Background
Autism is a complex neurodevelopmental disorder characterized by deficiencies in social interaction and communication, and repetitive and stereotyped behaviors. Autistic disorder, Asperger syndrome, and pervasive developmental disorder-not otherwise specified (PDD-NOS) comprise a heterogeneous group of neurodevelopmental disorders known as autism spectrum disorders (ASD). The abnormalities are usually identified in the early years of childhood and often coexist with impairments in cognitive functioning, learning, attention and sensory processing. According to a recent report, the prevalence of this disorder has risen to 1 in 110, with a male to female ratio of 4.5:1 [
1].
A growing body of evidence from biochemical and neuroimaging studies has suggested that a disturbed brain bioenergetic metabolism underlies the pathophysiology of autism in some cases. Magnetic resonance spectroscopy studies have shown, in the brain of autism patients, abnormal levels of metabolites relating to brain bioenergetics, such as decreased levels of phosphocreatine and
N-acetyl-aspartate, and elevated lactate [
2,
3].
Mitochondria serve as the energy powerhouses of eukaryotic cells, since they generate most of the adenosine triphosphate (ATP), the source of chemical energy in cells. The findings of abnormal brain bioenergetics, therefore, support an involvement of mitochondrial dysfunction (MtD) in the pathogenesis of ASD [
4]. Diminished levels of ATP have been observed in autism brain [
2]. Rats induced for MtD have been found to exhibit certain brain, behavioral and metabolic changes consistent with ASD, including microglial activation, reduced levels of glutathione, repetitive behaviors, social interaction deficits, hyperactivity and oxidative stress (OS) [
5‐
8].
In a systematic review and meta-analysis, MtD was observed in approximately five percent of children with ASD; developmental regression, seizures, motor delay and gastrointestinal abnormalities were found to be significantly more prevalent in children with ASD/MtD as compared with the general ASD population [
9]. Defective lymphocytic mitochondria [
10] and ultrastructural abnormalities of mitochondria [
11,
12] have been reported in autism. Nutritional supplements (for example, carnitine, vitamin B) and/or antioxidants (for example, co-enzyme Q10) have been found to be beneficial in the treatment of some children with ASD/MtD [
13‐
15].
Recent studies have reported brain region-specific deficits of mitochondrial electron transport chain complexes in autism [
16,
17]. Upregulated expression of the mitochondrial aspartate/glutamate carrier (SLC25A12) [
18,
19], and evidence of hypoxia, as measured by a reduction in the anti-apoptotic protein Bcl-2 and an increase in the pro-apoptotic protein p53 [
20,
21], has also been reported in autism brain.
Several of the previous studies of MtD in autism were restricted to the biomarkers of energy metabolism, while most of the genetic studies were based on mutations in the mitochondrial DNA (mtDNA). Despite the mtDNA, most of the proteins essential for mitochondrial replication and function are encoded by the genomic DNA; so far, there have been very few studies of those genes. We aimed at elucidating the role of MtD in the pathogenesis of autism. Using the postmortem brains of autism patients and healthy controls, we compared the expression of 84 genes involved in diverse functions of the mitochondria such as, biogenesis, transport, translocation and apoptosis. Furthermore, we analyzed the genetic association of three of these genes with autism, in two independent studies involving family-based samples of different origins.
Discussion
Our study of MtD in autism involves a wide array of genes related to diverse mitochondrial functions. We report brain region-specific alterations in the expression of these genes in autism. MTX2, NEFL and SLC25A27 showed consistently reduced expression in the ACG, MC and THL of autism patients. We also observed nominal genetic association of NEFL and SLC25A27 with autism.
Gene expression was analyzed in three brain regions: ACG, MC and THL. ACG has been found to be involved in emotion formation and processing, learning and memory [
27,
28]; MC in planning, control and execution of voluntary motor functions [
29]; and THL in the processing and relaying of sensory information [
30]. In autistic individuals, abnormalities of anterior cingulate have been found to be linked with impairments in cognitive control [
31], social orientation [
32], social target detection [
33], and response monitoring [
34]. Increased white matter volume of MC has been reported to be associated with motor impairments in autistic children [
35]. Impairments in auditory, tactile, and visual sensory stimuli processing, found in autistic individuals, have been attributed to THL abnormalities [
36]. Reduced thalamic volume has also been observed in autism [
37].
Brain samples from the cerebellum and cortices had been used in previous studies of MtD in autism. The brain regions (ACG, MC and THL) used in our study have not been reported elsewhere. The differences in the results of our study and other whole genome transcriptomic analyses of autism brain [
17,
38,
39] might be due to the differences in the regions of brain that were analyzed. The differences in metabolic demands or brain region-specific pathophysiology could affect the expression of mitochondrial genes. The etiological heterogeneity and criteria for sample selection might also have influenced the results since MtD is observed in only a subset of autistic individuals.
