Exploration of group II metabotropic glutamate receptor modulation in mouse models of Rett syndrome and MECP2 Duplication syndrome

Abstract

Rett syndrome (RTT) and MECP2 Duplication syndrome (MDS) have opposing molecular origins in relation to expression and function of the transcriptional regulator Methyl-CpG-binding protein 2 (MeCP2). Several clinical and preclinical phenotypes, however, are shared between these disorders. Modulation of MeCP2 levels has recently emerged as a potential treatment option for both of these diseases. However, toxicity concerns remain with these approaches. Here, we focus on pharmacologically modulating the group II metabotropic glutamate receptors (mGlu), mGlu₂ and mGlu₃, which are two downstream targets of MeCP2 that are bidirectionally affected in expression in RTT patients and mice (Mecp2^Null/+) versus an MDS mouse model (MECP2^Tg1/o). Mecp2^Null/+ and MECP2^Tg1/o animals also exhibit contrasting phenotypes in trace fear acquisition, a form of temporal associative learning and memory, with trace fear deficiency observed in Mecp2^Null/+ mice and abnormally enhanced trace fear acquisition in MECP2^Tg1/o animals. In Mecp2^Null/+ mice, treatment with the mGlu_2/3 agonist LY379268 reverses the deficit in trace fear acquisition, and mGlu_2/3 antagonism with LY341495 normalizes the abnormal trace fear learning and memory phenotype in MECP2^Tg1/o mice. Altogether, these data highlight the role of group II mGlu receptors in RTT and MDS and demonstrate that both mGlu₂ and mGlu₃ may be potential therapeutic targets for these disorders.

Graphical Abstract

1. Introduction

MECP2-associated disorders are X-linked monogenic neurodevelopmental diseases that are caused by abnormal expression and/or function of the protein Methyl-CpG-binding protein 2 (MeCP2), which is encoded on the X chromosome. Rett syndrome (RTT) is caused by loss-of-function (LOF) mutations in MECP2, and is observed most often in females (Amir et al., 1999; Lombardi et al., 2015). The multi-domain clinical symptoms of RTT, including motor dysfunction, impaired social skills, cognitive decline, and breathing abnormalities, overlap with another MECP2-associated disorder, MECP2 Duplication syndrome (MDS), which is caused by multiple copies of the MECP2 gene. Due to random X chromosome inactivation in females, patients diagnosed with MDS are predominantly male (Ramocki et al., 2009; Van Esch et al., 2005). The symptoms observed in RTT patients are recapitulated in mice that ubiquitously or cell type-specifically lack Mecp2 or express functionally mutated Mecp2 (Adachi et al., 2009; Ballinger et al., 2019; Brown et al., 2016; Chen et al., 2001; Collins et al., 2021; Fyffe et al., 2008; Gemelli et al., 2006; Goffin et al., 2011; Guy et al., 2001; Heckman et al., 2014; Jentarra et al., 2010; Kerr et al., 2008; Lamonica et al., 2017; Lawson-Yuen et al., 2007; Luikenhuis et al., 2004; Moretti et al., 2006, 2005; Pelka et al., 2006; Pitcher et al., 2015; Samaco et al., 2013, 2009, 2008; Schaevitz et al., 2013; Shahbazian et al., 2002; Vermudez et al., 2021). Similarly, clinical characteristics of MDS are observed in mouse models that either have global or neuronal-specific MECP2 overexpression (Collins et al., 2004; Na et al., 2014). Interestingly, many phenotypes in RTT mice are antiparallel to those observed in MDS animal models, including anxiety, motor abnormalities, and cognitive function. For example, RTT mice display deficits in associative learning and memory in a fear conditioning task; in contrast, MDS mice exhibit abnormally enhanced learning in this task in addition to impairments in extinction of fear-learned behavior (Collins et al., 2004; Moretti et al., 2006; Na et al., 2012; Stansley et al., 2018; Stearns et al., 2007).

Previous studies in RTT models have shown that genetic normalization of MECP2, even after disease onset, can rescue many phenotypes in Mecp2 mutant mice, supporting the feasibility of treating the disorder (Gadalla et al., 2017, 2013; Garg et al., 2013; Guy et al., 2007; Luoni et al., 2020; Matagne et al., 2021, 2017; Powers et al., 2019; Sinnett et al., 2021, 2017; Tillotson et al., 2017). Promisingly, recent studies have also demonstrated that genetic manipulations after symptom onset, including normalization of MeCP2 dosage with antisense oligonucleotides, can improve abnormal phenotypes in MDS mice (Koerner et al., 2018; Sztainberg et al., 2015). Pharmacological approaches targeting genes/pathways downstream of MeCP2 could also have potential in treating symptom domains of RTT and MDS. For example, in mice with MeCP2 over-expression in neurons (Tau-Mecp2), pharmacological antagonism of GABA_A receptors can alleviate symptoms (Na et al., 2014). Comparably, studies have determined that modulation of receptors involved in neurotransmission and that are sensitive to MeCP2 dosage can improve phenotypes in RTT mice (Bittolo et al., 2016; Degano et al., 2014; Gogliotti et al., 2018, 2017, 2016; Li et al., 2017; Ogier et al., 2007; Roux et al., 2007; Scaramuzza et al., 2021; Zanella et al., 2008). These include the Trkb, glutamatergic AMPA, metabotropic glutamate (mGlu), and muscarinic acetylcholine receptors, which were initially identified in expression studies in RTT patient and mouse samples (Bedogni et al., 2016; Ben-Shachar et al., 2009; Chahrour et al., 2008; Gogliotti et al., 2018, 2017, 2016; Lin et al., 2016; Pacheco et al., 2017). The association of these receptors with neurological and neurodevelopmental disorders, as well as the advent of selective modulators, make these receptors potential therapeutic targets for new drug candidates.

