Integrin β1/FAK/SRC signal pathway is involved in autism spectrum disorder in Tspan7 knockout rats

TSPAN7 knockout causes ASD-like phenotypes in rats through the integrin β1/FAK/SRC signal pathway, which provides novel insights into the role of TSPAN7 in psychiatric diseases, highlighting SRC as a potential target.


Introduction
Autism spectrum disorder (ASD) is a complex, highly genetically heterogeneous neurodevelopmental disorder with comorbidities, frequently associated with intellectual disability (ID) comorbidities, affecting 1-2% of children worldwide (Lord et al, 2018;Persson et al, 2020;Qin et al, 2021). ASD is diagnosed before the age of 3 yr and remains throughout life (Berkel et al, 2012). The main specific autistic characteristics are social communication deficits, interest deficits, repetitive stereotyped behavior, and mental retardation in some cases (Sharma et al, 2018). Even though a specific etiology can be identified in some individuals with ASD, most of the variation that increase likelihood of ASD is believed to have its origin in a complex interaction between genetic and environmental factors (Bai et al, 2019;Bottema-Beutel et al, 2020;Hegarty et al, 2020). Genetic factors included copy number variations, single-nucleotide polymorphisms, and epigenetic alterations of genes such as E3 ubiquitin ligase-A (Ube3a), neuroligin-3 (Nlgn3), SH3 and multiple ankyrin repeat domains 3 (Shank3), methyl-CpG-binding protein 2 (Mecp2), phosphatase and tensin homolog deleted on chromosome 10 (Pten), tuberous sclerosis complex subunit 1 (Tsc1), and tuberous sclerosis complex subunit 2 (Tsc2) (Jamain et al, 2003;Varghese et al, 2017;Xu et al, 2018;Lee et al, 2021). A variety of environmental factors had been found to increase the likelihood of ASD, including premature birth, fetal exposure to psychotropic drugs, or pesticides (Sharma et al, 2018). An X-linked gene Tspan7 (also named Tm4sf2) had been found to be associated to developing ASD, which encodes for tetraspanin7 (TSPAN7) protein (Noor et al, 2009;Piton et al, 2011).
Previous studies found that TSPAN7 was involved in spine maturation in cultured hippocampal neurons through a direct interaction with PICK1 (Bassani et al, 2012). Tspan7 knockout caused an ASD-related phenotype in mice, and the deletion of TSPAN7 altered voltage-gated ion channels function and decreased PKC-ERK signaling (Murru et al, 2021). Synaptic dysfunction had been revealed to be a hallmark of ASD (Guang et al, 2018). However, the underlying mechanisms of TSPAN7 in ASD at the synapse and synaptic integrity level remain unclear. To investigate the effect of TSPAN7 deficiency on brain development and its possible mechanisms, we generated a Tspan7 knockout rat line using CRISPR/Cas9 technology.
The Tspan7 −/− rats displayed ASD-like phenotypes, such as increased repetitive behaviors, anhedonic-like state, altered sociability, and impaired learning and memory similar to mice. Interestingly, Tspan7 −/− rats showed obvious brain structure alteration and pathological changes, such as decrease in hippocampal and cortical volume, loss of hippocampal pyramidal neurons, and the reduction in synaptic density, which was not reported in Tspan7 −/y mice. More importantly, we found that deletion of TSPAN7 protein interrupted the distribution of integrin β1 on the neuron membrane; subsequently reduced integrin β1/FAK/SRC signaling; and downregulated the expression of synaptic associated proteins, such as postsynaptic density 95 (PSD95), synaptophysin (SYN), and glutamate reportor 1/2 (GluR1/2). These founding provided a new clue to understand the function and the importance of TSPAN7 in the development of the brain and cerebral pathogenesis of ASD.

