While conditional deletion of Dag1 from pyramidal neurons causes a loss of CCK⁺/CB₁R⁺ IN innervation in the forebrain, this has not been examined in dystroglycanopathy relevant mouse models exhibiting more widespread loss of functional Dystroglycan. We therefore generated five distinct mouse models; three to provide mechanistic insight into Dystroglycan function and two of which model mild dystroglycanopathy (schematized in Fig. 1A). Since complete loss of Dag1 results in early embryonic lethality in mice, we generated forebrain-specific conditional knockouts by crossing Emx1^Crewith Dystroglycan floxed mice (Dag1^F/F), to drive recombination in neuroepithelial cells in the dorsal forebrain beginning at embryonic day 10.5 (E10.5)²⁹. We verified the recombination pattern of Emx1^Crewith the mCherry reporter Rosa26^{Lox-STOP-Lox-H2B:mCherry}. H2B:mCherry signal was present in all excitatory neurons and astrocytes throughout the forebrain (Supplementary Fig. 1A, B, D) but not microglia or interneurons (Supplementary Fig. 1C, E).

To model loss of Dystroglycan glycosylation, Emx1^Crewas crossed with POMT2^F/Fconditional mice to generate POMT2^cKOs. POMT2 (protein O-mannosyltransferase 2) is a glycosyltransferase that functions in a heterocomplex with POMT1 to add O-mannose at the beginning of the Dystroglycan glycan chain³⁰. Without the initial O-mannose, no additional sugar moieties can be added to the glycan chain, resulting in near complete loss of Dystroglycan glycosylation. This is lethal embryonically in a global POMT2 knockout, however Emx1^Cre;POMT2^cKO mice are viable and survive into adulthood^31–33.

In addition to binding extracellular ligands, Dystroglycan binds cytoskeletal proteins and signals through the intracellular tail of its β-subunit. To determine whether the intracellular domain of Dystroglycan is required for synaptic development and/or function, we examined mice in which one copy of Dag1 was deleted, and the other copy lacks the intracellular domain of β-Dag1 (Dag1^cyto/-). These mice develop muscular dystrophy but show normal neuronal migration and axon guidance in regions throughout the central nervous system where Dystroglycan glycosylation is required^26,34,35.

To model mild forms of dystroglycanopathy, we examined mice expressing missense mutations in B4GAT1 (β-1,4-glucuronyltransferase, B4GAT1^M155T) and FKRP (fukutin related protein, FKRP^P448L), two genes required for Dystroglycan glycosylation. B4GAT1^M155T mice were initially identified in a forward genetic screen and develop mild muscular dystrophy and have diminished ligand binding capacity due to reduced Dystroglycan glycosylation³⁶. The FKRP^P448L missense mutation models a mutation found in a patient with dystroglycanopathy³⁷. In mice, the FKRP^P448L mutation leads to reduced glycosylation and mild muscular dystrophy, but no gross brain or eye malformations¹⁸.

We first examined the pattern of Dystroglycan localization in pyramidal neurons in CA1 of hippocampus in each of the five models by immunostaining adult (P30) mice with the IIH6 antibody, which detects the terminal matriglycan repeats on the glycan chain on α-Dag1^38,39. In wild-type (WT) mice, punctate IIH6 immunoreactivity was evident in the somatic and perisomatic compartment of CA1 pyramidal neurons (Fig. 1A). Immunoreactivity was also present in blood vessels, where Dag1 is also expressed^4,40. Neuronal immunoreactivity was undetectable in Emx1^Cre;Dag1^cKOs and Emx1^Cre;POMT2^cKOs, whereas blood vessel expression was maintained, illustrating the specificity of the conditional deletion (Fig. 1A). Punctate perisomatic IIH6 immunoreactivity was present in Dag1^cyto/-, B4GAT1^M155T/M155T, and FKRP^P448L/P448L mice (Fig. 1A).

We next prepared WGA-enriched lysate from neonatal (P0) forebrain and immunoblotted for (1) IIH6, to quantify the degree of α-Dag1 glycosylation and (2) β-Dag1, to measure total Dystroglycan protein levels (Fig. 1B). Dag1 glycosylation was significantly reduced in Emx1^Cre;Dag1^cKOs, Emx1^Cre;POMT2^cKOs, Dag1^cyto/-, B4GAT1^M155T/M155T, and FKRP^P448L/P448Lmice; however, the reduction in FKRP^P448L/P448L mice was less severe than the other models (Fig. 1C). The reduction in glycosylation observed in the Dag1^cyto/-mice is surprising given that the mutation is restricted to the intracellular domain. Dystroglycan heterozygotes (Dag1^+/-) show no reduction in IIH6 levels compared to wild-types (data not shown), so the reduction in Dag1^cyto/- mice can be presumed to be due to the intracellular deletion. It is possible that the intracellular domain is required for the trafficking of Dystroglycan through the endoplasmic reticulum and/or Golgi apparatus, where Dystroglycan undergoes glycosylation, however additional work is needed to verify this. As expected, β-Dag1 immunoblotting was significantly reduced in Emx^Cre;Dag1^cKOsand absent in Dag1^cyto/- mice but normal in the glycosylation mutants (Fig. 1D). The residual β-Dag1 in Emx1^Cre;Dag1^cKO brain is likely due to Dag1 expression in unrecombined cells, such as blood vessels, as well as unrecombined tissue that remained after the forebrain dissection.

