The ProSAP/Shank family of postsynaptic scaffold proteins comprises three family members called Shank1, ProSAP1/Shank2 and ProSAP2/Shank3. All three molecules, which are predominantly found in the Postsynaptic Density (PSD) of excitatory glutamatergic synapses, essentially contribute to synapse formation, maintenance, morphology and plasticity [1-5]. ProSAP/Shank malfunction has thus been implicated in various neuropsychiatric disorders (for reviews see [6‐8]). Apart from the PROSAP2/SHANK3 gene, which happens to be disrupted/mutated in both Phelan-McDermid Syndrome (a.k.a. 22q13.3 deletion syndrome), a neurodevelopmental disorder whose patients most often exhibit autistic‐like traits, and in other cases of autism (for most recent reviews see [9‐11]), genetic alterations such as Copy Number Variations (CNVs) or point mutations of the PROSAP1/ SHANK2 and SHANK1 genes have been directly associated with autistic individuals in recent studies from the last two years [12‐16]. To better understand the molecular pathology caused by those alterations, it is crucial to study their impact on the morphology, molecular composition and physiology of excitatory synapses. This can partly be conceived in vitro by knockdown or overexpression of mutated Shank1 and/or ProSAP1/Shank2 in primary neurons or by the generation of induced Pluripotent Stem cells (iPS cells) from affected individuals to differentiate and characterize patient‐derived neurons. However, the impact on different brain regions, the neuronal circuit level and finally behavior, still remains elusive. Therefore, it is essential to study the impact of genetic Shank1 and ProSAP1/Shank2 disruption in vivo by the generation and careful analysis of animal models such as mutant rats and mice.

The first animal model with a genetic disruption of ProSAP/ Shank was the Shank1 knock‐out mouse generated in Morgan Sheng’s laboratory [17]. Interestingly, this model had been published four years before the human SHANK1 gene was actually linked to patients with ASDs [16] and is still the only Shank1 mouse model available so far. In this model, exons 14 and 15 (encoding the PDZ domain) of the murine Shank1 gene were deleted via homologous recombination resulting in a lack of all known major isoforms of Shank1. Homozygous mutants (Shank1^-/-) weigh less [18], but do not exhibit any obvious changes in survival, brain size or brain histology. Morphological characterization of postsynaptic specializations in the CA1 hippocampus (apical dendrites of pyramidal neurons) showed that Shank1^-/- animals exhibit fewer and smaller spines with thinner PSDs, although the morphological changes are rather mild. Biochemical analysis of forebrain PSDs revealed reduced levels of Homer and GKAP; immunostaining of primary hippocampal cultures showed more diffusely distributed Homer signals and a loss of GKAP‐positive synapses. Moreover, basal synaptic transmission (field excitatory postsynaptic potentials, fEPSPs) is decreased at Shank1^-/- Schaffer collateral CA3‐to‐CA1 synapses, while synaptic plasticity (long term potentiation and depression, LTP and LTD) remains intact. Apart from these analyses focused on the hippocampus and implying deficits in hippocampal synapse maturation [17], Shank1 mutants underwent a broad range of behavioral tests [17‐19]. Homozygous mutants of both sexes are hypoactive, show motor coordination deficits, have reduced neuromuscular strength and exhibit a mild anxiety‐like phenotype. Regarding cognitive function, the lack of Shank1 leads to impaired contextual fear conditioning, enhanced spatial learning, but impaired long‐term retention of spatial memory. These behavioral findings are indeed of special interest, because synaptic plasticity at hippocampal CA3‐to‐CA1 synapses-the most popular synaptic parameter for learning and memory function, remains unchanged in Shank1 mutants.

Jacqueline Crawley’s laboratory further addressed the three core features of autism in Shank1^-/- mutants: impairment in social interaction, impairment in communication, and occurrence of restricted repetitive and stereotyped patterns of behavior. Although there was no behavioral alteration in support of the first and third feature [19], they were finally able to elucidate deficits in communication: Shank1^-/- pups at P8 emit fewer Ultrasonic Vocalizations (USVs), while the calls are higher in peak frequency and less frequency modulated. Adult male Shank1^-/- mice emit fewer calls in the first two minutes after being confronted with female urine. Furthermore, they do not change their calling pattern despite prior interaction with a female, spend less time in close proximity to the female urine spot, and leave less scent marks in this area [18].

