To date, our study is the first to look for potential biomarkers of ASD using MR. On the basis of the strong evidence provided in this study, we found a significant positive association between PKCα level in plasma and the risk of ASD. However, no evidence supported the causality between the other mTOR-dependent circulating proteins (Akt, eIF4E2, eIF4EBP2, HIF1α, PKCζ and RPS6KB1) and ASD.
Overactivation of the mTOR signaling pathway is one of the critical mechanisms in the pathogenesis of ASD. On the one hand, many monogenic disorders with overactivation of the mTOR signaling pathway have been related to a high prevalence of ASD, such as tuberous sclerosis complex (TSC), fragile X syndrome (FXS), phosphatase and tensin homolog (PTEN) hamartoma tumor syndrome and neurofibromatosis type 1(NF1) [45]. ASD is a common neurological feature of these diseases which are all caused by mutations in genes upstream of mTORC1 (e.g., TSC1 and TSC2 in TSC [46], FMR1 in FXS [47], PTEN in PTEN hamartoma tumor syndrome [48] and NF1 in NF1 [49]). On the other hand, the mTOR signaling pathway participates in a number of pathological processes in patients with ASD, such as increased brain volume [50], increased dendritic spine density [51] and disrupted synaptic pruning [52]. Moreover, Inhibition of mTOR activity increases PI3K/Akt/mTOR-mediated autophagy pathway and improves social interactions in ASD mice [53, 54]. However, the causal association between blood levels of downstream signaling molecules in the mTOR pathway and ASD has still not been exhaustively elucidated.
Here, we summarized past studies on the relationship between seven mTOR-dependent proteins and ASD. Zhang et al. found that Akt phosphorylation increases significantly in the hippocampus of a sodium valproate (VPA)-induced baby rat model of ASD [53]. In contrast, Nicolini et al. found that Akt levels in the fusiform gyrus of both VPA-induced ASD rats and patients with idiopathic autism are decreased compared to controls [55]. Russo found that the phosphorylated Akt in white blood cells is significantly lower in patients with ASD [56]. From the perspective of blood transcriptomics, Tylee et al. found a reduction in the PI3K-AKT-mTOR signaling cascade in patients with ASD [57]. The role of eIF4E and eIF4EBP in ASD has been similarly controversial in past studies. Most studies support that eIF4E overexpression or downregulation of eIF4EBP1/2 in rodent brain leads to ASD-like behaviors [22, 58–61]. However, some studies have also found that the expression of eIF4E or eIF4EBP1 is not significantly different in the brain or in peripheral blood mononuclear cells (PBMCs) of participants with ASD compared to controls [55, 62]. Wiebe et al. even found that deletion of eIF4E2 in excitatory neurons in the brain also causes autistic behavior in mice [63]. Wang et al. demonstrated that HIF1α may induce autism-like behavior in offspring rats under hypoxia by modulating synaptic plasticity [24]. In contrast, Pan et al. found that intermittent hypobaric hypoxia (IHH) activates DRN serotonergic neurons through upregulation of HIF1α, thereby ameliorating the autism-like phenotypes in mice [64]. A study based on a comparison of serum proteins from 40 patients with ASD and 40 healthy children found that serum HIF1α is significantly lower in patients with ASD [65]. Several research have found that increased PKC activity may contribute to the pathogenesis of ASD and that PKC inhibition is beneficial in the treatment of ASD [66, 67]. Some studies have also suggested that PKC activation has an ameliorative effect on symptoms in animal models of ASD [68, 69]. Meanwhile, a proteomic study of cerebral organoids (four autistic patients and six controls) found that PKCα is downregulated in ASD-derived cerebral organoids [70]. As for RPS6KB1, although Nicolini et al. found that RPS6KB1 is reduced in the fusiform gyrus of patients with idiopathic autism and in the neocortex of VPA exposed rats [55], previous other studies possessed an almost consistent tendency that overexpression or overactivation of RPS6KB1 is associated with the onset of ASD [71–75]. In summary, there is still no unanimous conclusion about the role of these seven mTOR-dependent proteins in the pathogenesis of ASD, which is partly attributable to the lack of high-quality clinical studies.
In view of the importance of the mTOR signaling pathway in ASD and the lack of well-defined biomarkers for ASD, our study centered on finding potential blood biomarkers for ASD using MR analysis. In this study, we identified a causal association between the plasma level of PKCα and ASD. Previous preclinical studies have been controversial about the role of PKC in ASD [66–70], while the independent effects of PKCα are difficult to elucidate due to the lack of specific inhibitors for the different isoforms of PKC. Only one study clearly showed that PKCα is downregulated in ASD-derived cerebral organoids [70], which is opposite to the trend of our results. Such differences may be due to the different tissues tested (brain or plasma). A growing number of mouse models of autism have identified an underlying imbalance in excitatory/inhibitory transmission that is considered to play a core role in the etiology of autism [76, 77]. Meanwhile, a study by Bemben et al. indicated that PKC robustly phosphorylates the T707 site of endogenous neuroligin 4X, which greatly enhances neuroligin 4X-mediated excitatory synaptic transmission [78]. In contrast, a single point mutation in the cytoplasmic structural domain of neuroligin 4X in ASD, where arginine is mutated to cysteine (R704C), completely eliminates PKC-mediated phosphorylation of T707 and subsequently reduces excitatory transmission [78]. Thus, PKCα is likely to participate in the pathological process of ASD by affecting neuronal excitatory transmission.
Contrary to previous studies that focused more on the role of PKC in the pathogenesis of ASD in the brain, we paid more attention to the association between PKCα level in plasma and ASD. We hypothesize that PKCα is a very prospective biomarker for ASD and has value in the adjunctive diagnosis and disease monitoring of ASD. Furthermore, lowering the plasma level of PKCα may help prevent ASD when infants and young children undergo the risk factors of ASD, and even help manage ASD. However, it will take further work to uncover the potential pathologic mechanisms. It is essential to define standard values for the plasma level of PKCα that are diagnostic, preventive, or therapeutic for ASD.
This study has a couple of advantages. Genetic instruments were used to proxy for circulating protein levels, thereby minimizing confounding factors and bias. Moreover, we obtained outcome datasets from two different consortiums, set up discovery and validation cohorts, and summarized the results using meta-analysis, which enhanced the robustness of the findings.
There were also some limitations to this study. First, only partial mTOR-related proteins were discussed in our study, and the association of the rest of the downstream proteins with ASD needs to be investigated in future research. Second, the GWAS datasets of the proteins in our study were all derived from blood samples, and this study did not involve the exploration of corresponding proteins in brain tissue. Third, the genetic data in this study were all from European ancestry, which makes our findings lack some generalizability. Fourth, ASD possesses great heterogeneity, encompassing both typical and atypical autism, and our study did not perform subgroup analyses between the various clinical subtypes. Fifth, proteomics GWAS studies are unable to identify as many genome-wide important genetic variants as possible due to their relatively small study samples. Therefore, we abandoned the significance level p < 5e-8 in common use and instead chose the more looser threshold p < 5e-6 on the premise of balancing statistical efficacy.