466: A quieter MECP2 switch and the male bias in autism

[00:00:20] Welcome to Base by Base, the papercast that brings genomics to you wherever you are. Thanks for listening and don't forget to follow and rate us in your podcast app. [00:00:30] Base by Bass is now on YouTube too, at base by bass, where every episode gets a video with chapters and the full description. Come subscribe. [00:00:44] Picture a family A mother, healthy, with nothing unusual in her history. [00:00:49] Two sons. One of them is diagnosed with autism and with attention deficit hyperactivity disorder. His speech is normal. His IQ is in the average range. He has no seizures and no problems with movement. His brother is neurotypical. Genetic testing finds nothing to explain the difference. No new mutation in a known autism gene. No missing or duplicated stretch of a chromosome. On paper, the family looks ordinary. [00:01:19] Autism is diagnosed about 44 times more often in boys than in girls. Nobody fully knows why. [00:01:26] One idea points to the X chromosome. Girls carry two copies of it. Boys carry one. So if a gene on the X is damaged, a girl has a backup copy that can soften the blow. A boy does not. Whatever that one copy does, it does alone. [00:01:45] Here is a way to picture it. Think of a gene as a lamp and of the stretches of DNA around it as a panel of dimmer switches that set how bright the lamp shines. Break the lamp and the room goes dark. That is a severe disease. But nudge one of the switches so the lamp shines a little dimmer and the change is subtle. In a room lit by two lamps, you might not notice at all. In a room lit by one, you might. [00:02:12] So the question of today's episode is precise. [00:02:15] Could a small change in one of those switches, carried silently by a mother, dim a critical brain gene in her son just enough to contribute to autism? And could we find such switches and test them in human neurons? [00:02:35] Today we celebrate the work of Rebecca Meyer, Schuman, Huda Zoghbi, and colleagues from Baylor College of Medicine, the Jan and Dan Duncan Neurological Research Institute at Texas Children's Hospital, the University of Washington, Peking University, the University of California, San Francisco, and New York University who have advanced our understanding of how gene regulation on the X chromosome may shape autism. Their paper, A massively parallel reporter assay of MEXCP2 CIS regulatory elements reveals genetic candidates for male biased autism, was published in the American Journal of Human Genetics in November 2020. [00:03:13] Foreign let's start with the gene. MECP2 makes a protein called methyl CPG binding protein 2, which binds to chemically marked DNA in brain cells and helps tune the activity of many other genes. It sits on the X chromosome and it is famously sensitive to dose. [00:03:34] Lose its function in a girl and the result is Rett syndrome, a severe neurological disorder. [00:03:40] Lose it completely in a boy and the result is usually a severe and often fatal encephalopathy in infancy. Too much is also a problem. Boys with an extra copy develop MECP2 duplication syndrome. [00:03:56] The pattern holds for milder changes too. A known variant that only partly weakens the protein causes intellectual disability and psychiatric symptoms in boys. While the mothers who carry it are healthy or only minimally affected, mothers who carry the duplication are usually unaffected as well. [00:04:14] Why? Because in each cell 1x chromosome is switched off, and in these mothers the altered copy is the one usually silenced. The healthy copy does the work. [00:04:27] Mice suggest how narrow the safe range is. [00:04:30] Earlier work deleted a consistent conserved enhancer inside the mouse gene expression dropped by 30%, and the mice showed social deficits, anxiety and hyperactivity. [00:04:42] Deleting a conserved repressor raised expression by 50% instead. And those mice also had social deficits, along with problems in movement, learning and memory. [00:04:52] Too little or too much and behavior changes. [00:04:56] 30% not a broken gene, just a dimmer one there was a hint in people as well. A study of brain tissue from autistic individuals without any identifiable MECP2 variant found extra methylation on the promoter and about a 32% drop in MECP2 expression. [00:05:18] So the team's hypothesis was rare variants in the switches that control MECP2 might push its level out of the healthy range and do it in boys who have no second copy to compensate. [00:05:33] There is a broader puzzle behind this. Earlier work compared new coding mutations in boys and girls with neurodevelopmental disorders. Surprisingly, the most sex biased genes, all on the X chromosome and including MECP2, had more damaging mutations in groups girls, not boys. [00:05:52] The likely reason is that the same complete loss in a boy is so severe that those boys are rarely found in autism cohorts, which leaves room for milder variants. The dimmer switches to matter in boys. [00:06:10] So how do you find those switches? First, the team looked for open chromatin stretches of DNA that are accessible in the developing human