The expression of the group II mGlu receptors, mGlu₂ and mGlu₃, has been consistently demonstrated to be affected by MeCP2 dosage in patient and preclinical samples (Bedogni et al., 2016; Ben-Shachar et al., 2009; Chahrour et al., 2008; Gogliotti et al., 2018; Lin et al., 2016; Pacheco et al., 2017). Additionally, these dimeric receptors have been implicated in neuropsychiatric disorders such as schizophrenia and depression (Chaki, 2017; Egan et al., 2004; Harrison et al., 2008; Maksymetz et al., 2017; Saini et al., 2017). Studies using subtype-selective modulators, preclinical knockout mouse models, and clinical genetic associations have shown that mGlu₃ has a vital role in cognition, specifically in hippocampal and prefrontal cortical function (Dogra et al., 2021; Egan et al., 2004; Fujioka et al., 2014; Joffe et al., 2020, 2019; Lainiola et al., 2014; Pöschel et al., 2005; Rosenberg et al., 2016; Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2014; Walker et al., 2017, 2015). mGlu₂ has also been shown to be involved in cognition, as mGlu₂-selective positive allosteric modulators (PAMs) improve learning and memory in rodent models of schizophrenia (Griebel et al., 2016; Nikiforuk et al., 2010). Given the molecular studies demonstrating the relationship of MeCP2 and the group II mGlu receptors, as well as the association of these receptors with cognition, we posited that mGlu₂ and mGlu₃ may play critical roles in the etiology or treatment of the cognitive phenotypes observed in RTT and MDS.

In this study, we show that mGlu₂ and mGlu₃ receptor levels are decreased in temporal cortex samples from RTT patient autopsies. Expression of these receptors is also reciprocally decreased and increased in the hippocampus of Mecp2^Null/+ and MECP2^Tg1/o mice, respectively. Based on these findings, we tested the hypothesis that an mGlu_2/3 agonist or antagonist would positively affect behavior in Mecp2^Null/+ and MECP2^Tg1/o mice in a hippocampal-dependent behavioral cognitive assay. We show here that activation of mGlu_2/3 receptors reverses deficient trace fear acquisition in Mecp2^Null/+ animals, whereas mGlu_2/3 antagonism normalizes the abnormal enhanced trace fear acquisition phenotype in MECP2^Tg1/o mice. Collectively, these data demonstrate that both mGlu₂ and mGlu₃ receptors are implicated in the cognitive function of RTT and MDS model mice, and that modulation of mGlu_2/3 activity may be beneficial in alleviating cognitive symptoms of these two disorders.

2. Materials and Methods

2.1. Animals

All animals used in the present study were group housed with food and water given ad libitum and maintained on a 12hr light/dark cycle. Animals were cared for in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All studies were approved by the Vanderbilt Institutional Animal Care and Use Committee and took place during the light phase. MECP2^Tg1/o mice (FVB-Tg(MECP2)1Hzo/J, stock no. 008679) were cryorecovered and Mecp2^Null/+ (B6.129P2(C)-Mecp2^tm1.1Bird/J, stock no. 003890) were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Given that MECP2^Tg1/o mice on the FVB/N background are prone to retinal degeneration, we utilized the F1 generation or hybrid mice (FVB/N x C57BL/6) for our studies as described previously (Samaco et al., 2012; Zhang et al., 2017). These F1 hybrid mice were generated by crossing MECP2^Tg1/o mice (FVB/N background) with wild-type (WT) C57BL/6J mice (The Jackson Laboratory, stock no. 000664). To reflect the predominantly male clinical population in MDS (Ramocki et al., 2009), male MECP2^Tg1/o mice and WT littermates were used for all experiments, and at 8–12 weeks old, an age range at which MECP2^Tg1/o mice were previously observed to exhibit abnormal phenotypes (Samaco et al., 2012; Zhang et al., 2017). Similarly, as a reflection of the predominantly female RTT patient population, female Mecp2^Null/+ mice and WT littermates (Mecp2^+/+) were utilized and aged to at least 20 weeks of age prior to experiments to reflect the symptomatic age of female Mecp2^Null/+ animals (Guy et al., 2007, 2001). Mecp2^Null/+ animals were maintained on a C57BL/6J background by breeding Mecp2^Null/+ with WT C57BL/6J mice (The Jackson Laboratory, stock no. 000664).

2.2. Total Protein Preparation

The cortex and hippocampus were microdissected from 8–9-week-old male WT littermates and MECP2^Tg1/o mice, and 20–25-week-old female littermates, Mecp2^+/+ and Mecp2^Null/+ animals. Total protein was prepared as previously described in (Fisher et al., 2018). Briefly, tissue samples were homogenized using a hand-held motorized mortar and pestle in radioimmunoprecipitation assay buffer (RIPA) containing 10mM Tris-HCl, 150mM NaCl, 1mM ethylenediaminetetraacetic acid (EDTA), 0.1% sodium dodecyl sulfate (SDS), 1% Triton X-100, and 1% deoxycholate. After homogenization, samples were centrifuged and the supernatant was collected. Protein concentration was determined using a bicinchoninic acid (BCA) protein assay (Pierce).

2.3. SDS-Page and Western Blotting

As previously described in (Fisher et al., 2018), 50μg of total protein was electrophoretically separated using a 4–20% SDS polyacrylamide gel and transferred onto a nitrocellulose membrane (iBlot2, ThermoFisher (for mGlu₂ and mGlu₃); Criterion^™ Blotter, Bio-Rad (for MeCP2)). Membranes were blocked in TBS Odyssey blocking buffer (LI-COR) for 1hr at room temperature. Membranes were probed with primary antibodies overnight at 4°C: mouse anti-mGlu₂ (1:1000 Abcam, cat. no. ab15672), rabbit anti-mGlu₃ (1:1000, Alomone, cat. no. AGC_012), rabbit anti-MeCP2 (1:1000, Millipore, cat. no. 07–013), rabbit anti-vGlut2 (1:1000, Cell Signaling Technology, cat. no. 71555) and mouse anti-Gapdh (1:1000, ThermoFisher, cat. no. MA5–15738), followed by the fluorescent secondary antibodies: goat anti-rabbit (800nm, 1:5000, LI-COR, cat. no. 926–32211) and goat anti-mouse (680nm, 1:10,000, LI-COR, cat. no. 926–68020). Fluorescence was detected using the Odyssey (LI-COR) imaging system at the Vanderbilt University Medical Center Molecular Cell Biology Resource (MCBR) Core and then quantified using the Image Studio Lite software (LI-COR). Values were normalized to Gapdh and compared relative to littermate controls (WT littermates or Mecp2^+/+).

