SFARI Sex Differences Collaborations gather to share their progress

On January 30–31, SFARI convened its first annual meeting of the SFARI Sex Differences Collaboration (SSDC). Director Kelsey Martin welcomed 86 attendees, half of whom attended in person, with half attending virtually, to hear the first reports from this collaboration.

Funded in 2021, the SSDC collaboration consists of five groups of researchers who seek to understand the male overrepresentation in autism spectrum disorder (ASD) diagnoses; how this comes about could hold important insights into the nature of autism itself.

“This sex bias could be nature’s hint about something important to know about autism,” said Kelsey Martin, director of SFARI, when she welcomed the attendees.

The five collaborations follow different lines of inquiry, ranging from genomic to cellular to behavioral studies, in animal models and humans with a total funding commitment of $28M over 4 years. Each team consists of several principal investigators spread across multiple institutions. Although each team applied to the RFA separately, the goal is for teams to collaborate both within and across teams.

In her comments, Brigitta Gundersen, SFARI Senior Scientist who oversees the collaboration, stressed the importance of open science to its success, saying, “SFARI is working to set up a data-sharing platform that will allow the teams to share their findings first with each other, and then with the broader research community.”

Genetic and molecular dissection of autism sex differences

Aravinda Chakravarti of New York University School of Medicine introduced the scope of the first collaboration, which seeks to understand how sex differences in ASD might arise by looking for sex-biased penetrance of ASD gene variants through sex-biased gene regulation. Work in Chakravarti’s group is mapping regulatory elements in the genome, including those that are sex-hormone sensitive. Preliminary results find a complex picture, with over 200,000 cis-regulatory elements (CREs) identified in human prenatal cortical tissue. The effects of these CREs on gene expression are being evaluated in neuronal cell models in vitro.

Evan Eichler of the University of Washington described plans to analyze more than 50,000 families with autism to discover high-impact ASD-related genes that show sex bias1. His group is also generating long-read sequence and assembly data from more than 70 autism families where short-read sequencing has failed to identify a genetic cause. The X chromosome also remains an important suspect for sex bias, and a re-examination of the chromosome with long-read sequencing may prove fruitful. Long-read sequencing also has increased sensitivity for structural variants and variants within repetitive loci compared to short-read sequencing2.

Tomasz Nowakowski of the University of California, San Francisco is tracing how sexually dimorphic patterns of gene expression arise in the developing human brain. Preliminary results from a survey of over 1 million single-cell RNA sequences from human prenatal brain tissue has pinpointed over 300 differentially expressed genes between males and females. Some of these genes are established SFARI ASD risk genes, and over half are expressed solely in microglial cells, which suggests a role for this cell type in brain sexual dimorphism.

Huda Zoghbi of Baylor College of Medicine will assess the function of potential genes underlying the ASD sex bias found by the collaboration. Currently, she is examining regulators of MECP2, the dosage-sensitive X-linked gene of which too little expression produces Rett syndrome, and too much produces MECP2 duplication syndrome. This includes identification of noncoding regulatory elements that affect MECP2 levels3, transcription factors that interact with MeCP24, synonymous mutations that ultimately affect MeCP2 protein levels, and post-translational regulators.

Sex bias through epistasis

Joseph Dougherty of Washington University School of Medicine introduced his collaboration, which is looking for epistatic interactions between genes or between genes and environment, that may produce the ASD sex bias. A guiding hypothesis is that the same genetic variants may have sex-specific effects, which could increase male liability for ASD5. This is in contrast to the idea that risk variants produce the same effects in females and males, but that the liability threshold for autism is lower for males than females.

Dougherty described a strategy to systematically look for interactions between ASD-related risk variants and sex using an approach they have recently applied to study sex interactions in depression
. The approach, called massively parallel reporter assay (MPRA), measures the effects of hundreds of regulatory variants on gene expression in mouse neurons in vivo simultaneously. Other experiments will explore how sex hormones, which are released with different trajectories in males and females, interact with brain development.

Sven Sandin of the Icahn School of Medicine at Mount Sinai and Karolinska Institutet presented a population genetics approach to test the idea that a female protective effect underlies the ASD sex bias. Using multi-generational data from Swedish national registries, he looked for evidence of a female protective effect in sisters of affected probands: if there is a female protective effect, then their offspring would be expected to carry a higher genetic liability for ASD, and thus have a higher rate of ASD. However, the recurrence rate was not higher than expected, which does not support a female protective effect7. This suggests an alternative model in which males and females share the same liability threshold for ASD, but that an increased variance in liability in males may produce a sex bias.

Tychele Turner of Washington University School of Medicine presented computational approaches from her lab for finding and evaluating genomic variants relevant to autism. One of these methods, called Hare And Tortoise (HAT) speeds up the detection of de novo variants from sequencing data 100-fold, and works for both short and long-read sequencing data8. Turner also described a method to help interpret non-coding variation called GeneRealized clustering by Mutation Position (GRUMP), which is similar to her unbiased cluster method called CLUstering by Mutation Position (CLUMP)9, as well as a new method called 3D-CLUMP developed in collaboration with Rachel Karchin at Johns Hopkins University that tests for clustering of rare, de novo missense variants in the 3D structure of a protein, with genome-wide significance.