In neocortex, Dag1 expression in radial glia is required for proper migration of neurons, with Dag1 conditional deletion from neuroepithelial cells or radial glia resulting in Type II lissencephaly^25,26,41,42. This requires proper Dystroglycan glycosylation, but not its expression in neurons^36,43–45. To compare cortical migration across our five models of dystroglycanopathy, we performed immunostaining for the upper layer marker CUX1 (layers II/III-IV) and the deep layer marker TBR1 (layers III, VI) in P30 somatosensory cortex (Fig. 2A-B). Emx1^Cre;Dag1^cKOsand Emx1^Cre;POMT2^cKOs showed complete cortical dyslamination with 100% penetrance, whereas the cytoplasmic (Dag1^cyto/-) deletion mutants appeared normal (Fig. 2C-D). B4GAT1^M155T/M155T missense mutants showed a migration phenotype only at the cortical midline, while FKRP^P448L/P448Lmissense mutants did not show any cortical migration phenotype (Fig. 2 C-D, Supplementary Fig. 2A). These results indicate that cortical migration depends on Dystroglycan glycosylation but does not require its cytoplasmic domain. Furthermore, taken with the data in Fig. 1B-D, they illustrate that the severity of the cortical migration phenotype scales with the degree of Dystroglycan hypoglycosylation.

CCK⁺/CB₁R⁺ IN innervation is largely absent from the cortex and hippocampus when Dag1 is deleted selectively from pyramidal neurons using NEX^{Cre 27,28}. However, the development and function of CCK⁺/CB₁R⁺ INs has not been examined in mouse models that more broadly lack Dag1 throughout the central nervous system (CNS) and thus more accurately reflect the neuropathology of dystroglycanopathy. We focused our analysis on region CA1 of the hippocampus, as its overall architecture is grossly unaffected in each of our mouse models. Both Emx1^Cre;Dag1^cKO and Emx1^Cre;POMT2^cKOmice exhibited a mild granule cell migration phenotype in dentate gyrus (Fig. 3A, yellow arrows), however CA1-CA3 showed normal pyramidal neuron organization. In WT control mice, CCK⁺/CB₁R⁺ IN axon terminals were abundant throughout the hippocampus, with their highest innervation density in the CA1 pyramidal cell body layer (stratum pyramidale, SP) where they form characteristic basket synapses onto pyramidal neurons (Fig. 3A-C). In Emx1^Cre;Dag1^cKOmice, CCK⁺/CB₁R⁺ axons were present but failed to target the pyramidal cell layer (Fig. 3A-C), a marked difference from the phenotype observed in NEX^Cre;Dag1^cKO mice which lack CCK⁺/CB1R⁺ IN innervation entirely^27,28.

Emx1^Cre;POMT2^cKO mice fully phenocopied the aberrant Emx1^Cre;Dag1^cKO CB₁R⁺ immunoreactivity pattern (Fig. 3A-C), demonstrating that proper CCK⁺/CB₁R⁺ IN basket axon targeting requires Dystroglycan glycosylation. Both the B4GAT1 and FKRP mutants showed a normal distribution of CB₁R⁺ axon targeting to the somatodendritic compartment of CA1 neurons, but with reduced CB₁R intensity in SP (Fig. 3A-C). The cytoplasmic domain of Dystroglycan also plays a role in the appropriate targeting of CB₁R immunoreactive axons, as the distribution of axons was perturbed in the Dag1^cyto/-mutants, though with an intermediate phenotype in which the upper portion of SP appeared normal while the lower portion of SP showed loss of selective CB₁R⁺ axon targeting (Fig. 3A-C).

Due to the axonal targeting defect in the CCK⁺/CB₁R⁺ IN population, we next examined the parvalbumin (PV) population of interneurons in the hippocampus, as these cells also form perisomatic basket cell synapses on to CA1 pyramidal cells. The distribution of PV⁺ IN axons showed normal targeting to SP in all mouse models (Supplementary Fig. 3A-B), indicating that the axon targeting phenotype is specific to the CCK⁺/CB₁R⁺ IN population. Interestingly, Emx1^Cre;Dag1^cKO mice exhibited a slight increase in PV intensity in SP, perhaps indicating that there is a degree of compensation (Supplementary Fig. 3A-B).

Notably, CB₁R expression was also abnormal in brain regions outside of the hippocampus. In somatosensory cortex, CB₁R immunostaining reflected the dyslamination phenotype in both the Emx1^Cre;Dag1^cKOand Emx1^Cre;POMT2^cKO mice throughout neocortex and the B4GAT1^M155T/M155Tmice at midline, while it appeared normal in cortex of Dag1^cyto/- and FKRP^P448L/P448L mice (Supplementary Fig. 4A). CB₁R staining was also reduced and disorganized in the basolateral amygdala (BLA) in Emx1^Cre;Dag1^cKOand Emx1^Cre;POMT2^cKO mice (Supplementary Fig. 4B). Interestingly, CB₁R immunostaining in the inner molecular layer (IML) of dentate gyrus appears normal in all mutants (Fig. 3A). In the IML, CB₁R is present in excitatory mossy cell axons targeting dentate granule cells whereas in both cortex and amygdala CB₁R expression is restricted to GABAergic interneurons^46–48. Therefore, glycosylated Dystroglycan instructs the development of inhibitory CB₁R⁺ interneuron populations in multiple brain regions.