2.4. Total RNA Extraction and cDNA Synthesis

For total RNA extraction of human temporal cortex samples, frozen samples were obtained from the University of Maryland Brain and Tissue Bank and the Harvard Brain Tissue Resource Center, which are Brain and Tissue Repositories of the National Institutes of Health NeuroBioBank (neurobiobank.nih.gov). For mouse total RNA extraction, the cortex and hippocampus were microdissected from 8–9-week-old male WT littermates and MECP2^Tg1/o mice, and 20–25-week-old female littermates, Mecp2^+/+ and Mecp2^Null/+ animals. Total RNA was prepared from tissue samples using TRIzol Reagent (ThermoFisher) and isolated using a RNeasy Mini Kit (Qiagen) in accordance with manufacturer’s instructions. Total RNA was DNase-treated with RNase-Free DNase Set (Qiagen), and cDNA from 2μg of total RNA was synthesized using a SuperScript™ VILO™ cDNA Synthesis Kit (ThermoFisher, cat. no. 11754050).

2.5. Quantitative Real-Time PCR (qRT-PCR)

qRT-PCR (CFX96, Bio-Rad, Vanderbilt University Medical Center MCBR Core) on 50ng/9μL cDNA was run in duplicate using TaqMan™ Fast Universal PCR Master Mix (2X), no AmpErase™ UNG (Life Technologies, cat. no. 4352042) and Life Technologies gene expression assays for human GRM2 (Hs00968358_m1), GRM3 (Hs00932301_m1) and G6PD (Hs00166169_m1), and mouse Grm2 (Mm01235831_m1), Grm3 (Mm00725298_m1) and Gapdh (Mm99999915_g1). Similar qRT-PCR was performed using PowerUp™ SYBR™ Green Master Mix (ThermoFisher, cat. no. A25742) with the following primers (5’ to 3’): Mecp2 (exon 4, forward: ATGAGACTGTGCTCCCCATC, reverse: TTTTCTCACCAAGGGTGGAC) and Gapdh (exon 6, forward: CGACTTCAACAGCAACTCCC, reverse: GCCGTATTCATTGTCATACCAGG). All primers used for SYBR qRT-PCR were designed using Primer3 and constructed by Sigma through the Vanderbilt University Medical Center MCBR Core. Ct values for each sample were normalized to G6PD/Gapdh expression and analyzed using the delta–delta Ct method as described in (Gogliotti et al., 2017). Values exceeding two times the standard deviation were classified as outliers. Each value was compared to the average delta-Ct value acquired for control human samples or wild-type littermate control mice (WT littermate or Mecp2^+/+) and calculated as percent-relative to the average control delta-Ct.

2.6. Drugs

LY379268 (mGlu_2/3 agonist) and LY341495 (mGlu_2/3 antagonist) were purchased from Tocris (Minneapolis, MN). All drugs used for behavioral experiments were diluted in 10% Tween-80.

2.7. Behavioral Assays

All behavioral experiments were conducted at predicted symptomatic ages (8–12-week-old MECP2^Tg1/o and 20–25-week-old Mecp2^Null/+ mice) and using sex- and age-matched littermate controls (WT littermates or Mecp2^+/+ mice) at the Vanderbilt Mouse Neurobehavioral Lab (MNL) Core. All experiments were preceded by intraperitoneal (i.p.) injections of the following drugs (T_max in parentheses): vehicle (10% Tween-80), LY379268 (1mg/kg, 30 min) or LY341495 (3mg/kg, 20 min pre-LY379268 administration or 30 min). Quantification was performed either by a researcher blinded to the genotype and treatment and/or by automated software.

2.7.1. Open Field

Mice were placed in the activity chamber for 30 min and locomotor activity was quantified as beam breaks in the X, Y and Z axis using Activity Monitor software (Med Associates Inc).

2.7.2. Trace Fear Conditioning

Mice were habituated to the room for 1 hour before all tests. On acquisition or conditioning day, mice were treated with compounds or vehicle prior to being placed into an operant chamber with a shock grid (Med Associates Inc.) in the presence of a 10% vanilla odor cue. Modified from previous studies (Dogra et al., 2021; Xu et al., 2014), mice were acclimated for 1 min and exposed to a mild 1 sec foot shock (0.5 mA for WT littermate and MECP2^Tg1/o mice, and 0.7mA for Mecp2^+/+ and Mecp2^Null/+ animals) that was preceded by a 15 sec tone. A precise 30 sec interval or “trace” separated the tone and shock. Three or four tone-trace-shock pairings were applied, 240 sec apart, for the MECP2^Tg1/o or Mecp2^Null/+ mouse lines, respectively. Percentage of time spent freezing during each trace was measured by Video Freeze software (Med Associates Inc.).

2.8. Statistical Analyses

Statistics were carried out using Prism 9 (GraphPad) and Excel (Microsoft). All data shown represent mean ± SEM. Statistical significance between genotypes was determined using Student’s t-test, or 2-way ANOVA with Sidak’s or Tukey’s post-hoc. Sample size (denoted as “n”), statistical test and results of statistical analyses are specified in each figure legend.

3. Results

3.1. Group II mGlu receptor expression is decreased in clinical and preclinical RTT samples

We first investigated the expression of mGlu₂ and mGlu₃ in post-mortem brain tissue from female patients clinically diagnosed with RTT. Complementing our previous studies in the motor cortex and cerebellum (Gogliotti et al., 2018, 2017, 2016), we obtained fourteen temporal cortex samples (Brodmann area 20 or 38, BA 20 or BA 38) from female RTT patients characterized as having truncation mutations in MECP2 (R168X, R255X, and R270X) and fourteen age-, sex- and post-mortem interval (PMI)-matched controls (Supplementary Table 1). qRT-PCR analyses revealed decreased levels of GRM2 and GRM3 mRNA in the temporal cortex of RTT patients (Figure 1A), which is in agreement with previous transcriptomic analyses of other brain regions, including the cerebellum and frontal, motor and temporal cortices (Ben-Shachar et al., 2009; Gogliotti et al., 2018; Lin et al., 2016).