Bioinformatics analysis of TSPAN7
The TSPAN7 amino acid sequences among human, macaque, rat, mouse, clawed frog, and zebrafish were compared by the alignment Figure 1. Bioinformatics analysis of TSPAN7. (A) TSPAN7 protein sequences in different species were obtained from the GenBank database, and the identity of TSPAN7 proteins in various species of H. sapiens, M. mulatta, R. norvegicus, M. musculus, X. tropicalis, and D. rerio were analyzed using the alignment tool of DNAMAN and ClustalX. (B) mRNA expression of TSPAN7 in different tissues were compared among the three species of human, rats, and mice, where the data were obtained from the GenBank database and normalized by the data of brain tissues. (C, D, E, F) Differential expression of TSPAN7 in the brains of ASD, HD, AD, and PD patients were compared with control (CTL), where data were obtained from the Gene Expression Omnibus database. (G, H, I, J) mRNA level of Tspan7 and protein expression of TSPAN7 in the heart, liver, spleen, lung, kidney, and brain tissues from WT Wistar rats at age of 2 mo were detected by RT-PCR and Western blot and quantified by ImageJ. n = 3, *P < 0.05, **P < 0.01, ***P < 0.001 versus brain. Data information: In (C, D, E, F), data were analyzed by two-tailed unpaired t tests. Error bars represent SEM. *P < 0.05, **P < 0.01, n.s, not significant. Source data are available for this figure.
TSPAN7 is involved in ASD by SRC signal Pang et al.
https://doi.org/10.26508/lsa.202201616 vol 6 | no 3 | e202201616 tool in the database of BlastP, GenBank, PubMed, and Ensembl. Strikingly, 89.92% identity among the human and five kinds of chordates was delineated ( Fig 1A) and suggested that Tspan7 was an evolutionarily conserved gene. The expression data from database of GenBank were normalized by brain tissue and showed a similar expression pattern in different tissues among the three species of human, rats, and mice. Tspan7 was highly expressed in tissues of the brain and lung and lowly in tissues of the spleen, heart, kidney, and liver ( Fig 1B). The published datasets also showed that the expression of TSPAN7 was decreased in the brain tissues from patients of ASD, Huntington's disease (HD), Parkinson's disease (PD), and Alzheimer's disease (AD) (Fig 1C-F, *P < 0.05, **P < 0.01, n.s, not significant). Finally, the TSPAN7 expression in different tissues was confirmed by RT-PCR and Western blot, and the results showed that TSPAN7 expression was the highest in the rat brain ( Fig  1G-J, n = 3, *P < 0.05, **P < 0.01 and ***P < 0.001 versus brain).
Generation and general observation of Tspan7 2/2 rats Immunofluorescence analysis showed that TSPAN7 was expressed in NEUN-stained neurons of the rat brain (Fig 2A). The Western blot indicated high-level expression of TSPAN7 proteins in the rat brain from 0.5 to 4 mo of age and the relative lower expression of TSPAN7 at birth (Fig 2B and C, n = 6, *P < 0.01). We used the CRISPR/Cas9 technique to knock out the Tspan7 ortholog in rats and acquired two Tspan7 −/y founders. One Tspan7 −/y founder was selected to breed and obtain Tspan7 knockout rats, which had a deletion of a 546-bp fragment containing exon 2 and part of intron 2 of the Tspan7 ortholog ( Fig 2D). The deletion caused TSPAN7 protein ablation, and Western blot analysis confirmed the absence of TSPAN7 protein in the brains of Tspan7 KO rats n = 3). No obvious changes of body weight between WT and Tspan7 KO rats were detected (Figs 2H and S1G, n.s). However, the ratio between male (Tspan7 −/y ) to female (Tspan7 −/− ) rats was reduced significantly ( Fig 2I).
Tspan7 2/2 rats displayed ASD-like behaviors Because of the low birth rate of Tspan7 −/y male rats, we used Tspan7 −/− rats to perform the experiments in this current study. The five behavior tests were performed here. First, the Tspan7 −/− rats displayed abnormal self-grooming behavior indicated with Figure 2. Determination of the Tspan7 expression pattern in the rat brain and the generation of Tspan7 2/2 rats. (A) Paraffin sections of the brain from WT Wistar rats at 6 mo old were stained by double immunofluorescence with antibodies of anti-NEUN (green) and anti-TSPAN7 (red). The images showed that TSPAN7 was expressed in mature neurons of the rat brain (Scale bar = 10 μm). (B, C) Protein expression level of TSPAN7 in the whole-brain tissue from WT rats at age of 0 d, 0.5, 1, 2, and 4 mo was detected by Western blot with the anti-TSPAN7 and quantified with ImageJ. n = 6, *P < 0.05. (D) Exon 2 of Tspan7 was knocked out as shown in the strategy of the CRISPR/Cas9 method. (E) Positive pups were genotyped by tail DNA PCR. N was negative, indicating WT. Numbers 4, 7, and 8 are WT; numbers 2 and 3 are heterozygotes; numbers 1, 5, and 6 are homozygotes (n = 8). (F, G) Deletion of TSPAN7 protein in brain tissues of Tspan7 −/− rats was detected by Western blot with the anti-TSPAN7 and quantified with ImageJ (n = 3). (H, I) Body weights and birth rates were compared between WT and Tspan7 −/− rats at 6 mo old. Data information: In (G), data were analyzed by two-tailed unpaired t tests. Error bars represent SEM. ***P < 0.001, n.s, not significant. Source data are available for this figure. increase of self-grooming episodes and times and no significant change in the total distance ( Fig 3A-D, n = 13, *P < 0.05, ***P < 0.001, ns). Second, Tspan7 −/− rats reduced by 6.7% uptake of sucrose water (1%) compared with the WT rats, which naturally displayed a preference for respect 1% sucrose to the sole water ( Fig 3E and F, n = 13, **P < 0.01). Third, Tspan7 −/− rats displayed Female WT and Tspan7 −/− rats at 6 mo old were first applied to the self-grooming test. The self-grooming episodes, self-grooming times, and the total distance were recorded. (E, F) Rats were then applied to the sucrose preference test, and uptake percentage of 1% sucrose solution was calculated and compared between WT and Tspan7 −/− rats. (G, H, I, J) Three-chamber sociability test was performed. The discrimination index for sociability (stranger versus object cage), the total investigation time, and distance were recorded and analyzed. (K, L, M) Y-maze test was performed. The spontaneous alternation (%) and total distance were, respectively, calculated and compared between WT and Tspan7 −/− rats. (N, O, P, Q, R) Morris water maze test was performed, and the latency to find the hidden platform during the five training days, the platform frequency, duration in target quadrant, the total movement, and swimming track on day 6 were compared between WT and Tspan7 −/− rats. Data information: In (B, C, E, G, I, K, L, M), data were analyzed by two-tailed unpaired t tests. Error bars represent SEM. n = 13, *P < 0.05, **P < 0.01, ***P < 0.001, ns, not significant. Diagrams of (A, D, F, H, J) were created with BioRender.com. Source data are available for this figure. . Analysis of the pathological changes in the brain of Tspan7 2/2 rats. (A, B) Female WT and Tspan7 −/− rats at 6 mo old were anesthetized and examined on an MRI system, and the volume of the hippocampus, cortex, cerebellum, and cerebrum was analyzed on the working station; n = 6 in the WT group and n = 5 in the Tspan7 −/− group. (C) Rats were euthanized, and the whole brains were stripped out, and the ratios of brain/body weight were calculated; n = 12 each group. inactive social behavior indicated with spending less time investigating the stranger rats compared with that of WT rats in the three-chamber sociability test (Fig 3G and H, n = 13, **P < 0.01). However, there was no significant difference in the total investigation time and distance between the Tspan7 −/− rats and WT rats (Fig 3I and J,n = 13,ns). Moreover, Tspan7 −/− rats showed impaired working memory indicated with the decrease of spontaneous alternations in the Y-maze test (Fig 3K and L, n = 13, **P < 0.01) and did not change the total distance (Fig 3M,n = 13,ns). Finally, in the Morris water maze test, Tspan7 −/− rats displayed an increased latency to find the hidden platform during the acquisition phase ( Fig 3N, n = 13, *P < 0.05). During the probe test, Tspan7 −/− rats displayed abnormal spatial memory indicated with a significant reduction of the platform crossing number and target quadrant duration compared with WT rats (Fig 3O, P, and R, n = 13, *P < 0.05). However, there is no significant difference in the total movement between the Tspan7 −/− rats and WT rats (Fig 3Q, n = 13, ns).

