Studying the variance in the distribution of gene expression represents a new approach to gene expression analysis. Because changes in a limited number of genes could not account for a change in the shape of the entire gene expression distribution, changes in variance suggest alterations at global levels of gene expression regulation. This “birds-eye view” of gene expression levels contrasts with gene-centered approaches that seek to identify specific genes or pathways that differ between groups. While gene specific approaches are useful they may miss changes occurring at the global level of gene expression regulation. The two approaches, therefore, represent distinct yet potentially complimentary approaches to understanding the full magnitude of effects on gene expression regulation. Here we used our novel approach of studying the gene expression distribution to associate decreased variance in gene expression levels with the diagnosis of autism and a risk factor for autism, increased paternal age. We complimented this global approach with gene specific analyses that indicated that changes in global levels of gene expression regulation might be related to large-scale changes in transcriptional regulation. Finally, we reinforced the potential role of transcription regulation by demonstrating with a publically available dataset that inhibition of transcription could, in fact, lead to decreased variance in the distribution of gene expression levels.
Global levels of gene expression regulation may be particularly important in understanding the pathological basis of diseases such as autism where multiple systems are affected. This is true because alterations at the global level of gene expression regulation could be shared across systems even when the tissue specific genes affected by these global changes were different. For instance, in the current study, even though gene expression patterns in the brain and blood are known to be different, it is possible that in the brains of children with autism, variance in gene expression is decreased as it is in the blood. Thus, it is possible that a common mediator, a change at the global level of gene expression regulation, could offer a unifying explanation for multi-systemic effects of disease. By understanding that global levels of gene expression regulation can be disrupted in disease, it becomes clear that targeting global levels of gene expression regulation might also be therapeutic, an idea strongly supported by successful treatments such as HDAC inhibitors in mental health disorders and cancer
. Interestingly, HDAC inhibitors are known to affect immune, endocrine, and neurological systems, thus giving added support to the idea that globally targeting gene expression regulation can affect multiple systems 
The association of paternal age with overall variance in gene expression in the blood of healthy controls is interesting. In our study, decreased variance in gene expression was a risk factor for the diagnosis of autism. That another well established risk factor for autism, increased paternal age, was associated with a similar decrease in variance raises the interesting possibility that the associated increased risk of autism in children of older fathers could be mediated by changes in global levels of gene expression regulation or by paternally transmitted age-related factors that are linked to changes in the global regulation of gene expression. In other words, an effect on global levels of gene expression regulation transmitted by older fathers might act to increase the likelihood that neurodevelopmental abnormalities will occur and thus increase the rates of autism in children of older fathers. Determining whether this hypothesis is correct would require a large-scale prospective study in which statistical models assessed whether increased rates of autism in children of older fathers were statistically mediated by effects on variance in gene expression.
The association of variance in gene expression with paternal age is also interesting because it suggests that fathers may transmit factors influencing global levels of gene expression regulation in offspring. Paternal age has previously been associated with another genomic phenotype, increased telomere length, indicating that, in principal, global genomic properties can be sensitive to paternal age
. Possible mechanisms for the transmission of factors influencing global levels of gene expression regulation are suggested by the association of aging with de novo
germ line mutations and global changes in DNA methylation 
. Each of these properties, if paternally transmitted to offspring, could influence global regulation of gene expression. Given the vast number of gene expression changes that occur throughout development, understanding how global effects on gene expression regulation may be transmitted across the germ line to ultimately influence the overall variance in gene expression in differentiated cells could prove crucial to our understanding of how general genomic properties are transmitted to and influence the function of very different cell types.
The parallels of the current study to our work in mice also bear discussion. We found in mice that epigenetic factors acted during early brain development to elicit changes in adult behavior that were accompanied by changes in the variance in gene expression
. In a separate study, we found that the same hippocampal-dependent behaviors were paternally transmitted to offspring even though fathers were not present for rearing suggesting an epigenetic germ line transmission of factors that influenced brain development, behavior, and variance in gene expression 
. In autism, paternally transmitted epigenetic factors related to age have also been implicated 
and in the current study paternal age was associated with differences in the variance in gene expression in the general population. Though much further study is need, both lines of work suggest the presence of epigenetic paternally transmitted factors in both mice and humans that can influence brain development and global levels of gene expression regulation.
Alterations in the overall variance in gene expression in peripheral blood lymphocytes are also of interest. Children with autism have immunological abnormalities and it has been previously reported that gene expression differences were found in immune cells of children with autism 
. More surprising was the finding in this study that increased paternal age was associated with the same global gene expression phenotype and specific alternations in transcription-related biological pathways in healthy controls as was associated with the diagnosis of autism. As discussed, a relationship to paternal age suggests that factors influencing global levels of gene expression regulation may be transmitted across the germ line. Additionally, findings in the blood suggest the possibility for effects of paternal age on immune function. An association of increased paternal age with immune function is suggested by a common link between increased paternal age and mental disorders such as schizophrenia, bipolar disorder, and autism, and the association of these disorders with a common set of immune abnormalities that includes increases in pro-inflammatory cytokines, IL-1beta and IL-6 
. A more generalized effect of paternal age on immune function seems plausible. Increased paternal age has been associated with impairments in neurocognitive ability during infancy and childhood in the general population 
and with an increase in “externalizing” behaviors in the general population 
. Given the links between paternal age, neurodevelopomental disorders, and immune function, it seems that paternal age might also have a more generalized effect on immune function as it was found to have on neurodevelopment.
Finally, traditional approaches to gene expression analysis suggested a possible mechanism for the changes in gene expression variance. We found enrichment for genes involved in transcriptional regulation in genes that were down-regulated with respect to paternal age and autism. Analysis of the effects of transcription inhibition on gene variance was supportive of the hypothesis that down-regulation of transcriptional regulators may account for the decreased variance in gene expression in the current study. Also of interest we found enrichment for genes regulated by zinc, an important component of many transcription factors. Since it has been demonstrated that neuronal plasticity is often transcription dependent 
, it seems plausible that an alteration in the general regulation of transcription could adversely impact neurodevelopment. Though further study is needed, the suggestion by pathway analysis of transcription dysregulation in autism is consistent with genetic findings that mutations in general transcription factors are associated with neurodevelopmental disorders including autism 
. The possible importance of zinc is also of interest given the recent report of a decreased zinc/copper ratio in children with autism spectrum disorders 
Multiple lines of evidence including genetic linkage 
, pharmacological studies 
, and the associations observed in the current study suggest that variations in global levels of gene expression regulation may contribute to the risk for neurodevelopmental disorders including autism. An association of variance in gene expression with paternal age in healthy controls suggests that factors influencing global levels of gene expression regulation may be transmitted across the germ line. Findings in immune cells suggest that global levels of gene expression regulation may impact systems other than the brain. Finally, pathway analysis suggests that altered regulation of transcription may underlie decreased variance and may increase risk for autism. A more thorough understanding of the mechanisms and the biological impact of the reported associations with overall variance in gene expression represent important areas for future research.