Figure 1: mGlu₂ and mGlu₃ expression is decreased in RTT patients and mice.

(A) Compared to controls (black bars / white circles, n=11–12), GRM2 and GRM3 mRNA are both decreased in RTT patients in temporal cortex autopsy samples (red bars / white squares, n=14) (GRM2: t(23)=3.144, p=0.0045; GRM3: t(24)=2.196, p=0.0380). RTT patient samples used in this study have truncated MeCP2 mutations, specifically R168X, R255X and R270X (see Supplementary Table 1). (B-C) Compared to littermate controls, Mecp2^+/+ (white bars / black circles, n=4–5), Grm2 and/or Grm3 mRNA levels are decreased in 20–25-week-old Mecp2^Null/+ mice (red bars / squares, n=4–5) (cortex: Grm2: t(7)=2.897, p=0.0231; Grm3: t(7)=2.104, p=0.0734; hippocampus: Grm2: t(7)=2.599, p=0.0355; Grm3: t(7)=2.460, p=0.0435). (D) Representative immunoblots illustrating cortical dimeric (“D”, 200 kDa) form of mGlu₂ or mGlu₃ and Gapdh (37 kDa) loading control in Mecp2^+/+ (“WT”) and Mecp2^Null/+ (“Null/+”) animals. (E) Cortical mGlu₂ protein expression is significantly decreased in Mecp2^Null/+ mice (n=4–5) relative to Mecp2^+/+ animals (n=5–6) (mGlu₂: t(7)=2.715, p=0.0300; mGlu₃: t(9)=2.131, p=0.0619). (F) Representative immunoblots for hippocampal proteins as in (D) with the addition of the mGlu₂ or mGlu₃ monomeric protein (“M”, 100 kDa) and vGlut₂ (65 kDa). (G) Compared to Mecp2^+/+ animals (n=5–6), mGlu₂ and mGlu₃ proteins (total = monomer + dimer) are reduced in the hippocampus of Mecp2^Null/+ mice (n=5–6) (mGlu₂: Monomer: t(10)=2.109, p=0.0611; Dimer: t(10)=2.285, p=0.0454; Total: t(10)=2.250, p=0.0482; mGlu₃: Monomer: t(10)=2.108, p=0.0612; Dimer: t(10)=2.798, p=0.0188; Total: t(10)=2.811, p=0.0185). vGlut₂ is unchanged between genotypes (t(10)=0.04986, p=0.9612). Student’s t-test. ns (not significant), *p<0.05, **p<0.01.

Decreased expression of mGlu₂ and mGlu₃ mRNA has also been observed in preclinical RTT mouse models, particularly in the cortex of male Mecp2 null mice (Bedogni et al., 2016; Chahrour et al., 2008; Pacheco et al., 2017). Consistent with these transcriptomic studies, we found significantly reduced Grm2 mRNA expression in the cortex of naïve 20–25-week-old female Mecp2^Null/+ animals compared to the wild-type (WT) littermate controls, Mecp2^+/+ (Figure 1B). Interestingly, cortical Grm3 mRNA was not statistically different between the mutant and control mice. Given that mGlu₂ and mGlu₃ have both been linked to cognitive function, specifically hippocampal-dependent cognition (De Filippis et al., 2015; Lyon et al., 2011), we assessed Grm2 and Grm3 transcript levels in the hippocampus, and found that hippocampal Grm2 and Grm3 mRNA levels were also decreased in Mecp2^Null/+ animals (Figure 1C).

Next, we determined whether altered expression of mGlu₂ and mGlu₃ in Mecp2^Null/+ mice, which express ~50% less MeCP2 than littermate controls Mecp2^+/+ (Supplementary Figure 1), was maintained at the protein level. As illustrated in the representative immunoblots, only mGlu₂ was significantly decreased in the cortex of Mecp2^Null/+ animals compared to their WT counterparts, which is consistent with the transcript expression (Figure 1D-E, antibody validation in Supplementary Figure 2; mGlu₃ global knockout animals have been previously characterized in (Dogra et al., 2021); monomer level in the cortex of both mGlu₂ and mGlu₃ was too low to be accurately quantified). Also in agreement with the transcript expression data was the observed decreased expression of mGlu₂ and mGlu₃ protein in the hippocampus of Mecp2^Null/+ mice (Figure 1F-G). Notably, for both receptors, the dimer and total (sum of monomer and dimer) protein levels were significantly reduced in Mecp2^Null/+ animals without differences in the control synaptic protein vGlut2 (Figure 1G). These clinical and preclinical data provided rationale for investigating the role and therapeutic potential of group II mGlu receptors in RTT and related disorders.

3.2. Group II mGlu receptor expression is increased in MECP2^Tg1/o mice

We further explored the molecular relationship of mGlu₂ and mGlu₃ with MeCP2 using a rodent MDS model, MECP2^Tg1/o_, and tested the hypothesis that increased MeCP2 expression in the MECP2^Tg1/o context also enhances mGlu₂ and mGlu₃ expression. Similar to the use of symptomatic female Mecp2^Null/+ animals to reflect the clinical population, we utilized 8–9-week-old male MECP2^Tg1/o mice and WT littermates to assess protein and mRNA expression in the hippocampus. In agreement with previous reports (Collins et al., 2004; Fisher et al., 2018; Na et al., 2012; Sztainberg et al., 2015), MeCP2 mRNA and protein expression were increased in MECP2^Tg1/o mice compared to WT littermates (Figure 2A-C). Comparably, Grm2 and Grm3 mRNA levels were also significantly increased in MECP2^Tg1/o animals (Figure 2A). To determine if this effect was also observed at the protein level, we evaluated mGlu₂ and mGlu₃ protein expression. As illustrated in the representative immunoblots, expression of both receptors was increased in MECP2^Tg1/o animals (Figure 2B). In Figure 2C, quantification of expression showed that statistical significance was observed for the expression of the monomer, dimer and total (sum of monomer and dimer) protein for mGlu₂, and the level of the dimer and total mGlu₃ protein was significantly different between MECP2^Tg1/o mice and WT littermates. These overexpression data of group II mGlu receptors suggested that these proteins could potentially contribute to the abnormal phenotypes in MDS mice, which could be amenable to receptor modulation.