TSPAN7 deficiency caused brain structure changes
The geometry changes of the brain in Tspan7 −/− rats were first examined by magnetic resonance imaging (MRI) and showed a significant decrease in hippocampal and cortical volume in Tspan7 −/− brains compared with those of WT rats, whereas no obvious difference in cerebral and cerebellar volumes was detected ( Fig 4A and B, n = 6, *P < 0.05, n.s). The gross anatomy examination showed no significant change in the brain coefficient (brain weight/body weight) of Tspan7 −/− rats compared with WT rats ( The H&E staining showed the gross structure of the hippocampus and decreased pyramidal neuron numbers in CA1, CA2, and CA3 regions of Tspan7 −/− rat brains and WT rat brains ( Fig 4D-I, n = 6 in the WT group and n = 5 in the Tspan7 −/− group, *P < 0.05, **P < 0.01). Further Nissl staining showed that the neurons were more loosely arranged and hypochromic in the Tspan7 −/− brain of rats compared with those of WT rats (Fig 4E-J). The Nissl-positive neurons were significantly decreased in CA1, CA2, and CA3 regions of Tspan7 −/− brains in comparison with those of WT rats ( Fig  4J-M, n = 6 in the WT group and n = 5 in the Tspan7 −/− group, *P < 0.05, **P < 0.01), whereas no significant difference in the DG area of the hippocampus and cortex was observed (Fig 4J, N, and O, n = 6 in the WT group and n = 5 in the Tspan7 −/− group, n.s).

TSPAN7 deficiency altered expression of autism-related genes
To further explore the molecular events involved in TSPAN7, we analyzed the expression of nine ASD-related genes by qRT-PCR in the hippocampus and cortex regions of the rats. The results revealed that the expression of four synaptic-related genes of Ube3a, PSD95, Nlgn3, and Syn was significantly down-regulated in Tspan7 −/− rats (Fig 5A-D, n = 6, *P < 0.05). However, expression of the Shank3, Mecp2, Pten, Tsc1, and Tsc2 was not significant changed in Tspan7 −/− rats. (Fig 5E-H, n = 6, n.s). These results suggested that TSPAN7 could be involved in specific autistic features through a synaptic mechanism.