During early postnatal development, CCK⁺/CB₁R⁺ IN axons undergo a dramatic laminar rearrangement, progressing from more distal localization amongst pyramidal cell dendrites, to eventually target pyramidal neuron cell bodies in the hippocampus^28,49–51. We examined the developmental time course of CCK⁺/CB₁R⁺ IN axon targeting in our Emx1^Cre;Dag1^cKOmice beginning at P5, when the axons are first readily identifiable^52–55. At P5 in control Emx1^Cre;Dag1^Ctrl mice, CCK⁺/CB₁R⁺ axons were initially concentrated in the stratum radiatum (SR) of the hippocampus (Fig. 4A-C). Between P10 and P30, the CCK⁺/CB₁R⁺ axons underwent developmental reorganization, with reduced innervation of SR coinciding with a progressive increase in innervation of SP. In contrast, overall CCK⁺/CB₁R⁺ innervation was initially reduced in the hippocampus of mutant Emx1^Cre;Dag1^cKO mice at P5, and these axons failed to undergo laminar reorganization as they developed (Fig. 4A-C). By P30, after IN synapse formation and targeting are largely complete in control mice, the density of CCK⁺/CB₁R⁺ axons in Emx1^Cre;Dag1^cKOmice was uniform across all hippocampal lamina (Fig. 4A-C). Therefore, Dystroglycan plays a critical developmental role during the first two postnatal weeks, for the proper laminar distribution and perisomatic targeting of CCK⁺/CB₁R⁺ IN axons in the hippocampus.

Given the perturbed distribution of CCK⁺/CB₁R⁺ IN axons in the hippocampus, we next wanted to determine whether the remaining CCK⁺/CB₁R⁺ IN axons were capable of forming synapses in dystroglycanopathy models. Using VGAT as a marker of inhibitory presynaptic terminals, we saw no difference in total VGAT puncta density in SP in any of the mouse models, indicating that the total number of inhibitory synapses is normal (Fig. 5B-D). Immunostaining for CB₁R showed a significant decrease in CB₁R in SP of Emx1^Cre;Dag1^cKO, Emx1^Cre;POMT2^cKO, B4GAT1^M155T/M155T, and FKRP^P448L/P448L mutants, but not Dag1^cyto/- mutants (Fig. 5E). This suggests that the difference in CB₁R⁺ axon distribution described in SP of Dag1^cyto/-mutants in Fig. 3A-C likely reflects a change in CCK⁺/CB₁R⁺ IN axon targeting but not synapse formation, whereas Emx1^Cre;Dag1^cKO and all three glycosylation mutants exhibit a reduction in CCK⁺/CB₁R⁺ IN axon targeting and synapse number in SP. In contrast to CCK⁺/CB₁R⁺ INs, the PV⁺ population of basket interneurons showed no change in puncta density in SP in any of the models (Supplementary Fig. 6A-E; analysis of VGAT, CB₁R, and PV densities in SO and SR included in Supplementary Fig. 5A-B and Supplementary Fig. 7A.)

To better approximate the extent of basket synapse formation, we quantified the co-localization between VGAT and CB₁R or PV. In SP, the percent of CB₁R puncta co-localized with VGAT was reduced in the same models that showed a reduction in CB₁R density (Emx1^Cre;Dag1^cKO, Emx1^Cre;POMT2^cKO, B4GAT1^M155T/M155T, and FKRP^P448L/P448L mutants) but not Dag1^cyto/-mutants (Fig. 5C, F), suggesting that CCK⁺/CB₁R⁺ INs require postsynaptic glycosylated Dystroglycan in order to form synapses whereas the cytoplasmic domain is required for axon targeting but not synapse formation.

Interestingly, the percent of PV co-localized with VGAT increased in the SP of Emx1^Cre;Dag1^cKOsand Emx1^Cre;POMT2^cKOs mice, with no change in any of the other models (Supplementary Fig. 6C, E; analysis of co-localization in SO and SR included in Supplementary Fig. 5C, Supplementary Fig. 7B). It is possible that the reduction in inhibitory CCK⁺/CB1R⁺ synapses prompts homeostatic compensation through an increase in PV⁺ synapses. Alternatively, this may reflect competition between CCK⁺/CB₁R⁺ and PV⁺ INs for physical space on the perisomatic region of pyramidal cells, with the decrease in CCK⁺/CB₁R⁺ synapses in Emx1^Cre;Dag1^cKOsand Emx1^Cre;POMT2^cKOsallowing additional PV⁺ IN synapses to form.