Figure 2: Hippocampal mGlu₂ and mGlu₃ expression is increased in MDS mice.

(A) Compared to WT littermates (white bars / black circles, n=5–7), Mecp2/MECP2, Grm2 and Grm3 transcripts are increased in the hippocampus of 8–9-week-old MECP2^Tg1/o mice (blue bars / squares, n=4–5) (Mecp2/MECP2: t(10)=11.89, p<0.0001; Grm2: t(8)=2.476, p=0.0384; Grm3: t(7)=2.586, p=0.0362). (B) Representative immunoblots illustrating MeCP2 (72 kDa), mGlu₂ or mGlu₃ dimer (“D”, 200 kDa) and monomer (“M”, 100 kDa), vGlut2 (65 kDa), and Gapdh (37 kDa) loading control in hippocampal samples of WT littermates and MECP2^Tg1/o mice. (C) Protein expression of MeCP2, mGlu₂, and mGlu₃ (total = monomer + dimer) is increased in MECP2^Tg1/o mice (n=5–6) relative to WT littermates (n=6–7) (MeCP2: t(11)=2.269, p=0.0444; mGlu₂: Monomer: t(10)=4.669, p=0.0009; Dimer: t(9)=2.642, p=0.0268; Total: t(9)=4.028, p=0.0030; mGlu₃: Monomer: t(11)=4.140, p=0.0016; Dimer: t(11)=1.701, p=0.1169; Total: t(11)=2.597, p=0.0248). vGlut₂ is unchanged between genotypes (t(11)=1.433, p=0.1797). Student’s t-test. ns (not significant), *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

3.3. RTT and MDS mice exhibit abnormal and bidirectional phenotypes in trace fear acquisition

Numerous studies have characterized RTT model mice as exhibiting abnormal phenotypes in cognition, specifically in associative learning and memory, which is often assessed using a cued or contextual fear conditioning assay (Collins et al., 2021; Gogliotti et al., 2017, 2016; Lamonica et al., 2017; Merritt et al., 2020; Moretti et al., 2006; Pelka et al., 2006; Pitcher et al., 2015; Samaco et al., 2013; Schaevitz et al., 2013; Stearns et al., 2007; Vermudez et al., 2021). Interestingly, a recent study highlighted the role of mGlu₃ in mediating a temporal- and hippocampal-dependent associative learning and memory phenotype known as trace fear acquisition or conditioning (Dogra et al., 2021). Prior to investigating the relationship of mGlu_2/3 receptors and trace fear conditioning in a RTT mouse model, we first characterized the trace fear behavior of Mecp2^Null/+ animals. In this task, the mice were trained to associate their environment and presented tone with an aversive stimulus in the form of a mild foot shock (0.7 mA); the “trace” period that separates the tone and shock presentations is thought to enhance learning (Figure 3A). As shown in Figure 3B, increases in percent freezing in Mecp2^+/+ animals paralleled the increase in tone-trace-shock pairings (black symbols), reflecting learning behavior. However, we demonstrate here that Mecp2^Null/+ mice exhibit abnormal trace fear acquisition, as illustrated by the attenuated percent freezing compared to their Mecp2^+/+ counterparts. In particular, freezing behavior was significantly different between Mecp2^+/+ and Mecp2^Null/+ animals at the last trace period (T4), suggesting a deficit in trace fear acquisition.

Figure 3: Contrasting phenotypes in trace fear conditioning between Mecp2^Null/+ and MECP2^Tg1/o animals.

(A) Diagram illustrating the trace fear conditioning paradigm. Mice are trained to associate their environment (context and tone) with an aversive stimulus in the form of a mild foot shock (0.7 mA), with a temporal component (trace) separating the tone and shock. Four tone-trace-shock pairings are presented and percent freezing during the trace period is measured as a proxy for fear learning behavior. (B) Compared to littermate controls, Mecp2^+/+ (black circles, n=17), 20–25-week-old Mecp2^Null/+ animals (red squares, n=12) display attenuated percent freezing during the fourth trace period (T4) (F(4,108)=4.276; p=0.0016). (C) Diagram illustrating the trace fear conditioning paradigm in WT littermates and MECP2^Tg1/o animals. The protocol mentioned above in RTT mice is followed with notable differences in the foot shock intensity (0.5 mA) and the number of tone-trace-shock pairings (three). (D) Compared to WT littermates (black circles, n=11), 8–12-week-old MECP2^Tg1/o animals (blue squares, n=10) exhibit increased percent freezing during the third trace period (T3) (F(3,57)=35.14; p<0.0001). 2-way ANOVA with Sidak’s post-hoc test. **p<0.01, ****p<0.0001.

Similar to RTT model mice, MDS mice have been shown to exhibit abnormal phenotypes in cued or contextual fear associative learning and memory (Collins et al., 2004; Fisher et al., 2018; Na et al., 2014, 2012; Stansley et al., 2018). However, as in RTT model animals, phenotypes in trace fear conditioning have not been assessed in mice modeling MDS. WT littermate and MECP2^Tg1/o animals were subjected to a similar trace fear acquisition paradigm as RTT mice, with the notable differences of a milder foot shock intensity (0.5 mA) and reduced number of tone-trace-shock pairings (Figure 3C). These changes were performed to account for the high percent freezing phenotype that MDS mice exhibit in fear conditioning assays (Collins et al., 2004; Fisher et al., 2018; Na et al., 2014, 2012; Stansley et al., 2018). Compared to WT littermates, MECP2^Tg1/o mice displayed enhanced trace fear acquisition during the last trace period measured (T3, Figure 3D). This abnormal enhancement in trace fear acquisition was in contrast to the deficient phenotype in Mecp2^Null/+ animals; coupled with the increases in expression of mGlu₂ and mGlu₃ in the hippocampus of MECP2^Tg1/o mice, we hypothesized that these changes in trace fear behavior might be sensitive to mGlu_2/3 receptor modulation.