TSPAN7 deficiency impaired the neuronal morphology
To obtain a complete understanding of the changes in neuronal morphology and dendritic spines in Tspan7 −/− rats, we performed Golgi-Cox staining on the hippocampus and cortex of Tspan7 −/− rats and WT rats. We found that the dendritic and axonal branching was decreased in the Tspan7 −/− brain ( Fig 6A, n = 3). Then, the number of intersections of dendrites of neurons at 40-140 μm in the Tspan7 −/− brain was significantly reduced compared with that of the WT brain (Fig 6B, n = 3, *P < 0.05, ***P < 0.001).
The dendritic spines were divided into five types: filopodia spines, long-thin spines, thin spines, stubby spines, and mushroom spines according to its length and width described by the previous report (Risher et al, 2014). Among them, the length of filopodia was >2 μm, the length of long-thin spine <2 μm, the length of thin spine <1 μm, the ratio of length/width of stubby spine <1, and the width of mushroom spine >0.6 μm ( Fig 6C). The total number of spines at both the apical dendrites and basal dendrites was significantly reduced in the Tspan7 −/− brain compared with that of the WT brain. The reduction of spines mainly existed in mushroom spine, stubby spine, thin spine, and long-thin spine, whereas the filopodia was unchanged obviously (Fig 6D-F, n = 3, *P < 0.05, **P < 0.01, ***P < 0.001, n.s). The integrity of synapses was detected by immunofluorescent staining with PSD95 antibody and SYN antibody. The overlapping puncta of PSD95-immunolabelled postsynaptic and SYN-immunolabelled presynaptic puncta were analyzed, and the results showed that the overlapping puncta were significantly decreased in the hippocampus and cortex of the Tspan7 −/− brain compared with the WT brain (Fig 7A-J, n = 3, *P < 0.05).

TSPAN7 deficiency inactivated the integrin β1/FAK/SRC signal pathway
The binding of TSPAN7 with integrin β1 in primary neurons was indicated in previous report (Bassani et al, 2012) and raised a question that how TSPAN7 regulated the integrin β1 signal pathway. We first observed the decreased levels of integrin β1 by immunofluorescence staining in primary Tspan7 −/− neurons of the hippocampus and cortex. (Fig 8A and C, n = 9, **P < 0.01). A significant reduction of integrin β1 in total lysates and plasma membrane fraction in the brain tissues of Tspan7 −/− rats was then confirmed by Western blot, whereas the levels of integrin β1 in cytosol fraction were not changed (Fig 8B and D-F, n = 6, **P < 0.01, n.s).
FAK and SRC are the central downstream mediators of the integrin-dependent signal pathway (Kalia et al, 2004;Ribeiro et al, 2013). The phosphorylation of FAK and SRC were also inhibited corresponding to the abnormity of integrin β1 in the brain tissues of Tspan7 −/− rats (Fig 8B, G, and H, n = 6, *P < 0.05). SRC appeared to be involved in regulation of synaptic function (Theus et al, 2006;Cho et al, 2013;Zhang et al, 2018). Corresponding to the inactivation of SRC, the expression of PSD95, SYN, GluR1, and GluR2 was significantly inhibited in brain tissues of Tspan7 −/− rats compared with that of WT rats (Fig 8B and I-L, n = 6, **P < 0.01, ***P < 0.001). Our results suggested that the TSPAN7 deficiency, as a binding protein, interrupted integrin β1 localization on the plasma membrane and then affected the activation FAK and SRC. The inhibition of SRC could be resulted by the down-regulation of synapse-related genes and impairment of synaptic integrity.

Recovery of SRC activity improved synaptic damage in primary cultured neurons
We then isolated primary neurons from the brains of WT and Tspan7 −/− rats to confirm the results of the roles of SRC in regulating synapse-related genes and synaptic integrity in brain tissues. The WT neurons were treated with or without an SRC inhibitor (bosutinib) and an SRC activator (EPQpYEEIPIYL), and the Tspan7 −/− neurons were treated with or without the SRC activator. In WT neurons, when the SRC phosphorylation was inhibited by bosutinib, the expression of PSD95, SYN, GluR1, and GluR2 was significantly inhibited. Like the inhibitor, in Tspan7 −/− neurons, Tspan7 knockout also caused the inhibition of SRC signal and the down-regulation of the synaptic proteins and receptors. But when the SRC activator was administrated to the cells, the SRC phosphorylation, as well as the expression of PSD95, SYN, GluR1, and GluR2, was significantly enhanced or restored in the WT and Tspan7 −/− neurons (Fig 9A-F, n = 3, *P < 0.05, **P < 0.01). Furthermore, PSD95 and SYN double immunofluorescence confirmed the impairment effect of SRC inhibitor and Tspan7 knockout on synaptic structure proteins and the protection effect of SRC activator on synaptic structure proteins in WT and Tspan7 −/− neurons (Fig 9G and H, n = 9, *P < 0.05, ***P < 0.001). The results suggested that TSPAN7 knockout caused the impairment of synaptic integrity through regulating the integrin β1/FAK/SRC signal pathway.