NEFL,
SLC25A27 and
MTX2 showed reduced expression in all the three brain regions of autism patients.
NEFL is located in 8p21.2, which has been suggested as a susceptible region for autism in a genome-wide association study [
40]. Moreover, 8p is known as a potential hub for developmental neuropsychiatric disorders [
41]. Being a major constituent of neurofilaments, NEFL plays a pivotal function in the assembly and maintenance of axonal cytoskeleton [
42]. Knocking out of
Nefl has been found to reduce axonal caliber and conduction velocity in mice [
43]. Sensorimotor impairments and reversal learning deficits have been observed in
Nefl transgenic mice [
44]. NEFL has also been found to have a vital role in regulating mitochondrial morphology, fusion, and motility in neurons [
45,
46]. Reduced NEFL expression may thus restrict mitochondrial translocation to areas of the cell requiring energy. We observed a nominal association of an
NEFL SNP with autism in the AGRE samples. However, this SNP is located in the UTR of exon 4 and might not have a functional significance.
SLC25A27, also known as uncoupling protein 4 (UCP4), belongs to the large family of mitochondrial anion carrier proteins that are located on the inner mitochondrial membrane. It is expressed predominantly in the central nervous system (CNS) [
47]. It has also been suggested to have roles in the reduction of reactive oxygen species [
48], neuroprotection against OS and ATP deficiency [
49], inhibition of apoptosis [
50], neuronal cell differentiation [
51], mitochondrial biogenesis [
52], and mitochondrial calcium homeostasis [
53]. Downregulation of
SLC25A27 could thus have detrimental effects on these processes. Pharmacological targeting of neuronal uncoupling proteins (UCPs) represents an important avenue to combat MtD. Fatty acids have been reported to activate UCPs [
54,
55]. Consequently, a ketogenic diet has been found to increase the protein levels and activities of UCPs, including that of SLC25A27 [
56]. We observed a nominal association of
SLC25A27 with autism in Japanese samples. However, rs6901178, the SNP that showed association, is located in intron 4 and might not have a functional significance.
MTX2, located on the cytosolic face of the outer mitochondrial membrane, has been suggested to function as an import receptor for mitochondrial preproteins, a crucial process for cell survival [
57,
58]. It also plays a major role in the regulation of apoptosis [
59]. In this study, we observed a downregulation of
MTX2 in the ACG, MC and THL of autism patients; however, we did not observe an association of this gene with autism.
We also observed, in autism brains, region-specific alterations in the expression of several other mitochondria-related genes (Table
2). These genes fall into the ten functional groups as described in Additional file
3 and presented below:
1)
Membrane polarization and potential (MPP): MPP plays a crucial role in energy production, maintenance of calcium homeostasis, protein import and cell survival [
60,
61]. We observed, in the ACG of autism patients, an elevated expression of
BCL2 and
TP53, which are involved in the maintenance of MPP.
2)
Mitochondrial transport: In brain, the proper localization of mitochondria in the neurons is necessary for the generation of synaptic and action potentials, regulation of intracellular calcium dynamics and ATP synthesis [
62,
63]. In various regions of autism brains, we observed alterations in the expression of several genes related to mitochondrial transport, such as,
AIP,
BCL2,
DNAJC19,
HSP90AA1,
MFN2,
MIPEP,
TP53 and
TSPO. The expression of
DNAJC19 was downregulated in the MC and THL of autism patients.
3)
Small molecule transport, SLC25A family: The expression of several members of SLC25A solute carrier family was altered, with most of them being downregulated, in autism. Mitochondrial solute carriers transport a variety of solutes (di- and tri-carboxylates, keto acids, amino acids, nucleotides and coenzymes/cofactors) across the inner mitochondrial membrane [
64]. We observed a reduced expression of
SLC25A12 and
SLC25A14 in the ACG and MC of autism patients. However, upregulated expression of
SLC25A12 has been observed in some prior studies [
18,
19]. The brain regions used in this study were different from those in the aforementioned studies. The variation in metabolic demands of different brain regions could consequently affect the expression of mitochondrial genes. There are also conflicting reports about the association of
SLC25A12 with autism [
65‐
67]. The expression of
SLC25A24 was reduced in the MC and THL of autism patients.
4)
Targeting proteins to mitochondria.
5)
Mitochondria protein import: Of the hundreds of proteins that are found within the mitochondria, the mitochondrial genome encodes only 13, and the rest must be imported from the cytosol [
68]. The nuclear-encoded, cytoplasmically synthesized proteins should be precisely targeted and imported to the mitochondria. In this study, the expression of several genes involved in protein targeting and import were found to be altered, with the majority of them being downregulated, in autism brains. Among these,
DNAJC19 was downregulated in the MC and THL of autism patients.
6)
Outer membrane translocation.
