The interaction of Shank proteins with their various partners can be regulated by several mechanisms. First, alternative exon splicing results in multiple splice variants of Shank3 in the human brain (17). Splice sites are present downstream of the SH3 and PDZ domains and within the PRC and SAM domains of Shank3 (25) and the human brain shows higher expression of exons encoding these functional regions (60). Second, Shank3 has six intragenic promoters resulting in differential expression of isoforms, named Shank3a through Shank3f, during brain development (25, 61, 62). Finally, the subcellular and tissue-specific patterns of expression and turnover of Shank isoforms during development and synaptic activity are also regulated by methylation (63–65) and ubiquitination (66).
The Shank3 rat model used here demonstrated attentional deficit and reduced hippocampus-to-mPFC signaling in both Shank3-Het and Shank3-KO rats compared to WT (55), pointing towards a synaptic deficit in the mPFC. The synaptic deficits may be due to either changes in synapse density or decreased recruitment of proteins to the PSD that would be visible as ultrastructural changes of the PSD or spine head. Our results show no significant change in the total density of excitatory synapses among the three experimental groups at 5 weeks of age. Although few studies have assessed synapse density per se, our results are corroborated by previous reports showing no change in spine density in the neocortex of Shank3-Het mice (67) and in the hippocampus of Shank3-KO mice with an exon 21 deletion (68). However, other studies have reported a significant decrease in spine density in Shank3-deficient models compared to controls, as observed in neurons differentiated from human induced pluripotent stem cells carrying SHANK3 mutations (69), the PFC of a macaque model with deletions in exons 6 and 12 (70), the hippocampus of 5-week-old Shank3-KO rats with deletion of exons 11–21 and loss of all the Shank3 isoforms (71), the striatum of 5-week-old Shank3-KO mice lacking Shank3b following deletion of exons 13–16 containing the PDZ domain (72), and the cerebellum of Shank3-Het mice with an exon 21 deletion (73). However, Shank3-KO mice with an exon 4–9 deletion and loss of the Shank3a and Shank3b isoforms bearing the ANK and SH3 domains showed decreased spine density in the hippocampus at 4 weeks of age, but no change at 10 weeks (74). Shank3-KO mice with exons 4–22 deleted and a complete loss of all Shank3 isoforms and splice variants showed decreased spine density only in the striatum, but not the hippocampus, at 8 weeks of age (75). In vitro studies have shown that full-length Shank3 was localized in dendritic spines, whereas a C-terminal truncated isoform was diffusely distributed in dendrites and axons (61) and the isoform length determined its effects on dendritic spine density (62). Thus, the age of the animal, the extent of Shank3 deletion, and the brain regions examined, all determined whether synaptic density was altered in Shank3-deficient animals.
Synapses show different structures, with perforated or non-perforated PSDs according to changing neurotransmission efficacy in response to activity (76, 77) and function (76–79). We did not observe any differences in density of non-perforated or perforated synapses in the PFC among the three genotypes of rats. However, in the hippocampus of 5-week old Shank3-Het mice where the ANK domain was deleted, non-perforated synapse density was unchanged but perforated synapse density was higher as compared to the Shank3-KO and WT groups; this change was not present at 3 months of age (59). The observed difference in density of the perforated synapses between the two Shank3 models may be due to different effects of Shank3 in the two brain areas studied.
Maximal PSD length was unchanged, but PSD area and HD were increased in the Shank3-Het group and not in the Shank3-KO as compared to WT rats. The PSD area was calculated using the PSD length across the series of sections where the synapse was visible. We did not find a significant increase in maximal PSD length, leading us to suppose that the observed change in PSD area is small but significant when adding the PSD length across the thickness of the dendritic head to find the PSD area. The observed change in size of the spine head could also modify the PSD area. Similar to our results, PSD length and thickness were unchanged in hippocampal CA1 neurons of Shank3-KO mice with an exon 4–9 deletion (74). In a different model with a similar deletion, PSD length and area as well as HD were unchanged in the hippocampus of both Shank3-Het and Shank3-KO mice compared to WT at 5 weeks and 3 months of age (59). These results were confirmed in the hippocampus of Shank3-KO mice with exon 4–21 deletion, but in the striatum of these mice, decreased PSD length and thickness were seen at 8 weeks of age (75). Of note, the increase in PSD area and HD that we observed was present only in the Shank3-Het rats compared to WT and no change in the synaptic ultrastructure was observed in the Shank3-KO group. The Shank3-Het rats carry one copy of undeleted Shank3 that can express the full-length isoforms of the protein, though at lower levels than the WT. Shank3-KO rats express no full-length Shank3 and the truncated isoforms in these animals may not be sufficient to recruit the Shank3 binding partners in the PSD that potentially compensate for this lack in the Shank3-Het rats. The Shank3-KO rats may be able to maintain the structure of dendritic spines and the PSD comparable to WT through the shorter Shank3 isoforms or by recruiting the other Shanks to the PSD. Although the ANK domain of Shank3 is deleted in our model, the synapse-targeting SAM domain as well as the major binding sites on Shank3 for recruiting the NMDA and AMPA receptors to the PSD and for binding cytoskeletal proteins essential for spine morphology are preserved in the Shank3-KO and may be sufficient to maintain synaptic morphology at a level similar to that in the WT.