Discussion
Tspan7 gene mutations are correlated to ID and ASD (Piton et al, 2011;Bassani et al, 2013;Penzes et al, 2013). It has been indicated that TSPAN7 deficiency altered synapse structure and functionally impaired learning and memory (Murru et al, 2017). The summary of data from BlastP, GenBank, PubMed, and Ensembl showed that Tspan7 was an evolutionarily conserved gene in vertebrates (Fig 1A). The Tspan7 exhibited a similar expression pattern in human, rats, and mice, which was highly expressed in the brain and lung but lowly in tissues of the spleen, heart, kidney, and liver (Fig 1B, G, and  H). Tspan7 was also found to be down-regulated in patients' brains in a variety of diseases including ASD, HD, PD, and AD (Fig 1C-F, *P < 0.05, **P < 0.01, n.s), implying the important role in brain disease.
Consistent with Tspan7 −/y mice, the Tspan7 −/− rats showed both of ASD-like and ID-like phenotypes with decreased sociability, increased self-grooming, decreased sucrose preference, and impaired spatial memory and working memory compared with WT rats (Fig 3, n = 13, *P < 0.05, **P < 0.01 ***P < 0.001). Except for the similar behaviors between Tspan7 −/y mice and Tspan7 −/− rats, we found obvious pathological changes in Tspan7 −/− brains, which was not observed in Tspan7 −/y mice, including the decreased volume of the hippocampus and cortex (Fig 4A and B, n = 6, *P < 0.05); decreased number of neurons; and Nissl-positive neurons in CA1, CA2, and CA3 regions (Fig 4D-H, n = 5, *P < 0.05, **P < 0.01). The hypochromic hippocampus was also observed in Tspan7 −/− brains by Nissl staining (Fig 4E-J), which detected specifically the Nissl bodies (endoplasmic reticulum) in the cytoplasm of neurons and indirectly report the status of protein synthesis in neurons (Kádár et al, 2009). These results suggested that the TSPAN7 deficiency resulted in the weakness of hippocampal neurons compared with that of WT rats. Furthermore, we detected expression of several ASD-related genes in rat brains, which was involved in various biological processes, such as an abnormal mTOR signaling pathway, cell growth, cell proliferation, and chromatin remodeling (Kwon et al, 2006;Tsai et al, 2012;Yin & Schaaf, 2017). Previous studies found that total brain weight and cortical volume were significantly reduced in Ube3a-and NLGN3-deficient mice (Varghese et al 2017). In the Tspan7 −/− rat model, we found that the volume of the cortex and hippocampus of the Tspan7 −/− rats was significantly reduced. The expression of Ube3a, PSD95, Nlgn3, and Syn were significantly down-regulated in the Tspan7 −/− hippocampus and cortex compared with those of WT rats (Fig 5A-D, n = 6, *P < 0.05), which are important for synapse formation and remodeling (Kim et al, 2008;Liu et al, 2018;Xu et al, 2018;Chen et al, 2020). It suggested that the down-regulation of synapse-associated genes such as Ube3a, PSD95, Nlgn3, and Syn in the hippocampus and cortex was closely associated with the decrease in hippocampal and cortical volume caused by TSPAN7 knockout, and TSPAN7 deficiency could be involved in pathogenesis of ASD through a synaptic mechanism.
We then analyzed the changes of neuron morphology and dendritic spines and found that neuronal complexity and dendritic spines were significantly decreased in the Tspan7 −/− brain compared with the WT brain (Fig 6, n = 3, *P < 0.05 **P < 0.01, ***P < 0.001).
TSPAN7 is the member of the tetraspannin family, which regulated proteins interacting with C kinase 1 (PICK1), and the PICK1 guides the AMPA receptor trafficking by its PDZ domain in primary neurons (Bassani et al, 2012). However, activation of PKC-ERK, rather than the changes of PICK1, PSD95, and SYN, was detected in Tspan7 −/y mice (Murru et al, 2017(Murru et al, , 2021. Unexpectedly, we did not observe any significant changes of phosphorylation on ERK, PKCε (PKC epsilon), and PKCα and expression levels of ERK, PKCε, PKCα, PICK1, and NMDAR1 in Tspan7 −/− brains compared with those of WT rats (Fig S1, n = 5).
Besides the PICK1, integrin β1 has been identified as another binding protein of TSPAN7 (Bassani et al, 2012), suggesting that TSPAN7 might be involved in integrin β1-initiated signaling. We found impaired integrin β1 distribution in primary neurons of In (E, F), data were analyzed by two-tailed unpaired t tests. Error bars represent SEM, n = 3. Each point represents data from one spine, and at least six spines were analyzed for each rat, *P < 0.05, **P < 0.01, ***P < 0.001, n.s, not significant. Diagram of (C) was created with BioRender.com. Source data are available for this figure.
TSPAN7 is involved in ASD by SRC signal Pang et al.
Integrin β1 is localized to the excitatory-synapse postsynaptic membrane and is critical for synaptic integrity (Mortillo et al, 2012;Figure 8. Determination of the signal pathways in brain tissues. (A) Primary neurons from embryonic day 17. WT rats and Tspan7 −/− rats were cultured for 6 d and then were stained by double immunofluorescence with antibodies of anti-MAP2 (green) and anti-integrin β1 (red). Scale bar = 10 μm. (B) Plasma membrane fractions, cytosol fractions, and total lysates of brain tissues from female WT and Tspan7 −/− rats at 6 mo old were prepared and separated on SDS-PAGE, respectively. Integrin β1 in the membrane, cytosol fractions, and total lysates and phosphorylated and total FAK, SRC, PSD95, SYN, GluR1, and GluR2 in total lysates were detected with Western blot. (C) Integrin β1-immunolabelled area in primary neurons was quantified. (D, E, F, G, H, I, J, K, L) Expression levels of integrin β1 in the membrane, cytosol fractions, and total lysates and phosphorylated FAK, SRC, PSD95, SYN, GluR1, and GluR2 in total lysates were quantified with a software of ImageJ. n = 6. Data information: In (C, D, E, F, G, H, I, J, K), data were analyzed by two-tailed unpaired t tests. Error bars represent SEM, *P < 0.05, **P < 0.01, ***P < 0.001, n.s, not significant. (C) Each point represents data from one cell in (C), and at least nine cells were analyzed. Source data are available for this figure.  Heintz et al, 2014;Wang et al, 2016). The activated or clustered integrin β1 in the cell membrane recruits cytoskeletal proteins, RhoGTPases, and non-receptor tyrosine kinases, including FAK and SRC (Mitra et al, 2005). The recruited FAK is autophosphorylated at tyrosine 397 and then causes SRC recruitment and activation (Arold, 2011). FAK is a major protein that integrates integrin β1 signaling and function in neurons (Ribeiro et al, 2013). SRC is widely expressed in neurons and implicated in proliferation and differentiation during the development of the central nervous system (Kalia et al, 2004). SRC is also a crucial element and appears to be involved in regulation of synaptic integrity, which can indirectly regulate the expression of PSD95 (Cho et al, 2013), glutamate receptors , and synaptophysin (Theus et al, 2006).
Our results showed that inhibition of SRC phosphorylation by bosutinib reduced the expression of PSD95, SYN, and GluR1/2 and impaired synaptic integrity in primary WT neurons, whereas the level of synaptic proteins and glutamate receptors were enhanced by the recovery of SRC activity with an activator in Tspan7 −/− neurons (Fig 9, n = 9, *P < 0.05, ***P < 0.001). The results suggested that SRC was essential for regulation of PSD95, SYN, and GluR1/2 expression via TSPAN7/integrin β1/FAK signaling.
In conclusion, our finding indicated that TSPAN7 deficiency caused both of ASD-like and ID-like phenotypes in rats and obvious pathological changes in brain tissues including neuron loss and synaptic impairment in the hippocampus. The integrin β1/FAK/SRC signal pathway was interrupted by TSPAN7 deficiency and resulted in the down-regulation of PSD95, SYN, and GluR1/2, which were essential proteins of synaptic integrity. Our results suggested that TSPAN7/integrin β1/FAK/SRC was a new signal pathway involved in pathogenesis of ASD and ID related to TSPAN7.

