A genomic lifespan program that reorganises the young adult brain is targeted in schizophrenia
- University of Edinburgh, United Kingdom
Figures
Figure 1 with 1 supplement
Gene expression trajectories can be classified and quantified based on the characteristics of their turning points (TTTPs).
(a) Simple trajectories (upper panel) do not contain TTTPs (red dots) whereas complex trajectories (lower panel) contain a TTTP. Complex trajectories are further classified according into those that …
Figure 1—figure supplement 1
ALiGeT scoring assigns a score to each gene for each year of age.
The value decays further with increasing distance from the year at which the turning point occurs. This diagram shows the equations and graphs for how the scores (dashed red line) decay for a …
Figure 2 with 3 supplements
Trajectories and turning points characterise brain age.
(a) Percentage of TTTPs at each age of human lifespan. Mean age is 26.0 years for males and 27.5 years for females. (b) Percentage of TTTPs at each age of human lifespan using three different …
Figure 2—figure supplement 1
Distribution of ALiGeT scores at 15 (a), 25 (b), 35 (c) and 55 years (d) along with examples of trajectories, their scores and expression data.
The top row of panels shows the frequency distribution of ALiGeT scores. The second row of panels shows the ten interpolated trajectories with highest scores that upregulate prior to turning. …
Figure 2—figure supplement 2
the mouse samples are divided amongst two background strains (C57Bl/6 and 129s5) and two sexes, and range from 58 to 600 days of age, whilst human samples are from both sexes and span almost 80 years.
(a) The distribution of mice from different background strains is balanced with respect to age, whilst (b) shows the same for sex. (c) The age of the Braincloud human samples from each sex.
Figure 2—figure supplement 3
The optimal number of probes to use for age predictions was 40 in humans (a) and 100 in mice (b).
These graphs show the Sum Squared Errors found when 10 training/test runs are performed for each value of Ncutoff.
Figure 3
Genes associated with high levels of expression in particular cell types are enriched at different stages of the human and mouse lifespan.
(a) Expression Weighted Cell-type Enrichments (EWCE) for the top 10% of genes with largest DeGeT scores within each age window for the human prefrontal cortex. Enrichments were calculated separately …
Figure 4 with 1 supplement
Post-synaptic density (PSD) genes are associated with TTTPs during windows in adolescence and early adulthood.
(a) PeGeT scores for human PSD (hPSD) genes compared with scores from 100 randomly sampled gene lists. For each of the 100 random lists, the ith largest score is plotted against the ith largest hPSD …
Figure 4—figure supplement 1
DeGeT enrichment for the human post-synaptic density gene set was repeated in both the (a) Somel and (b) BrainSpan datasets.
Boxplots show the mean bootstrapped scores (10,000 random lists) and the red cross marks the score of the hPSD list for that age. *, Black asterisks mark significance with bonferroni corrected p<0.05.
Figure 5 with 8 supplements
Schizophrenia susceptibility associated genes are associated with TTTPs during windows in adolescence and early adulthood.
(a) Schizophrenia susceptibility genes show a window (22–26 years) of increased ALiGeT scores during young adulthood in the Braincloud dataset. Beneath the red horizontal line marks the point of …
Figure 5—figure supplement 1
DeGeT enrichment for the schizophrenia associated genes in the two additional human frontal cortex datasets.
Results are for (a) combined schizophrenia gene set in the Somel dataset (b) combined schizophrenia gene set in brainspan dataset. Boxplots show the mean bootstrapped scores (10,000 random lists) …
Figure 5—figure supplement 2
Early adult peak of turning points as well as significant windows for disease and synapse ALiGeT increases are robust against changes in the model used to fit the data.
Examples of how alternate methods fitting the data affects individual trajectories are shown in the first two rows using the data for Plp1 as an example. The following models are shown: cubic spline …
Figure 5—figure supplement 3
Down-sampling sensitivity analysis indicates that DeGeT enrichments are stronger in earlier adulthood for schizophrenia associated genes than for the hPSD.