brain using two independent datasets from fetal prefrontal cortex around MECP2, they found five candidate regulatory elements the promoter, the two elements already studied in mice called element 2 and element 6 and two new ones which they named element A and element B. [00:06:37] Unlike the mouse elements, these two new ones are conserved in primates but not in mice. [00:06:45] Next, the variants the team analyzed whole genomes from 30,916 people in 8,245 families with autism across 12 cohorts. They kept only variants that fell in one of the five elements that were private to a single family, that were found only in male probands, and that were inherited from an unaffected mother. That filter left 22 variants. [00:07:11] As a comparison, they added 19 variants from GNOME AD, each seen in at least 50 males presumed to be neurotypical. [00:07:20] The central technique is called a massively parallel reporter assay. It works like A short piece of A regulatory element is placed in front of a reporter gene. A unique barcode, a short random tag of DNA, is attached to it. [00:07:36] This is done for thousands of pieces at once, and the whole library goes into cells. Then the team reads how much RNA each barcode produces compared with how much DNA was delivered. More RNA per copy means a stronger switch in our LAMP picture. It is like testing thousands of versions of the dimmer panel at the same time, each with a label and reading how bright each one makes the bulb. The team cut each element into overlapping tiles of 270 base pairs shifted by 70 in every tile. They predicted every possible binding site for transcription factors, the proteins that grip DNA and turn genes up or down. [00:08:20] Then they scrambled each predicted site twice independently. That added over 1,000 mutations to each element. [00:08:30] Where did all this happen? Not in a generic cell line. Mecp2 matters most in neurons. So the team used human induced pluripotent stem cells turned into cortical excitatory neurons in a dish. They ran four replicates, reached a median of about 135 barcodes per element, and the replicates agreed Closely known high activity and low activity sequences behaved as expected. [00:08:56] The idea is simple. If scrambling a site lowers activity, something there was turning the gene up. If it raises activity, something there was holding it down. [00:09:11] So what did they find? [00:09:13] Four of the five elements had at least one hotspot, a place where scrambling consistently changed activity. [00:09:20] The promoter had an activating hotspot. Element 2, the known enhancer, had four activating sites and three repressing ones. Element 6, the known repressor, had two strong repressing sites and one activating site. [00:09:35] Element B showed no active sites in neurons, which fits its lower accessibility there. It may work earlier in neural stem cells. And the new element A had three repressing sites and no activating one. [00:09:50] Can a map like this tell you which proteins sit on each switch? The team tried for the strongest activating site in element 2. They matched the scrambled sequence to known motifs, and the best candidates were two transcription factors TCF3 and ZNF281. Then they tested them. Knocking down TCF3 in neurons lowered MECP2 by 20% after one week and by about 55% two weeks later. Knocking down the second factor lowered it by 47% at the later time point. [00:10:27] The tiles also revealed teamwork. One activating site in element two worked no matter which neighbors surrounded it, but a second site nearby. A run of AD needs only worked when the first site was all also in the tile. Move the window so the first site falls out and the second loses its effect. [00:10:46] Put another way, some switches only work when a partner switch is close by. A map of single sites is not the whole story. [00:10:57] Now the variants from the families in element A, the new primate specific repressor. Four autism variants were tested next to five genome AD controls. [00:11:08] One control variant modestly raised activity, which shows how much variation healthy people tolerate. Two autism variants changed activity in one tile, but less than that control one did nothing, and one sitting right inside the strongest repressing hotspot raised activity beyond the control. [00:11:27] A full length luciferase test confirmed it. [00:11:31] Element A normally pushes expression below an empty vector with this variant it pushed less. [00:11:40] That variant came from a family with two affected brothers. Both inherited it from their unaffected mother. Neither carried a new coding mutation or a copy number change in a known autism gene, though both carried inherited missense variants of uncertain significance in other neurodevelopmental genes. [00:11:58] The SPOT is conserved in primates and and other large mammals, and a separate prediction tool flagged the same repressor motif. The scrambling had a loosened break on a switch that only primates have in the promoter. Seven autism variants were tested with two NOMAD controls that had no effect. [00:12:20] Four of the seven changed activity in one tile and three of those more