The relative impermeability to molecules that do not freely diffuse across membranes is perhaps the most widely recognized feature of the BBB, and is essential for maintaining brain homeostasis. Disruption of the BBB, which we define here as loss of impermeability to molecules due to leakage, can be induced by a number of immune factors including cytokines and chemokines [
15,
34], microbes and their components [
35‐
37], complement proteins [
38], acute phase proteins [
39], etc. A few methods are available to evaluate BBB disruption in vitro, which we discuss in-depth in this section. Further, we provide some examples in the literature on use of in vitro models to evaluate BBB disruption in response to immune factors.
Mechanisms and measurement of BBB disruption in vitro
Just as BBB impermeability depends on acquisition of intercellular tight junctions, loss of fenestrae, and decreased pinocytosis, so in theory, BBB disruption can occur by paracellular and transcellular/transcytotic pathways. BBB disruption is generally considered to be a pathological process that results in increased paracellular and/or transcellular diffusion of substances due to reduced functions of TJPs and/or increased vesicular mechanisms, respectively [
1]. However, some workers in the field feel that there may be a physiological variation in tightness of the barrier. It has been shown recently that there are at least regional variations in BBB features among brain microvascular segments, with small brain arterioles having an abundance of caveolar vesicles that are virtually absent in brain capillaries [
40]. These arteriolar vesicles were shown to mediate neurovascular coupling, and thus have physiological functions. In some instances of more severe injuries, BBB disruption also occurs due to degeneration and loss of brain endothelial cells, resulting in cerebral microbleeds [
1].
In conventional 2D culture systems where endothelial cells are grown on transwells, disruption can be measured by evaluating TEER, as well as the permeability of inert tracers across the monolayer in the blood-to-brain direction. TEER evaluates the electrical resistance of the monolayer, which requires a patent channel between the luminal and abluminal transwell chambers. It is mainly influenced by tight junction protein integrity and can be measured using a voltohmmeter, or through impedance spectroscopy [
41]. Voltohmeters that are typically used to measure TEER (e.g. the EVOM2) apply an alternating current of ± 10 μA at a single frequency of 12.5 Hz. Specialized electrodes are used that simultaneously apply current and measure voltage and come in two formats: the STX2 or “chopstick” electrode pair and the EndOhm meters. The STX2 is manually positioned with one electrode in the upper and one electrode in the lower chamber for TEER measurement. TEER readings with STX2 electrodes depend on the electrode positions, and the uniformity of the current density can have a substantial effect on TEER measurements—overestimation of TEER may occur as a result of uneven current density when larger transwell inserts are used [
41]. The STX electrodes must also be carefully handled to ensure consistent placement, and no cell damage during measurement. EndOhm chambers serve as an alternative to STX electrodes—these chambers have electrodes with fixed geometries, and the transwell is placed into the chamber for TEER measurement, resulting in more uniform current density across the membrane. As a result, there is lower variation in measurement of a given sample in the EndOhm chamber when compared with STX2 electrodes [
41]. In general, measurement of TEER using voltohmmeter requires performing measurements outside of the cell incubator, and so fluctuations in medium temperature may contribute to variability of the measurements if the plates are not equilibrated. Impedance measurement systems use a different approach for measuring TEER, and can also evaluate properties like cell membrane capacitance which is an indicator of cell attachment and growth. In these systems, AC voltage is applied with a frequency sweep and the amplitude and phase of the resulting AC current is measured. Electrical impedance is then calculated as the ratio of the voltage–time function and the current–time function [
41,
42]. Like resistance, impedance is expressed in ohms and can be normalized to the growth surface area. An additional benefit of commercially available impedance measurement systems is that they offer real-time measurements without the need to remove the cells from the incubator.