Two mouse models lacking all known major isoforms of ProSAP1/ Shank2 were published in 2012, two years after the first association between human PROSAP1/SHANK2 and ASDs [12]. In the first model, exon 7 was targeted for deletion (Shank2^e7-/-, [20]), in the second one, exons 6 and 7 were targeted (Shank2^e6+7-/- [21]). Most interestingly, in these two distinct disruptions of the murine ProSAP1/Shank2 gene (both target the region encoding the PDZ domain), each resemble a different de novo microdeletion of human PROSAP1/SHANK2 identified in two unrelated autistic patients by Berkel et al. [12] in 2010-a 120 kb deletion resulting in a loss of exon 7 causing a frameshift, and a 69 kb deletion resulting in a loss of exons 6 and 7 also causing a frameshift. Both individuals exhibit ASD of comparative severity combined with mild to moderate mental retardation based on the Autism Diagnostic Observation Schedule (ADOS) and IQ measures. In line with these clinical observations, both mouse models show a comparable behavioral phenotype with minor variability among each other. Regarding core autistic‐like features, adult Shank2^e7-/- and Shank2^e6+7-/- mice show similar impairments in social interaction as measured by almost analogical behavioral tests (Resident‐intruder vs. home‐cage social interaction, three chamber social interaction assay). The same holds true for impairments in communication. Male mutants from both models take longer to emit the first USV when allowed to interact with a novel wild‐type female mouse. The total call rate was reduced only in Shank2^e6+7-/- males during the test, but assessment of call types in Shank2^e7-/- males revealed shorter and more unstructured calls (USVs during same‐sex interactions have only been evaluated in Shank2^e7-/- mutants so far; both males and females take longer for the first USV in this scenario; the call rate is reduced in females, who utter shorter, less mixed and more unstructured calls). Furthermore, regarding repetitive and stereotypical behaviors, Shank2^e7-/- and Shank2^e6+7-/- mutants, both extremely hyperactive (a robust phenotype also occuring in heterozygous mutants of both models), show enhanced jumping behavior mixed with upright scrabbling (for the Shank2^e7-/- model, this feature was not included in Schmeisser et al. [20], but is a commonly observed behavior (according to unpublished data by M.J. Schmeisser and T.M. Boeckers) and show decreased digging behavior in strangerfree home cages. Enhanced grooming behavior was observed in isolated Shank2^e7-/- females in home cages and in Shank2^e6+7-/- mutants during the novel object recognition test. As anxiety disorders frequently appear in autistic individuals, it is important to mention that both Shank2^e7-/- and Shank2^e6+7-/- mutants exhibit increased anxiety‐like behavior. With respect to learning and memory disabilities, Shank2^e7-/- mutants have normal working memory assessed by the Y‐Maze paradigm, while Shank2^e6+7-/- mutants performed worse than wild‐type littermates in the Morris Water Maze paradigm indicating spatial memory defects related to hippocampal dysfunction (spatial memory is still under investigation in Shank2^e7-/-mutants). No changes were observed in olfaction and novel object recognition in both Shank2^e7-/- and Shank2^e6+7-/- mutants; nesting and pup retrieval behavior was only assessed in Shank2^e6+7-/- (both impaired); body weight (reduced), survival (reduced, but registered only within the first month after birth), motor coordination (rotarod, unchanged) and pup USVs (higher call rate in isolated mutant females at P4 and P10) only in Shank2^e7-/-, which also show hind‐limb clasping when lifted by the tail.