7)
Inner membrane translocation: The TIMM/TOMM translocases are involved in the translocation of nuclear DNA-encoded mitochondrial proteins across the outer and inner mitochondrial membranes [
69]. Several genes belonging to the TIMM/TOMM family showed altered expression in autism brain.
TOMM20 showed a reduced expression in the ACG and MC of autism patients.
8)
Mitochondrial fission and fusion: The expression of
MFN2, one of the genes involved in the regulation of mitochondrial fission and fusion was found to be downregulated in the ACG of autism patients. Mitochondrial fission and fusion are crucial in maintaining the integrity of mitochondria, electrical and biochemical connectivity, turnover of mitochondria, segregation and protection of mtDNA, and programmed cell death [
70]. In the neurons, this is involved in the formation and function of synapses in the dendritic spines and axons [
71,
72].
9)
Mitochondrial localization: We observed reduced expression of
DNM1L,
LRPPRC,
MFN2 and
RHOT2, localized predominantly in the mitochondria. These genes are involved in the biogenesis, maintenance of morphology and integrity, trafficking, and homeostasis of mitochondria [
73‐
75]. The expression of
DNM1L and
LRPPRC were reduced in the ACG and MC of autism patients.
10)
Apoptosis: The expression of apoptotic genes were altered, with most of them being upregulated, in the brain of autism patients. Recent studies have demonstrated a possible association between neural cell death and autism [
76,
77].
A two-way ANOVA showed that the expression of all the genes that were differentially expressed in two or more brain regions of autism were dependent on the disease status rather than being region-specific (data not shown).
It is not yet clear if MtD is the cause or effect of autism. ASD patients have often been found to manifest biochemical or neuropathological traits linked with altered mitochondrial function. Since mitochondrial abnormalities often result in CNS dysfunction, leading to developmental regression, learning disability, and various behavioral disturbances, ASD could be an important clinical presentation of MtD [
78]. However, the clinical features, and the biochemical and genetic abnormalities in ASD patients with an underlying MtD have been found to be heterogeneous. In addition, several of the biochemical abnormalities indicative of MtD may occur in the absence of any relevant genetic alterations [
79]. On the contrary, mitochondrial abnormalities might also manifest as a secondary to certain pathophysiological processes involved in autism, such as immune dysregulation, OS and altered calcium homeostasis [
79]. Even though it is possible that a greater proportion of individuals with ASD might have MtD at the genetic level, it may not be manifested clinically.
We observed only nominal association of
NEFL and
SLC25A27 with autism. Recent studies have indicated that only a subset of autism may be associated with the biochemical endophenotype of mitochondrial energy deficiency [
80]. Therefore, related genes might not show a strong association with the disorder. Considering the highly heterogeneous nature of autism, nominal associations of genes with subtle effects on the disease phenotype should not be ignored. The small sample size of the Japanese trios is, however, a serious limitation of this study.
MTX2,
NEFL and
SLC25A27 were selected for genetic association studies since their expression was reduced in all of the three brain regions analyzed. Nevertheless, there would have been other important genes directly impacting mitochondrial functions, albeit differential expression in just one or two brain regions of autism patients. However, a detailed study involving several genes was not possible due to financial constraints.
Factors inherent in postmortem brain studies, and beyond the investigator’s control, might have influenced our results. We did not have sufficient data regarding brain pH. However, large-scale gene analysis have shown that brain pH or PMI has no significant correlation with RNA integrity [
81,
82]. The pH could be lower in the postmortem brains of individuals who suffered prolonged agonal states, such as in respiratory arrest, multi-organ failure and coma [
83]. However, the cause of death was sudden for most of the subjects included in our study. So, we assume that brain pH might not have affected the gene expression. The other concern is the effect of medication; antidepressants, antipsychotics and selective serotonin re-uptake inhibitors are known to inhibit mitochondrial activities [
84,
85]. In this study, medication status was available for only three autism patients, two of whom had received more than two classes of drugs (drug doses unknown). Therefore, it was difficult to examine the effects of medication on gene expression. Another matter of concern is that the cause of death for a majority of the autism patients was seizure or drowning, where the latter could also have been due to seizures. Seizure activity has been known to impair mitochondrial energy production by altering the activity of mitochondrial enzymes involved in ATP production [
86,
87]. In this study, we have not examined the expression of any genes directly involved in mitochondrial energy production. Therefore, we assume that the cause of death might not have influenced our results. Moreover, it is not yet clear if MtD is the cause or effect of seizures.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AA was involved in conception, design, conducting experiments, data analysis and drafting of article. IT, KY, YI and TT were involved in analysis and interpretation of data. HM, TM, SY, MT, KJT, KM, YIwata, KS and HI were involved in drafting the article. KN, TS, TY and NM were involved in revising the article critically. All authors read and approved the final manuscript.