Perisomatic inhibitory basket cell synapses powerfully control activity in the hippocampal circuit⁵⁶. Previous studies in NEX^Cre;Dag1^cKO mice, in which CCK⁺/CB₁R⁺ INs are absent, demonstrated reduced inhibitory synaptic function²⁷. In the current study, however, CCK⁺/CB₁R⁺ INs are present but mistargeted. Thus, we wanted to determine whether with the changes in CCK⁺/CB₁R⁺ basket synapse localization in our mouse models were associated with altered inhibitory synaptic function. CCK⁺/CB₁R⁺ IN basket cells can be selectively activated by muscarinic receptor activation, which increases the rate of spontaneous inhibitory post-synaptic currents (sIPSCs) in nearby pyramidal cells^27,57. Thus, to assay function at CCK⁺/CB₁R⁺ IN synapses, we performed whole cell patch clamp electrophysiology from CA1 pyramidal neurons in slices from control and mutant mice. After recording 5 minutes of baseline sIPSCs, the cholinergic receptor agonist Carbachol (CCh) was added to the bath and an additional 5 minutes of sIPSCs were recorded. While both Emx1^Cre;Dag1^Ctrland Emx1^Cre;Dag1^cKO cells displayed a CCh-mediated increase in sIPSC frequency, this response was dramatically attenuated in Emx1^Cre;Dag1^cKOs mice compared to Emx1^Cre;Dag1^Ctrl mice (Fig. 6A-C). Furthermore, in Emx1^Cre;Dag1^Ctrls, 19/21 cells (90.5%) responded to CCh application (defined as a >20% increase in sIPSC frequency), whereas only 13/22 cells (59.1%) responded in Emx1^Cre;Dag1^cKOs(Supplementary Fig. 8B). Proper Dystroglycan glycosylation was also required for CCK⁺/CB₁R⁺ IN synapse function, as Emx1^Cre;POMT2^cKOmice exhibited the same phenotype as Emx1^Cre;Dag1^cKOs: a reduced response to CCh overall, and a reduced proportion of responsive cells (Fig. 6A-C, Supplementary Fig. 8B). CCh also increased the mean sIPSC amplitude in each of the controls (Supplementary Fig. 8A), which may reflect an increased contribution of larger-amplitude action potential-mediated perisomatic events elicited by CCh^27,57. Consistent with the decreased function of CCK⁺/CB₁R⁺ IN synapses, a CCh-mediated sIPSC amplitude was also absent in each of these models (Supplementary Fig. 8A). Together, these data indicate that the altered perisomatic CCK⁺/CB₁R⁺ IN synaptic localization in CA1 is associated with a functional deficit in synaptic signaling.

Dag1^cyto/- mutants also had dramatically attenuated sIPSC increase in CCh compared to WT controls (Fig. 6C). Notably, even baseline sIPSC frequency was reduced in Dag1^cyto/-mutants (2.27±1.70 Hz) compared to WT controls (4.46±2.04 Hz, p = 0.002), whereas baseline sIPSC frequencies appeared normal in all other mutants when compared to their respective controls. Together with the finding that these mutants contain a normal number of CCK⁺/CB₁R⁺ basket synapses (as measured using immunohistochemistry; Fig. 5A-D), these results indicate that the cytoplasmic domain of Dystroglycan may play a critical role in mediating the assembly of functional postsynaptic signaling/receptor complexes at these synapses.

Neither of the more mildly hypoglycosylated mutants (B4GAT1^M155T/M155T, FKRP^P448L/P448L) were different from their respective littermate controls in terms of the magnitude of the CCh effect on sIPSC frequency (Fig. 6C), although the B4GAT1^WT mice appeared to possess a reduced effect of CCh compared to other control conditions (Fig. 6A-C). The B4GAT1 line is of a mixed genetic background, which could possibly explain the difference in CCh response. This finding is of unclear significance and may have obscured potential differences. Importantly, however, the marked functional synaptic differences observed between the Emx1^Cre;POMT2^cKO, Emx1^Cre;Dag1^cKO and Dag1^cyto/- mice when compared with each of their respective controls described above was not seen in either of these phenotypically milder mutants.

Together, these results suggest that Dystroglycan is required for the function of CCK⁺/CB₁R⁺ IN perisomatic basket synapses in a glycosylation-dependent manner, as evidenced by the Emx1^Cre;Dag1^cKOand Emx1^Cre;POMT2^cKO synaptic phenotypes, and that the intracellular domain of Dystroglycan is also required for normal CCK⁺/CB₁R⁺ IN basket synapse function. In contrast, B4GAT1^M155T/M155T and FKRP^P448L/P448L hypomorphic mutants both appear to retain sufficient Dystroglycan glycosylation to maintain normal synapse function.

Human patients with dystroglycanopathy have an increased risk of seizures and epilepsy^58–61, however the underlying cause has yet to be determined. The observed defects in inhibitory basket synapse function suggest that alterations in neuronal circuit inhibition could potentially predispose mutant mice to seizures. To test whether mouse models of dystroglycanopathy exhibit a reduced seizure threshold, we exposed mice to the volatile chemoconvulsant flurothyl and measured the latency to generalized tonic-clonic seizure (TCS)⁶².

The latency to TCS was significantly faster in Emx1^Cre;Dag1^cKOmice than their littermate controls (a 40.9% reduction on average, Fig. 7A), with no difference in seizure latency between sexes in either group (Supplementary Fig. 9C). Emx1^Cre;POMT2^cKOsand Dag1^cyto/- mutants also had a significantly shorter latency to TCS than littermate controls (42.9% and 33.6% reductions, respectively; Fig. 7A), indicating that the mechanism underlying Dystroglycan’s role in seizure susceptibility requires both extracellular glycosylation and intracellular interactions. B4GAT1^M155T/M155T mutants showed a small but significant reduction (16%) in seizure latency, despite exhibiting no detectable functional deficit by electrophysiology (Fig. 6A-C, Fig. 7A). Finally, FKRP^P448L/P448L mutants showed no significant change in seizure susceptibility (Fig. 7A). These results demonstrate that disruptions in Dystroglycan function, including both its extracellular glycosylation and intracellular interactions, increase sensitivity to seizures.

