The Brain Vasculome. What Is It and What Does It Do?

What is the Brain Vasculome?

The brain's blood vessels are special, and different from those in other tissues. For instance, their external surface is surrounded by a patchwork coating of cellular processes termed end-feet, formed by astroglial cells. Further, brain endothelial cells express a specialised transcriptomic profile, including tight junction proteins (notably, claudin-5) and transcellular transporters, to form the blood-brain barrier. Brain vessels return fluid and solutes to the circulation via perivascular or intramural clearance pathways (as the brain lacks lymphatic vessels). And they participate in the blood flow autoregulation mechanisms that ensure an adequate supply of blood to active neurones. No surprise then, that the functionally-specialised cells residing in and around a cerebral blood vessel – endothelia, myocytes, myofibroblasts, pericytes (on the capillaries), astrocytes and perivascular macrophages – have characteristic profiles of gene expression. The transcriptome and proteome that reflects this brain vascular specialism among different cell types, is collectively termed the brain vasculome^{1, 2}, or "Neurovasculome". A recent AHA/ASA Scientific Statement reviews this concept and discusses the neurovasculome in relation to brain health and cognitive impairment (Iadecola et al. in press).

Brain cell atlases

This exercise is enabled by the emergence of state-of-the-art technologies to catalogue at high speed the RNAs present in single cells or single nuclei. Thus, a tissue sample the size of a pea, after biochemical digestion to form a cell suspension, can yield an enormous list of the RNAs present in each cell (typically, 5,000-10,000 well-defined RNAs). Identifying an established pattern of gene expression allows each cell to be assigned to a cell-type (myocyte, astrocyte, neuron, etc). For example, a cell containing mRNAs for α-actin and smooth muscle myosin, and lacking those for endothelial markers such as CD31/PECAM1, is defined as a vascular myocyte. This approach has generated so-called brain cell atlases, mapping the transcriptomic profile of each cell-type^3-8. These shed new light on brain vascular function. They can also be used to catalogue transcriptional changes in disease states, including AD^{3, 4, 7}, Huntington's disease⁶ and vascular malformations⁵.

The Neurovascular Unit (NVU)

The concept of the NVU arose from the recognition that local interactions between brain cells (neurons, glia) and vascular components are central to brain function and disease^{9, 10}. The notion of a NVU helpfully demolishes the barrier separating traditional disciplines of neuroscience and vascular biology, and no less helpfully that between clinical specialities of Neurology and Stroke Medicine. The principal functions of the NVU are: formation of the blood-brain barrier; trafficking fluid, via perivascular and intra-mural clearance pathways; regulation of local cerebral blood flow; immune cell access from the blood (and back into it); angiogenesis and microvascular remodelling. Gene expression profiles across the neurovasculome reflect these functions of the NVU¹⁰.

Regional specialization

Single cell RNA profiling has made us aware of subpopulations of vascular cells along the neurovascular tree^{5, 6, 8}. As a simple example, Iadecola and colleagues noted from a mouse brain atlas¹¹ that arterial endothelial cells preferentially express genes for cellular plasticity (e.g. Igfbp4), whereas capillary endothelia feature genes for trans-endothelial transport and O₂-response (Mfsd2a, encoding a lysolipid transporter; Fmo2) and venous endothelia, inflammation-related genes (Cfh, encoding complement factor H)¹¹. Iadecola and colleagues highlight that advances in brain proteomics, epigenomics, lipidomics and metabolomics are likely to elucidate the functional roles of the multiple vascular cell subtypes identified by single cell RNA profiling. There will be variations with brain region, in particular grey matter versus white matter differences, with age and with disease state. These may be revealed by spatially resolved ‘omic methods such as imaging mass cytometry (a form of single cell proteomics)¹².

Of Mice and Men

A large fraction of the knowledge base regarding the NVU and neurovasculome comes from studies in mice. As Iadecola et al. make clear, the neurovasculome of M. musculus differs substantially from that of H. sapiens. This is apparent in single cell RNA profiling of samples from mouse brain and human brain^4-6. For example, microglial differentially-expressed genes identified in human AD (P2Y12, TMEM119) differed substantially from the microglial profile of a much-used murine model of AD (5XFAD transgenic strain)⁴. This should come as no surprise, considering the anatomical and physiological differences between these two species. An adult mouse is nocturnal, quadrupedal, with chronological age between 12 weeks and 3 years and a resting pulse rate over 600 s^-1. Its brain fits comfortably on a fingernail, lacks the folds that characterise a higher mammal's brain and contains radically different white matter conformation and content (about one tenth by volume, whereas humans have almost 50%)¹³. Zooming in to the NVU, murine penetrating arteries lack an adventitial coating of collagen-IV-rich basement membrane, and may also lack a perivascular space, both of which are notable features of the human NVU. Thus, the cellular signalling cues received by a particular cell-type in mouse brain (a vascular myocyte, for example) will differ fundamentally from the equivalent cell-type in human brain. Hence it is to be anticipated that their transcriptional responses also differ.

The Neurovasculome in Human Stroke and Dementia

Failure of NVU function can lead to brain disease¹⁰. Iadecola et al refer to a recent Lancet Commission¹⁴, which identified twelve modifiable risk factors for potential prevention of dementia. Of these, nine are cardiovascular (notably hypertension, diabetes mellitus, smoking, obesity, hypercholesterolemia). Though this does not predicate a direct causal link to changes in the neurovasculome, such changes might be predicted and they are indeed identified. In AD brains, cell type-specific changes in gene expression include increased APOE in microglia^{3, 4} and, in oligodendrocytes, reduced expression of genes controlling axon guidance (SEMA3B) and maturation of myelin-forming cells (MIR219A2)^{3, 4}. Among vascular myocytes, the AD-related gene APOE was zonated to venous, rather than arterial, myocytes⁶. Within human brain vascular cell populations, inflammatory cytokine signalling (including the IL1 receptor, IL1R1) was preferentially zonated to cells of the venous compartment (surprisingly, not arteries or capillaries)⁷. Transcriptional changes related to brain vascular abnormalities (arterio-venous malformations) included enhanced endothelial expression of well-known inflammatory pathways (including VEGF and TGF-beta signalling) as well as novel ones (increased SPP1, encoding osteopontin, and CD99)⁵. A sub-population of monocytes, expressing the marker GPNMB, were identified as being related to myocyte depletion and increased likelihood of haemorrhagic stroke⁵. These findings are fertile ground for generating or refining mechanistic hypotheses. The single cell expression data that are emerging will not only widen our knowledge of the molecular phenotype of cells within the NVU, but also highlight how they change in disease. That in turn has potential to uncover much-needed pharmacological targets in stroke and dementia.

Conclusion

Many years of careful research at the interface between blood and brain are now being extended by novel high-throughput platforms for single cell transcriptomic and proteomic screening. We are witnessing an explosion of molecular data, relevant to physiological brain function as well as to pathophysiological mechanisms. There is a cynical true-ism that "genomics has contributed nothing to CNS drug discovery". Rapidly accumulating novel data on the neurovasculome may allow us to change that.

I am grateful for helpful comments from Stuart M Allan, Hilda MTC Mulrooney and Daniel N Meijles.