrane potential, sag, and rebound (Table 2) between FX and wild type OLM interneurons. There was also no significant difference in the number of action potentials fired (two-way ANOVA, main effect of genotype F(1,30)=0.83; p=0.369; Figure 4G-H) or the maximum firing rate (WT: 58.1 ± 6.81 Hz; FX: 56.0 ± 4.41 Hz; unpaired t-test; t=0.2587; df = 30; p=0.79) between wild type and FX OLM interneurons.

Figure 4: The lower intrinsic excitability of FX LTS interneurons is restricted to the LTH population.

A) Voltage traces of WT and FX LTH interneurons in response to family of hyperpolarizing current steps. B) FX LTH interneurons have lower steady state input resistance compared to WT LTH interneurons (unpaired t-test; df = 22; t=3.575; p=0.001). C) Action potential firing in response to a 75 pA current step in WT and FX LTH interneurons. D) FX LTH interneurons fired significantly fewer action potentials across a range of current injections compared to WT (repeated measures ANOVA, main effect of genotype (F(1,19) = 7.971; p = 0.01). E) Voltage traces of WT and FX OLM interneurons in response to a family of hyperpolarizing current steps. F) There was no significant difference in steady-state input resistance between WT and FX OLM interneurons (unpaired- t-test, df =29, t=1.87, p= 0.07). G) Action potential firing in response to a 50 pA current injection in WT and FX OLM interneurons. H) There was no significant difference in action potential firing output between WT and FX OLM interneurons (repeated measures, two-way ANOVA, main effect of genotype F(1,30)=0.83, p=0.369). LTH interneurons (A-D): WT n=8 cells/8 mice; FX n=16 cells/12 mice; OLM interneurons (E-H): WT n=10 cells/10 mice; FX n=22 cells/15 mice.

Table 2: No difference in sag and rebound response between wild type and FX LTH or OLM neurons.

Resting membrane potential (RMP), sag, and rebound responses of wild type and FX LTH and OLM neurons. No difference in RMP, sag or rebound response between wild type and FX LTH neurons (RMP: unpaired- t-test, df =33, t=0.089, p= 0.9295; sag: unpaired t-test, df=33, t=0.161, p=0.8732; rebound: unpaired t-test, df=16, t=0.42, p=0.67) No difference in RMP, sag or rebound response between wild type and FX OLM neurons (RMP: unpaired- t-test, df =30, t=0.45, p= 0.65; sag: unpaired t-test, df = 29, t=1.52, p=0.14; rebound: unpaired t-test, df = 11, t=2.13, p=0.06). Data are presented as the mean ± SEM.

	wild type	FX	p-value
LTH (n = 16 wt, 19 ko)
RMP (mV)	−51.4 ± 1.67	−51.2 ± 0.957	0.9295
Sag ratio	1.19 ± 0.03	1.19 ± 0.03	0.8732
Rebound (mV/mV)	−0.28 ± 0.06	0.31 ± 0.06	0.67
OLM (n = 10 wt, 22 ko)
RMP (mV)	−51 ± 1.71	−52 ± 1.71	0.65
Sag ratio	1.04 ± 0.05	1.12 ± 0.05	0.14
Rebound (mV/mV)	−0.04 ± 0.01	−0.07 ± 0.01	0.06

Action potential threshold is not different in LTH interneurons.

