Myelination selectively modulates BOLD signal in white matter

There is increasing recognition that blood oxygenation level dependent (BOLD) signals are detectable in white matter (WM) and reflect an important, heretofore overlooked functional activity in the brain, but their biophysical origins remain understudied and poorly understood. By integrating several disparate, multimodal data sets, we established the associations of resting state BOLD signals with key microstructural, hemodynamic and metabolic features in WM. In particular, we identified the roles of myelination and fiber type in modulating BOLD effects, and derived relationships between measurements of BOLD signal power and cerebral blood volume, flow, oxygen extraction and metabolic rate of oxygen consumption, which are predicted using a simple theory and then verified empirically. Our findings demonstrate that myelin selectively influences the fractional amplitude of low-frequency fluctuations (fALFF) in BOLD signals, and that differences in myelin content account for variations in their temporal spectra and hemodynamic response functions, but these in turn are qualitatively different in association versus projection fibers. Other determinants of BOLD in WM are further revealed by converging biological, genomic and neurochemical evidence, including measurements of neurite and mitochondrial densities. Moreover, analyses of images of the optic nerve from human subjects confirm that BOLD activations evoked by visual stimuli are preferentially localized to unmyelinated portions, with minimal responses in fully myelinated regions of the same nerve, suggesting a myelin-dependent requirement for BOLD effects in WM.