Data extraction
The TSPAN7 protein sequence of different species was obtained from the GenBank database. Multiple sequence alignment of TSPAN7 sequences of human (H. sapiens; NP_004606), macaque (M. mulatta; NP_001248645), rat (R. norvegicus; NP_001385666), mouse (M. musculus; NP_062608), clawed frog (X. tropicalis; NP_989350), and zebrafish (D. rerio; NP_999939) was performed by tools like DNAMAN and Clustal, and the identities of the amino acid sequence were analyzed. The expression of TSPAN7 in different tissues in human, rats, and mice was obtained from the GenBank database (https:// www.ncbi.nlm.nih.gov/gene/7102, https://www.ncbi.nlm.nih.gov/gene/ 363447, https://www.ncbi.nlm.nih.gov/gene/21912). The different expression data of TSPAN7 in human diseases were obtained from the NCBI Gene Expression Omnibus database. "Homo sapiens" AND "tissue" AND "disease" were first searched in the Gene Expression Omnibus datasets, and then the data with small sample sizes (n < 3) or no control group were filtered out. Finally, the data were identified for analysis, including ASD (GSE30573), HD (GSE129473), AD (GSE36980), and PD (GSE20295).

Animals
The Tspan7 knockout rats were generated using CRISPR/Cas9 following our previously described procedure (Ma et al, 2016). Briefly, sgRNA (CCACCATGGTGGGCATGTACT) and sgRNA (CCTGGCACAGTAAACAGAGTC) were designed to target exon 2 of Tspan7. The mixture containing Cas9 protein (30 ng/μl) and sgRNAs (10 ng/μl each) was microinjected into the cytoplasm and male pronucleus of the fertilized eggs from Wistar rats. The microinjected eggs were then transferred to pseudopregnant Wistar rats who gave birth to pups. The litters were genotyped by PCR with primers of 59-GCAAACTACCACATCATTCCCAG-39 and 59-CTGTGTCCATCAGCTCAGATCAG-39. The PCR fragments were analyzed by DNA sequencing to confirm the knockout DNA fragment. All rats used in this study were housed in groups and bred in an AAALAC-accredited facility, and all rat experiments were approved by the Animal Care and Use Committees of the Institute of Laboratory Animal Science of Peking Union Medical College (Approval no. MYW22004).

Quantitative real-time PCR
The fresh hippocampal and cortical tissues were sampled from WT female Wistar rats (WT) and female Tspan7 knockout rats (Tspan7 −/− ). Total RNA in brain tissues was isolated by TRIzol reagent (15596018; Invitrogen). The cDNA was produced with a reverse-transcription kit (RR047A; Takara). Quantitative real-time PCR (qRT-PCR) was applyed using SYBGreen real-time PCR kits (RR820A; Takara) on a PCR equipment (QuanStudioTM3 Real-Time PCR instrument; Thermo Fisher Scientific). The mRNA expression of nine genes related with autism was detected and normalized to Gapdh, including Tspan7, Ube3a, PSD95, Nlgn3, Shank3, Mecp2, Pten, Tsc1, and Tsc2. The primers for the quantitative real-time PCR are summarized in Table 1.