Previous studies found differences in action potential threshold in multiple cell types in FX mice (17, 24, 44, 45). To test if changes in action potential threshold could also contribute to the difference in action potential firing between wild type and FX interneurons, we analyzed the first elicited action potential during the smallest current injection to trigger action potentials. There was no significant difference in threshold for the first action potential between wild type and FX LTH interneurons (WT: −40.3 ± 0.77 mV; FX: −41.5 ± 0.92 mV; unpaired- t-test, df =25, t=0.9849, p= 0.3341; Figure 5A-B) or OLM interneurons (WT: −41.4 ± 0.84 mV; FX: −40.2 ± 0.87 mV; unpaired- t-test, df =30, t=0.9536, p= 0.3479; Figure 5D-E). Further analysis of the action potential waveform found no significant differences in amplitude, half-width, afterhyperpolarization, maximum rate of rise, or maximum rate of decay (Table 3). During repetitive firing, the threshold for action potential firing becomes more depolarized due changes in the availability for voltage-gated sodium channels (e.g., cumulative inactivation) (46). An increase in cumulative inactivation in FX LTH interneurons would decrease action potential firing by depolarizing threshold during repetitive firing. Consistent with previous results, we found that action potential threshold depolarized during a train of 25 action potentials in both wild type and FX LTH interneurons. There was, however, no significant difference in the change in action potential threshold during repetitive firing between wild type and FX LTH interneurons (Figure 5C; two-way ANOVA, main effect of genotype F(1,16)=0.53; p=0.47) or OLM interneurons (Figure 5D-E; two-way ANOVA, main effect of genotype F(1,13)=0.02, p=0.88). These results suggest that the lower action potential firing rate in FX LTH interneurons is not due to differences in action potential properties.

Figure 5: Action potential threshold in LTH and OLM interneurons is not different between WT and FX mice.

A) Representative voltage traces of the first action potential elicited by the smallest current injection in WT and FX LTH interneurons. B) Action potential threshold is not different between WT and FX LTH interneurons (unpaired- t-test, df =21, t=1.02, p= 0.31). C) The change in action potential threshold during repetitive firing is not different between WT and FX LTH interneurons (two-way ANOVA, main effect of genotype F(1,16)=0.53; p=0.47). D) Representative voltage traces of the first action potential elicited by the smallest current injection in WT and FX OLM interneurons. E) AP threshold is not different between WT and FX OLM interneurons (unpaired t-test, df=18 t=0.28 p= 0.78). F) The change in action potential threshold during repetitive firing is not different between WT and FX OLM interneurons (repeated measures, two-way ANOVA, main effect of genotype F(1,13)=0.02, p=0.88). LTH interneurons: WT n=8 cells/8 mice; FX n=16 cells/12 mice; OLM interneurons: WT n=10 cells/10 mice; FX n=22 cells/15 mice.

Table 3: Action potential properties in LTH neurons are not different between wild type and FX mice.

Analysis of single AP properties showed no difference between wild type and FX LTH neurons. AP Amplitude: (unpaired t-test, df = 16, t=1.17, p=0.26); AP half-width: (unpaired t-test, df = 16, t=0.71, p=0.49); AHP: (unpaired t-test, df = 16, t=0.15, p=0.88; Max rate of rise: (unpaired t-test, df = 16, t=0.75, p=0.46); Max rate of fall: (unpaired t-test, df = 16, t=0.52, p=0.58). Data are presented as the mean ± SEM.

	wild type (n=11)	FX (n=7)	p-value
AP amplitude (mV)	101.7 ± 3.03	98.1 ± 3.03	0.26
AP half-width (ms)	0.76 ± 0.06	0.71 ± 0.06	0.49
AHP (mV)	−21.1 ± 2.6	−20.89 ± 2.6	0.88
Max. rate of rise (mV/ms)	285.6 ± 21.4	269.5 ± 21.4	0.46
Max. rate of fall (mV/ms)	−137 ± 13.4	−144.4 ± 13.4	0.58

Block of I_h normalizes input resistance between wild type and FX LTH interneurons.