3.4. mGlu_2/3 activation reverses deficits in trace fear acquisition in Mecp2^Null/+ mice

Detection and quantitation of a trace fear acquisition deficit in RTT mice allowed us to then test the hypothesis that administration of a nonselective mGlu_2/3 agonist, LY379268, to increase receptor activity would improve this abnormal cognitive phenotype in Mecp2^Null/+ animals. Animals were intraperitoneally administered with either vehicle (10% Tween-80) or 1mg/kg LY379268 30 minutes prior to trace fear conditioning (Figure 4A). LY379268 treatment reversed the trace fear acquisition deficit in vehicle-treated Mecp2^Null/+ mice with statistical significance observed between vehicle- and LY379268-treated Mecp2^Null/+ mice at trace periods 2 (T2) and 4 (T4) (Figure 4B-D). The increased freezing behavior at period T2 was only observed in Mecp2^Null/+ animals (Figure 4C), suggesting an increased sensitivity to LY379268 in the mutant mice compared to their WT counterparts. In contrast, Mecp2^+/+ animals did not exhibit a significant change in response to LY379268 at any of the trace periods measured. To eliminate the possibility that the increased percent freezing in Mecp2^Null/+ animals was due to hypolocomotive effects of LY379268, as has previously been shown in WT animals (Imre, 2007; Imre et al., 2006; Woolley et al., 2008), we placed the mice in an open field chamber and monitored spontaneous locomotor activity. As illustrated in Supplementary Figure 2, 1mg/kg LY379268 did not change the distance traveled of Mecp2^+/+ or Mecp2^Null/+ animals, which already have reduced locomotion at baseline.

Figure 4: Attenuated trace fear acquisition in Mecp2^Null/+ mice is improved with group II mGlu receptor activation.

(A) Diagram illustrating timeline of drug administration prior to trace fear conditioning. Mice are treated intraperitoneally (i.p.) with vehicle (10% Tween-80) or 1mg/kg LY379268 (mGlu_2/3 agonist) 30 minutes prior to placement in the fear conditioning box. (B-D) Treatment with LY379268 increases trace fear acquisition in Mecp2^Null/+ mice (red bars, vehicle: closed red squares, n=15; LY379268: open red squares, n=7). Freezing behavior in littermate controls, Mecp2^+/+(black bars) is not different between the treatment groups, vehicle (closed black circles, n=15) and LY379268 (open black circles, n=8). Analysis of trace periods T2 and T4 (dark grey bars in B) are shown in (C) and (D), respectively. (F(12,148)=3.314; ***p=0.0001 (T2, LY379268: Mecp2^+/+ vs Mecp2^Null/+), ^####p<0.0001 (T2, Mecp2^Null/+: Vehicle vs LY379268), **p=0.0034 (T4, Vehicle: Mecp2^+/+ vs Mecp2^Null/+), ^###p=0.0005 (T4, Mecp2^Null/+: Vehicle vs LY379268)). (E) Diagram illustrating timeline of administration of LY341495 (mGlu_2/3 antagonist) and LY379268 prior to trace fear conditioning. LY341495 (3mg/kg, i.p.) is administered 20 minutes prior to LY379268 (1mg/kg, i.p.). (F-I) Co-administration of LY341495 and LY379268 blocked the reversal effect of LY379268 in Mecp2^Null/+ animals (red bars, vehicle: closed red squares, n=12; LY341495 + LY379268: pink squares, n=6) at trace periods T2-T4. Percent freezing in Mecp2^+/+ mice (black bars) is decreased with co-administration of LY341495 and LY379268, particularly at trace periods T3-T4 (vehicle: closed black circles, n=17; LY341495 + LY379268: grey circles, n=6). Analysis of trace periods T2-T4 (dark grey) is shown in (G-I). (F(20,200)=3.423; ^###p=0.0005 (T2, Mecp2^Null/+: Vehicle vs LY379268), ^####p<0.0001 (T2, Mecp2^Null/+: LY37968 vs LY379268 + LY341495), ^##p=0.0044 (T3, Mecp2^+/+: Vehicle vs LY379268 + LY341495), ^##p=0.0043 (T3, Mecp2^Null/+: LY379268 vs LY379268 + LY341495), **p=0.0042 (T4, Vehicle: Mecp2^+/+ vs Mecp2^Null/+), ^##p=0.0011 (T4, Mecp2^+/+: Vehicle vs LY379268 + LY341495), ^###p=0.0008 (T4, Mecp2^+/+: LY379268 vs LY379268 + LY341495), ^##p=0.0020 (T4, Mecp2^Null/+: Vehicle vs LY379268), ^###p=0.0010 (T4, Mecp2^Null/+: LY37968 vs LY379268 + LY341495)). 20–25-week-old mice. 2-way ANOVA with Tukey’s post-hoc test. *between-genotypes. ^#within-genotypes. **p<0.01, ***p<0.001, ^##p<0.01, ^###p<0.001, ^####p<0.0001.

To further validate that the reversal effects in trace fear conditioning were due to mGlu_2/3 modulation by LY379268 and not an off-target result, we administered the mGlu_2/3 orthosteric antagonist LY341495 (3mg/kg) prior to LY379268 treatment (Figure 4E). LY341495 blocked the effect of LY379268 in Mecp2^Null/+ mice, specifically at trace periods T2-T4 (Figure 4F-I). In these trace periods, the freezing behavior of vehicle-treated Mecp2^Null/+ animals was indistinguishable from Mecp2^Null/+ mice that received both LY341495 and LY379268. Importantly, the ability of LY341495 to block LY379268’s increased freezing effect at period T2 in Mecp2^Null/+ animals suggests that the left-shifted response of the mutant animals to LY379268 is a consequence of mGlu_2/3 modulation (Figure 4G). Interestingly, co-administration of LY379268 and LY341495 to Mecp2^+/+ animals decreased percent freezing at trace periods T3-T4 (Figure 4H-I). Again, this could potentially be due to alterations in locomotor activity; however, assessment of distance traveled in an open field assay revealed no locomotor effects of LY341495 and LY379268 co-administration in either the Mecp2^+/+ and Mecp2^+/− mice (Supplementary Figure 3). Overall, these data support the ability of mGlu_2/3 activation to enhance the trace fear acquisition response in Mecp2^Null/+ animals.

