Science1 publisher2 min readPublished
Seventeen autism mouse lines sorted into two mirrored molecular states in the prefrontal cortex
An Institute for Basic Science team read more than 1,000 mouse brain transcriptomes from 17 autism lines and found two opposing gene-expression profiles that responded differently to fluoxetine and lithium. In seven lines, sex decided which one.
The Scientist · Science desk
What happened
- A team led by Professor Kim Eunjoon at the Institute for Basic Science analyzed more than 1,000 mouse brain transcriptomes across 17 genetically engineered lines carrying autism-linked mutations.
- The lines converged on two opposing molecular states in the prefrontal cortex, one with suppressed synaptic genes and raised gene-regulatory activity, the other its mirror image.
- Early postnatal fluoxetine or lithium moved Group 1 animals more consistently toward control gene expression patterns than it moved Group 2 animals.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- constraint The same mutation was observed on both sides of the divide. A null result from one line, one sex, one age now covers only that combination.
- capability A drug screen now has a stratification variable other than the genotype, and it can be measured in the tissue before a compound is chosen.
- decision Choosing a brain region becomes a design decision with consequences: the structure was clear in prefrontal cortex and weak in hippocampus, so the wrong region hides the grouping entirely.
- exposure Published preclinical effects that report a genotype and a compound without sex, age and region are open to the question of which of the two states they measured.
Group membership is not a property of the genotype. In seven of the 17 lines (41 percent of the lines tested, 7 divided by 17 [1]), males and females carrying the identical mutation landed in opposite molecular groups [7]. Four lines followed across maturation changed assignment as the animals aged [8].
Autism has been linked to variations in more than 1,200 risk genes [3]. "Genetic discoveries have revealed extraordinary diversity in autism, but diversity alone does not explain the biology," said Dr. Bae Mihyun, a co-corresponding author on the work [13]. The two-state divide came out of three analyses run on the same brains: expression levels, alternative RNA splicing, and co-expression networks [17]. Group 1 showed reduced expression of synaptic communication genes with raised chromatin remodeling and RNA processing activity; Group 2 was the mirror image [5][6].
The obvious competing explanation for a two-state split is one vulnerable cell type driving everything. Single-nucleus sequencing of roughly one million nuclei from 205 mice, about 4,900 per animal [3], does not support that reading: the signatures came from coordinated shifts across many neuronal and glial populations [10]. Group 1 brains also varied more in the relative proportions of particular cell types [11].
The divide was clear in the prefrontal cortex and far less pronounced in the hippocampus [9]. Region matters for the sampling too. More than 1,000 transcriptomes spread across 17 lines averages about 59 per line [2], and those 59 are then divided again by sex, by developmental stage, and by brain region [7][8][9].
The drug arm dosed cohorts with fluoxetine or lithium during early postnatal development [16]. Group 1 animals shifted more consistently toward control gene expression patterns than Group 2 animals did [12]. The readout is transcriptome distance from controls. A screen that runs one compound in one line, of one sex, at one age is reading one point on a grid the same mutation can occupy from either side [7], and on this evidence the answer it returns depends on which side the cohort happened to sit. Every animal here is a mouse [1].
What to watch
- Whether the two-state divide reproduces in human postmortem prefrontal cortex, or stays a mouse transcriptome finding.
- Whether group assignment predicts a behavioral drug response, beyond a shift in gene expression toward controls.
- Whether the drug difference survives dosing outside the early postnatal window used here.