Shared Synaptic Pathophysiology in Syndromic and Nonsyndromic Rodent Models of Autism

See allHide authors and affiliations

Science  05 Oct 2012:
Vol. 338, Issue 6103, pp. 128-132
DOI: 10.1126/science.1224159

Reversing Autism in Mice

Autism comprises a heterogeneous group of neurodevelopmental disorders characterized by defects in communication and social inter action. A group of nonsyndromic forms of autism is associated with mutations in the neuroligin genes, which encode postsynaptic adhesion molecules. Using a reversible knockout approach, Baudouin et al. (p. 128, published online 13 September) investigated the in vivo functions of neuroligin-3 in the mouse cerebellum. Mutant mice showed a major defect in metabotropic glutamate receptor–dependent, long-term potentiation; disrupted heterosynaptic competition; and ectopic synapse formation in vivo. These synaptic defects could be rescued by reactivation of the neuroligin gene in the adult.


The genetic heterogeneity of autism poses a major challenge for identifying mechanism-based treatments. A number of rare mutations are associated with autism, and it is unclear whether these result in common neuronal alterations. Monogenic syndromes, such as fragile X, include autism as one of their multifaceted symptoms and have revealed specific defects in synaptic plasticity. We discovered an unexpected convergence of synaptic pathophysiology in a nonsyndromic form of autism with those in fragile X syndrome. Neuroligin-3 knockout mice (a model for nonsyndromic autism) exhibited disrupted heterosynaptic competition and perturbed metabotropic glutamate receptor–dependent synaptic plasticity, a hallmark of fragile X. These phenotypes could be rescued by reexpression of neuroligin-3 in juvenile mice, highlighting the possibility of reverting neuronal circuit alterations in autism after the completion of development.

Autism comprises a heterogeneous group of neurodevelopmental disorders characterized by variations in social interactions and communication and the manifestation of ritualistic behaviors (1). A large number of rare high-impact mutations have been identified in autistic patients (24). However, most insights into the synaptic pathophysiology of autism are derived from models of monogenic syndromes, such as fragile X syndrome, in which about 25% of patients meet diagnostic criteria of autism (5). In fragile X, the key defect in synaptic transmission is elevated group I metabotropic glutamate receptor–dependent long-term depression (mGluR-LTD). However, most cases of autism are nonsyndromic, and it is unclear whether these share pathophysiology with fragile X. One class of nonsyndromic forms of autism is associated with mutations in the neuroligin genes (Nlgn1, -2, -3, and -4), which encode postsynaptic adhesion molecules involved in synapse assembly (proteins NL1, -2, -3, and -4) (68). For Nlgn3, a R451C point mutation (7) and deletions (4, 9) have been identified in several patients with autism. The R451C point mutation results in NL3 trafficking defects (10), whereas the deletions remove the entire Nlgn3 coding sequence (4, 9). Nlgn3R451C knockin and Nlgn3KO (KO, knockout) mice exhibit impairments in social interactions, social memory, ultrasonic vocalization, and olfaction [(11, 12) but see (13)]. In Nlgn3R451C mice, synaptic transmission is altered in the somatosensory cortex and hippocampus (12, 14). However, the subcellular localization of NL3 protein in vivo and synaptic defects resulting from Nlgn3 ablation are unclear (15, 16). To understand the pathophysiology of Nlgn3 deletions, we focused on the synaptic connectivity in the cerebellum, because cerebellar activation is altered in autistic patients (17), and cerebellar lesions result in behavioral changes reminiscent of autism (18, 19). Using NL3-specific antibodies (16), we detected strong NL3 immune reactivity in the molecular layer and surrounding mossy fiber glomeruli of the inner granular layer (Fig. 1, A and B). Antibody reactivity was largely abolished in mutant mice carrying a STOP cassette inserted after the transcriptional start site (20), which resulted in a complete loss of NL3 expression (Nlgn3KO) (Fig. 1A and fig. S1, A and B). In the inner granular layer, NL3 was detected at glutamatergic as well as γ-aminobuyteric acid–responsive (GABAergic) synapses, whereas NL3 in the molecular layer was primarily detected at parallel fiber synapses (Fig. 1, C and D). In the same preparations, we detected only little apposition of NL3 immune reactivity with markers of climbing fiber and interneuron synapses (fig. S1C). Similar observations were made in Nlgn3PC mice, carrying a knockout allele where NL3 is selectively reexpressed in Purkinje cells (PCs) under the control of the tetracycline transactivator (Fig. 1, A and B, and fig. S1D).