TEER is conventionally expressed as the resistance in ohms (Ω) multiplied by the area of the growth surface in cm
2, and the TEER of a cell-free transwell in cell culture medium is subtracted to calculate the TEER of the cell monolayer [
41]. In vivo, the BBB has been estimated to have very high TEER of around 8000 Ω*cm
2 based on studies that evaluated ion flux from blood to brain in situ [
43]. Conventional models of the BBB typically do not achieve TEER values this high, however it has been found using bovine brain endothelial cells that when TEER exceeds about 131 Ω*cm
2, subsequent increases in TEER result in such minor decreases in permeability of small inert tracers such as sodium fluorescein (376.3 Daltons) that the relation becomes flat (at a permeability threshold of about 0.5 × 10
−6 cm/s; more details on permeability measurements provided below) [
33]. A more recent study using porcine brain cerebral endothelial cells found an inflection point where sucrose (342.3 Daltons) permeability stabilized at a similar level of about 0.5 × 10
−6 cm/s, but at a higher TEER threshold of about 600 Ω*cm
2 [
44]. In iPSC-derived brain endothelial cells, a higher TEER threshold of around 500 Ω*cm
2 is needed before the TEER-permeability relation flattens for sodium fluorescein, with the permeability measure stabilizing around 0.5 × 10
−6 cm/s [
45]. The reason for these differences in models is presently unclear, but highlights that similarly designed studies may need to be done as quality controls to evaluate model-specific thresholds of permeability and TEER that indicate sufficient barrier properties for the studies being performed. Conventional 2D BBB models using isolated primary rodent brain endothelial cells can typically achieve TEERs of 100–300 Ω*cm
2, but TEER values of up to 800 Ω*cm
2 have been reported under optimized conditions in monoculture (e.g. in the presence of hydrocortisone) and when co-cultured with other cell types of the neurovascular unit such as astrocytes or pericytes or their conditioned medium [
25,
46]. Primary cultures of brain endothelial cells from larger animals such as cows or pigs typically have TEERs of around 400–1000 Ω*cm
2 or higher [
25], but apparent species differences may reflect technical factors such as BEC yield or isolation and culture conditions rather than differences in physical tightness of the barrier in vivo.
Although a low TEER indicates a connection between the luminal and abluminal chambers, it does not provide information on the diameter of that connection [
25]. Assessment of size-selective BBB leakage in vitro can be done by measuring permeability of inert, hydrophilic tracers that are fluorescent or radiolabeled. Generally, this is done by applying tracers of varying molecular weights to the luminal chamber, and then measuring the amount of tracers that reach the abluminal side over time. The flux of substances across in vitro BBB monolayers is conventionally expressed as the permeability-surface area coefficient of the endothelial monolayer (abbreviated as Pe), expressed in units of cm/min or cm/s [
47]. Occasionally, other abbreviations such as Pc and Papp are used to express permeability [
44,
48]. Although TEER and Pe tend to be correlated up to a threshold, as described above, TEER is heavily influenced by the path of lowest resistance in the monolayer, whereas Pe depends on the sum of the transport across all pathways. Pe of inert tracers varies by size, with a large tracer like albumin (66.5 kDa) showing about a 30-fold lower Pe value than that of sucrose (340 Da) [
35]. Typical Pe values for sucrose in conventional in vitro BBB models is around 0.5–10 × 10
−6 cm/s in primary cells where TEER ranges from about 300–1000, and tends to be an order of magnitude higher or more in commonly used immortalized cell lines [
25].
BBB disruption may follow different modalities, and the abilities of TEER and Pe to detect these modalities should be considered. We note that the discussion that follows is theoretical, and based on principles of Pe and TEER measurements described above rather than on formally published studies. We first posit that TEER does not readily measure BBB disruption that is caused by transcytotic mechanisms such as pinocytosis since these mechanisms do not typically result in a patent channel. As the vesicles are in the 80–100 nm range [
40], transcytosis does not display the size-dependent leakage discussed above with Pe, although larger sized molecular tracers can more sensitively detect BBB disruption across transcytotic pathways vs. smaller tracers because they are less permeant at baseline. Conversely, smaller tracers can more readily penetrate paracellular openings and thus tend to correlate with TEER. The degree to which a pathological event results in BBB disruption that is due to paracellular, transcellular, or transcytotic processes can at best be inferred from comparison of measures by TEER and permeability of molecular weight markers. Verification of the predominating disruptive pathways requires their microscopic evaluation. Immunofluorescent imaging of TJPs, for example, can provide insight on their expression and proper localization at cell–cell contacts. However, TJP changes do not always correlate with measures of BBB disruption [
49], and so a more comprehensive method of evaluating structural change in BECs, such as electron microscopy, may be required for concluding which mode of disruption predominates.