Both excitatory and inhibitory neurons express HCN channels (I_h) which contribute to the resting properties of the neuron (47–51). Given the high relative contribution of I_h to the resting properties of LTH interneurons (36), we hypothesized that higher I_h contributes to the lower R_N in FX LTH interneurons. As a first test of this hypothesis, we made current clamp recordings from wild type and FX LTH interneurons before and after extracellular application of 50 µM ZD7288 to block HCN channels (Figure 6A). In agreement with our earlier results, the input resistance of FX LTH interneurons was significantly lower than wild type LTH interneurons (two-way repeated measures ANOVA: main effect of genotype, F(1,11)=9.477; p=0.0491). Consistent with the contribution of I_h to R_N in LTH interneurons, ZD7288 significantly increased the R_N of LTH interneurons in both wild type (pre-ZD: 315 ± 25 MΩ, post-ZD: 532 ± 52 MΩ) and FX mice (pre-ZD: 192 ± 27 MΩ, post-ZD: 325 ± 53 MΩ) (two-way repeated measures ANOVA: main effect of ZD7288, F(1,11)=46.85; p<0.001) (Figure 6A-B). As expected, ZD7288 also significantly decreased both voltage sag (pre-ZD: −1.15 ± 0.038, post-ZD: 1.0 ± 0.003) and FX mice (pre-ZD: 1.24 ± 0.06, post-ZD: 1.02 ± 0.017) (two-way repeated measures ANOVA: main effect of ZD7288, F(1,11)=37.99; p<0.001; Figure 6C) and rebound wild type (pre-ZD: −0.212 ± 0.035 mV/mV, post-ZD: 0.007 ± 0.014 mV/mV) and FX mice (pre-ZD: −0.264 ± 0.03 mV/mV, post-ZD: 0.002 ± 0.02 mV/mV) (two-way repeated measures ANOVA: main effect of ZD7288, F(1,11)=117.2; p<0.001; Figure 6D) in wild type and FX LTH interneurons. The maximum action potential firing rate in both wild type and LTH interneurons was significantly increased after application of ZD7288 (two-way repeated measures ANOVA: main effect of ZD7288, F(1,12)=10.41; p=0.0073; Figure 6E-F). Interestingly, while ZD7288 increased the number of action potentials fired in wild type LTH interneurons (two-way repeated measures ANOVA: main effect of ZD7288, F(1,12)=11.42; p=0.0045; Figure 6G), the effect of ZD7288 on action potential was not significant in FX LTH interneurons (two-way repeated measures ANOVA: main effect of ZD7288, F(1,12)=3.98; p=0.074; Figure 6H). Taken together, these results suggest that higher I_h contributes to the lower input resistance in FX LTH interneurons compared to wild type although ZD7288 did not however, normalize action potential firing between wild type and FX LTH interneurons (Figure 6F).

Figure 6: Block of I_h with ZD7288 increases input resistance and AP firing in both WT and FX LTH interneurons.

A) Representative voltage responses to a hyperpolarizing step before and after 50 µM ZD7288 for WT (black, baseline; grey ZD7288) and FX (red, baseline; pink, ZD7288) LTH interneurons. B) ZD7288 significantly increased steady-state input resistance in both WT and FX LTH interneurons. C-D) ZD7288 significantly decreases voltage sag (C) and rebound (D) in WT and FX LTH interneurons. E) Representative action potential firing in response to a depolarizing step before and after ZD7288 for WT (black, baseline; grey ZD7288) and FX (red, baseline; pink, ZD7288) LTH interneurons. F) ZD7288 significantly increased the maximum firing rate in both WT and FX LTH interneurons. G-H) Application of ZD7288 significantly increased the number of action potentials fired across a range of current injections in WT (G, two-way repeated measures ANOVA: main effect of ZD7288, F(1,12)=11.42; p=0.0045) but not FX LTH interneurons (H, two-way repeated measures ANOVA: main effect of ZD7288, F(1,12)=3.98; p=0.074). WT n=11 cells/10 mice; FX n=9 cells/7 mice.

LTH interneurons in FX mice have higher I_h

Given the effect of ZD7288 on input resistance, we used a previously published current clamp, voltage clamp protocol to directly measure I_h in LTH interneurons (36, 47). First, we measured the ISI ratio of action potential firing using current clamp to identify the interneuron as FS, OLM, or LTH. Following interneuron classification, we applied a combination of pharmacological agents to block voltage-gated Na⁺, K⁺, and Ca²⁺ channels (Figure 7A). After 20 minutes, we switched to voltage clamp and measured the current in response to a family of hyperpolarizing voltage steps from a holding potential of −30 mV before and after application of 50 µM ZD7288. The current in the presence of ZD was subtracted from the baseline current (Figure 7B). A slowly activating ZD-sensitive inward current consistent with I_h in LTH interneurons (Figure 7C; Hewitt et al., 2021) was recorded from both wild type and FX mice. The ZD-sensitive current significantly increased with hyperpolarization in both wild type and FX LTH interneurons (two-way repeated measures ANOVA: main effect of voltage, F(4,24)=10.93; p<0.001; Figure 7D). In agreement with our hypothesis, the ZD-sensitive current was significantly larger in FX LTH interneurons compared to wild type at hyperpolarized membrane potentials (for −130 mV, WT: −39 ± 12.9 pA; FX: −136 ± 49.1 pA; mixed factor ANOVA, interaction between voltage and genotype: F(4,24)=3.529, p=0.021; Figure 7D). These results support the hypothesis that the lower input resistance in FX LTH interneurons is due in part to higher I_h.