3.5. mGlu_2/3 antagonism normalizes the abnormal trace fear acquisition phenotype in MECP2^Tg1/o mice

Given the beneficial effects of activating mGlu_2/3 in RTT mice, which display opposite trace fear conditioning phenotypes from MDS mice, we then posited the converse hypothesis: that inhibiting mGlu_2/3 could alleviate the abnormal increase in trace fear behavior in MECP2^Tg/o animals. Prior to trace fear conditioning, animals were intraperitoneally administered either vehicle (10% Tween-80) or the nonselective orthosteric mGlu_2/3 antagonist LY341495 (3mg/kg) (Figure 5A). LY341495 treatment did not affect trace fear behavior in WT littermates (Figure 5B-C). However, MECP2^Tg1/o animals administered LY341495 exhibited significantly reduced percent freezing at the last tone-trace-shock pairing (T3) compared to vehicle-treated MECP2^Tg1/o mice (Figure 5C). Notably, LY341495-treated MECP2^Tg1/o mice displayed comparable freezing behavior to vehicle-treated WT counterparts. To rule out potential locomotor effects of mGlu_2/3 inhibition with the nonselective antagonist, we placed vehicle- and LY341495-treated animals in an open field chamber. At baseline, MECP2^Tg1/o mice presented with hypolocomotor activity compared to WT littermates (Supplementary Figure 4). Importantly, regardless of genotype, LY341495 administration did not affect distance traveled in the open field task, which is in agreement with previous studies using systemic administration of LY341495 (Fukumoto et al., 2016; Gleason et al., 2013). Overall, these data suggest that mGlu_2/3 antagonism has positive effects in normalizing enhanced trace fear behavior in MECP2^Tg1/o animals.

Figure 5: Group II mGlu receptor antagonism normalizes enhanced trace fear acquisition in MECP2^Tg1/o mice.

(A) Diagram illustrating the timeline of LY341495 administration prior to trace fear conditioning. Mice are treated intraperitoneally (i.p.) with vehicle (10% Tween-80) or 3mg/kg LY341495 (mGlu_2/3 antagonist) 35 minutes prior to placement in the fear conditioning box. (B-C) LY341495 treatment significantly decreases the enhanced trace fear acquisition phenotype in MECP2^Tg1/o mice (blue bars, vehicle: closed blue squares, n=11; LY341495: open blue squares, n=5) but not in WT littermates (white bars, vehicle: closed black circles, n=10; LY341495: open black circles, n=7). Analysis of trace period T3 (dark grey) is shown in (C). (F(9,87)=16.87; ****p<0.0001 (Vehicle: WT vs MECP2^Tg1/o), ***p=0.0008 (LY341495: WT vs MECP2^Tg1/o), ^####p<0.0001 (MECP2^Tg1/o: Vehicle vs LY341495)). 8–12-week-old mice. 2-way ANOVA with Tukey’s post-hoc test. *between-genotypes. ^#within-genotypes. ***p<0.001, ****p<0.0001, ^####p<0.0001.

4. Discussion

Our understanding of MECP2-associated disorders has developed tremendously in the past 20 years. However, despite the monogenicity of these disorders and the link of MeCP2 protein expression and/or function to each disease, no effective treatment is currently available for RTT or MDS. Promisingly, recent studies have demonstrated that genetic manipulations of MECP2 and pharmacological modulations of downstream targets of the protein improve phenotypes in preclinical rodent models of RTT (Bittolo et al., 2016; Degano et al., 2014; Gadalla et al., 2017, 2013; Garg et al., 2013; Gogliotti et al., 2018, 2017, 2016; Li et al., 2017; Luoni et al., 2020; Matagne et al., 2021, 2017; Ogier et al., 2007; Powers et al., 2019; Roux et al., 2007; Scaramuzza et al., 2021; Sinnett et al., 2017; Tillotson et al., 2017; Zanella et al., 2008) and MDS (Koerner et al., 2018; Na et al., 2014; Sztainberg et al., 2015). In this study, we aimed to capitalize on previous reports by investigating two potential therapeutic targets, mGlu₂ and mGlu₃, and employing pharmacological modulators of these receptors to evaluate effects in RTT (Mecp2^Null/+) and MDS (MECP2^Tg1/o) mouse models.

mGlu₂ and mGlu₃ are group II mGlu receptors that have been robustly implicated in neuropsychiatric disorders (Chaki, 2017; Joffe and Conn, 2019; Maksymetz et al., 2017; Swanson et al., 2005; Walker and Conn, 2015). Their role in neurodevelopmental disorders is limited; however, mGlu₃ has been shown to be involved in the cognitive phenotypes of Fragile X syndrome (FXS) (Choi et al. 2011) and mGlu_2/3 receptors have been implicated in autism-like phenotypes in a rat model of autism (Chen et al. 2014). Moreover, expression studies demonstrate that both mGlu₂ and mGlu₃ are decreased in male Mecp2^Null/y mice and patients diagnosed with RTT (Bedogni et al. 2016; Chahrour et al. 2008; Pacheco et al. 2017; Ben-Shachar et al. 2009; Gogliotti et al. 2018; Lin et al. 2016). Here, we further support these data by showing attenuated mGlu₂ and mGlu₃ expression in the temporal cortex of RTT autopsy samples from patients with truncating mutations in MeCP2. Altered levels of these receptors are conserved in a female RTT mouse model, Mecp2^Null/+, specifically in the cortex and hippocampus. Since RTT and MDS are due to loss-of-function mutations and multiple copies of MECP2, respectively, and given the role of MeCP2 as a transcriptional activator (Chahrour et al. 2008), we posited that mGlu₂ and mGlu₃ expression would be increased in MDS. In the hippocampus of MECP2^Tg1/o mice, we confirmed previous studies showing increased MeCP2 expression (Collins et al. 2004; Fisher et al. 2018; Na et al. 2012) and also observed increased mGlu₂ and mGlu₃ mRNA and protein expression. These findings suggest that group II mGlu receptors may be targets of MeCP2 transcriptional regulation, whether through direct or indirect mechanisms.

