Background
Autism spectrum disorder (ASD) is one of a number of neurodevelopmental disorders that display sexual dimorphism, occurring more frequently in males, which affect brain structure, gene expression, pathways, function, and ultimately behaviors that will require individualized treatments [
1‐
3]. Extensive evidence demonstrates sex differences in ASD brain [
4‐
7]. Sex chromosomes may play a role, with the Y chromosome being a possible risk factor for ASD and X chromosome perhaps having a protective effect [
3]. Females appear to have a higher threshold for being affected by genetic factors than males, thus requiring a greater genetic burden, and may have greater brain plasticity [
8,
9]. Females carry a higher proportion of de novo CNVs (copy number variants) than males, the CNVs in females disrupt a larger number of genes than in males, and females carry a greater number of de novo single nucleotide variants (SNVs) than males [
10,
11]. Environmental and hormone factors that differ between the sexes, like testosterone, may impact the time course and severity of symptoms [
12].
Few molecular studies of ASD brain tissue to date have considered potential sexual dimorphism often because of limited tissue availability of female cases. In typical brain development, male-biased gene expression changes are enriched for extracellular matrix, immune response, chromatin, and cell cytoskeleton pathways that have been implicated in ASD [
13]. Sex differences in microRNA (miRNA) expression in the frontal cortex have also been described in typical neurodevelopment [
14]. Specific genetic mechanisms, such as the expression of retinoic acid-related orphan receptor alpha (RORA) in the frontal cortex, which regulates CYFIP1, may be related to elevated testosterone levels and a potential contributor to the sex bias [
15]. There is no clear evidence to date for systematic sex-differential expression of ASD risk genes in human brain; however, genes expressed at higher levels in males are significantly enriched for genes upregulated in postmortem autistic brain, including astrocyte and microglia markers [
16].
Here, we extend our analyses of previously published microarray data [
17,
18] to examine sexual dimorphism of microRNA and other small noncoding RNA (sncRNA) in male and female ASD and control brain tissue. We focused on two temporal cortical regions: the superior temporal sulcus (STS), a region implicated in social impairments in ASD [
19‐
22], and the primary auditory cortex (PAC). Predicted miRNA targets were identified as well as pathways in which they over-populated. We then evaluated our findings in relation to the SFARI database and two gene expression studies in ASD brain: Ziats and Rennert, 2013 [
13] and Werling et al., 2015 [
16].
Discussion
This short report underscores the importance of considering biological sex as a factor when interpreting gene expression studies of postmortem brains from individuals with ASD. Our findings from this small sample suggest that the expression of miRNA and other sncRNA is sexually dimorphic in the temporal cortex of ASD individuals. There are generally more dysregulated sncRNA, miRNA target genes, and pathways in ASD females compared to ASD males. There is a significant overlap between the male miRNA putatively dysregulated pathways in our study and that which occurs during normal male neural development [
13]. Though there is an overlap between the female ASD regional STS-PAC miRNA targets and the SFARI genes, most of the sncRNA-regulated sexually dimorphic target genes are not enriched for autism risk genes, similar to the findings of Werling et al. [
16]. The sexually dimorphic over-enrichment of miRNA target genes in the immune and nervous system pathways is consistent with prior gene expression studies of ASD brain [
24,
25] that postulated that the immune pathways were related to environment and the neuronal pathways were related to genetics [
24].
The greater number of dysregulated sncRNA, miRNA target genes, and pathways in females compared to male ASD subjects supports a body of evidence, suggesting that there is a greater genetic load in ASD females [
9] [
3,
26], and is consistent with a possible female protective effect [
3,
27,
28]. The greater sncRNA dysregulation in females in our study might also support a recent proposal that female brains are less vulnerable to ASD because they are more plastic [
8]. Our data could be interpreted to mean that the female ASD brain mounts a greater protective molecular response compared to males, factors that may contribute to the so-called “female camouflage effect” [
29].