Synaptogenesis requires multiple distinct steps: (1) synaptic partner recognition, (2) recruitment and assembly of core pre- and post-synaptic machinery, (3) differentiation and maturation of synaptic identity, and (4) synaptic maintenance⁶³. Based on data from this study (Fig. 4) and previous work from our group and others, mice lacking Dystroglycan exhibit defects in CCK⁺/CB₁R⁺ IN development at the earliest time point they can be reliably identified (P0-P5), before the peak phase of inhibitory synapse formation (P9), suggesting that Dystroglycan functions at the earliest stages of synaptogenesis such as synaptic partner recognition⁶⁴. Determining the precise onset of synapse targeting and formation for most IN subtypes, including CCK⁺/CB₁R⁺INs, is limited by a lack of genetic tools for visualizing and manipulating IN subtypes during developmental stages.

The impairment in CCK⁺/CB₁R⁺ IN development throughout the forebrain suggests a trans-synaptic role for Dystroglycan (Fig. 3, Supplementary Fig. 4). The identity of the trans-synaptic binding partner between Dystroglycan-expressing cells and CCK⁺/CB₁R⁺ INs remains unknown. Our data in Emx1^Cre;POMT2^cKO mice point to a critical role for the glycan chains on Dystroglycan mediating this binding. All proteins that bind to the glycan chains on Dystroglycan do so through at least one Laminin G (LG) domain. There are over 25 LG-domain containing extracellular or transmembrane proteins expressed in the hippocampus. Neurexins, a family of highly alternatively spliced synaptic cell-adhesion molecules (NRXN1-3) which each contain multiple LG domains, bind Dystroglycan in a glycosylation-dependent manner^65–68. The specific splice isoforms of Nrxns that bind Dystroglycan are expressed by CCK⁺/CB₁R⁺ INs^66,69. Neurexin-3 conditional knockout (targeting all Nrxn3 isoforms) and CRISPR-mediated Dag1 knockout both result in similar synaptic deficits in olfactory bulb and prefrontal cortex⁷⁰. While a Dystroglycan knock-in mouse with reduced glycosylation that impairs Neurexin binding (Dag1^T190M) shows normal CCK⁺/CB₁R⁺ synapse formation by immunohistochemistry, the functionality of these synapses was not assessed by electrophysiology²⁷. Similar to B4GAT1^M155T/M155T and FKRP^P448L/P448Lmutants, the Dag1^T190Mmutation does not fully eliminate Dystroglycan glycosylation, and therefore does not rule out the possibility that Neurexins play a role at CCK⁺/CB₁R⁺ synapses. It is also possible that a yet undescribed Dystroglycan interacting protein is required for initial synapse recognition, and Nrxn-Dag1 interactions are required for subsequent synapse maturation and maintenance only. Indeed, the majority of studies indicate that Neurexins are not required for the initial formation of synapses, but rather regulate the maturation and structural maintenance of synapses after they have formed^70–74. Interestingly, while Dystroglycan localizes to both PV⁺ and CCK⁺/CB₁R⁺ inhibitory basket synapses in CA1, only the CCK⁺/CB₁R⁺ IN population was affected in the dystroglycanopathy models²⁷. Presumably, PV⁺ INs have a distinct developmental program independent of Dystroglycan and likely require a different postsynaptic recognition partner.

In brain and muscle tissue, Dystroglycan forms a complex with Dystrophin and several other proteins, collectively known as the Dystrophin Glycoprotein Complex (DGC). Like Dystroglycan, Dystrophin is also expressed throughout the forebrain and is associated with inhibitory synapses in multiple brain regions⁷⁵. Patients with mutations in Dystrophin develop Duchenne Muscular Dystrophy (DMD), and frequently exhibit cognitive impairments in the absence of brain malformations, suggesting a general role for the DGC in synapse development and function^76–78. A mouse model of DMD lacking all neuronal Dystrophin isoforms (mdx) exhibits defects in CCK⁺/CB₁R⁺ IN synapse development and abnormal innervation in the hippocampus, resembling the innervation pattern we observed in Emx1^Cre;Dag1^cKOand Emx1^Cre;POMT2^cKO mice in this study⁷⁹. Since Dystroglycan interacts with Dystrophin through its intracellular domain, we expected to observe similar phenotypes in mice lacking the intracellular domain of Dystroglycan (Dag1^cyto/-). However, Dag1^cyto/- showed a milder axon targeting defect than Emx1^Cre;Dag1^cKOor mdx mice. In addition, IIH6 puncta were normally localized to the somatodendritic compartment in Dag1^cyto/-mutants, suggesting that Dystroglycan does not require interactions with Dystrophin for is localization to somatodendritic synapses. However, Dystroglycan’s synaptic localization has not been examined in the mdx mutants. Clearly, additional work is required to better understand the relationship between Dystroglycan and Dystrophin at synapses in the brain.

