Quantitative MRI maps of human neocortex explored using cell type-specific gene expression analysis

Abstract

Quantitative magnetic resonance imaging (qMRI) allows extraction of reproducible and robust parameter maps. However, the connection to underlying biological substrates remains murky, especially in the complex, densely packed cortex. We investigated associations in human neocortex between qMRI parameters and neocortical cell types by comparing the spatial distribution of the qMRI parameters longitudinal relaxation rate (${R_{1}}$), effective transverse relaxation rate (${R_{2}}^{\ast }$), and magnetization transfer saturation (MTsat) to gene expression from the Allen Human Brain Atlas, then combining this with lists of genes enriched in specific cell types found in the human brain. As qMRI parameters are magnetic field strength-dependent, the analysis was performed on MRI data at 3T and 7T. All qMRI parameters significantly covaried with genes enriched in GABA- and glutamatergic neurons, i.e. they were associated with cytoarchitecture. The qMRI parameters also significantly covaried with the distribution of genes enriched in astrocytes (${R_{2}}^{\ast }$ at 3T, ${R_{1}}$ at 7T), endothelial cells (${R_{1}}$ and MTsat at 3T), microglia (${R_{1}}$ and MTsat at 3T, ${R_{1}}$ at 7T), and oligodendrocytes and oligodendrocyte precursor cells (${R_{1}}$ at 7T). These results advance the potential use of qMRI parameters as biomarkers for specific cell types.

Keywords: hMRI; isocortex; magnetic resonance imaging; myelin; relaxometry.

Fig. 1

The left hemisphere spatial distribution of the qMRI parameters at each magnetic field strength (top row) and of the respective first and second PLS components (bottom two rows) projected on the inflated FreeSurfer fsaverage brain. The qMRI parameter plots show the mean over vertices and subjects in each area of the HCP-MMP1.0 atlas (units: MTsat; ; ). The PLS component plots show the score-vectors (Rosipal et al. 2006) of the matrix for each qMRI parameter, giving a visual representation of the latent PLS variables (in arbitrary units). PLS components are only plotted when they explain of the spatial variance of a qMRI parameter (Table 1). In each case, top: lateral view, bottom: medial view. A: anterior, P: posterior, I: inferior, S: superior. The regions marked in grey represent areas with no data, i.e. non-cortical tissue (mostly corpus callosum), regions without robust cortical samples in the AHBA, and the potentially artefact affected areas in the 7T case. Lower and upper limits of the colour maps in each plot are the 5th and 95th percentiles of the data, respectively. Colours from http://colorbrewer.org by Cynthia A. Brewer, Geography, Pennsylvania State University via https://github.com/DrosteEffect/BrewerMap.

Fig. 2

Summary of the significant, replicated associations found between cell type-specific gene expression in the genes associated with each qMRI parameter. Replications at the level of the top 5% of genes associated with each qMRI parameter (robust associations) are shown in black, with replications at lower levels in shades of grey. Non-significant (n.s.) and non-replicating (n.r.) associations are in white.

Fig. 3

EWCE results showing the cell type associations of the top 5% of genes associated with at 3T and 7T (first PLS component only). Plotted are the number of standard deviations (stds) by which the EWCE value deviated from the mean value over bootstrapped target lists. Results from the two cell type-specific datasets are plotted in different colors: SMART-seq in black, DroNc-seq in grey. Top: 3T. Bottom: 7T. Bars are only plotted when FDR-corrected . *: FDR-corrected . Significant cell-type associations which replicated between both cell type-specific datasets (robust results) are underlined and in bold.

Fig. 4

EWCE results showing the cell-type associations of the top 5% of genes associated with MTsat at 3T. Plotted are the number of standard deviations (stds) by which the EWCE value deviated from the mean value over bootstrapped target lists. Results from the two cell type-specific datasets are plotted in different shades: SMART-seq in black, DroNc-seq in grey. Left: First component of the PLS. Right: Second component of the PLS. Bars are only plotted when FDR-corrected . *: FDR-corrected P. Significant cell-type associations that replicated between both cell type-specific datasets (robust results) are underlined and in bold.

Fig. 5

EWCE results showing the cell-type associations of the top 5% of genes associated with at 3T and 7T. Plotted are the number of standard deviations (stds) by which the EWCE value deviated from the mean value over bootstrapped target lists. Results from the two cell type-specific datasets are plotted in different shades: SMART-seq in black, DroNc-seq in grey. Top: 3T. Bottom: 7T. Left: First component of the PLS. Right: Second component of the PLS. Bars are only plotted when FDR-corrected . *: FDR-corrected . Significant cell-type associations which replicated between both cell type-specific datasets (robust results) are underlined and in bold.