Fig. 1

Cell type–specific synaptic localization of NL3. (A) Nlgn3KO (Nlgn3STOP-tetO) mice lack NL3 protein expression, and Nlgn3PC mice (Nlgn3STOP-tetO::Pcp2tTA) show reexpression in cerebellar lysates. (B) Mossy fiber inputs (MF) in the inner granular layer (IGL) are relayed to PCs via parallel fibers (PF). Climbing fibers (CF) form synapses directly onto the proximal PC dendritic arborization. In Nlgn3PC mice, NL3 immunoreactivity is increased on PC dendrites (calbindin, CBP) and abolished in the IGL. (C) NL3 is apposed to vGluT1+ PF boutons and colocalizes with mGluR1α in CBP+ spines, similar to GluD2. (D) In glomeruli, NL3 is detected at PSD95+ and GABA-A receptor α2+ synapses and colocalizes with GFP+ terminals in GlyT2GFP transgenic mice.

Ultrastructural analysis of parallel fiber synapses in Nlgn3KO mice did not reveal a dramatic difference in several morphological parameters (Fig. 2A). Miniature excitatory postsynaptic current (mEPSC) recordings in Nlgn3KO PCs identified a small but significant reduction in mEPSC amplitude (Fig. 2B), whereas paired-pulse facilitation of parallel fiber synapses was not detectably altered (Fig. 2C and fig. S2A). Parallel fiber synapses exhibit a marked LTD that is thought to contribute to forms of cerebellar function and learning [(21) but see (22)]. Simultaneous activation of group I mGluRs and AMPA receptors in the postsynaptic compartment results in protein kinase C activation, AMPA receptor (GluA2 subunit) phosphorylation at serine 880, and subsequent dissociation from the postsynaptic scaffold and endocytosis (21, 23). Given the reduction in mEPSC amplitudes, we tested whether mGluR-dependent LTD was modified in Nlgn3KO PCs. In 2- to 3-month-old wild-type mice, application of the group I mGluR agonist DHPG resulted in a persistent reduction of EPSC amplitudes. No depression was observed in Nlgn3KO cerebellar slices (Fig. 2D and fig. S2B). This loss in mGluR-LTD may be a consequence of an impaired LTD expression or an occlusion due to constitutive activation. We observed a twofold increase in basal GluA2 phosphorylation at serine 880 in Nlgn3KO as compared to the wild-type. Upon DHPG stimulation, phospho-GluA2 levels were increased in wild-type but decreased in Nlgn3KO mice, probably due to mGluR-stimulated degradation of the phosphorylated form of GluA2 (Fig. 2E) (24).

Fig. 2

Occlusion of mGluR-LTD in Nlgn3KO mice. (A) PF synaptic ultrastructure in Nlgn3KO mice by transmission electron microscopy (n ≥ 4 animals, 200 synapses per animal). (B) Cumulative distribution of mEPSCs (***P < 0.001, Kolmogorov-Smirnov test) and mean amplitude (n = 9 cells for wild-type and n = 21 cells for Nlgn3KO mice; ***P < 0.001, Mann-Whitney test). (C) mEPSC inter-event intervals, paired pulse ratios (n = 9 wild-type and n = 15 Nlgn3KO cells). (D) mGluR-LTD induced by 10 min of 50 μM DHPG (n ≥ 8 cells) and representative traces before (black) and after (red) DHPG stimulation. (E) Quantitative Western blot of basal and DHPG-induced phospho-GluA2 (normalized to GluA2/3 protein level, n ≥ 4 mice, *P < 0.03, t test). DHPG-induced phosphorylation is expressed as the ratio of treated to untreated samples (n ≥ 4 mice, **P < 0.01, t test).

