We have assessed sensory processing, behavioural phenotypes and EEG recordings in the NRXN1 and SHANK3 genetic mouse models at various development stages and performed behavioural profiling of PTCHD1 knock-out mice. The NRXN1 and SHANK3 models reveal different trajectories of sensory information processing deficits, which provides a basis to establish readouts for pathogenic pathways. Sensory hyperexcitability in NRXN-1 knock-out mice is associated with increased wakefulness that may be linked to disturbed circadian systems. In the SHANK3 model, we induced inflammation during pregnancy and found that this moderated the influence of genetic vulnerability on autism symptoms, confirming the two-hit model.
We completed work on a translational battery of tests that span six bio-behavioural domains (social processing, social/non-social reward, emotional reactivity, predictability, attention and executive functioning, and sensory processing). The tests have been matched conceptually and content wise for administration to mice and humans. The SHANK3 knock-out mice show altered social exploration, more anxiety, and greater sensitivity to unpredictable noise, and changes in the expression of NMDA and AMPA receptor subunits in the striatum in adulthood. We have also created a new set-up that combines physiological measures with behaviour in freely moving animals in complex behavioural tasks. The tablet test battery for preschoolers demonstrates good feasibility and psychometric properties and provides initial evidence of biological convergence in multivariate cluster analyses that showed meaningful associated with variation in brain structure and clinical variation in ADHD symptoms severity.
Our immune profiling work shows that potential immune-based clusters do not align clearly with diagnostic categories such as autism, ADHD, or ID. The analysis of the microbiome data shows that thirteen bacterial genera are robustly associated with the broader group of NDCs and a partially overlapping list of seven genera with autism. A meta-analysis compared the gut microbiota landscape between adults with ADHD and controls and found specific genera to be associated with ADHD. All these bacterial genera play a role in inflammatory processes.
The recruitment and data collection for wave-1 and wave-2 of the PIP-CANDY clinical study has been completed, with wave-3 data collection still ongoing. Quality control, preprocessing and data analyses are under way and working groups have formed to establish data-analytic and publication strategies for each data domain and across-domains. We successfully evaluated the efficacy of the novel AI tool, SynthSR that employs a convolutional neural network to enhance scan quality, effectively correcting motion artifacts within PIP preschool MRI scans.
The recruitment and data collection of the Multiplex clinical study has been completed, reaching the recruitment target of 104 Multiplex families. We have accomplished the whole genome sequencing of the families and made an operative list of genes associated with NDCs and conducted a large-scale study on more than 13,000 individuals with autism and 210,000 non-autistic individuals. We found several differences in the genetic architecture of NDCs between Simplex (SPX) and Multiplex (MPX) families such that de novo mutations in high confidence NDC genes were less frequently observed in MPX compared to SPX families. In contrast, MPX have higher levels of autism and ADHD polygenic scores compared to SPX families.
We have finalised OWEY v3 as a consortium-wide database and delivered multiple new statistical techniques as implementable and usable tools (extensions of the normative modelling approach; SPADE - Spatial Analysis for Discriminative Estimation; SupBigFLICA - semi-supervised and integrated estimation of multi-modal data with behavioural/clinical data; and SPADE-strat (multi-group stratification using SPADE).