Gross brain morphology is unaltered in both Shank2^e6+7-/- and Shank2^e6+7-/- adult mutants, as it is the ultrastructure of spiny synapses in the hippocampal CA1 region. With respect to CA1 dendritic spines and synapses, Shank2^e6+7-/- mutants exhibit a small reduction of spine numbers (evaluated by Golgi staining), while in Shank2^e6+7-/- mutants, no significant changes in synapse density were observed (evaluated by immunohistochemistry and electron microscopy). On the protein level, it is more difficult to compare the models, because molecular analyses of Shank2^e6+7-/- mice were predominantly conducted in region‐specific synaptosomal fractions and molecular analyses of Shank2^e6+7-/- mice in whole brain homogenates. A grossly overlapping finding in both models, though, is an increase in GluN1 levels, the obligatory subunit of the NMDA Receptor (NMDAR). GluN1 up‐regulation occurs in whole brain PSD fractions, hippocampal and striatal synaptosomes of adult Shank2^e6+7-/- (P70) and hippocampal synaptosomes of juvenile Shank2^e6+7-/- (P25) animals on the one hand-and in whole brain homogenates of juvenile Shank2^e6+7-/- (3‐4 weeks) animals on the other hand. Most interestingly, GluN2A and GluN2B levels are not changed in either whole brain PSD fractions from adult Shank2^e6+7-/- mutants or whole brain homogenates of juvenile (in case of GluN2A, also adult) Shank2^e6+7-/- mutants. However, analysis of region‐specific synaptosomes from juvenile and adult Shank2^e6+7-/- mice revealed a clear up‐regulation of GluN2B subunits in mutant hippocampus and both GluN2A and GluN2B in mutant striatum, while no changes were seen in mutant cortex. These biochemical results argue for a biochemically detectable region‐specific synaptic up‐regulation of distinct NMDAR subtypes in the Shank2^e6+7-/- brain, a phenomenon that might be masked due to the analysis of whole brain biochemistry of Shank2^e6+7-/- mutants if everything is present in this model. Biochemical analysis of juvenile Shank2^e6+7-/- whole brain homogenates further revealed a reduction of NMDAR‐associated signaling, supported by reduced levels of the phosphorylated forms of CamKIIα/β, ERK1/2, p38 and GluA1 (S831 and S845).

A second molecular aspect to discuss at this point is the upregulation and compensation by other ProSAP/Shank family members. In Shank2^e6+7-/- mutants, this aspect was only once addressed by biochemical analysis of juvenile whole brain homogenates (3‐4 weeks). Neither Shank1 nor ProSAP2/Shank3 are up‐regulated in this scenario. The same holds true for ProSAP2/Shank3 in cortical, hippocampal and striatal synaptosomes of Shank2^e6+7-/- animals at the same age (P25) and in whole brain homogenates of adult mutants. However, a strong increase in ProSAP2/Shank3 protein, but not RNA levels, is present in whole brain PSD fractions and striatal synaptosomes of adult Shank2^e6+7-/- animals–once more supporting subcellular and brain‐region specific, but also developmental effects. In line with these diverse molecular phenomena, it is to mention that Shank1 is decreased only in hippocampal, but not striatal synaptosomes of juvenile Shank2^e6+7-/- animals while its levels in synaptosomes from cortex, hippocampus and striatum at the adult stage, remain unchanged.

Contrary to the behavioral and morphological/molecular data, Shank2^e6+7-/- and Shank2e6+7-/- mutants show opposite phenotypes in neurophysiology based on recordings from acute hippocampal slices at 3‐4 weeks (Shank2^e6+7-/-) and 4‐5 weeks (Shank2e6+7-/-), respectively. Basal synaptic transmission (fEPSPs) at Schaffer collateral CA3‐to‐ CA1 synapses is strongly decreased in Shank2^e6+7-/- mutants (40%, also detectable in heterozygous mutants and in homozygous mutants at 3 months of age), while no change was observed in Shank2e6+7-/- animals. The same holds true for the frequency of Miniature Excitatory Postsynaptic potentials (mEPSCs) from CA1 pyramidal neuronsreduced in Shank2^e6+7-/-, unaltered in Shank2e6+7-/- (the mEPSC amplitude remains unchanged in both models). Further parameters differ among the models: NMDA/AMPA ratio and LTP-both increased in Shank2^e6+7-/-, but decreased in Shank2^e6+7-/-, or LTD-unchanged in Shank2^e6+7-/-, abolished in Shank2^e6+7-/-. Other neurophysiological parameters were only assessed in either of the models: excitability of presynaptic fibres, intrinsic firing threshold and whole‐cell input resistance of CA1 neurons, AMPA-mediated whole cell currents and GABAergic synaptic transmission at CA3‐to‐CA1 synapses in Shank2^e6+7-/- (for all parameters: no marked changes); paired‐pulse ratio, postsynaptic excitability, Metabotropic Glutamate Receptor (mGluR)‐LTD, NMDAR EPSCs including GluN2B‐mediated EPSCs at CA3‐to‐CA1 synapses and NMDA/AMPA ratio in the medial prefrontal cortex in Shank2^e6+7-/- (for all parameters: no marked changes). Despite the aforementioned discrepancies in hippocampal neurophysiology-which might be related to the two distinct genetic targeting strategies-selective alterations of NMDAR levels and/or functions seem to be a core molecular feature of ProSAP1/Shank2 deletions and most probably contribute to the ASD‐like phenotype. This conclusion is strongly supported by the fact that the decreased NMDA/AMPA ratio and impaired LTP and LTD at hippocampal CA3‐to‐CA1 synapses, the NMDAR signaling defects in whole brain homogenates, and the social interaction deficits of Shank2^e6+7-/- mutants were all successfully restored by application or intraperitoneal injection of a membrane‐permeable positive allosteric modulator of mGluR5 called CDPPB (3‐cyano‐N‐(1,3‐diphenyl‐1Hpyrazol‐ 5‐yl)benzamide), which is known to increase the responsiveness of mGluR5 to glutamate and to enhance NMDAR function [22].