These molecular data provided our rationale to further explore mGlu₂ and mGlu₃ in the behavioral phenotypes of RTT and MDS, focusing on the learning and memory phenotypes given the role of group II mGlu receptors in cognition. Numerous studies have established that RTT and MDS mice have contrasting characteristics in contextual fear conditioning, a hippocampal-dependent learning and memory paradigm (Na et al. 2012; Na et al. 2014; Collins et al. 2004; Fisher et al. 2018; Gogliotti et al. 2017; Stansley et al. 2018; Moretti et al. 2006; Samaco et al. 2012; Stearns et al. 2007). In particular, RTT mice exhibit deficits in contextual fear conditioning, whereas MDS mice display an abnormal enhancement of this behavior. Several studies have described a similar hippocampal-dependent cognitive task that relies on both the spatial and temporal factors of learning and memory called trace fear conditioning (Huerta et al. 2000; Xu et al. 2014; Zhao et al. 2005; McEchron et al. 1998). Excitingly, our group has demonstrated a specific role for mGlu₃ in trace fear acquisition (Dogra et al. 2021). Before exploring the relationship of mGlu_2/3 receptors and trace fear conditioning in mouse models of MECP2-associated disorders, we first characterized the behavior of both Mecp2^Null/+ and MECP2^Tg1/o animals in this cognitive task, which has not yet been investigated. Here, we extend previous studies in cognitive domains to show that Mecp2^Null/+ mice exhibit a deficit in trace fear acquisition. Conversely, MECP2^Tg1/o animals have an abnormal enhanced trace fear acquisition phenotype, which supports the bidirectional behavioral phenotypes of cognition in mouse models of these two disorders.

These data prompted us to test our hypothesis that mGlu_2/3 modulation could improve abnormal cognitive phenotypes in Mecp2^Null/+ and MECP2^Tg1/o mice using the trace fear paradigm. Previous preclinical studies have shown that mGlu₃ activation and mGlu₂ positive allosteric modulators improve deficits in cognitive tasks (Griebel et al. 2016; Nikiforuk et al. 2010; Walker et al. 2017). Given that both receptors are decreased in Mecp2^Null/+ mice, it is likely that the function of both mGlu₂ and mGlu₃ is also disrupted. Therefore, we activated both receptors using a nonselective mGlu_2/3 agonist, LY379268. Excitingly, in Mecp2^Null/+ animals, LY379268 rescued deficiencies in trace fear acquisition. Interestingly, we observed a left-shifted response to LY379268 that was specific to Mecp2^Null/+ mice. This could be attributed to a ceiling effect of the trace fear paradigm, for example the strong shock intensity, on the baseline trace fear behavior of Mecp2^+/+ animals, which has also been previously observed in WT animals (Dogra et al. 2021), or to a potential sensitivity of the receptors to agonist in the Mecp2^Null/+ context. Importantly, LY379268’s effect in augmenting freezing behavior in Mecp2^Null/+ was completely blocked by the mGlu_2/3 antagonist, LY341495, suggesting that LY379268 mediates its effects via mGlu₂ and mGlu₃ in the trace fear acquisition paradigm in RTT mice.

In contrast to the effects seen in Mecp2^Null/+ animals, inhibition of mGlu_2/3 receptors with LY341495 reversed the abnormally enhanced trace fear acquisition of MECP2^Tg1/o mice. The lack of LY341495’s effect in WT animals suggests that the heightened trace fear phenotype in MECP2^Tg1/o mice is driven, in part, by the increased expression of the mGlu₂ and/or mGlu₃ receptors. The bidirectionality in mGlu_2/3 modulation to alleviate the abnormal behavioral cognition in Mecp2^Null/+ and MECP2^Tg1/o animals parallels the contrasting trace fear phenotypes. These together may indicate that, in both the RTT and MDS contexts where both receptors are affected, one or both of the group II mGlu receptors are needed to mediate this form of associative learning and memory.

In MECP2^Tg1/o and Mecp2^Null/+ mice, the functional consequences of altered mGlu₂ and mGlu₃ receptor expression, such as in impacting long-term synaptic plasticity in the hippocampal Schaffer collateral-CA1 (SC-CA1) synapse, remains to be elucidated. Previous reports have demonstrated that SC-CA1 long-term potentiation (LTP) is bidirectionally altered in Mecp2 mutant and MECP2^Tg1/o animals (Asaka et al. 2006; Collins et al. 2004; Guy et al. 2007; Moretti et al. 2006; Sztainberg et al. 2015). Additionally, in Mecp2 mutant mice, NMDA antagonism has been shown to reverse LTP deficits (Weng et al. 2011). Although no studies have yet shown whether NMDA function is implicated in MDS, NMDA dysregulation has been associated with the enhanced LTP phenotype of mice modeling a related disorder, Pitt-Hopkins syndrome (Kennedy et al. 2016; Thaxton et al. 2018). Coupling these data with previous evidence illustrating that mGlu_2/3 receptors play a role in NMDA-mediated LTP at the SC-CA1 synapses (Rosenberg et al. 2016), it is therefore possible that mGlu_2/3 modulation may correct abnormal SC-CA1 LTP in RTT and MDS animals. However, empirical studies are needed to test this theory.

In conclusion, we have demonstrated that the expression of mGlu₂ and mGlu₃ is altered in RTT clinical samples and bidirectionally affected in Mecp2^Null/+ and MECP2^Tg1/o animals, indicating a potential role of group II mGlu receptors in RTT and MDS phenotypes. Correspondingly, we establish that Mecp2^Null/+ and MECP2^Tg1/o mice present abnormal and contrasting phenotypes in trace fear conditioning, a temporal-dependent form of associative learning and memory. Pharmacological mGlu_2/3 activation or antagonism exerted efficacy in modulating trace fear behavior of Mecp2^Null/+ and MECP2^Tg1/o mice, respectively. Altogether, our data provide novel evidence that mGlu₂ and mGlu₃ are implicated in neurodevelopmental disorders, particularly in RTT and MDS.

Highlights.

• mGlu₂/mGlu₃ receptors are decreased in Rett syndrome patients and mice, Mecp2^Null/+

• mGlu₂/mGlu₃ receptors are increased in MECP2 Duplication syndrome mice, MECP2^Tg1/o

• mGlu_2/3 activation rescues trace fear acquisition deficits in Mecp2^Null/+ mice

• mGlu_2/3 inhibition reverses heightened trace fear acquisition in MECP2^Tg1/o mice