There are a number of sexually dimorphic miRNAs from our study that have been associated with ASD-relevant diseases and processes (Table
3). For example, one study of miRNA in serum of children with ASD [
30] found two miRNAs, miR-151 and miR-181 that were also differentially expressed in the current study. miR-181 promotes synaptogenesis and decreases in axon growth [
31,
32]. It is expressed in the brain and previously associated with autistic phenotypes [
33] and schizophrenia [
34]. The microRNA is also involved in an inflammatory response [
35], influences apoptosis and mitochondrial function [
36] in astrocytes, and targets GABA receptors [
37].
Table 3
Relevant literature on example miRNAs with sexual dimorphism in ASD relative
miR-151 (female STS) | Regulated in serum of children with autism [ 30] |
miR-181 (female STS vs PAC) | Regulated in serum of children with ASD [ 30]; expressed in brain, promotes synaptogenesis and decreases axon growth [ 31, 32]; associated with ASD phenotypes [ 33] and schizophrenia [ 34]; associated with inflammatory responses of astrocytes [ 35]; influences apoptosis and mitochondrial function in astrocytes [ 36]; targets GABA receptors [ 37]. |
miR-219 (female STS) | Regulates oligodendrocyte differentiation and likely myelin production [ 38]; regulates neural progenitors by dampening apical Par protein-dependent Hedgehog signaling [ 57]; polymorphisms in miR-219 affect genes involved in NMDAR signaling and schizophrenia [ 58]. Young age and environmental enrichment increase serum exosomes containing miR-219 that promote CNS myelination [ 59]; human endometrial-derived stromal stem cells (EnSCs) can be programmed into pre-oligodendrocyte cells by overexpression of miR-219 or miR-338 [ 60, 61]. |
miR-338 (female STS) | Regulates oligodendrocyte differentiation and likely myelin production [ 38]; attenuates cortical neuronal outgrowth by modulating expression of axon guidance genes and axonal mitochondrial genes [ 39‐ 41] |
miR-488 (female PAC) | Associated with panic disorder and regulate several anxiety candidate genes and related pathways [ 62] |
miR-125 (female STS) | Differentially expressed in male vs female frontal lobe regions during normal neurodevelopment [ 14]; neuronal differentiation, and specifically promotes the generation of neurons of dopaminergic fate and possibly other types of neurons [ 63] |
In STS of ASD female compared to control female, miR-219 and miR-338 showed the highest level of downregulation. Both miRNAs are involved in regulating oligodendrocyte differentiation and likely myelin production [
38]. Though abnormalities of white matter have been observed in ASD brains for some time, recent studies point to important sex differences. For example, a recent anatomical study of autism brains showed large white matter regions showing significant sex × diagnosis interactions [
5]. This was supported by sex differences found in the corpus callosum of young children with ASD [
6]. While no alterations were observed in ASD males compared to control males, mean diffusivity, axial diffusivity, and radial diffusivity were all increased in ASD females compared to control females [
6].
miR-219 and miR-338 both promote oligodendrocyte differentiation [
38]. Inhibition of both miRNAs inhibits oligodendrocyte maturation and function in part by directly repressing negative regulators of oligodendrocyte differentiation, including transcription factors Sox6 and Hes5 [
38]. miR-338 also attenuates cortical neuronal outgrowth by modulating expression of axon guidance genes and axonal mitochondrial genes [
39‐
41]. However, it is the role of miR-219 and miR-338 on differentiation of oligodendrocyte precursors during development (with downregulation over fourfold in STS of ASD females, but not of ASD males) that may significantly contribute to the sexually dimorphic changes of white matter tracts including the corpus callosum seen in ASD brain [
5,
6]. It is notable that decreased miR-219 and miR-338 were detected in female STS of ASD brains and not PAC, an important finding since STS is an association cortex implicated in social behavior, which is a core symptom of ASD, whereas PAC is not generally associated with ASD [
42]. It will be important to examine miR-219 and miR-338 in amygdala and other brain regions implicated in ASD core symptoms, particularly since the number of oligodendrocytes in amygdala is decreased in ASD brain [
43]. It will also be important to quantify oligodendrocyte numbers in STS of ASD brain given the current miRNA results.
