The 39,XY*O mouse model exhibits face and construct validity for neurodevelopmental disorders, showing behavioural endophenotypes associated with such conditions, and lacking two genes (Sts and Asmt) whose human orthologues have been implicated in ADHD and autism pathogenesis respectively. In this study, we investigated the neurobiology of this model by two methods with a view to identifying mechanisms by which loss of function of these genes might contribute towards behavioural pathology; this work is important given the current lack of availability of single gene knockout models for Sts and Asmt.
Microarray and quantitative PCR analyses comparing 40,XY and 39,XY*O brain tissue identified a surprisingly small number of robust gene expression differences between the groups. Our inability to verify differential microarray expression calls at P < 0.05 for multiple transcripts of relevance to autism and ADHD suggests that we are likely to have successfully identified all the genes that truly differ in their expression across the whole brain in 40,XY and 39,XY*O mice. However, it is possible that there are further group differences in gene expression within specific brain regions; indeed, our microarray analysis did not indicate altered expression of Htr2c (encoding the serotonin 2c receptor) which we have previously shown to be upregulated in the 39,XY*O hippocampus .
Erdr1 gene expression was significantly reduced in 39,XY*O whole brain, and in hippocampal and striatal dissections; this gene is retained in the 39,XY*O mouse and is located adjacent to the fusion point of the X and Y chromosomes . Our current results suggest that in addition to deleting of the Sts and Asmt genes, the lesion in 39,XY*O mice disrupts a genetic element that enhances Erdr1 expression. Theoretically, in concert with loss of Sts and Asmt, reduced expression of the widely-expressed, but poorly-characterised, erythroid differentiation regulator 1 protein encoded by Erdr1 could contribute towards downstream gene expression changes, abnormal striatal and hippocampal monoamine neurochemistry, and behavioural phenotypes in the 39,XY*O model; brain and behavioural investigations in mice in which the function of this gene alone is disrupted will help clarify the extent and specificity of this contribution.
Our microarray and qPCR analyses also showed that the expression of C1qc (encoding the protein complement component 1, q subcomponent, c chain) was upregulated in the brains of young, behaviourally-naïve 39,XY*O mice, but downregulated in the striata and hippocampi of older, behaviourally-trained mutant animals. These data suggest a potential basis for the neurochemical and behavioural abnormalities seen in the 39,XY*O mouse that may be sensitive to spatiotemporal or environmental regulation, and further suggest the possibility that the neurobehavioural pathology in individuals lacking functional STS and/or ASMT proteins may be due, in part, to altered C1QC levels. Previous animal and clinical studies have implicated aberrant expression of C1q family members in developmental and behavioural phenotypes. In rodents, C1q deletion results in altered synaptic elimination [33, 34], C1qc expression levels are altered in a model of developmental hippocampal pathology , and C1qc expression is associated with behavioural phenotypes (notably the consumption of ethanol relative to water) . In man, individuals with autism can exhibit elevated C1QC serum levels  and altered gastrointestinal C1q deposition [38, 39]. Whether the 39,XY*O mouse exhibits alterations in synaptic structure/function or hippocampal structure, or heightened alcohol preference remains to be investigated. It also remains to be seen whether individuals with elevated C1QC levels and autism possess genetic mutations in either STS or ASMT, and whether individuals lacking functional STS and/or ASMT genes are at increased risk of alcohol dependence.
By the same logic, the present findings further indicate that disrupted expression of Metap1d and/or Sfi1 could play a role in 39,XY*O phenotypes, and in developmental phenotypes associated with Xp22.3 mutations, but probably not through influencing striatal or hippocampal physiology. There is some evidence for an association between a linkage block at 2q31.1 containing METAP1D and autism , whilst copy number variants encompassing SFI1 have previously been identified in autism and related developmental disorders [41–45].
We also showed that the genetic mutation in 39,XY*O mice resulted in reduced hippocampal expression of the Dhcr7 gene. DHCR7 is a known modulator of serotonergic system development in mammals ; therefore, its reduced expression represents a strong candidate mechanism for abnormal serotonin levels in the 39,XY*O hippocampus and associated behavioural phenotypes [19, 22]. In man, defects in the DHCR7 enzyme underlie Smith-Lemli-Opitz syndrome (SLOS). Individuals with SLOS exhibit a range of behavioural symptoms with some overlap with autism, including: hyperactivity, aggression, insomnia, self-injurious behaviour, sensory hypersensitivity and repetitive behaviours ; interestingly, several of these behavioural abnormalities are also observed in the 39,XY*O mouse [19, 21, 22], indicating that hippocampal loss of DHCR7 function may underlie key SLOS phenotypes, and suggesting the 39,XY*O mouse as a potential novel model for aspects of the syndrome.
Genetic and functional work in mice has indicated a link between steroid sulphatase and aggressive behaviour [46, 47]. Consistent with this, 39,XY*O mice , and mice co-administered COUMATE and DHEAS , exhibit elevated levels of aggression. Previous rodent studies have demonstrated that major urinary proteins may elicit aggressive behaviour through their actions at sensory neurons expressing Vmn2r putative pheromone receptors . Of the 124 genes within the Vmn2r family, the expression of just one, Vmn2r86, was significantly altered in 39,XY*O brain; this increased expression thus represents an excellent candidate mechanism underlying aggression in these mutant mice. The fact that there is no human orthologue of Vmn2r86 may explain why individuals with Xp22.3 deletions encompassing STS and/or ASMT do not consistently show obvious aggressive tendencies.
Acute administration of COUMATE, a specific steroid sulphatase inhibitor, given at a dose known to induce behavioural changes, did not recapitulate any of the whole brain gene expression changes seen in the 39,XY*O mouse. There are two obvious possibilities why this might be the case: i) the expression changes in the 39,XY*O mouse result from abnormal developmental expression of STS and/or ii) the gene expression changes in the 39,XY*O mouse are the result of loss of function of ASMT, or reduced expression of Erdr1. A previous study found no effect of acute administration of COUMATE on the concentrations of endogenous DHEAS or DHEA in whole mouse brain, although the drug did reduce entry of systemic DHEAS into the brain . Thus, the molecular basis of COUMATE-induced behavioural changes remains obscure; it is plausible that the drug induces brain region-specific gene expression changes that we were unable to detect, and this possibility remains to be investigated.
Our current analyses provide, for the first time, a systematic profile of the steroid milieu in the mouse brain. There was substantial overlap between the free and sulphated steroids that were detectable in the adult male mouse brain (present data), and those that were most readily detectable in the adult male rat brain  consistent with a degree of cross-species homology, although, interestingly, concentrations of most compounds tended to be higher in mouse brain. We found no significant differences in the concentrations of the detectable compounds between 40,XY and 39,XY*O brains, consistent with an absence of large between-group differences in steroid brain biochemistry. This finding, taken together with our previous data showing reduced serum DHEA levels in 39,XY*O mice  suggests the possibility that where STS is absent developmentally, as in 39,XY*O mice, a compensatory mechanism is recruited to cleave sulphated steroid esters in brain, but not in peripheral tissues. Due to the difficulty of generating 39,XY*O mice and precisely genetically-matched controls, and the apparent variability in brain steroid levels in the mutant group, our study had limited power, with several steroids below the limit of detection. As such, we cannot completely exclude the possibility that there are subtle differences in levels of one or more steroids within 40,XY and 39,XY*O brain tissue.