Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2013 May;14(2):143-52.
doi: 10.1007/s10048-013-0363-z. Epub 2013 Apr 28.

Peripheral Blood Gene Expression Signature Differentiates Children With Autism From Unaffected Siblings

Affiliations
Free PMC article

Peripheral Blood Gene Expression Signature Differentiates Children With Autism From Unaffected Siblings

S W Kong et al. Neurogenetics. .
Free PMC article

Abstract

Autism spectrum disorder (ASD) is one of the most prevalent neurodevelopmental disorders with high heritability, yet a majority of genetic contribution to pathophysiology is not known. Siblings of individuals with ASD are at increased risk for ASD and autistic traits, but the genetic contribution for simplex families is estimated to be less when compared to multiplex families. To explore the genomic (dis-) similarity between proband and unaffected sibling in simplex families, we used genome-wide gene expression profiles of blood from 20 proband-unaffected sibling pairs and 18 unrelated control individuals. The global gene expression profiles of unaffected siblings were more similar to those from probands as they shared genetic and environmental background. A total of 189 genes were significantly differentially expressed between proband-sib pairs (nominal p < 0.01) after controlling for age, sex, and family effects. Probands and siblings were distinguished into two groups by cluster analysis with these genes. Overall, unaffected siblings were equally distant from the centroid of probands and from that of unrelated controls with the differentially expressed genes. Interestingly, five of 20 siblings had gene expression profiles that were more similar to unrelated controls than to their matched probands. In summary, we found a set of genes that distinguished probands from the unaffected siblings, and a subgroup of unaffected siblings who were more similar to probands. The pathways that characterized probands compared to siblings using peripheral blood gene expression profiles were the up-regulation of ribosomal, spliceosomal, and mitochondrial pathways, and the down-regulation of neuroreceptor-ligand, immune response and calcium signaling pathways. Further integrative study with structural genetic variations such as de novo mutations, rare variants, and copy number variations would clarify whether these transcriptomic changes are structural or environmental in origin.

