DMSXL mice exhibit altered spatial and/or temporal distribution of glutamate and GABA neurotransmitters in the hippocampus, a brain region with abundant RNA foci. Using microdialysis, glutamate uptake was significantly reduced in the whole DMSXL hippocampus, relative to WT controls. However, this sampling technique failed to detect higher levels of extracellular glutamate in the hippocampus, unlike the frontal cortex of DMSXL mice, where we previously reported increased levels of extracellular glutamate in association with GLT1 downregulation ^[27][30]. Despite the unchanged global glutamate levels found in the DMSXL hippocampus by microdialysis, it is conceivable that the spatial resolution of this sampling technique is not sufficient to target specific subregions of the hippocampus (DG vs. CA1). In addition, individual synaptic compartments or membrane microdomains in this complex and heterogeneous brain region cannot be sampled by this approach ^[28][34]. To overcome this limitation, we used ex vivo patch clamp techniques on hippocampal slices to specifically investigate events on the membrane of CA1 and DG neurons. We found an increase in tonically activated NMDARs in CA1 and DG, likely due to an elevation of synaptic and extrasynaptic glutamate in both hippocampal areas of DMSXL mice. Interestingly, the increase in glutamate tonic currents detected by patch clamp in the DMSXL hippocampus is similar to the findings previously reported in aged rats ^[29][35]. Increased glutamate tonic currents could not be accounted for by higher spontaneous release of this neurotransmitter, which remained unchanged in both hippocampal structures, suggesting that the mechanisms of synaptic glutamate spontaneous release are relatively unaffected in DMSXL mice. It is estimated that 95% of extracellular glutamate reuptake relies on the glial glutamate transporters GLT1/EAAT2 ^[29][35]. Hence, our data suggest that the DMSXL increase in tonic current can be attributed to an elevation of extrasynaptic glutamate levels, which results from the ~50% downregulation of the astrocyte-specific GLT1 glutamate transporter and reduced glutamate uptake. It appears, however, that the alterations in glutamate levels are spatially confined to the extrasynaptic space and insufficient to have an impact on the global extracellular glutamate content measured by microdialysis. Importantly, higher ambient glutamate levels are in line with the neuronal hyperexcitability previously reported in DMSXL mice [23] and the increased susceptibility of this line to seizures induced by GABA antagonists [26].

Interestingly, in vivo microdialysis revealed an unexpected elevation of extracellular GABA in the hippocampus of DMSXL mice, which was further corroborated by increased tonic GABA currents in CA1 and DG. In contrast to glutamate, higher GABA tonic currents were accompanied by a higher frequency of spontaneous GABA release. In the CNS, the GABAergic system acts in close synergy with the glutamatergic system to control the excitation/inhibition balance. The increase in GABA in DMSXL mice represents an attempt to compensate for glutamate elevation resulting from the downregulation of GTL1 and lower glutamate uptake. However, in contrast with WT mice, extracellular levels of GABA and glutamate did not correlate in DMSXL mice, suggesting a selective presynaptic impact of the DM1 mutation on GABAergic interneurons and a perturbation of the fine equilibrium between neuronal excitation and inhibition, key for proper brain function. Interestingly, we previously reported enhanced basal neurosecretion in cell culture models of DM1, in the absence of stimulation, in association with the abnormal expression and phosphorylation of key synaptic proteins [20]. It is still unclear whether these molecular abnormalities contribute to changes in nonevoked GABA release and how they would specifically affect this neurotransmitter. 

3. Neurotransmission and Synaptic Plasticity in DMSXL Mice

Electrophysiological analysis of synaptic plasticity in the CA1 and DG hippocampal areas of DMSXL mice was performed to gain insight into the mechanisms of DM1 brain disease. We first demonstrated that PPF and PPD ratios were significantly higher in CA1 and DG, respectively. The paired-pulse ratio is considered a form of short-term plasticity sustained by different mechanisms ^[30][36]. Facilitation is prominent at synapses with a low probability of release, meaning that the first stimulus of the pair will not lead to a complete emptying of the readily releasable pool (RRP) of neurotransmitters. When a second stimulus follows shortly (between 20 and 50 ms) after the first, an additional release allows for the increase in the second response due to residual calcium in the presynaptic bouton. The PPF ratio of fEPSPs measured in DMSXL mice was increased, indicating a decrease in the release probability of excitatory glutamate during the first fEPSP. This may reflect changes in the mechanisms underlying PPF, such as a modification of quantal content, or a change in residual presynaptic calcium. 

The glutamate transporter inhibitor TBOA induced a depression of fEPSPs in both mice genotypes, probably due to extrasynaptic NMDAR activation and glutamate spillage, but this chemical depression was stronger in DMSXL mice, in agreement with the altered glutamate clearance. In addition, the time course of LTP was significantly perturbed in DMSXL CA1 and DG, particularly in the short time window that followed tetanic stimulation, suggesting that synapses are less prone to short-term plasticity. As induction of LTP is dependent on optimal synaptic NMDAR activation, we studied pharmacologically isolated NMDA-dependent fEPSPs. No alteration of these responses was observed, suggesting that LTP deficits were not due to a modification in NMDAR activation, as suggested, for instance, in the aging hippocampus ^[31][32][37,38]. A shift in the LTP/LTD threshold in favor of LTD, often associated with a loss of synapses, could also explain the reduced LTP ^[33][39]. However, we found no evidence for an increase in electrically induced LTD in DMSXL mice. The electrophysiological phenotypes of young DMSXL mice studied here differ from the milder defects reported in older animals at 4–7 months of age [20], likely due to the reduction of transgene expression with age ^[34][40] and the high mortality of the most severely affected animals during the first month [21].

