The blood-brain barrier: an engineering perspective

Abstract

It has been more than 100 years since Paul Ehrlich reported that various water-soluble dyes injected into the circulation did not enter the brain. Since Ehrlich's first experiments, only a small number of molecules, such as alcohol and caffeine have been found to cross the blood-brain barrier, and this selective permeability remains the major roadblock to treatment of many central nervous system diseases. At the same time, many central nervous system diseases are associated with disruption of the blood-brain barrier that can lead to changes in permeability, modulation of immune cell transport, and trafficking of pathogens into the brain. Therefore, advances in our understanding of the structure and function of the blood-brain barrier are key to developing effective treatments for a wide range of central nervous system diseases. Over the past 10 years it has become recognized that the blood-brain barrier is a complex, dynamic system that involves biomechanical and biochemical signaling between the vascular system and the brain. Here we reconstruct the structure, function, and transport properties of the blood-brain barrier from an engineering perspective. New insight into the physics of the blood-brain barrier could ultimately lead to clinical advances in the treatment of central nervous system diseases.

Keywords: blood-brain barrier; capillary; microvasculature; neurovascular unit; transport.

Figure 1

The blood-brain barrier. (A) Schematic illustration of transport across the brain microvascular endothelial cells (BMECs) that form the lumen of brain capillaries. Paracellular transport is severely restricted due to the formation of tight junctions between endothelial cells. Metabolic nutrients and other essential molecules are transported across the luminal and abluminal membranes by channels, pumps, or mediated transport systems. Small lipophilic molecules can passively diffuse across the lipid bilayer but, in many cases, are returned to the blood by efflux pumps. (B) Proposed molecular interactions at tight junctions. Lateral association of claudins (cis-interaction) results in the formation of oligomers whereas association of claudins on opposing membranes (trans-interaction) results in tight junction formation. Multiple regions of trans-interactions appear as particles in electron microscopy images.

Figure 2

The neurovascular unit. The microvascular endothelial cells that form the lumen of brain capillaries are partially covered by pericytes and basement membrane, and almost completely surrounded by the end feet of astrocytes. Functional interactions between BMECs, astrocytes, pericytes, other glial cells, and neurons are key to regulating brain homeostasis. Blood flow is associated both biomechanical and biochemical signaling mediated by multiple cell types and soluble factors. The brain microvascular endothelial cells function in a cylindrical geometry with high curvature and experience shear stress resulting from blood flow. The BMECs and pericytes are surrounded by basement membrane consisting primarily of fibronectin, laminin 1, and collagen type IV. The extra-cellular matrix in the brain is based on hyaluronic acid.

Figure 3

Transport systems at the blood-brain barrier. (1) Small ions and water molecules can cross the blood-brain barrier through ion channels. (2) Small lipophilic molecules that are soluble in the hydrophobic core of the cell membrane can be transported passively across the cell. (3) Essential polar molecules that cannot diffuse through the cell membrane are shuttled across the cell membranes by carrier-mediated transport. These solute carriers may be directional, in or out of the cell, or bidirectional. Other molecules can be actively transported across endothelial cell membranes by carrier-mediated transporters, receptor-mediated transporters, adsorption-mediated transcytosis, or efflux pumps.

Figure 4

Schematic illustration of (a) in vitro and (b) in vivo transport measurements. (A) In the 2D transwell assay, a monolayer of cells is formed on a porous membrane separating two compartments. Astrocytes and/or pericytes may be seed on the opposite side of the membrane or in the output chamber. (B) In vivo studies, a solute is injected into the blood of an animal model, and the penetration into the brain measured using a suitable chemical detection assay or imaging technique.

Figure 5

Schematic illustration of analysis of diffusion transport in 2D and 3D. In 2D: c_in is the solute concentration on the input side in a transwell assay, A is the area of the cell monolayer (cm²), N is the number of moles of solute measured on the output side in volume V (cm³), and k_{in, 2D} is the 2D rate constant (cm³ s⁻¹). As long as c_in = constant then k_{in, 2D} = P_2DA. In 3D: c_pl is the solute concentration in plasma (g_s cm⁻³), S is the normalized surface area of the lumen (cm² g_br⁻¹), and F is the normalized flow rate (cm³ g_br⁻¹ s⁻¹), Q_br is the amount of solute transported to the brain (g_s g_br⁻¹), and k_{in, 3D} is the 3D rate constant (cm s⁻¹ g_br⁻¹).

Figure 6

Kinetics of solute transport across a 2D monolayer. c_out/c_in is plotted as a function of time t, with A = 1 cm², and V = 1 cm³ for (■) P_2D = 10⁻⁵ cm s⁻¹, and (▲) 10⁻⁶ cm s⁻¹, and (•) P_2D = 10⁻⁷ cm s⁻¹, At short times (inset) where P_2DAt/V « 1, the slope is P_2DA/V and the rate constant can be obtained from k_in,2D = P_2DA.

Figure 7

(A) Schematic illustration of a resected brain capillary defined by a layer of endothelial cells immersed in a bath. k_m is the rate constant for passive diffusion across a cell membrane. It is implicitly assumed that passive diffusion across the apical and luminal membranes is the same. k_pgp is the rate constant for active transport via efflux pumps from the cell to the lumen. It is assumed that active transport at the apical membrane is negligible. k_in and k_out represent the overall rate constants for transport from bath to lumen and lumen to bath, respectively. (B) Accumulation of solute in the lumen for a resected capillary with diameter d = 5 μm, P_3D = 3 × 10⁻⁷ cm s⁻¹, k_pgp/k_m = 0, 5, 10, 20.

Figure 8

(A) Permeability of tracers, nutrients, and drugs obtained from in situ rat brain perfusion vs. lipophilicity. (•) Summerfield et al. (2007), (▲) Takasato et al. (1984), (▼) Youdim et al. (2004), (■) Liu et al. (2004) (B) Comparison of permeability of various CNS drugs obtained from transwell assays on monolayers of MDR1-MDCK (P_2D) and in situ rat brain perfusions (P_3D). P_3D values were obtained from in situ rat brain perfusion measurements reported in the literature. For data reported as the permeability surface area products (P_3DS, cm³ s⁻¹ g_br⁻¹) we take S = 150 cm² g_br⁻¹. Values of P_3D where S ≠ 150 cm² g_br⁻¹ were recalculated with S = 150 cm² g_br⁻¹. Corresponding literature values for logP_oct were obtained from calculation Liu et al. (2004), Summerfield et al. (2007), and Youdim et al. (2004) or direct measurement of solute partitioning into octanol and water phases [Takasato et al. (1984)].

Figure 9

Schematic illustration of solute transport from the vascular system into the brain. The solute may bind with proteins or other components in blood that may reduce the amount that can enter the brain. Solute transported across the endothelium may be partitioned between the interstitial fluid and intracellular fluid in neurons and glial cells. Solute in the interstitial fluid may be bound to the ECM, reducing the amount available for uptake by cells.