than either control, but none of them repeated the effect in other tiles. Context dependent or false alarms? The luciferase test settled it. Two of the three made no difference. The third clearly lowered activity. It sits 17 base pairs from the activating hotspot and 120 base pairs upstream of where transcription starts. [00:12:48] Who carried it? A boy from the Simon Simplex company collection. Diagnosed with autism and adhd, he inherited it from his unaffected mother. His neurotypical brother did not. He also carried a missense variant of uncertain significance in another gene, but his unaffected brother carried that one too. [00:13:07] Because a variant this close to the core promoter was a surprise, the team checked his records for RET features. Full scale IQ of 97, normal verbal skills, no motor deficits, no seizures. [00:13:22] The variant changes a highly conserved letter and it is predicted to break a binding site for nfy, a three part protein complex that helps open promoters and start transcription. NFY is already known to bind the MECP2 promoter in other cell types, so the team used CRISPR to write this single letter change into the native gene of stem cells, then compared two edited clones with two unedited clones that went through the same steps. [00:13:52] First binding Using a method called cut and tag, they measured where NFY sits across the genome in the neurons. Then they ranked the strength of binding at every bound promoter. In the normal neurons the MECP2 promoter ranked around the 51st percentile. In the edited neurons it fell to the 9th, a clear loss of grip caused by one letter. [00:14:18] In four week old neurons, both known forms of the MECP2 message dropped by about 30%. [00:14:26] The protein dropped by about 30% too. [00:14:29] 30% the same size of drop that in mice produced social deficits, anxiety and hyperactivity downstream. Some genes moved only slightly, but one EBF3, a gene itself linked to neurodevelopmental disorders, fell by about half. [00:14:48] The lamp was not broken. It was dimmed. [00:14:57] What does this add up to? [00:14:59] The authors describe this as the first first link between non coding MECP2 variants and male biased autism. [00:15:07] Both variants that changed activity came from unaffected mothers and were found in affected sons as the hypothesis predicts, and both fell within or close to a strong binding site from the map. [00:15:19] That map is the second contribution. As whole genome sequencing spreads into research and clinics, it can help decide which non coding variants to deserve a closer look. [00:15:32] There is a twist in where the promoter variant sits. It is 17 base pairs away from the strongest activating site, not on it. The authors suggest this may be no accident. A variant hitting the activating motif directly might dim the gene so much that it causes RET like features or worse in a boy. And that boy would not be in an idiopathic autism cohort. The mild cases are the ones that survive the filter. [00:16:03] The idea also reaches beyond one gene. Other X linked genes are just as sensitive to dose and the same strategy map the switches, then test the variants found in boys could be applied to them. [00:16:16] The authors frame it as a way to explain part of the autism's missing heritability and part of the male bias. While noting that most genes linked to neurodevelopmental disorders sit on other chromosomes. So the X is unlikely to be the whole answer. [00:16:33] Now the limits the evidence rests on two families. A reporter assay takes a short piece of DNA about 300 letters hereout of its native setting so it can make miss long range loops and chromatin effects. One known insulator site barely registered in the assay, a reminder of that blind spot. [00:16:54] Neurons in a dish are young and simple and may hide subtle effects. Motifs can match several related factors. The parent cell line also carries a deletion in another autism linked gene, AUTS2, controlled for by using matched clones. [00:17:15] So here is where this leaves us around a gene known for severe disease. The team mapped the switches that set its level in human neurons. Among 22 rare variants passed from healthy mothers to autistic sons, two changed those switches. One loosened a break on a primate only element. The other in the promoter, weakened a binding site and dimmed the gene by about 30% in a boy with autism and average IQ, remember the lamp and its dimmer panel? In a room with two lamps, a dimmer bulb goes unnoticed. In a room with one, it may change everything. [00:17:56] If part of the male bias in autism lives in these quiet switches, what does this mean for the future of how we read the non coding genome of every boy who is tested? [00:18:13] This episode was based on an Open Access article under the CCBY 4.0 license. You can find a direct link to the paper and the license in our episode description. If you enjoyed this, follow or subscribe in your podcast app and leave a five star rating. If you'd like to support our work, use the donation link in the description. Thanks for listening and join us next time as we explore more science base by base.