Ser/Thr-phosphorylation forms of occludin are found concentrated at TJs whereas dephosphorylated occludin is rather detected on basolateral membranes and associated with disrupted TJs in epithelial cells [
142,
143] as well as in brain endothelial cells in experimental autoimmune encephalomyelitis, a murine model of multiple sclerosis characterized by brain inflammation [
144]. Regarding claudin-5, phosphorylation of its C-terminal domain on Thr207 residue in response to PKA or Rho kinase activation [
120,
123,
145] generally affected TJ integrity in brain endothelial cells and increased permeability.
Differential regulation of TJs by Protein Kinases C (PKCs)
PKC-dependent pathways have been involved in endothelial barrier disruption, as reported following treatment by pertussis toxin, an inhibitor of Gαi heterotrimeric G proteins [
146], or in response to the pro-inflammatory cytokine interleukin-6 (IL-6) which plays a critical role during hypoxia [
147]. However, early reports had clearly established that PKC activity was crucial for BBB integrity in epithelial cells, inasmuch as PKC inhibitors completely blocked the formation of TJs [
148,
149]; in addition, PKC-mediated phosphorylation of occludin (on residue Ser338) was involved in occludin targeting to TJs and TJ stabilization in epithelial MDCK cells [
148].
At least part of the interpretation of these apparently conflicting data may be found in the heterogeneity of the PKC family. The Ser/Thr-kinases PKCs are indeed classified into conventional (cPKC: α, βI, βII and γ), novel (nPKC: δ, ε, θ, η, μ) and atypical (aPKC: λ,ζ) PKC isozymes [
150] according to their modes of regulation. Accumulating evidence has pointed to a differential capacity of PKC isozymes to regulate BBB permeability. Indeed, activation of nPKC-θ and aPKC-ζ signaling by hypoxia-mediated TJ proteins results in relocalization (such as claudin-5, occludin, ZO-1) and increased BBB permeability in rat brain microvascular endothelial cells (
in vitro and
in vivo) [
121,
151]. In human brain microvascular endothelial cells, cPKCα, cPKC βII and aPKCλ/ζ isoforms were activated by HIV-1 gp120 envelope protein, leading to BBB disruption, intracellular calcium increase and monocyte migration across cell monolayer [
122]. Interestingly, when cPKC-α was found to contribute to TJ disassembly, nPKC-ε activation mediated TJ formation in epithelial MDCK cells [
152]. In line with this observation, over-expression of cPKC-α in rat epididymal microvascular endothelial cells was reported to enhance thrombin-induced permeability, whereas nPKC-δ expression promoted barrier function [
153]. By contrast, IL-25, expressed by mouse brain capillary endothelial cells, was shown to prevent inflammation-induced BBB disruption and down-regulation of TJ proteins (occludin, claudin-5, JAMs) through activation of the nPKC-ε pathway [
154]. Altogether, these observations strongly suggest that nPKC-selective activation generally contributes to maintaining barrier integrity, whereas cPKC activation has the opposite effect, both in polarized epithelium and endothelium (Table
1).
Regarding aPKC isoforms (λ and ζ), they have been shown to contribute to the establishment of epithelial cell polarity, via participation in the PCP complex together with Par-3 and Par-6 [
63,
68,
155]. As mentioned above, the PCP complex is recruited to endothelial TJs by Par-3 binding to JAM proteins [
64‐
66]. Over-expression of a dominant negative mutant of aPKC causes mislocalization of Par-3 and affects the biogenesis of the TJs in epithelial cells [
67], suggesting that Par-3 is a substrate of aPKC and that its localization in epithelial cells is dependent upon its phosphorylation. In the same line, the VE-cadherin/CCM1 (a protein encoded by the CCM1 gene which is mutated in a large proportion of patients affected by cerebral cavernous malformation) complex controls aPKC-ζ activation and Par-3 localization during early steps of brain endothelial cell polarization [
98]. The participation of this PCP complex to TJ integrity was further illustrated by the recent observation that meningococcal adhesion to human cerebral endothelial cells recruited Par-3, Par-6 and aPKC-ζ under bacterial colonies and induced disruption of cell-cell junctions [
156]. Surprisingly, a distinct Par-3/Par-6 complex, directly associated with VE-cadherin and lacking aPKC, has also been identified in endothelial cells [
157]. Finally, although additional polarity complexes are known in epithelial cells (the apical Crumbs complex and the basolateral Scribble complex) where they also contribute to TJ formation and regulation, no similar observations have been reported, to our knowledge, in brain endothelial cells.
The RhoA GTPase signaling pathway, activated by several membrane receptors, has been extensively documented in various cell types to induce actin cytoskeleton rearrangements involved in cell migration and proliferation. In brain endothelial cells, RhoA activation increased permeability, in response to inflammatory stimuli, through one of its major effectors Rho kinase (ROCK) [
158,
159]. Among these inflammatory stimuli, chemokines like MCP-1/CCL2, acting via their seven transmembrane-domain receptors, are known to activate the RhoA/ROCK pathway in mouse brain endothelial cells, to induce occludin, claudin-5 and ZO-1 Ser/Thr-phosphorylation, followed by their delocalization from TJs, ultimately leading to increased barrier permeability [
124,
160]. Similarly, enhanced monocyte migration across human brain endothelial cells was observed in an HIV-1 encephalitis model [
123,
161]. Also, adhesion molecules like ICAM-1 and VCAM-1 were shown, in response to lymphocyte/monocyte adhesion, to transduce signals in rat brain endothelial cell lines including activation of the RhoA/ ROCK pathway [
162,
163]: activation of this pathway ultimately leads to enhanced lymphocyte migration, suggesting that this process may be involved in the massive infiltration of immune cells into the CNS observed in multiple sclerosis. It must be mentioned, however, that lymphocyte migration across the BBB may also happen via a transcellular pathway, leaving intact endothelial TJs [
164].
Rearrangements of the actin cytoskeleton have long been recognized to be regulated, not only by the RhoA/ROCK pathway, but also, often in a coordinated manner, by the myosin light chain kinase (MLCK): MLCK directly phosphorylates the myosin light chain, leading to actomyosin contraction and endothelial barrier disruption [
165‐
167]. In the same line, inhibition of MLCK in bovine brain endothelial cells was more recently reported to prevent hypoxia-induced BBB disruption [
127], whereas alcohol increased human brain endothelial cell permeability via activation of MLCK and phosphorylation of occludin and claudin-5 [
125,
126]. Recently, pro-inflammatory cytokines (IL1β and TNFα), secreted by lymphocytes chronically infected by the HTLV-1 retrovirus, were reported to induce barrier disruption in the human brain endothelial cell line hCMEC/D3, associated with loss of occludin and ZO-1 through activation of the MLCK pathway [
140].
In conclusion, as summarized in Table
1, inflammation- or infection-induced actin cytoskeleton rearrangements in brain endothelial cells, mediated by the RhoA/ROCK and/or MLCK pathways, are associated with the phosphorylation, followed by delocalization or degradation of TJ proteins, and BBB disruption.