Autism, RORA, and testosterone

Autism, RORA, and testosterone

Autism and high testosterone (or androgen) levels have been linked for a long time in part because autism appears to be so much more prevalent among boys compared to girls (more on that later). Thanks to this apparent male bias, the connection between androgens and autism, the “extreme male brain hypothesis” of autism, gets a lot of attention. Adding to this attention is recent intriguing work from Valerie Hu and colleagues, published in PLoSOne, connecting an androgen to the regulation of a specific gene, called RORA, that may be associated with autism.

The RORA finding has the news media, well, ROR-ing about it with headlines promising that “Testosterone may bump autism rates in males” and “Faulty testosterone cycle may explain male autism bias.” This latter story even concludes in the lede that the association demonstrates that it is more likely that “males will accumulate testosterone in the dangerous amounts that are thought to lead to autism.” A bit of a leap, that, especially as the researchers did not strictly use testosterone in this work and did not show a male bias in RORA levels.

A pause for a few observations before we move on

1. This study was an in vitro study in which a nerve cell line participated through direct exposure to hormones in a culture dish and a protein from the brain tissues from autistic and non-autistic people was measured.

2. None of the work was done in a living organism. That’s not an indictment of the work, just a caution against leaping to conclusions. That’d be you, news media.

3. The male brain in mammals forms as male and males act like males in part thanks to estradiol, a kind of estrogen that an enzyme called aromatase makes from testosterone. Aromatase also converts androstenedione into another estrogen, estrone (see figure).

4. Another enzyme, 5-alpha-reductase, also can act on testosterone to turn it into dihydrotestosterone (DHT), a powerful androgen that binds the androgen receptor about three times more tightly than testosterone itself.

5. Hormone pathways are usually tightly regulated, engaged in a number of feedback loops to maintain the hormone levels within a very narrow concentration range. Many proteins are involved in this maintenance, some grabbing hormones and making them unavailable and others converting excess hormone into another form until levels return to a balance.

RORA and autism

In previous work, Hu and her group assessed the gene expression patterns between sets of identical twins, one of whom had autism. They found that RORA (retinoic acid-related orphan receptor alpha) stood out as a gene tagged more often for shutdown in the autistic twins, so it wasn’t used as much. They looked at brain slices from autistic people and compared RORA expression levels in these slices to those from non-autistic people and found that RORA expression was lower in the autistic brains. Not all autistic brains, but most. RORA was officially identified as a “candidate gene” in autism.

RORA and aromatase

Their current work involved both a nerve cell line (a neuroblastoma cell line) growing in a dish and brain tissue samples from autistic people and sex- and age-matched controls. To evaluate how hormones affect RORA expression, Hu’s team exposed the nerve cells either to dihydrotestosterone (DHT) or estradiol. A hormone called 5-alpha-reductase is required to make DHT, which is a more powerful androgen than testosterone, and a hormone called aromatase is required to make estradiol. The DHT and estradiol receptors, the proteins that receive the hormone messages, both have the job of turning genes off or on. In this study, DHT caused its receptor to inhibit the RORA gene, while estradiol caused its receptor to activate RORA. Finally, RORA itself also turns genes on or off. Hu’s team showed that one of RORA’s targets is the aromatase gene.

Low RORA because of too much androgen, or too much androgen because of low RORA?

Quote from the Discussion:

On the other hand, we observed reductions in RORA in brain tissues from both male and female subjects with ASD, suggesting that RORA deficiency is not gender-specific. Interestingly, RORA and ER share a consensus binding site on DNA (AGGTCA) and consequently common target genes. The existence of shared gene targets may explain why females, with higher levels of estrogens, are less susceptible to autism. That is, estrogens may not only protect females against autism by increasing the level of RORA expression, but also by inducing shared target genes of RORA through ER, thus compensating in part for RORA deficiency.

Hu’s previous work identified differences in RORA chemical tagging between identical twins with and without autism. Their current work looks at levels of RORA expression. A chicken–egg question arises for anyone focused on an androgen framework: Do people with autism appear to have low RORA because of too much androgen, or do they have too much androgen because of low RORA? Their statement above seems to allow for both scenarios. Yet RORA levels didn’t differ between autistic males and females, and without a sex bias in RORA and aromatase levels, these findings aren’t easily used to explain the presumed sex bias in autism. Unless you change the question.

Do people with autism have too much brain androgen…or too little brain estrogen?

When Hu’s team looked at the brains of people with autism, the RORA decrease didn’t differ between males and females, and we can only assume that both also shared low aromatase levels (the authors do not specify). The authors note findings suggesting that a female’s more abundant estrogens could engage in crosstalk with a RORA pathway to make up for any deficiency.

Their cell culture outcomes indicate that an estrogen deficiency would lead to less RORA activation, further reduced aromatase, and worsened estrogen deficiency. Even normal androgen levels could exacerbate this decrease if aromatase isn’t making enough estrogen to balance androgen action. Hu’s group doesn’t seem to have tested both hormones at once. The brain tissue findings suggest that both males and females have low RORA and aromatase, which points to an autism association based on estrogen deficiency with no confirmation of excess androgen.

An estrogen deficiency in the presence of normal androgen levels, which studies indicate is a reasonable scenario, would result in the overall systemic normalcy of autistic children (e.g., no hyperandrogenized sex characteristics). Yet it would leave the potential for more prominent androgen-based outcomes in the sex that already has less estrogen: the males. If females do have a compensatory mechanism because of overall higher estrogen levels in the first place (thanks to ovaries), then the “cross-talk” the authors suggest would likely result in a lesser severity of autism symptoms in many girls.

Brains are funny things

The features of a brain-related disorder or difference in a male can vary considerably from the features of the same disorder in a female, which takes me to another observation: It is possible that one reason for this male bias is underdiagnosis among girls? Females often manifest autism in more subtle ways and are more likely to be misdiagnosed with depression or personality disorder, are less likely to be referred for help at an early age, and show less aggression that might lead to a referral or diagnosis. Again, this scenario falls right into the framework of an estrogen deficiency hypothesis in which girls compensate to some extent with their naturally higher starting levels of estrogen.

I’m not suggesting that retinoic acid-related orphan receptor alpha isn’t a candidate gene to pursue further as part of one of the many potential paths that lead to autism; in fact, I just spent about 1300 words focusing on how it might work along with aromatase to reduce estrogen levels. In the PLoSOne paper and in interviews, Valerie Hu and her team also made sure to note that RORA was just one among many possibilities. That didn’t stop the dramatized headlines, though. As always with any results having to do with a candidate autism gene, if it turns out to be involved in autism, it’s likely a part of only one of several pathways that are.