The constitutive increase in phospho-GluA2 in the Nlgn3KO cerebellum indicated a gain of function in the synaptic plasticity pathway. Parallel fiber LTD is regulated by several postsynaptic receptors, including GluA2, GluD2, and mGluR1α (21). Using quantitative Western blotting, we discovered a selective increase in the mGluR1α protein in the Nlgn3KO cerebellum, whereas the mRNA level was unchanged (Fig. 3, A and B, and fig. S3; there is a similar mGluR1α protein increase in the thalamus). mGluR2 and mGluR7, two additional metabotropic receptors expressed in the cerebellum, were unaltered (fig. S3B). Endogenous mGluR1α and NL3 colocalized in the heads of Purkinje dendritic spines (Fig. 1D). In Nlgn3KO mice, synaptic mGluR1α levels were increased but were unaltered in the Nlgn1KO cerebellum (Fig. 3, C and D). This mGluR1α deregulation was a cell-autonomous consequence of NL3 loss of function, because reexpression of NL3 specifically in PCs (Nlgn3PC) reduced the mGluR1α protein level and its synaptic abundance back to the wild-type level (Fig. 3, B and D). Moreover, DHPG-induced phospho-GluA2 signals were returned to wild-type levels in mice that selectively reexpressed NL3 protein in PCs (Nlgn3PC, Fig. 3E).

Fig. 3

Elevation of synaptic mGluR1α levels in Nlgn3KO mice. (A) Quantitative Western blot. Protein levels were normalized to tubulin and expressed relative to the wild type. (B) Quantitative Western blot of mGluR1α, GluD2, and GluA2/3 levels in the cerebellum (normalized to tubulin and expressed relative to the wild type, n ≥ 4, **P = 0.002, t test). (C) Quantitative line scan on cerebellar sections stained with antibodies to mGluR1α, Homer 1, and calbindin. (D) Line scan intensity ratios of mGluR1α/CBP and Homer1/CBP (n = 6 mice, 1200 synapses per genotype; mGluR1α, *P = 0.04; Homer, P = 0.6; t test). Reexpression of NL3 in PCs restores mGluR1α level (n = 4 mice, P = 0.4). There was no difference in Nlgn1KO mice (n = 6 mice, P = 0.9). (E) DHPG-induced phosphorylation of GluA2 expressed as the ratio of treated to untreated samples (normalized to GluA2/3 protein level, n ≥ 5 mice, * P < 0.05, t test).

We next examined whether the loss of NL3 results in wiring alterations in the cerebellar network. In the mature cerebellum, each PC is innervated by a single climbing fiber, and synaptic competition between parallel fiber and climbing fiber inputs excludes climbing fiber inputs from the PC distal dendrites (21, 25). In Nlgn3KO cerebella, we observed a significant invasion of vGluT2+ terminals into the distal molecular layer (Fig. 4, A and B, and fig. S4A). Ectopic climbing fiber synapses were not observed in Nlgn1KO cerebella, and the expression of Nlgn3 exclusively in PCs was sufficient to suppress ectopic synapse formation in Nlgn3PC mice. Synapse density along the entire length of the overshooting climbing fibers was unaltered, resulting in an increased total number of climbing fiber synapses (Fig. 4, A and B, and fig. S4A). Consistent with this observation, evoked climbing fiber transmission was elevated in the Nlgn3KO mice (Fig. 4C; there is a normal regression of multi-innervation, fig. S4B).

Fig. 4

Disrupted heterosynaptic competition and adult phenotypic rescue in Nlgn3KO mice. (A) vGluT2 puncta on distal PC dendritic tree (upper 20%, dashed line, anticarbonic anhydrase-related protein VIII, Car8) marked by green arrowheads (vGluT2+ clusters per 100 μm2, n = 6 mice per genotype, **P = 0.0023 for wild-type versus Nlgn3KO mice, t test). (B) Density of CF synapses along individual fibers (CRF+, n ≥ 3 mice, P = 0.6). (C) Evoked CF-PC transmission assessed by extracellular stimulation (n ≥ 7 cells from ≥ 4 mice, **P = 0.003, t test). (D) Counted runs, step times, and missed steps on the Erasmus ladder [n ≥ 10, *P < 0.05, **P < 0.01, analysis of variance (ANOVA), post hoc Tukey test]. The performance of Nlgn3PC mice over four training days is indistinguishable from that of wild-type mice (P = 0.2, ANOVA, post hoc Tukey test). (E) Adult reexpression of NL3. Doxycycline (100 μg/ml from embryonic day 0) was removed at P30, and mice were killed at P60. (F) (Left) Quantitative Western blot of mGluR1α in Nlgn3PCadult mice (at P60, n ≥ 5 mice, **P = 0.004). (Right) Quantification of vGluT2 ectopic puncta (at P60, n ≥3 mice, **P = 0.003, *P = 0.03).