Sex differences and microglia

David Page of Massachusetts Institute of Technology introduced a collaboration exploring how the ASD sex bias may arise molecularly, and whether microglia are key mediators. Page proposed that the inactivated X chromosome — which females have, but males do not — may contribute to the ASD sex bias. His lab has found that the inactivated X chromosome can modulate gene expression on the active X, in cell lines made from people with X chromosome aneuploidies10.

Many of the modulated genes are established SFARI autism risk genes, which suggests that males lack a level of control of these genes that females have.

Specific cell types may mediate the sex bias in ASD, which has led Christopher Glass of the University of California, San Diego to explore a role for microglia. Though microglia make up only a small proportion of cells in the nervous system, they are present within the brain early in development, and have a sensitivity to the environment that might put them in a place to mediate risk for disease. Glass is examining the microglial gene expression phenotypes in fetal and postnatal human brain tissue because these may display sex differences and may be governed by specific enhancers that control cell-specific and stage-specific gene expression11.

Nicole G. Coufal of the University California, San Diego has been investigating both autism genes and sex differences in human stell cell–derived microglia. The ASD-related gene MEF2C is highly active in developing human microglia. These microglia show increased signs of inflammation, lipid accumulation and dystrophy both in vitro and in vivo using a xenotransplantation model. Experiments that engraft these deficient microglia into the brains of mice are exploring their contribution to behaviors associated with autism. These techniques and approaches are being applied to interrogate the contribution of the sex chromosomes to microglial behavior, since microglia are among the brain’s most sexually dimorphic cells.

APEX - Autism and Prenatal Sex Differences

Simon Baron-Cohen of the University of Cambridge introduced the next collaboration, which is focused on interactions between genetics and prenatal sex steroid hormones, including those from the placenta and adrenal glands, particularly using two legacy novel biobanks they are creating: the Autism Placental Biobank and the Israel Amniocentesis Biobank. They were designed in consultation with the autism community. Using induced pluripotent stem cell derived neurons from autism cell lines and post-mortem fetal brain data, the team will be testing autism-associated gene-hormone interaction effects on neuronal development.

Studies of ASD incidence have found most boys receive diagnoses between 2 to 4 years of age and most girls receive diagnoses in adolescence. To see if this reflects differences in behavioral development, Varun Warrier of the University of Cambridge is studying several birth cohorts that follow children longitudinally. Using a data-driven computational approach to model data from the behavioral Strengths and Difficulties Questionnaire (SDQ), Warrier has found that the SDQ scores track with early and late ASD diagnoses, which suggests behavioral trajectories toward ASD differ between males and females.

Mahmoud Musa, a postdoctoral fellow with Hilary Martin at the Wellcome Sanger Institute, is also using birth cohorts to get at the genetics potentially underlying the ASD sex bias. The Avon Longitudinal Study of Parents and Children (ALSPAC) is a particularly rich resource, as it contains detailed phenotype information from several time points, and whole exome sequencing data from over 8,000 participants. Musa is exploring whether the burden of rare genetic variation associated with ASD is related to the development of social, communication and motor difficulties, and whether these burdens have differential sex effects.

Given known sex differences in brain structure, Dorothea Floris of the University of Cambridge is investigating whether any of these are related to ASD. In a recent study, she found that a computer model trained to distinguish between male and female brain scans of controls could then more accurately classify brain scans of males with ASD than those of females with ASD12. This suggests that male-specific properties of brain structure are accentuated in ASD, and these may originate from prenatal gene expression patterns.

If early events in human cortical development are key to ASD, then it will be important to understand how this typically unfolds at a molecular level. Jonathan Mill of the University of Exeter presented work on gene regulation during cortical development, with a focus on the DNA methylation patterns that govern gene expression. DNA methylation patterns differ across age, cell types and sex, and affect established ASD risk genes. Mill finds that over 1,200 genomic sites show sex differences in methylation in fetal cortex, and display tremendous transcriptional diversity, with novel and rare isoforms and instances of alternative splicing.

Deepak Srivastava of King’s College London described his studies of how hormones affect sex differences in cellular models. Preliminary results using neurons derived from three individuals carrying deletions of the ASD-related gene NRXN1, showing signs of sex-specific differences in transcription and splicing. In other experiments, Srivastava is exposing human-derived neurons and organoid models in vitro to sex hormones to understand the molecular outcomes.

Searching for the loci of sex bias

Stephan Sanders of the University of California, San Francisco introduced the idea guiding the final collaboration: this group hypothesizes that the ASD sex bias stems from an interaction between cells that differ between sex and cells impacted by ASD risk. That is, the cellular locations of sex differences do not necessarily match the locations vulnerable to ASD risk. Their approach will find the brain regions, circuits and cell types that are sexually dimorphic, then ask how these are influenced by ASD risk.