While the density of CCK⁺/CB₁R⁺ IN synaptic puncta was normal in Dag1^cyto/- mice, synaptic function was impaired to the same level as Emx1^Cre;Dag^cKO and Emx1^Cre;POMT2^cKO mice, and seizure latency was reduced. Given that Dag1^cyto/- and B4GAT1^M155T/M155T mutants show a similar reduction in Dystroglycan glycosylation (Fig. 1B-C), our observation that the functional synaptic phenotype is restricted to the Dag1^cyto/-mutant reinforces the notion that the intracellular domain of Dystroglycan plays an active role in organizing essential postsynaptic signaling elements. One possibility is that the intracellular domain of Dystroglycan is required to recruit additional postsynaptic scaffolding elements and receptors necessary for CCK⁺/CB₁R⁺ basket synapse function⁸⁰. Importantly, Dag1^cyto/- mice did not show a cortical migration phenotype (Fig. 2A-D), indicating that the functional synaptic deficits and reduced seizure latency occurred independent of cortical malformation.

In addition to muscular atrophy and hypotonia, dystroglycanopathy patients often present with central nervous system symptoms. Patients with the most severe forms of dystroglycanopathy (FCMD, muscle-eye-brain disease, and Walker-Warburg Syndrome) exhibit structural changes including hypoplasia of the retina, brainstem and spinal cord, cerebellar cysts, hydrocephalus, type II lissencephaly, and microcephaly, associated with seizures and cognitive disability^81,82. Patients with milder forms of dystroglycanopathy may show cognitive disability and/or seizures without gross brain malformations, suggesting that there may be synaptic deficits independent of early neurodevelopmental processes (e.g. neuronal migration, axon guidance)^58,82. Mutations in any of the genes involved in the glycosylation of Dystroglycan can result in dystroglycanopathy with seizures, but the incidence and severity of seizures is higher in patients with brain malformations^58,82,83.

Mice have been used to model dystroglycanopathy for decades, however to our knowledge the present study is the first to investigate seizure susceptibility in mouse models of dystroglycanopathy. It is probable that the CCK⁺/CB₁R⁺ interneuron axon targeting and synapse phenotypes in the mouse models described in the present study contribute to their seizure susceptibility and open the possibility that defective inhibitory synaptic signaling mechanisms may underlie seizures in dystroglycanopathy patients. Although severe neuronal migration phenotypes In Emx1^Cre;Dag1^cKOand Emx1^Cre;POMT2^cKOmice may contribute to seizure activity, our observation that Dag1^cyto/- mutants showed both abnormal CCK⁺/CB₁R⁺synaptic function and reduced seizure latency, with intact cortical migration, indicates that the seizure phenotype is likely associated with synaptic defects.

While our B4GAT1^M155T/M155T mutants showed only a slightly reduced seizure latency, the mutants experienced more severe seizures than the other mouse models, resulting in death in 50% of cases (4/8 mutants compared to 0/6 fatalities among littermate controls) (Supplementary Fig. 9A). Flurothyl-induced seizures are typically generalized forebrain seizures; however in seizure-prone mouse models or in mice exposed to higher concentrations of flurothyl, mice can experience a suppression of brainstem oscillations followed by sudden death^84,85. The B4GAT1^M155T/M155T mutation was originally identified based on a hindbrain axon guidance phenotype, suggesting they may have currently unknown defects in brainstem development or circuitry that could render them more susceptible to fatal brainstem seizures³⁶. Because the Dag1 and POMT2 mutants are forebrain-specific conditional knockouts, (Supplementary Fig. 1A), we would not anticipate abnormal axon guidance in the brainstem or hindbrain of these mutants. Further research on the nature and progression of seizures observed in mouse models may have a profound impact on our understanding of dystroglycanopathy and potential therapeutic interventions.

Most patients with dystroglycanopathy present with mutations in one of the 19 genes required for the glycosylation of Dystroglycan, resulting in hypoglycosylated Dystroglycan. We have demonstrated that a mild reduction in the glycosylation of Dystroglycan, as seen in FKRP^P448L/P448L and B4GAT1^M155T/M155Tmutants, does not significantly disrupt synapse function. This suggests that glycosylation may not need to be restored to wild-type levels in order to achieve normal synapse function. Gene replacement therapy may be well suited to treat certain forms of dystroglycanopathy by rescuing glycosylation. AAV-mediated delivery of fully functional glycosyltransferases has been shown to significantly improve muscle pathology and function in dystrophic mice, however synaptic phenotypes have not been examined⁸⁶. Supplementation with (CDP)-ribitol, which is synthesized by CRPPA (previously known as ISPD), can restore functional Dystroglycan glycosylation and improve muscle function in mouse models with hypomorphic mutations in CRPPA or FKRP⁸⁷. In mice lacking functional CRPPA or FKRP in skeletal muscle, (CDP)-ribitol can further enhance the therapeutic impact of gene restoration⁸⁸. However, whether (CDP)-ribitol treatment can improve Dystroglycan function in other models of dystroglycanopathy, or is capable of restoring Dystroglycan glycosylation and synaptic function in the nervous system, remains untested.