Figure 7: The hyperpolarization activated current I_h is larger in FX LTH interneurons compared to WT.

A) Experimental procedure for voltage-clamp recordings of I_h in WT and FX LTH interneurons. B) Current measured in response to a voltage step to −130 mV in WT (top) and FX (bottom) LTH interneurons before and after application of 50 µM ZD7288. C) Representative traces of ZD-sensitive current in response to step to −130 mV from WT (black) and FX (red) LTH interneurons. D) FX LTH interneurons have more I_h compared to WT (mixed factor ANOVA, interaction between voltage and genotype: F(4,24)=3.529, p=0.021). WT n=4 cells/4 mice; FX n=4 cells/4 mice.

Discussion

Over the last decade, many studies identified changes in voltage-gated ion channels in excitatory neurons in the mouse model of Fragile X syndrome. By contrast, there are surprisingly few studies on the neuronal properties of inhibitory interneurons in FX. We investigated the intrinsic physiological properties of fast spiking and low threshold spiking inhibitory interneurons in the stratum oriens of area CA1 in the hippocampus from wild type and FX mice.

We found no significant differences in intrinsic subthreshold properties or action potential firing between wild type and FX fast spiking interneurons in stratum oriens of area CA1. We found no significant difference in the intrinsic excitability of FS interneurons between wild type and FX mice. There are well known brain-region and cell-type specific changes in ion channel function in FX mice (for review see (52). In layer 4 of somatosensory cortex, FS interneurons in FX mice have higher input resistance and action potential firing rate compared to wild type (53). In primary visual cortex, spatial selectivity of cortical FS, PV-positive interneurons is decreased in FX mice (54). It is possible that the loss of FMRP has differential effects on FS interneurons between the cortex and hippocampus.

We found that low threshold spiking interneurons in FX mice had a lower input resistance and fired fewer action potentials compared to wild type. We recently showed that stratum oriens LTS interneurons may be subdivided into putative OLM and LTH interneurons (36). Current clamp experiments revealed that it is the LTH, but not OLM, population that had a significantly lower input resistance and fired fewer action potentials in FX mice. Application of the h-channel blocker ZD7288 increased the input resistance in wild type and FX LTH interneurons suggesting that this difference was due in part to higher I_h. Voltage clamp experiments demonstrated that FX LTH interneurons have more I_h compared to wild type.

Changes in voltage-gated ion channel function and expression, including HCN channels, were identified in FX excitatory pyramidal neurons (19, 21, 24, 37, 38, 55, 56). Here we show FX LTH interneurons have a lower input resistance due in part to a larger h-channel current. While other stereotypical effects of higher I_h were not observed in FX LTH interneurons (e.g., depolarized RMP, higher sag and rebound), this could be due in part to differences in the kinetics of h-channels and/or the localization of the channels (dendritic vs somatic expression) in FX LTH interneurons. To properly investigate this question, patch clamp recordings would be necessary to address the space clamp limitations of whole-cell voltage clamp. Cell-attached patch clamp experiments would require first using current clamp (with a separate electrode) to physiologically identify the interneuron type. Although outside-patch clamp is potentially feasible, it is not clear what the subcellular localization of HCN channels is in LTS interneurons. If there is a strong dendritic enrichment of HCN channels (or a redistribution in FX), visualization of and electrode placement on the small dendrites of stratum oriens interneurons would make this approach difficult.