To work out precisely where sex hormones act in the brain, Jessica Tollkuhn of Cold Spring Harbor Laboratories has been mapping the DNA binding sites of estrogen receptor alpha in the mouse brain, specifically in the sexually dimorphic bed nucleus of the stria terminalis (BNST)13. These receptors activate sexually dimorphic patterns of male-biased gene expression, and a similar mechanism may well mediate hormone-sensitive changes in brain development in humans.

Xin Jin of Scripps Research is working to generate comprehensive brain-wide maps of hormone receptors in the mouse brain throughout brain development. While adult versions of these maps exist, it is crucial to have atlases for different developmental time points with high resolution. Preliminary results find robust estrogen receptor alpha expression in many brain regions important for social behavior in newborn mice. This helps connect the sex differential hormonal signaling to autism, which manifests at young age with a developmental origin.

Spatial transcriptomics, a method to classify cell types based on their transcriptome in situ, also promises to highlight potential cell types important in sexual differentiation and ASD risk. Using this approach to survey multiple regions in the mouse brain, David Young of the University of California, San Francisco reports clusters of cells with enriched expression of ASD genes. The approach holds promise in its ability to pinpoint cells particularly vulnerable to ASD risk genes throughout the brain through the generation of brain-wide atlases of enrichment.

To understand how ASD risk integrates with sex differences, Donna Werling of the University of Wisconsin-Madison is studying the development of sexually dimorphic areas like the BNST in mouse models of ASD, such as those deficient in SHANK3. The gene expression in cells of these regions will be evaluated for interactions between sex and ASD risk.

Identifying relevant cell types can be important for studying sex-specific behavior. As an example of this, Mengyu Liu of David Anderson’s lab at the California Institute of Technology presented work focused on understanding how female mice switch their behavior toward male mice, from sexual receptivity in virgins to aggression when lactating. Using activity-dependent single-cell RNA sequencing, she found that two distinct cell types in the ventromedial hypothalamus (VMHvl) controlled these behaviors14. The relevant cell types may differ between males and females, however; while the cell types mediating aggression in females overlap and belong to subsets of aggression cell types in males, the cell type mediating mating in females does not overlap with those in males.

In studies of how neurons transmit information about social behavior, Vikaas Sohal of the University of California, San Francisco and colleagues have found that synergistic firing between neurons is important while mice engage in social interactions15. To get at sex differences, his group is mapping projections from the BNST to the medial prefrontal cortex to see if there are sex differences in input during social behavior.

Devanand Manoli of the University of California, San Francisco described work to understand sex differences in social attachment behaviors, as studied in prairie voles. Prairie voles form pair bonds, which consist of both a preference for the partner vole and a rejection of strangers. To understand how these behaviors are mediated, Manoli has developed tools to genetically manipulate prairie voles. Preliminary results find sexually dimorphic effects on pair-bonding behavior in animals carrying ASD-related mutations.


Final discussions ranged from how the different approaches outlined may or may not intersect; whether subtle differences in cell populations can be sufficient to cause sexually dimorphic behavior, or the sex bias in ASD; the possibility that sexually dimorphic behaviors could reflect operation of different circuits (rather than the same circuits acting differently); and the importance of comparative studies to find conserved behavioral motifs across species.

In closing, Kelsey Martin observed that the collaborations were asking the same question through different approaches, and that it would be essential to have continuous communication to expedite the research. “Stay in touch,” she added.


  1. Wang T. et al. Proc. Natl. Acad. Sci. U.S.A. 119, e2203491119 (2022) PubMed
  2. Noyes M.D. et al. Am. J. Hum. Genet. 109, 631-646 (2022) PubMed
  3. Shao Y. et al. Genes Dev. 35, 489-494 (2021) PubMed
  4. Zhou J. et al. Proc. Natl. Acad. Sci. U.S.A. 119, e2119078119 (2022) PubMed
  5. Dougherty J.D. et al. Neuron 110, 3243-3262 (2022) PubMed
  6. Mulvey B. et al. bioRxiv (2021) Preprint
  7. Dan B. et al. Biol. Psychiatry 88, 480-487 (2020) PubMed
  8. Ng J. and Turner T., bioRxiv (2023) Preprint
  9. Turner T.N. et al. Hum. Mol. Genet. 24, 5995-6002 (2015) PubMed
  10. San Roman A.K. et al. Cell Genom. 3, 100259 (2023) PubMed
  11. Nott A. et al. Science 366, 1134-1139 (2019) PubMed
  12. Floris D.L. et al. Am. J. Psychiatry 180, 50-64 (2023) PubMed
  13. Gegenhuber B. et al. Nature 606, 153-159 (2022) PubMed
  14. Liu M. et al. Neuron 110, 841-856 (2022) PubMed
  15. Frost N.A. et al. PLoS Biol. 19, e3001235 (2022) PubMed
Recent Workshop and Meeting Reports