Both the hPSD and schizophrenia (combined) gene sets were down sampled repeatedly (20 times for each gene set length) and Bonferroni corrected DeGeT enrichments calculated. Using subsets of 650 …
Figure 5—figure supplement 4
Enrichment of Schizophrenia associated genes does not occur as a side effect of turning points in genes with long transcript lengths or GC-content.
(a) Schizophrenia susceptibility genes still show windows of increased ALiGeT scores during young adulthood, when bootstrapped gene lists are selected to match the GC-content and transcript lengths …
Figure 5—figure supplement 5
:The ALiGeT scoring function had a single fixed parameter, which controls the extent to which genes which turn proximally to the target year receive high scores: higher values for the parameter result in increasingly rapid decay.
We show here that the main results are robust against a range of values for this parameter, with the primary effect being on the width of the enrichment windows. ALiGeT windows exist for …
Figure 5—figure supplement 6
Replication of results with all fetal samples are dropped from the Braincloud dataset.
(a) Percentage of TTTPs at each age of human lifespan. Peak age is 26.0 years for males and 27.5 years for females. (b) DeGeT enrichments for human post-synaptic density genes. (c) DeGeT enrichments …
Figure 5—figure supplement 7
DeGeT enrichment for the additional de novo gene schizophrenia gene set from the Gulsuner et al paper.
Results are for (a) additional de novo genes only against the Braincloud dataset (b) combined schizophrenia gene set and the additional de novo genes against the braincloud dataset. Boxplots show …
Figure 5—figure supplement 8
Schizophrenia heritability enrichments (calculated with GWAS summary statistics instead of associated gene set) are confirmed in the Somel dataset.
(a) DeGeT scores at 15—16 years are enriched for Schizophrenia heritability. Bootstrapping was performed by shuffling gene level association z-scores which had been calculated using MAGMA. Boxplots …
Figure 6 with 3 supplements
Trajectories of schizophrenia associated genes show further distinctions in their cell-type and trajectories.
(a) Schizophrenia susceptibility genes show a window of increased ALiGeT scores when the analysis is restricted to the 5000 most neuron specific genes, but not the 5000 most glial specific genes. …
Figure 6—figure supplement 1
Enrichments for schizophrenia susceptibility genes are specific to particular cell types.
(a) Schizophrenia susceptibility genes show significantly increased ALiGeT scores, when the analysis is restricted to the the 1000th—8000th most neuron specific genes. The x-axis shows the number of …
Figure 6—figure supplement 2
The synaptic proteome was found to show extensive changes during early adulthood, with particularly strong impact on proteins associated with adult-onset mental disorders.
Shown are the expression profiles for the differentially expressed proteins which are associated with adult-onset neurological disorders. Expression values shown are the mean for each month, divided …
Figure 6—figure supplement 3
No cognitive disorders other than Schizophrenia showed larger PeGeT scores than expected by chance.
PeGeT scores for gene lists associated with six disorders [autism (a,b), intellectual disability (c), Multiple Sclerosis (d), Amyotrophic Lateral Sclerosis (e), Parkinson’s disease (f) and …
Figure 7
Summary of age windows for synaptic and cellular processes and diseases across the human lifespan.
The intensity of the shaded boxes indicates the enrichment in relevant genes. The TTTP-peak in young adults (vertical dotted lines show approximate boundary) coincides with synaptic and neuronal …
Additional files
-
Supplementary file 1
(a) Human PEGET scores. (b) Mouse PEGET scores. (c) Set of genes that were found to have the top PEGET scores in both human and mouse. (d) Post-synaptic proteome gene sets. (e) Annotated table of proteins differentially expressed with age as well as the assigned functional classes used to determine enrichments. (f) Disease gene sets
- https://doi.org/10.7554/eLife.17915.026
-
Transparent reporting form
- https://doi.org/10.7554/eLife.17915.027