It is important to note that sex differences in the miRNAome are prominent in health and disease [
44‐
47]. Several miRNAs we identified in this study are also sexually dimorphic in other tissues and disease conditions. For example, miR-100, miR-196a, and miR-31 are sexually dimorphic in human amniotic mesenchymal stem cells (hA-MSCs) from obese versus normal weight women who gave birth to females, but not males [
48]. This suggests that these miRNAs may be involved in metabolic changes. Additionally, miR-31 was sexually dimorphic in an animal model of systemic lupus erythematosus (SLE) [
49]. Several mature miRNAs map to cytoband 14q32, where there is an imprinted miRNA cluster. In our study, miRNA-151b was downregulated in ASD-STS female vs typical STS female, while miR-342-3p, miR-432, and miR-485-5p were upregulated in ASD male in the inter-regional analysis compared to typical male. Loss of imprinting in this region has been associated with multiple diseases [
50,
51]. Our results suggest involvement of miRNAs within or near this cluster in ASD as well.
As we have reported previously [
17], there are changes of expression of sncRNA in ASD-PAC, though there are more changes in ASD-STS. In fact, some of the miRNAs we previously reported are included in the present study, indicating they may have been driven by sex differences as well. Interestingly, even though there were attenuated regional differences (STS vs PAC) in both male and female ASD compared to control (Fig.
2), the greatest number of miRNA targets by far was identified in the regional comparisons of STS to PAC in both females and males, though there were significantly more targets in females. These large differences of expression between regions, particularly between primary sensory cortices like PAC and association cortices like STS, have been noted previously in neurotypical adult and developing human brain [
52,
53] and are consistent with the present data both for ASD and for controls. Integrative functional analyses of ASD risk genes implicate cortical layers II/III and V/VI pyramidal neurons [
54], and co-expression network analyses of ASD risk genes also implicate layers V/VI cortical projection neurons [
55]. The current findings of dysregulated miRNA that control oligodendrocyte differentiation in female STS, but not in male STS, adds complexity to the picture particularly since miR-219 and miR-338 are not dysregulated in either females or males in PAC. Thus, if projection neurons are involved in ASD, they are likely selectively affected in specific cortical regions and it is possible that there are regional differences of miRNA-mediated oligodendrocyte-myelination of the projection neuron axons that are also sexually dimorphic.
In general, there is little overlap of the
specific miRNA regulated in ASD and control brains in this study and the miRNA described by Ziats and Rennert for typically developing human brain [
14]. However, there is tremendous overlap of the predicted pathways that are sexually dimorphic. There is significant overlap for the pathways for males in our regional analyses and male pathways in neurotypical brains [
13] and a trend for overlap between the sexually dimorphic pathways reported by Werling et al. [
16] and our study.
There are important limitations to consider when interpreting this data that are common to studies of postmortem human brains, including small sample size and variation in cause of death, postmortem interval, age at death, agonal state, postmortem RNA integrity, and tissue preparation at different brain banks. In addition, the changes in miRNA and predicted targets and pathways studied here could lead to some aspects of ASD, but they could just as well be a consequence of the condition and even represent compensatory mechanisms. Future studies will need to assess both miRNA and mRNA in the same samples, so that one can determine if specific miRNA target mRNA are present and are regulated in an inverse direction as occurs for most miRNA-mRNA interactions. Small sample size is also limitation of this report and findings need to be confirmed in larger cohorts and with alternative gene expression profiling techniques such as qRT-PCR. However, this study clearly indicates that sex needs to be considered when interpreting data on postmortem human brains in ASD.
Acknowledgements
The authors acknowledge the technical assistance of Alicja Omanska and Dr. Nicole Barger of the UC Davis MIND Institute, as well as Ryan Davis and Stephanie Liu of the Genomics Shared Resource at UC Davis. The authors also express their gratitude to the brain donors and their families who made this research possible.