Although input resistance was increased by block of I_h with ZD7288, FX LTH interneurons fired significantly fewer action potentials compared to wild type. The absence of sag and rebound after application of ZD7288 in those experiments suggests that I_h was blocked. Thus, this result suggests that other ion channels may be altered in FX LTH interneurons. Changes in both voltage-gated Na⁺ and K⁺ channels were reported in both excitatory and inhibitory neurons in FX mice (18, 20, 22, 24, 38). We did not observe any changes in action potential properties (Figure 5; Table 2) that would suggest significant changes in either voltage-gated Na⁺ or K⁺ channel function. A higher resting K⁺ conductance in FX LTH interneurons could contribute to the lower action potential firing in the presence of ZD7288. Due to the incredible diversity and limited knowledge of interneuron specific potassium channels however, it would be difficult to isolate which channels may contribute to this observation (for review: (43, 57).

Inhibitory deficits in Fragile X syndrome

Inhibitory interneurons play a critical role in balancing excitatory activity (7, 43). Proper inhibitory signaling promotes intact plasticity and guides network wide oscillatory events that are critical for learning and memory (6, 58). Several studies on cortical inhibition in FX mice highlighted changes in both fast spiking and low threshold spiking interneurons (53, 54, 59). Reduced activation of low threshold spiking interneurons by metabotropic glutamate receptor signaling and decreased synchronization in synaptic inhibition suggest weakened LTS signaling in somatosensory cortex (12). Network hyperexcitability, measured as prolong UP states, in somatosensory cortex is due in part to decreased inhibitory activity from FS interneurons (59). These studies demonstrate the importance of investigating specific types of inhibitory interneurons to understand how changes in inhibition contribute to FX pathology.

Although there are several known hippocampal functions that require appropriate stratum oriens inhibitory interneuron signaling (28, 29), there are surprisingly few investigations that directly investigate the hippocampal inhibitory interneurons of this region in FX models. While we saw no significant difference in intrinsic properties of FS interneurons between wild type and FX mice, changes in the excitatory synaptic drive onto FS interneurons or the inhibitory synaptic drive onto downstream neurons may be different in FX mice. We found that LTH interneurons in FX mice were less excitable compared to wild type mice. While the exact axon target in the hippocampus of CA1 LTH interneurons remains unknown, the lower intrinsic excitability would reduce inhibitory drive of hippocampal circuits. If like OLM interneurons, LTH interneurons provide feedback inhibition onto CA1 pyramidal neurons, the reduced excitability of FX LTH interneurons would make the synaptic excitation from CA1 neurons less effective, thus reducing feedback inhibition. Prior studies reported reduced inhibitory drive onto pyramidal neurons in CA1 in the FX hippocampus (60, 61). It is unclear however, which type of inhibitory interneuron(s) were the source of this inhibition. Additionally, a reduction in inhibitory drive from decreased intrinsic excitability from FX LTH interneurons could result in poor modulation of CA1 neuron output. FX mice have been shown to exhibit unstable theta-gamma coupling in the CA1 hippocampus (62). This discoordination of hippocampal rhythms could be a result of decreased LTH interneuron output, and therefore unreliable control of CA1 pyramidal neuron activity. Additionally, previous work found homeostatic compensation in the FX brain in response to a shift in E/I balance (17, 63, 64). Due to the regulatory nature of I_h on neuron excitability, the larger I_h expression in FX LTH interneurons could be a homeostatic response to the increased excitability seen in the FX hippocampus.

FXS patients display several behavioral deficits associated with hippocampal dysfunction such as epilepsy and sensory hypersensitivity. One of the leading hypotheses of these behavioral deficits in FXS is an imbalance of excitatory/inhibitory activity in the hippocampus. Each inhibitory interneuron subtype plays a specific role in orchestrating the flow of excitatory activity through the hippocampus and a deficit in even one subtype in FXS could contribute to hippocampal dysfunction. We found that LTH inhibitory interneurons, a recently identified subclass of low threshold spiking interneurons, in the CA1 stratum oriens of the hippocampus in FX mice have higher expression of the hyperpolarization activated cation current, I_h. The higher I_h in FX LTH interneurons contributed to lower input resistance compared to WT LTH interneurons. These changes contribute to lower intrinsic excitability of CA1 stratum oriens LTH interneurons in FX mice, which can potentially alter the E/I balance of the CA1 hippocampal circuit. This study highlights the importance of investigating hippocampal interneurons in a subtype and brain region specific fashion which is critical to understanding the cellular basis for neurological deficits in FX.