The ECM is crucial for the culture of brain microcapillary ECs and the establishment of tight junctions . In most in-vitro BBB models the surface of the transwell inserts is coated with rat tail type I collagen, to support and promote EC monolayer formation [205-207]. However, other ECM molecules, such as type IV collagen, fibronectin and laminin, either alone or in combination, were also shown significantly to elevate the TEER of low-resistance porcine brain endothelial monolayers . Unfortunately, the precise profile of integrin receptors in microvascular ECs of different brain regions and the composition of the ECM in these specific regions has not yet been reported. Therefore, is not surprising that in the available in-vitro BBB models ECM proteins coating the transwell inserts have been used according to availability, rather than based on physiological criteria. In order to generate refined, more realistic in-vitro BBB pharmacological models, there is an urgent need for a detailed analysis of the (region-specific?) composition of the brain capillary ECM and for translating this new knowledge into the technology of engineering BBB models. Alternatively, the emergence of new generations of synthetic biomaterials for use as 3D extracellular microenvironments, which closely mimic natural ECMs , may result in novel, synthetic, brain-biocompatible, ECM-like substrates that might help to tighten in-vitro BBB models. For example, hyaluronic acid hydrogels modified with laminin , procyanidolic oligomers  and similar compounds may be used for decreasing the permeability and increasing the TEER values of in-vitro BBB models. Recently, the first attempt to generate an in-vitro BBB chip was made by co-culturing for > 14 days bovine brain ECs and rat astrocytes on ultrathin, highly porous nanofabricated silicon nitride membranes. These membranes were about 10-fold thinner and twofold more porous than commercial membrane inserts, and also included a spun-on crosslinked collagen layer, which helped to improve astrocyte attachment . However, due either to the lack of primary isolates of the cells or to the "wrong" type of collagen coating, these ultrathin constructs still yielded TEER values orders of magnitude lower than in vivo. Nevertheless, we strongly believe that such synthetic biomaterials will be indispensable as biomimetic matrices, providing appropriate instructive microenvironmental cues for tissue engineering of future in-vitro BBB models.
Today, the pharmaceutical industry is seeking novel, high-fidelity in-vitro BBB pharmacological models which may be used as a first preclinical step in screening and characterizing the BBB-permeability properties ofnovel candidate compounds. Ideally, as a validation step, any high-fidelity in-vitro BBB model will demonstrate a close correlation of the permeability of known drugs with their known brain extraction pharmacokinetic properties in vivo.
The first in-vitro BBB model in which transcellular permeability was measured by the movement of 14C-labeled sucrose was established from bovine brain gray matter . The bovine brain contains large amounts of microcapillary ECs which should, in theory, suffice for a large-scale, high-throughput pharmacological screening .
One of the best in-vitro BBB models, a co-culture of cloned early-passaged bovine capillary ECs and rat glia, demonstrated TEER values of500-800 Q X cm-and low sucrose/mannitol permeability . The highest experimental TEER value (~2000 Q X cm- ), close to the physiological values measured in situ, was obtained in a bovine system; however, no concomitant studies on the monolayer permeability were performed .
Mouse brains yield low numbers of microcapillary ECs, and consequently few murine in-vitro BBB models have been reported . The availability of an in-vitro mouse BBB model would provide a unique opportunity for using transgenic and knockout brain capillary ECs (as well as glia and neurons) to investigate the contribution of different cellular proteins to TEER and permeability. Permeability measurements in rat brain-derived in-vitro BBB models have also been published . Similarly to mice, the rat brains yield limited numbers of ECs. According to published reports, the barrier properties of rat brain-derived BBB models are much better than those of mouse origin, but are still lower than some of the near-physiological levels obtained with bovine and porcine models.
To date, the best in-vitro BBB models, with high barrier resistance (18002000 Q X cm-2) and low sucrose permeability (Papp = 0.2-1.8 X 10-6 cm s-1), utilize porcine brain microcapillary ECs grown in serum-free culture conditions and treated with corticosteroids [216, 217]. The large amounts of EC that can be obtained from a porcine brain, and the close similarity between porcine and human cardiovascular physiology, suggest that, in the absence of a readily available human in-vitro BBB model, the porcine system might represent a reasonable in-vitro model for high-throughput drug screening.
In spite of the limited access to human brain tissues, a few human in-vitro BBB models have been established . These models were less robust then porcine or bovine BBB models, and provided relatively low barrier electrical resistance and high permeability values; the highest reported TEER value was ~500 Q, X cm-2 and the lowest Papp value for inulin 3.6 X 10-6 cm s-1 . However, recent biotechnological developments have allowed the isolation and culture of stem cells from human brain, bone marrow or umbilical cord blood, which then can be differentiated towards endothelial and, possibly also astrocytic and neuronal phenotypes, thereby raising the expectation that these cells will provide better cellular partners for engineering a human in-vitro BBB model for pharmaceutical purposes.
Whilst all of the above-mentioned biological and pharmaceutical in-vitro BBB models are cultured under static conditions, it must be remembered that in vivo the BBB, like every other endothelium, is exposed to blood flow and hydrostatic pressure. These "mechanical" parameters significantly influence endothelium physiology. Any truly high-fidelity in-vitro pharmacological model of the BBB must therefore incorporate these hemodynamic forces, both physiological as well pathological, such as hypertension. As a first dynamic model, Janigro and co-workers recently incorporated fluid shear stress into their in-vitro BBB model [219, 220], which is based on the principle of hollow-fiber bioreactors. Here, an endothelial cell monolayer, grown on the inside of a porous hollow fiber, is continually exposed to fluid-shear stress and interacts with astrocytes seeded onto the outside of the fibers. Thus, this model, for the first time, approximates the tubular morphology and hemodynamic forces of a brain capillary forming the BBB in vivo. Importantly, the TEER values in this system, though still lower than in situ, were increased by fluid shear stress. However, while this approach demonstrated the need for a dynamic concept, the permeability of drugs in this model has not yet been investigated and, in its present design, the model is ill-suited to high-throughput drug screening.
Was this article helpful?