Behavior analysis
The female WT and Tspan7 −/− rats at 6 mo old were applied to five kinds of behavior analysis to estimate the levels of autism and cognitive damage of Tspan7 −/− rats.
For grooming behavior, the procedure was performed as previously described (Murru et al, 2021). Rats were individually placed into an open-field arena (80 × 80 cm) and recorded for selfgrooming behavior. The experiment was recorded for a total of 20 min, and the total time spent grooming and the total number of times the rats were groomed during the last 10 min were measured. Self-grooming was defined as licking paws, washing the nose and face, or scratching fur with any foot.
The sucrose preference test was performed according to a previous study (Schiavi et al, 2019). Rats were given two bottles, one of sucrose water (1%) and one of tap water. The amount of sucrose water and pure water consumed was evaluated at every 24 h. The preference for the sucrose solution was calculated as the percentage of sucrose solution ingested relative to the total amount of liquid consumed.
Social interaction analysis was performed according to the reported methods (Murru et al, 2021). A stranger rat (stimulus rat) and an object are, respectively, placed inside a cage in the left or right compartment. The testing rat was placed in the middle compartment and let it explore around randomly. The time of the testing rat spent on the stimulus rat and object was monitored for 10 min. The discrimination index was measured as the difference in time spent with the stimulus rat and the objective, expressed as the percentage ratio of the total time spent exploring both the stimulus animal and the objective.
The Y-maze test was performed according to reported methods (Li et al, 2020), in a rat Y-maze apparatus (arm length 90 cm). Briefly, after one rat was placed at the end of one arm, the animal was allowed to explore freely in the three arms of the maze for 5 min. The activity of the rats was acquired and analyzed using a Visu Track system (XR-VT; xinruan). The sequence number of arm entries was recorded to calculate the percentage of alternation.
The Morris water maze test was modified from a protocol of our previous study (Pang et al, 2020). Briefly, the rats were placed in a large round pool with a radius of 75 cm, and the animals were allowed to swim for 60 s once and trained to find the hidden platform twice 1 d for five straight days. The time to find the platform was recorded as latency. On day 6, the rats were placed in a contralateral quadrant to swim for 120 s in the maze without the platform. The platform crossing number and target quadrant duration were recorded using a video-tracking system as above.

Histological analysis
The brain tissues were first fixed in formalin (10%) for 2 d at room temperature, and the 4-µm thick coronal paraffin sections were prepared as a standard histological procedure.
The dewaxed and rehydrated sections were first stained with H&E as previously described (Lu et al, 2018). The stained sections were then scanned under a digital slide scanner (Pannoramic 250 FLASH; 3DHISTECH), and the images were captured by CaseViewer. The images were quantified using ImageJ software. For Nissl staining, the dewaxed and rehydrated coronal sections were stained in toluidine blue solution (G1436; Solarbio) for 30 min at 55°C. After clearing in distilled water, the slides were gradually dehydrated and given dimethylbenzene. Then the slides were coverslipped with neutral balsam. Finally, the numbers of surviving neurons per ×400 field within the hippocampal CA1 were counted.
For immunofluorescent staining, the growing adherent cells on coverslips were fixed in paraformaldehyde (4%) for 20 min at room temperature and penetrated with 0.1% Triton X-100 for 5 min. The coverslips or the dewaxed and rehydrated sections were applied to immunofluorescent staining as previously described (Li et al, 2012;Dong et al, 2021). Briefly, the primary neurons were fixed in paraformaldehyde (4%) for 20 min at room temperature and penetrated with 0.1% Triton X-100 for 5 min. Then the slices washed 3 times with PBS for 5 min. The sections were blocked in 1% BSA at room temperature for 1 h and then incubated with primary antibodies overnight at 4°C. The sections were incubated with secondary fluorescent antibodies for 1 h in a dark box at room temperature and then washed off by PBS. The sections were mounted with DAPI buffer (ZLI-9557; ZSGB BIO) and observed by two-photon confocal laser scanning microscopy (Leica). The images were quantified using ImageJ software. Images were first converted to 8-bit gray scale and binary threshold to highlight a positive staining. Finally, Table 1. Primers used in this study.

Gene
Forward sequences Reverse sequences the percentage of the area positive for integrin-β1 was measured by ImageJ.
The primary antibodies and secondary antibodies used in this work were listed as Table 2.

Golgi-Cox staining
The manufacturer's instructions were strictly followed (FD Rapid GolgiStain Kit; PK401; FD Neurotechnologies, INC.). The fresh hippocampal and cortical tissues were rinsed with double-distilled water. The tissues were sequentially placed in a mixture of solutions A and B and solution C in the dark in the certain time. Frozen coronal sections of 90-μm thickness were prepared on a freezing microtome (POLAR-D-JC; SAKURA SEIKI) at −22°C and transferred to microscope slides onto a few drops of solution C. The slides were dried at RT in the dark overnight. Serial sections were stained with mixture of D and E solutions for 10 min, then dehydrated, cleared, and covered with a glass coverslip using neutral resin. Single-plane images were captured using a digital slide scanner (Pannoramic 250 FLASH; 3DHISTECH).