All animals were housed and cared for by the Department of Comparative Medicine (DCM) at Oregon Health and Science University (OHSU), an AAALAC-accredited institution. Animal procedures were approved by OHSU Institutional Animal Care and Use Committee (Protocol # IS00000539), adhered to the NIH Guide for the care and use of laboratory animals, and provided with 24-hour veterinary care. Animal facilities are regulated for temperature and humidity and maintained on a 12-hour light-dark cycle and were provided food and water ad libitum. Animals older than postnatal day 6 (P6) were euthanized by administration of CO₂, animals <P6 were euthanized by rapid decapitation,

The day of birth was designated postnatal day 0 (P0). Ages of mice used for each analysis are indicated in the figure and figure legends. Mouse strains used in this study have been previously described and were obtained from Jackson Labs, unless otherwise indicated (Table 1)^{35,36,43,44,89–91}. Breeding schemas are as described in Table 2. Where possible, mice were maintained on a C57BL/6 background. Dag1^cyto/- mice occurred at a frequency lower than Mendelian, suggesting that a proportion of progeny die embryonically. To increase viability of pups, the Dag1^cyto line was outcrossed to a CD-1 background for one generation. The B4GAT1 line has a mixed genetic background: it was founded on a C3H/He background and then crossed on to C57BL/6 for future generations. Genomic DNA extracted from toe or tail samples (Quanta BioSciences) was used to genotype animals. Primers for genotyping can be found on the JAX webpage or originating article. For each mouse strain, littermate controls were used for comparison with mutant mice.

Brains from mice younger than P15 were dissected and drop fixed in 5 mLs of 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS) overnight for 18-24 hours at 4°C. Mice P15 and older were deeply anesthetized using CO2 and transcardially perfused with ice cold 0.1M PBS for two minutes to clear blood from the brain, followed by 15 mLs of ice cold 4% PFA in PBS. After perfusion, brains were dissected and post-fixed in 4% PFA for 30 minutes at room temperature. Brains were rinsed with PBS, embedded in 4% low-melt agarose (Fisher cat. no. 16520100), and sectioned at 50μm using a vibratome (VT1200S, Leica Microsystems Inc., Buffalo Grove, IL) into 24-well plates containing 1 ml of 0.1M PBS with Sodium Azide.

Single and multiple immunofluorescence detection of antigens was performed as follows: free-floating vibratome sections (50μm) were briefly rinsed with PBS, then blocked for 1 hour in PBS containing 0.2% Triton-X (PBST) plus 10% normal goat or donkey serum. Sections were incubated with primary antibodies (Table 3) diluted in blocking solution at 4°C for 48-72 hours. For staining of Dystroglycan synaptic puncta, an antigen retrieval step was performed prior to incubation in primary antibody. Briefly, sections were incubated in sodium citrate solution for 15 min at 95 degrees in a water bath. Following incubation in primary antibody, sections were rinsed with PBS then washed with PBST three times for 20 min each. Sections were then incubated with a cocktail of secondary antibodies (1:500, Alexa Fluor 488, 546, 647) in blocking solution overnight at room temperature. Sections were washed with PBS three times for 20 min each and counterstained with Hoechst 33342 (1:10,000, Life Technologies, Cat# H3570) for 20 min to visualize nuclei. Finally, sections were mounted on slides using Fluoromount-G (SouthernBiotech) and sealed using nail polish.

Imaging was performed on either a Zeiss Axio Imager M2 fluorescence upright microscope equipped with an Apotome.2 module or a Zeiss LSM 980 laser scanning confocal build around a motorized Zeiss Axio Observer Z1 inverted microscope with a Piezo stage. The Axio Imager M2 uses a metal halide light source (HXP 200 C), Axiocam 506 mono camera, and 10X/0.3 NA EC Plan-Neofluar, 20X/0.8 NA Plan-Apochromat objectives. The LSM 980 confocal light path has two multi-alkali PMTs and two GaAsP PMTs for four track imaging. Confocal images were acquired using a 63X/1.4 NA Plan-Apochromat Oil DIC M27 objective. Z-stack images were acquired and analyzed offline in ImageJ/FIJI⁹² or Imaris 9.8 (Oxford Instruments). Images used for quantification between genotypes were acquired using the same exposure times. Brightness and contrast were adjusted in FIJI to improve visibility of images for publication. Figures were composed in Adobe Illustrator 2023 (Adobe Systems).

For imaging experiments, 4-8 images were acquired from 2-4 coronal sections per animal, and at least three animals per genotype were used for analysis.

Cortex and hippocampus were dissected from P0 brains and solubilized in 1mL of lysis buffer containing 100mM NaCl, 50nM Tris, 2.5mM CaCl₂, 1% Triton X-100, 1% n-Octyl-β-D-glucopyranoside, and protease inhibitors. Lysate was incubated at 4°C for 1 hour and then spun at 12,500g for 25 minutes. Supernatant containing 3,000μg of protein (as determined by Pierce BCA Protein Assay) was applied to agarose-bound Wheat Germ Agglutinin (WGA) (Vector Labs) overnight at 4°C. Beads were washed 3X in TBS and boiled in 1X LDS sample buffer with 2-Mercaptoethanol (1:100) for 5 minutes. Samples were run on a 4-15% gradient polyacrylamide gel at 100V for 75 minutes and then transferred to a PVDF membrane (100V for 100 minutes). For immunoblotting, membranes were blocked in 5% milk TBST and then incubated overnight at 4°C in 5% milk TBST containing primary antibody. Antibodies used: α-Dystroglycan (IIH6C4) (Millipore cat. no. 05-593, mouse IgM, 1:500), MANDAG2 (DSHB cat. no. 7D11, mouse IgG1, 1:500), β3-tubulin (Cell Signaling Technology cat. no. 5568, rabbit, 1:2000). Membranes were washed 3X in TBST and incubated on fluorescent IRDye secondary antibody (1:10,000, LI-COR) in 5% milk TBST for 1 hour at room temperature. Membranes were imaged using a LI-COR Odyssey CLx 0918 imager and signal analyzed using LI-COR Image Studio Lite version 5.2.