LTP induction and maintenance require optimal extracellular glutamate concentration, which is secured by glutamate transporters, such as GLT1 ^[35][41]. Homozygous GLT1 knockout mice exhibit impaired LTP, which is overcome by low concentrations of an NMDAR antagonist ^[36][33]. Given the GLT1 downregulation in DMSXL mice, we also tested low doses of D-AP5 on the LTP. However, light NMDAR blockade did not change the time course of LTP in DMSXL animals, suggesting that synaptic NMDARs were relatively spared in these mice, and that the elevation of ambient glutamate was not sufficient to activate extrasynaptic NMDARs. Finally, it has been shown that abnormal spine morphology is associated with a decrease in LTP in mouse models of Alzheimer’s disease (APPxPS1-KI mice) ^[37][42] and schizophrenia (Schnurri-2 knockout mice) ^[38][43]. Incidentally, altered neuronal morphogenesis was reported in mouse and cell models of DM1 ^[39][40][41][44,45,46], together with prevalent defects in transcripts and proteins associated with cytoskeleton and cell morphogenesis ^[42][43][44][47,48,49]. Hence, a detailed morphological analysis of dendritic spines should help elucidate the contribution of ultrastructural neuronal abnormalities to synaptic plasticity and presynaptic function (such as PPF) in DMSXL mice.

4. RNA Spliceopathy in DMSXL Mouse Brains and Possible Effect on Altered Synaptic Dysfunction

A sustained reduction in LTP was previously reported in a conditional DM1 mouse model expressing high levels of interrupted CUG RNA repeats in the brain ^[45][50], and in Mbnl2 knockouts, in association with altered NMDAR activity and modest mis-splicing of exon 5 of Grin1 (glutamate ionotropic receptor NMDA-type subunit, also named Nmdar1) [26], which regulates receptor potentiation ^[46][51]. The higher expression of the interrupted transgene in the conditional mice and the total absence of the MBNL2 protein in the knockout line may explain the more pronounced phenotypes of these DM1 models when compared with our mice. DMSXL mice exhibit normal splicing of the regulatory Grin1/Nmdar1 exon 5 in CA and DG, but reduced inclusion of exon 22 in DG. Although Grin1/Nmdar1 exon 22 appears to participate in protein localization to the membrane and in the interaction with other regulatory protein partners ^[46][51], NMDAR activity does not appear to be abnormally increased in DMSXL, in contrast with homozygous Glt1 knockout mice ^[36][33]. The functional consequences of exon 22 mis-splicing requires further investigation.

Among the genes mis-spliced in the DMSXL hippocampus, many are involved in the control of synaptic function. Cacna1d/Cav1.3 (a calcium channel) and Rhot1 (a mitochondrial GTPase) regulate calcium homeostasis, an essential aspect behind neuronal activity ^[47][48][52,53]. Similarly, Kcnd3/Kv4.3 regulates neuronal excitability and neurotransmitter release through voltage-dependent outward potassium currents ^[49][54]. Presynaptic Add3 and Stx2 facilitate the localization of synaptic vesicles at the active zone ^[50][51][55,56]. In the postsynaptic membrane, Camk2b and Grip1 regulate the intracellular trafficking of AMPA receptors ^[52][53][54][57,58,59], Sorbs1 participates in the clustering of acetylcholine receptors ^[55][60], and Gabrg2 regulates the GABAergic signaling ^[56][61]. Fermt2 encodes a scaffolding protein that regulates axonal growth, PPF, and LTP ^[57][62]. Many of the affected genes (e.g., Add3, Cacna1d, Camk2b, Fermt2, Gabrg2, Grip1, Kcnd3, Sorbs1, and Stx2) are regulated by MBNL proteins, and they are also misregulated in other DM1 mouse models and human tissues ^{[42][58][59][60][61][62]}[47,63,64,65,66,67]. These data corroborate the involvement of the mis-spliced events studied in DM1 neuropathogenesis, particularly in DG, where the spliceopathy appears to be more pronounced, possibly contributing to the more prominent synaptic dysfunction in this area of the hippocampus relative to CA1. Importantly, DG is a brain region where adult neurogenesis takes place to add new neurons to hippocampal circuits ^[63][68]. Defective DG neurogenesis may contribute to brain disease ^[64][69], and its role in DM1 neurodysfunction deserves further attention in future investigations.

5. Conclusion

In conclusion, using a transgenic mouse model of DM1, we found evidence of defective synaptic plasticity and abnormal glutamate and GABA neurotransmission. Neurochemical and electrophysiological defects were associated with localized splicing dysregulation, which was more pronounced in restricted areas of the hippocampus. RNA mis-splicing affected components of the presynaptic membrane (e.g., Stx2), postsynaptic receptors (e.g., Gabrg2, Grin1), and cytoskeleton and scaffold proteins (e.g., Fermt2, Itga6, and Sorbs1). Taken together, the molecular and functional changes in the synaptic environment may represent multiple factors of vulnerability to neuronal dysfunction, which contribute to cognitive impairment and behavioral changes typical of this debilitating disease. Deciphering the relative contribution of misregulated events to the neuropathophysiology of DM1 will require splicing modulation of alternative exons in DMSXL and/or WT mouse brains. The investigation of the physiological consequences of individual mis-splicing events will help identify priority disease targets for therapeutic intervention.