Quantification of dendritic spines
Images of shole neurons were obtained six cells per brain slice at ×20 magnification. Two parameters (dendritic complexity and dendritic spines) were analyzed. First, dendritic complexity was analyzed by the concentric circle method reported by Sholl using ImageJ (Sholl & Uttley, 1953). Then the dendritic spines were analyzed as previously mentioned (Shibata et al, 2021). Briefly, nearly 20 images could be obtained per z-stack to cover the entire dendrite thickness. The spine counts were analyzed by Reconstruct 1.1.0.0. software, which is publicly available at http://www.bu.edu/ neural/Reconstruct.html. We had divided the entire dendrite into 10-μm segments and calculated the number of spines, which were traced for length and width. The length and width ratios were used to define the spine subtype as previously described (Risher et al, 2014).

Immunoblot
For protein assay, the brain tissues were, respectively, homogenized in lysis buffer containing RIPA buffer (50 mM Tris; 150 mM NaCl; 1% Triton X-100; 1% sodium deoxycholate; 0.1% SDS, P0013B; Beyotime), a protease inhibitor mixture (87785; Thermo Fisher Scientific), a phosphatase inhibitor cocktail (78420; Thermo Fisher Scientific), and 1 mM phenylmethyl-sulfonylfluoride (PMSF; 36978; Thermo Fisher Scientific) as previously reported (Pang et al, 2021). The proteins in the cell membrane fraction and cytoplasm fraction were extracted using a ProteoExtract Subcellular Proteome Extraction Kit (539790; CALBIOCHEM). Then samples were, respectively, separated on a SDS-PAGE, and the proteins were electroeluted onto a nitrocellulose membrane (Immobilon NC; Millipore). Immunoblots were incubated in primary antibodies and then incubated in HRPlinked secondary antibodies for 1 h. Proteins were visualized using a chemiluminescent detection system (Western Blotting Luminal Reagent; Santa Cruz Biotechnology, Inc.) and analyzed using Quantity One software (version 3.0; Bio-Rad Laboratories, Inc.). The antibodies and working dilution are summarized in Table 2. Isolation and primary culture of rat neurons Dissociated primary brain cultures were performed using a modified protocol (Kaech & Banker, 2006). Briefly, the embryonic hippocampal and cortical tissues at embryonic day 17 (E17) from WT and Tspan7 −/− rats were dissected and dissociated in 2 mg/ml papain (G8430; Solarbio) and 1% DNase (2270A; Takara). Dissociated cells were suspended in DMEM-F-12 medium (2276628; Gibco) with FBS (10099-141C; Gibco) and then plated at a density of 200,000 cells/cm 2 in a six-well plate coated with poly-D-lysin (D6790; Solarbio) or at 50,000 cells/cm 2 in a 12-well plate within coverslips coated with poly-D-lysin-coated. 4 h after plating, the DMEM-F-12 medium was changed to neurobasal medium containing 2% B27 (2450340; Gibco), 0.5 Mm L-GlutaMAX (35050061; Gibco), and 0.05 mg/ml antibiotic penicillin streptomycin (15070063; Gibco). The cells were cultured for 6 d. The WT neurons were then treated with or without 10 μM SRC inhibitor (Bosutinib HY-10158; MedChemExpress) for 4 h. As previously reported (Park et al, 2015), the WT Tspan7 −/− neurons were starved using DMEM-F-12 medium supplemented with 0.1% BSA for 2 h followed by stimulation with serum or a 10 μM Src activator (EPQpYEEIPIYL; HY-P3279; MedChemExpress) for 1 h. The cells in the plate were sampled in lysis buffer for Western blot, and the cells on coverslips were fixed in paraformaldehyde (4%) at room temperature for immunofluorescent staining.

Quantification of postsynaptic and presynaptic puncta marked by immunostaining
The postsynaptic and presynaptic puncta were analyzed as previously described (Ippolito & Eroglu, 2010;Block et al, 2022). For each area of the brain, using both the 488-nm channel to detect SYN and the 555-nm channel to detect the PSD95, a z-stack for each SC section was collected in LAS X software for a total depth of 5 μm (15 × 0.33 μm optical sections) at 63× magnification using multiphoton laser scanning microscopy. Maximum image projections were generated for groups of three consecutive optical sections yielding five maximum image projections/section each representing 1 μm of depth. Maximum projections of three consecutive optical sections were generated using ImageJ. The Puncta Analyzer Plugin (available at: https://doi.org/10.5281/zenodo.6800214) for ImageJ was used to count the number of colocalized synaptic puncta and the number of synaptic puncta produced by Puncta Analyzer.

Statistical analysis
The data were analyzed using two-tailed unpaired t tests and twoway ANOVA with Sidak's multiple comparison method. Data are presented as mean ± SEM. P < 0.05 was considered to indicate a statistically significant difference.