Figures

Fig. 1
Fig. 1
Differences in global gene expression profiles using squared Mahalanobis distances from the centroids of probands and unrelated controls. Overall gene expression profiles were not different between probands, siblings, and unrelated controls by principal component analysis (Supplementary Figure). We calculated a multivariate distance, i.e., squared Mahalanobis distance (MD), of each individual sample from the centroid of unrelated controls and that of probands. The density plot on the top of scatter plot show that probands and siblings are equally distant from unrelated controls. Interestingly, the distributions of MDs from the centroid of probands show significant difference (p-value 6.6×10−9). This result suggests that there seems to be subtle but significant difference in transcriptomic signature between proband and siblings.
Fig. 2
Fig. 2
Cluster analysis with the 189 differentially expressed genes. A) Unsupervised hierarchical clustering of probands and siblings with the 189 significant genes successfully clusters probands and siblings except for the one sibling who is clustered with probands. The up (red) and down (blue) regulated genes in probands compared to siblings are also grouped in the heatmap. B) Principal component analysis of 20 proband-sib pairs also reveals that two groups are well separated with the 189 genes, and no significant bias presents for gender or subtype of autism spectrum disorders.
Fig. 2
Fig. 2
Cluster analysis with the 189 differentially expressed genes. A) Unsupervised hierarchical clustering of probands and siblings with the 189 significant genes successfully clusters probands and siblings except for the one sibling who is clustered with probands. The up (red) and down (blue) regulated genes in probands compared to siblings are also grouped in the heatmap. B) Principal component analysis of 20 proband-sib pairs also reveals that two groups are well separated with the 189 genes, and no significant bias presents for gender or subtype of autism spectrum disorders.
Fig. 3
Fig. 3
Identification of proband-like and control-like siblings in gene expression patterns. A) Boxplot showing the squared Mahalanobis distances (MDs) from unaffected siblings to probands (red circles) and unrelated controls (blue circles). A pair of red and blue circles connected by line represents the distances to probands and unrelated controls for each unaffected sibling. That is, each pair of connected red and blue circles represents one sibling sample. Five control-like siblings (CLS) are equidistant to unrelated controls and matched probands (highlighted in green arrow heads and solid connecting lines), whereas fifteen proband-like siblings (PLS) are closer to matched probands (black connecting lines). B) Five CLS individuals are highlighted in green on the Fig 2B, where CLS group is on the opposite side of probands relative to PLS. C) Venn diagram of overlapping significant genes from 3 comparisons. D) Principal components analysis of probands and siblings with the differentially expressed genes from CLS vs. probands and PLS vs. probands. Ellipsoids are the minimum spanning ellipsoid for CLS and matched probands.
Fig. 3
Fig. 3
Identification of proband-like and control-like siblings in gene expression patterns. A) Boxplot showing the squared Mahalanobis distances (MDs) from unaffected siblings to probands (red circles) and unrelated controls (blue circles). A pair of red and blue circles connected by line represents the distances to probands and unrelated controls for each unaffected sibling. That is, each pair of connected red and blue circles represents one sibling sample. Five control-like siblings (CLS) are equidistant to unrelated controls and matched probands (highlighted in green arrow heads and solid connecting lines), whereas fifteen proband-like siblings (PLS) are closer to matched probands (black connecting lines). B) Five CLS individuals are highlighted in green on the Fig 2B, where CLS group is on the opposite side of probands relative to PLS. C) Venn diagram of overlapping significant genes from 3 comparisons. D) Principal components analysis of probands and siblings with the differentially expressed genes from CLS vs. probands and PLS vs. probands. Ellipsoids are the minimum spanning ellipsoid for CLS and matched probands.
Fig. 3
Fig. 3
Identification of proband-like and control-like siblings in gene expression patterns. A) Boxplot showing the squared Mahalanobis distances (MDs) from unaffected siblings to probands (red circles) and unrelated controls (blue circles). A pair of red and blue circles connected by line represents the distances to probands and unrelated controls for each unaffected sibling. That is, each pair of connected red and blue circles represents one sibling sample. Five control-like siblings (CLS) are equidistant to unrelated controls and matched probands (highlighted in green arrow heads and solid connecting lines), whereas fifteen proband-like siblings (PLS) are closer to matched probands (black connecting lines). B) Five CLS individuals are highlighted in green on the Fig 2B, where CLS group is on the opposite side of probands relative to PLS. C) Venn diagram of overlapping significant genes from 3 comparisons. D) Principal components analysis of probands and siblings with the differentially expressed genes from CLS vs. probands and PLS vs. probands. Ellipsoids are the minimum spanning ellipsoid for CLS and matched probands.
Fig. 3
Fig. 3
Identification of proband-like and control-like siblings in gene expression patterns. A) Boxplot showing the squared Mahalanobis distances (MDs) from unaffected siblings to probands (red circles) and unrelated controls (blue circles). A pair of red and blue circles connected by line represents the distances to probands and unrelated controls for each unaffected sibling. That is, each pair of connected red and blue circles represents one sibling sample. Five control-like siblings (CLS) are equidistant to unrelated controls and matched probands (highlighted in green arrow heads and solid connecting lines), whereas fifteen proband-like siblings (PLS) are closer to matched probands (black connecting lines). B) Five CLS individuals are highlighted in green on the Fig 2B, where CLS group is on the opposite side of probands relative to PLS. C) Venn diagram of overlapping significant genes from 3 comparisons. D) Principal components analysis of probands and siblings with the differentially expressed genes from CLS vs. probands and PLS vs. probands. Ellipsoids are the minimum spanning ellipsoid for CLS and matched probands.
Fig. 4
Fig. 4
Significant biological pathways between probands and siblings. A) KEGG pathways identified by GSEA for the three comparisons at nominal p-value < 0.01 (see Methods) for the comparisons of probands (prob) and siblings (sib), control-like siblings (cls) vs. matched prob, proband-like siblings (pls) vs. matched prob, and pls vs. cls. The pathways that are significant in any of 4 comparisons are listed and red (up-regulated in prob or pls) and blue (down-regulated) boxes represent significantly changes. Ribosomal genes were consistently up-regulated in all 4 comparisons (see Results). B) The Venn diagram shows the overlap of significant pathways listed in Fig 4A.
Fig. 4
Fig. 4
Significant biological pathways between probands and siblings. A) KEGG pathways identified by GSEA for the three comparisons at nominal p-value < 0.01 (see Methods) for the comparisons of probands (prob) and siblings (sib), control-like siblings (cls) vs. matched prob, proband-like siblings (pls) vs. matched prob, and pls vs. cls. The pathways that are significant in any of 4 comparisons are listed and red (up-regulated in prob or pls) and blue (down-regulated) boxes represent significantly changes. Ribosomal genes were consistently up-regulated in all 4 comparisons (see Results). B) The Venn diagram shows the overlap of significant pathways listed in Fig 4A.

Similar articles

See all similar articles

Cited by 16 articles

See all "Cited by" articles

Publication types

LinkOut - more resources

Feedback