For acute slice preparation, mice were deeply anesthetized in 4% isoflurane and subsequently injected with a lethal dose of 2% 2, 2, 2-Tribromoethanol in sterile water followed by transcardial perfusion with 10mL ice cold cutting solution containing the following (in mM): 93 NMDG, 2.5 KCl, 1.2 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 24 glucose, 5 Na Ascorbate, 2 Thiourea, 3 Na Pyruvate, 13 N-Acetyl Cysteine, 1 Kynurenic acid, 10 MgSO₄, 0.5 CaCl₂; pH 7.3, 300-340mmol/kg. After rapid decapitation, the brain was briefly submerged in ice cold cut solution bubbled with carbogen (95% oxygen, 5% CO2) and then sectioned into 300μm sagittal sections (Leica VT1200S vibratome) in bubbled ice-cold cut solution. Slices were recovered in 37°C cut solution, bubbled, for 15 minutes followed by 1 hour in room temperature recording ACSF (containing, in mM: 125 NaCl, 25 NaHCO₃, 1.25 NaH₂PO₄, 3 KCl, 25 D-Glucose, 2 CaCl₂, 1 MgCl₂) with an osmolarity of 310-320mmol/kg and supplemented with 1.5mM Na Ascorbate, bubbled.

CA1 pyramidal cells were patched in whole cell configuration using 3-5MΩ borosilicate glass pipettes filled with high chloride internal solution containing the following (in mM): 125 CsCl, 2.5 MgCl₂, 0.5 EGTA, 10 HEPES, 2 Mg-ATP, 0.3 Na-GTP, 5 QX-314; pH 7.2, 300mmol/kg. Pipettes were wrapped in parafilm to reduce capacitive currents. Cells were voltage clamped at -70mV and continuously superfused with 2-3 mL/min bubbled recording ACSF (310-320mmol/kg) containing 10μM NBQX to block excitatory transmission. Recordings were performed at 34°C. After 5 minutes of spontaneous IPSC (sIPSC) recording, 10μM Carbachol was added to the perfusate and another 5 minutes of sIPSC were recorded. Signals were amplified with an AxoPatch 200B amplifier (Molecular Devices), low-pass filtered at 5 kHz, and digitized and sampled at 10 kHz with a NIDAQ analog-to-digital board (National Instruments). Data were acquired and analyzed using a custom script in Igor Pro 8 (Wavemetrics). A hyperpolarizing step of -10mV was applied before each sweep to monitor input resistance, series resistance, and measure cell capacitance. Series resistance was not compensated and was maintained below 20MΩ. Cells were excluded if series resistance changed by more than 25%.

To calculate the proportion of cells that responded to Carbachol, cells were sorted into “responsive” and “non-responsive” categories. Cells were categorized as responsive if sIPSC frequency increased by 20% or more with the addition of Carbachol. If sIPSC frequency in a cell changed by less than 20% or less than 0.5Hz it was deemed non-responsive.

Mice aged P40-P55 were used for the flurothyl-induced seizure susceptibility assay to determine seizure threshold. Briefly, mice were placed in an enclosed glass chamber equipped with a vaporization chamber out of reach of the mouse. Volatile liquid 10% Bis (2,2,2-Trifluorotheyl) Ether (Millipore Sigma cat. no 287571) in 95% EtOH was delivered to the vaporization chamber at a rate of 6mL/hour. Seizure latency was determined as the amount of time until generalized tonic-clonic seizure (TCS). Upon exhibiting TCS, animals were immediately removed from the chamber and returned to their home cage, whereupon seizures ceased rapidly. Sample size was determined using power analysis as described below. The Emx1^Cre;Dag1 experimental groups were powered sufficiently to determine sex differences. Because no sex difference was found (Supplementary Fig. 9C), sexes were pooled for the remaining experiments. For statistical tests, outliers were excluded. Outliers were calculated as follows: first, the interquartile range (IQR) was calculated and multiplied by 1.5. This 1.5*IQR value was subtracted from the 25% quartile (Q1) and added to the 75% quartile (Q3). Points outside of the range Q1 - 1.5*IQR < x < Q3 + 1.5*IQR were categorized as outliers and indicated as such on all graphs. This was done to remove bias from extreme outliers observed in this experiment.

Phenotypic analyses were conducted using tissue collected from at least three mice per genotype from at least two independent litters. The number of mice and replicates used for each analysis (“N”) are indicated in Supplementary Table 1. Power analysis using pilot data was used to determine samples sizes with α = 0.05 and β = 0.80. Phenotypes were indistinguishable between male and female mice and were analyzed together. In many cases, highly penetrant phenotypes revealed the genotypes of the mice and no blinding could be performed. For comparisons between two groups, significance was determined using a two-tailed Students t-test. For comparisons between more than two groups, significance was determined using a 2-way ANOVA with Tukey HSD post-hoc analysis. Statistical significance was set at α = 0.05 (p < 0.05) and data presented as means ± SEM. Statistical analyses and data visualization were performed in R (version 4.0.2).