We used the Erasmus ladder (26) to explore motor coordination in Nlgn3KO mice. In this behavioral assay, mice cross a ladder consisting of pressure sensors that measure motor output. Wild-type and Nlgn3KO mice completed the same number of valid runs on the ladder. Step times for Nlgn3KO mice were significantly elevated, indicating a perturbation of motor coordination (Fig. 4D). The occurrence of missteps during ladder crossing was unaltered in the Nlgn3KO mice and declined similarly as for wild-type mice over several days of training. NL3 expression specifically in PCs rescued the elevated step times progressively after several training days (Fig. 4D). Nlgn3PC mice did exhibit a higher number of missteps (Fig. 4D), most likely because the reexpression in PCs exceeds the endogenous NL3 level and results in a gain-of-function phenotype.

The extensive ectopic synapse formation in Nlgn3KO mice raises the question of whether such structural defects can be corrected after the completion of development. We used the tetracycline transactivator to temporally control the reexpression of NL3 in PCs of Nlgn3PC mice (Fig. 4E). Nlgn3PC mice were raised in the presence of doxycycline [transactivator inactive, NL3 expression off (Fig. 4E)]. At 30 days after birth (P30), doxycycline was removed, permitting the reexpression of NL3 selectively in PCs. Reexpression restored wild-type mGluR1α protein levels and resulted in the removal of ectopic synapses from the distal PC dendritic tree (Fig. 4F and fig. S4, C and D). Therefore, neurodevelopmental phenotypes and ectopic synapse formation arising from Nlgn3 deletion can be corrected in the mature cerebellar system.

Our results provide insight into the synaptic pathophysiology of a model of nonsyndromic autism. Phenotypes observed in Nlgn3KO mice represent a surprising parallel to the synaptic pathophysiology in Fmr1 and Tsc2 mutant mice (27). Like these syndromic autism models, Nlgn3KO mice exhibit a deregulation of mGluR-LTD. In the Fmr1KO mice, mGluR-LTD in the forebrain and cerebellar neurons is increased, and this phenotype can be suppressed by reduction of group I mGluR activity (5, 28, 29). Nlgn3KO mice exhibit an occlusion in mGluR-LTD due to increased mGluR1α expression, indicating a common core pathway of synaptic dysfunction. mGluR5 antagonists can revert cellular phenotypes in Fmr1KO mice and may show therapeutic benefit in some fragile X patients (5, 30). Our work indicates that group I mGluR antagonists hold promise for designing treatment strategies for nonsyndromic autism. Our genetic experiments in Nlgn3KO mice reveal that not only functional alterations but also ectopic synapse formation can be reversed, highlighting the fact that structural neurodevelopmental phenotypes can be rescued by intervention after the completion of development.

Supplementary Materials

Materials and Methods

Figs. S1 to S4


References and Notes

  1. Acknowledgments: We are grateful to our colleagues Arber, Barde, Gosh, Sylwestrak, Witte, and Xiao for comments on the manuscript; to Ponti, Genoud, Sauder, Ahrné, Ehrenfeuchter, and Stiefvater for technical help; and Zeilhofer, Fritschy, Brose, and Sans for sharing reagents. C.D.Z. was supported by the Dutch Organization for Medical Sciences (ZonMw); Life Sciences (ALW); Senter (NeuroBasic); Prinses Beatrix Fonds; and the European Research Council–advanced, CEREBNET, and C7 programs (European Community). This work was supported by a SystemsX grant (P.S. and W.S.); by EU-AIMS (European Autism Interventions), which receives support from the Innovative Medicines Initiative Joint Undertaking under grant agreement no. 115300, the resources of which are composed of financial contributions from the European Union's Seventh Framework Programme (grant FP7/2007-2013), from the European Federation of Pharmaceutical Industries and Associations companies’ in-kind contributions, and from Autism Speaks, resulting in a total of €29.6 million; by NCCR Synapsy; the Swiss National Science Foundation; and Kanton Basel-Stadt.
View Abstract

Navigate This Article