In addition to the two deletion models, the human SHANK2‐ R462X mutation has also been expressed in mouse forebrain neurons in vivo [23]. In a study about the impact of three PROSAP1/SHANK2 mutations on neuronal morphology, Berkel et al. [23] included data on mice expressing the SHANK2‐R462X stop mutation in neuronal populations after viral infection of the forebrain with GFP‐tagged constructs at either P0 or P42. Interestingly, the recombinant GFPSHANK2‐ R462X protein does primarily localize to the cell bodies of hippocampal and cortical neurons rather than showing a synaptic expression pattern, a phenomenon most probably due to the disruption of the ProSAP1/Shank2 C‐terminus in GFP‐SHANK2‐R462X, which is crucial for synaptic targeting [24]. Both P0 and P42-infected animals exhibit the same physiological consequences at the adult stage (2‐3 months), a reduction in mEPSC amplitude on CA1 pyramidal neurons, respectively. The P0‐infected animals were more intensely studied and additionally exhibit reduced mEPSC amplitudes in cortical layer 2/3, enhanced number and size of GluA1 clusters in the CA1 hippocampus and filopodia‐like structures among dendrites of CA1 pyramidal neurons. These synaptic alterations might lead to the observed cognitive behavioral phenotype: less interest in novel object exploration and impaired executive functions in a puzzle box paradigm.

Over the last decade, autism spectrum disorders caused by the genetic disruption of any ProSAP/Shank family member have not only emerged to play a central role in the field, but sustainably strengthened the involvement of key synaptic proteins in neurodevelopmental disease (for comment/opinion see [25] and for review see [10] and [26]). For this and other reasons, mutant animal models of all three ProSAP/ Shank family members have been generated and characterization is ongoing. For Shank1 and ProSAP1/Shank2 mutants (main features of both models are summarized in Table 1), one future direction should hence include brain region, neuronal subpopulation, layer and pathway‐specific analyses, combined with attempts to selectively reverse phenotypes at different developmental stages. This will certainly tell us more about the origin of the pathological features caused by Shank1 and/or ProSAP1/Shank2 disruption and their reversibility throughout the lifespan. Another direction should address compensatory effects by other family members. These in particular seem to occur only in subcellular compartments and specific brain regions at certain stages of postnatal development, rather than being pleiotropic effects [20], thus giving rise to elucidating putative local post-translational regulatory mechanisms of the ProSAP/Shank family. Furthermore, the differences in hippocampal neurophysiology between Shank2^e7-/- and Shank2^e6+7-/- mutants are indeed very interesting and of special interest for closer investigation. This is because of two reasons: 1) Overlapping, but also differing parameters have been observed among the hitherto published mutants of closely related family member Shank3 that are all based on different genetic targeting strategies (for more details see Yong‐hui Jiang’s review in this issue), 2) It could well be that distinct deletions/gene targeting result in distinct phenotypes, which would have an important meaning for the future development of therapeutic strategies. To conclude, the final aim is to identify therapeutic targets, and this might be the biggest challenge. One prerequisite certainly is the basic neurobiological and neurophysiological characterization of each model on a more region‐ and circuit‐specific level to clearly pinpoint the underlying defects finally causing the behavioral changes remniscient of ASD.

Table 1: Main features of Shank1 and ProSAP1/Shank2 deletion mouse models.

The authors are supported by the Deutsche Forschungsgemeinschaft (DFG 1718/3-1 and 4-1 to TMB) and Ulm University (Baustein L.SBN.0081 to MJS and the International Graduate School in Molecular Medicine)