Effects of immune factors on BBB disruption in conventional in vitro BBB models
BECs express cytokine and chemokine receptors, as well as receptors such as toll-like receptor 4 (TLR4) for pathogen-associated molecular pattern molecules like bacterial lipopolysaccharide (LPS), permitting them to respond to these factors [
1]. Although there are numerous studies in vivo that have provided insight on how inflammatory molecules contribute to BBB disruption, we limit consideration here to processes that have been specifically explored using in vitro BBB models. In primary BECs, LPS can cause BBB disruption that is associated with TJP dysfunction [
35,
50‐
52] and may cause BEC apoptosis at a higher dose [
53]. The effects of LPS on BBB disruption are mitigated by indomethacin, indicating that cyclooxygenases contribute to LPS-induced BBB disruption [
35]. Conflicting results have been reported regarding how supportive cells of the NVU influence BEC responses to LPS. Whereas bovine BECs were protected from LPS-induced BBB disruption during co-culture with astrocytes or mixed glia from rats [
50], mouse BECs were not protected from LPS-induced BBB disruption when in co-culture with mouse astrocytes, pericytes, or both [
35]. Other microbial factors such as HIV-1 and its proteins Tat and gp120 can also contribute to BBB disruption, via mechanisms that involve oxidative stress and modifications of TJPs [
36,
54‐
56] and sensitization to other microbial factors like LPS [
57]. Pro-inflammatory cytokines such as tumor necrosis factor-alpha (TNF-α), interleukin-1 beta (IL-1β), and interleukin-6 (IL-6) have been shown to lower TEER [
58]. In primary rat BEC cultures in astrocyte conditioned medium, TNF-α and IL-1β were shown to induce transient reductions in TEER by 60 min that recovered after about 210 min. Recovery was attributed, in part, to degradation of the applied cytokines. In contrast, application of IL-6 caused TEER reductions that persisted for at least 300 min. The TEER-reducing effects of all three cytokines were cyclooxygenase-dependent [
58]. In bovine BECs co-cultured with astrocytes, increases in permeability to sucrose and inulin induced by TNF-α treatment were delayed, with no increases in permeability being evident until 16-h post-treatment [
59]. Primary human microvascular BECs also showed peak reductions in TEER following application of TNF-α, interferon-gamma (IFN-γ), or IL-1β after about 18 h post-treatment [
52]. Studies on mechanisms contributing to BBB disruption in inflammatory contexts have largely focused on dysregulation of TJPs, but recent works have also evaluated participation of cellular prion protein [
60] and micro-RNA [
61]. Chemokines like C–C motif chemokine ligand (CCL) 2 can also mediate BBB disruption under physiological and pathological conditions, such as oxygen–glucose deprivation [
34,
62]. It was shown that CCL2 causes subcellular redistribution of claudin-5, occludin, ZO-1, and ZO-2 via serine phosphorylation through a RhoA/protein kinase C-α pathway [
63], and also induces claudin-5 and occludin internalization via caveolae-dependent endocytosis [
64]. Other inflammatory mediators such as bradykinin, histamine, serotonin, arachidonic acid, and ATP can also increase the paracellular permeability of the BBB [
65].
An interesting speculation is whether the BBB and BECs respond to baseline levels of LPS in the circulation. It is clear that under certain non-pathologic conditions such as a high fat meal or jogging that LPS can leak from the gut and contribute to increased LPS concentrations in the blood [
66,
67], and there is some speculation that low levels of LPS may circulate physiologically. If so, LPS and the immune system may play roles in the physiologic regulation of the BBB.
In summary, in vitro models of the BBB have provided a useful tool to better understand direct interactions of immune factors such as cytokines and chemokines, LPS, and others that cause BBB disruption. We have highlighted some limitations and technical considerations associated with use of in vitro models to study BBB disruption. BECs in vitro have provided much information on paracellular mechanisms of BBB disruption that are attributed to the modification of TJPs, but much less is known regarding use of in vitro BBB models to evaluate vesicular/transcellular mechanisms of disruption. Transcellular BBB disruption does occur in vivo following inflammatory insults [
68,
69], and thus may be an important future avenue of investigation.