Research ArticleNEUROSCIENCE

Dopaminergic organization of striatum is linked to cortical activity and brain expression of genes associated with psychiatric illness

See allHide authors and affiliations

Science Advances  09 Jun 2021:
Vol. 7, no. 24, eabg1512
DOI: 10.1126/sciadv.abg1512
  • Fig. 1 Overview of methods.

    (A) CCA is used to investigate the relationship between D2/3 receptor availability and CBF. The first step involves principal components analysis (PCA) to reduce the dimensionality of the data. CCA is then performed on these PCA matrices (number of rows = number of scans; number of columns = number of PCA-derived components). CCA calculates the canonical weight vectors U and V that maximize the correlation between canonical variates PU and CV. Each point in the scatter graph represents a single scan. (B) A striatal covariation influence map is calculated by correlating the canonical variate PU at each voxel of the maps of striatal D2/3 receptor availability. Gene expression data from the Allen Human Brain Atlas (AHBA) is normalized to Montreal Neurological Institute (MNI) space. Partial least squares (PLS) analysis is then used on voxels for which gene expression data exist to determine which genes show patterns of gene expression that most closely track the map of covariation influence.

  • Fig. 2 Striatal D2/D3 receptor availability and CBF measured with simultaneous PET-MRI.

    (A) Mean striatal BPND for placebo session scans. Before averaging across individuals, each individual participant scan was corrected for its mean value. (B) Mean CBF for placebo session scans. Before averaging across individuals, each individual participant scan was corrected for its mean value. (C) Change in mean value–corrected BPND, Z scores represent results of a paired t test (amphetamine > placebo), and positive values (red) indicate greater relative BPND during the amphetamine scan. (D) Change in mean value–corrected CBF, Z scores represent results of a paired t test (amphetamine > placebo), and positive values (red) indicate greater relative CBF during the amphetamine scan.

  • Fig. 3 CCA identifies associations between striatal dopamine and cortical blood flow.

    (A) A scatterplot illustrating the mode of covariation between cortical blood flow and striatal D2/3 receptor availability, where each point represents an individual scan, and CCA has been performed using the entire sample. (B) The out-of-sample correlation predicted by the mode of covariation is statistically significant. The orange coloring represents results, where scan pairs were not allowed to split over train test partitions of the data, and the vertical line represents the observed correlation coefficient compared to a null distribution generated from 10,000 random subject-level permutations (rp = 0.41, P < 0.001). The blue line represents the case in which permutations of the data used to calculate a null distribution maintained the linked dependency structure resulting from repeated scans (rp = 0.51, P = 0.002). (C) Striatal voxels were colored according to their association, with the mode of covariation illustrated in (A), red indicates greater D2/3 receptor availability, and blue indicates lower D2/3 receptor availability, for high-scoring subjects (and vice versa for low-scoring subjects). (D) Cortical voxels were colored according to their association with the mode of covariation illustrated in (A); red indicates greater blood flow, and blue indicates lower blood flow, for high-scoring subjects (and vice versa for low-scoring subjects). An individual who had a pattern of cortical blood flow similar to (D) would be expected to show a pattern of striatal D2/3 receptor availability similar to (C). (E) The change in striatal D2/3 CCA scores following amphetamine administration is associated with a change in CBF CCA scores (rs = 0.49, P = 0.01). (F) Average CCA scores (mean of striatal and cortical score) are not associated with the scores on the subjective effects of amphetamine scale in the sample as a whole (rs = 0.26, P = 0.07) and when restricted to the amphetamine sessions (rs = 0.34, P = 0.11).

  • Fig. 4 Gene expression predicts the influence of striatal tissue samples on the corticostriatal mode of covariation.

    (A) Partial least squares analysis identifies a pattern of gene expression associated with striatal D2/3 covariation influence (see Fig. 2C); each point represents a striatal voxel, with predicted striatal D2/3 covariation influence plotted against actual influence. (B) Average predicted out-of-sample striatal D2/3 covariation influence across 10 separate fivefold cross-validated partial least squares models compared to observed covariation influence; each point represents a striatal voxel. (C) The out-of-sample correlation between observed striatal D2/3 covariation influence and that predicted by partial least squares analysis of gene expression is represented by the vertical line. This is statistically significant compared to null models generated by 10,000 random permutations of striatal voxels (P < 0.001) and permutations where spatial autocorrelation is preserved (P = 0.044). (D) Out-of-sample prediction accuracy is greatest when using six partial least squares components, and this model is used in subsequent analyses. (E) Heatmap illustrating spatial patterns of gene expression for the 200 genes most strongly positively and negatively associated with striatal D2/3 covariation influence (y axis) against striatal voxels vectorized and arranged from left to right in order of covariation influence (x axis). Values smoothed with a 3-mm gaussian kernel in the x direction and clustered so as to visualize the positive-negative pattern of gene expression.

  • Fig. 5 Genes associated with the corticostriatal mode of covariation are enriched for genes associated with synaptic proteins and with schizophrenia.

    (A) Enriched Gene Ontology terms associated with genes linked to the corticostriatal mode of covariation. Only terms with FDR-corrected significance at P < 0.05 and REViGO dispensability < 0.3 are shown. BP, biological process; CC, cellular component; MF, molecular function. (B) Figure on left shows that genes up-regulated in schizophrenia (P < 0.001), bipolar disorder (P = 0.004), and autism spectrum disorder (P < 0.001) are enriched among the genes associated with striatal D2/3 covariation influence. Figure on right shows that genes down-regulated in bipolar (P = 0.005) are enriched among the genes associated with striatal D2/3 covariation influence. Vertical lines indicate true median rank of disorder-related genes compared to null distributions representing 10,000 randomly selected gene sets. *P < 0.05.

Supplementary Materials

  • Supplementary Materials

    Dopaminergic organization of striatum is linked to cortical activity and brain expression of genes associated with psychiatric illness

    Robert A. McCutcheon, Kirsten Brown, Matthew M. Nour, Stephen M. Smith, Mattia Veronese, Fernando Zelaya, Martin Osugo, Sameer Jauhar, William Hallett, Mitul M. Mehta, Oliver D. Howes

    Download Supplement

    The PDF file includes:

    • Figs. S1 to S3
    • Tables S1 and S3
    • Legend for Table S2
    • References

    Other Supplementary Material for this manuscript includes the following:

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article