In the Shank3-Het rats that carry only one copy of the gene and can express full-length Shank3 together with the shorter isoforms, the other Shank family proteins may compensate for the loss of Shank3, resulting in the observed increase in size of the dendritic head and PSD area. Shank2, that shares both a homologous PRC domain and a synapse-targeting C-terminal region with Shank3 (22, 28), seems to be the better candidate to compensate for reduced Shank3 expression. Indeed, Shank2 with an intact PRC domain can rescue reduced head diameter of dendritic spines in hippocampal neurons, induced by a knockdown of all three Shank proteins (80). Overexpression of Shank1 containing the PRC domain also results in enlargement of spine head size (81). In contrast to our results, decreased spine head volume was observed in human neurons differentiated from induced pluripotent stem cells sourced from subjects with Shank3 mutations (69). Furthermore, a deletion of the ANK-SH3 domains of Shank3 results in reduced spine head area in mouse hippocampal neurons (53). In our Shank3-Het rat model, not only are low levels of the full-length Shank3 protein expressed, but also shorter isoforms lacking the ANK domain but containing the PDZ domain which recruits NMDA and AMPA receptors during spine maturation (53), the PRC domain where Homer1 and cortactin bind (20, 21, 25), and the synapse-targeting SAM domain (61). Thus, overcompensation by the full-length protein recruiting the shorter Shank3 isoforms or other Shank proteins and giving an enlargement of the HD and PSD area seems probable.
PSD fractions from the neocortex of Shank3-Het and Shank3-KO rats with exons 11–21 deleted in Shank3 (71) and hippocampal neurons after knockdown of four major isoforms of Shank3 in vitro show no change in Shank1 or 2 expression (82), but other compensatory mechanisms may occur in vivo in the PFC. A study in the mouse brain found that proteins interacting with Shank3 in the PFC are different from those in the hippocampus and striatum (83). Dimethylation of histone lysine 9 following higher expression of euchromatic histone methyltransferases is found in the PFC, but not striatum or hippocampus, of 5- to 6-week-old Shank3-Het mice with an exon 21 deletion and also in BA9 from postmortem ASD brains, suggesting a brain region-specific mechanism for regulation of protein expression (84). The same mouse model also shows low levels of histone acetylation in the PFC as compared to WT mice (85). Notably, the expression of Shank3 is highest in the PFC compared to other regions in the macaque brain (70), whereas Shank3 isoforms containing ANK domain are highly expressed in the upper cortical layers, hippocampus and striatum of the mouse brain (86). Multimer formation of Shank proteins may enable their recruitment to the synapse without an overall increase in protein concentration. Shank1 forms multimers via interactions between ANK repeats and SH3 domain (87) or PDZ domains (88). The same is true of Shank3 via its PRC and SAM domains (20), although these interactions remain to be proven endogenously in neurons. Interestingly, the interaction of ANK repeats and SH3 domains in Shank1 is reported to require all the repeats in the ANK domain (87) and, if this extends to Shank3, the lack of the ANK domain in Shank3 of the Shank3-KO rats could explain their difference from the Shank3-Het rats in our study. Despite these potential compensatory mechanisms for some functions when Shank3 is deficient, the presence of clinical phenotypes due to Shank3 haploinsufficiency in humans carrying SHANK3 mutations (6, 7, 9, 13) and of behavioral deficits in animal models with Shank3 deficiency (55, 67, 71, 89) suggests such compensation may not be fully effective to rescue the functional phenotype but may be able to partially preserve the ultrastructure of synapses.
Limitations:
Although the prevalence of PMS is equal in both sexes (90), ASD occurs four times more frequently in males than females (1). This study used only male mice to uncover synaptic changes in the PFC relevant to both ASD and PMS. The deletion in exon 6 of the Shank3 gene would affect the N-terminal ANK domain, truncating the full-length protein, but shorter isoforms of Shank3 may still be expressed in the Shank3-Het and Shank3-KO rats. Thus, the changes observed using this rat model would more accurately reflect the effects of mutations affecting the N-terminal of the protein than those resulting from deletion of the gene and the loss of all isoforms of the protein.