PLVAP, which is the only known molecular component of SDs and FDs (14,28), forms homodimers that are cross-linked in situ (5,14,28). The upregulation of PLVAP expression upon treatment of EC cultures with PMA was associated with the de novo formation of SDs and FDs that were demonstrated to contain PLVAP (14). In addition, silencing of PLVAP mRNA expression inhibited de novo diaphragm formation in caveolae, TECs and fenestrate (14,15), and knockout of the PLVAP gene in mice resulted in the complete absence of SDs and FDs (10). These findings suggested that PLVAP is required for the formation of SDs and FDs in ECs.

It has been hypothesized that PLVAP is responsible for the formation of the radial fibrils that constitute both SDs and FDs (31). To elaborate, these diaphragms consist of thin fibrils originating at the inner surfaces that then intertwine into a knot at the lumenal center of caveolae, fenestrae or TECs (4). At present, it is unclear whether PLVAP is the only protein present in the diaphragms (31). Various functional groups at the C-terminus of PLVAP potentially offer binding sites for other PLVAPs, causing the fibrils to interweave in the observed manner (31). In addition, the polysaccharides occupying ~15% of the molecular weight of PLVAP near the membrane may support the membrane and maintain the fibrils in their correct position (4). Furthermore, interactions between PLVAP C-termini and the existence of a rigid structure attaching PLVAP to the cytoskeleton may contribute to the stabilization of the diaphragms (4).

PLVAP is restricted to a subset of capillary endothelial linings, as demonstrated by immunofluorescence using anti-PLVAP antibodies (5,28). On a C57BL/6N genetic background, PLVAP-null embryos died before birth and suffered from edema, hemorrhages, and defects to the vascular wall of subcutaneous capillaries (36). Furthermore, diaphragms were missing from caveolae of the subcutaneous capillaries and endocardium ECs of PLVAP-null mice (36). In addition, capillaries of the pancreas and kidney peritubular interstitium of PLVAP-null mice demonstrated a complete absence of diaphragms in fenestrae, caveolae, and TECs (36). Transmission electron microscopy confirmed the lack of endothelial integrity, detecting extensive defects in the endothelial lining of capillaries (36).

Similarly, Stan et al (10) demonstrated that the deletion of the PLVAP gene in mice of a mixed BALB/c-C57Bl/6J-129Sv/J background resulted in decreased survival due to the absence of diaphragms. In particular, the loss of diaphragms disrupted the barrier function of fenestrated capillaries, causing a major leak of plasma proteins, and leading to hypoproteinemia, hypertriglyceridemia, and an increased plasma concentration of chylomicron remnants (10). As a result, PLVAP-null mice suffered from premature death due to severe enteropathy and edema of the intestine, kidneys, and pancreas (10). In addition, the results of an Evans Blue dye extravasation assay indicated that the intestinal capillaries displayed the highest rate of protein leakage (10). Conversely, endothelial reconstitution of PLVAP by breeding with PLVAP transgenic mice rescued the PLVAP-null phenotype and restored the diaphragms of fenestrae and caveolae (10). As human influenza hemagglutinin (HA)-tagging of PLVAP did not alter its behavior in cultured cells, VEC-PV1HA-HA transgenic mice were generated that expressed PV1-HA under the control of the vascular endothelial-cadherin promoter and a 5′-intronic enhancer element (10,37). These mice demonstrated ~30–50% reconstitution of PLVAP expression, in addition to the restoration of FDs/SDs in the lungs, adrenal glands, kidneys, pancreas, thyroid and intestine, which led to improved survival (10). Therefore, PLVAP is crucial for the regulation of vascular permeability (Fig. 3).

The migration of leukocytes from the bloodstream into tissues via a transcellular pathway through the endothelium is one of the central paradigms of inflammation and immunity (19,38). In vitro transmigration experiments have demonstrated that rings containing PLVAP and caveolin-1 surround lymphocytes during their migration (19). PLVAP was also reported to redistribute in the presence of and partially colocalize with vimentin and caveolin-1 (19). In capillary flow assays, tumor necrosis factor-α activated HUVECs grown in glass capillaries and incubated with anti-PLVAP antibody (19). As a result, the transmigration of peripheral blood mononuclear cells through the HUVEC layer was significantly inhibited, although the rolling and adhesion functions were unaffected (19). Blockage of PLVAP using a MECA-32 antibody in an acute peritonitis model decreased leukocyte migration by ~85% (19). Therefore, PLVAP potentially aggravates inflammation by increasing leukocyte trafficking.

Lymph nodes are specialized for efficient interaction of peripheral antigens with lymphocytes (39,40). Caveolae, TECs and vesiculo-vacuolar organelles (VVOs) with diaphragms were identified in the subcapsular sinus lymphatic ECs (LECs) of lymph nodes (41). PLVAP is synthesized in LECs, which line the sinuses and cover the distal lymphatic vessels (41). Diaphragms consisting of PLVAP fibrils in TECs act as physical sieves that are responsible for regulating the entry of soluble antigens and lymphocytes into the parenchyma (41). PLVAP-null mice demonstrated an increase in lymphocyte transmigration through the sinus floor, in addition to non-selective antigen entry into the lymph system (41). Similarly, in the absence of PLVAP diaphragms on the sinusoidal floor LECs, the migration of lymph-generated lymphocytes to the parenchyma of peripheral lymph nodes (PLNs) is increased (41). When PLVAP-containing diaphragms interacted with antibodies, the entry of lymphocytes into the PLNs was prohibited at the sinus-cortex interface (41). These observations suggested that the selective entry of lymphocytes and antigens into the lymphatic nodes requires PLVAP (41).

In a previous study, migrating lymphocytes were demonstrated to produce F-actin-rich podosomes that probe the lymphatic sinus floor (41). The large PLVAP-positive patches observed on the LEC floor at points of transendothelial diapedesis in the sinus were colocalized with the areas populated by F-actin-rich podosomes (41). Electron microscopy revealed that, lymphocytes typically made contact with the wild-type LEC membrane in regions containing caveolae, VVOs, TECs and diaphragms (41). Therefore, PLVAP-positive diaphragms guarding the pores, tubular structures and channels may offer the path of least resistance through the sinusoidal floor LECs (41).

With regard to cancerous tissues, PLVAP was first observed in malignant glioma microvasculature associated with increased fenestration, malperfusion and hyperpermeability (16). Subsequently, PLVAP was shown to be upregulated in ECs in tumors of the brain, lungs, breasts, stomach, liver, pancreas, colon, small intestine, kidneys, ovaries, prostate, uterus, skin and lymph nodes (16,33,42). Furthermore, upregulation of PLVAP was reported to be similar in primary and metastatic tumors (18). Increased expression levels of PLVAP in tumors have been associated with angiogenesis (17). PLVAP is typically induced in large, well-vascularized tumors, in which PLVAP colocalizes with the vascular endothelial markers, cluster of differentiation 31 (CD31) and von Willebrand factor (16,18,43). Furthermore, PLVAP was reported to be induced in vitro following exposure to VEGF-enriched medium and tumor cell lines (17).

As the presence of PLVAP facilitates vascular growth in cancer, it is considered a novel target for cancer therapy (16,17). Downregulation of PLVAP using small hairpin RNA prevented the development of pancreatic adenocarcinoma in xenografts (32). In a previous study, infusion of recombinant monoclonal anti-PLVAP antigen-binding fragment co-expressed with the extracellular domain of human tissue factor into the feeding artery of hepatocellular carcinoma (HCC) led to vascular thrombosis and extensive necrosis of HCC (27). Furthermore, the suppression of tumor growth and minimal systemic toxicity indicated that PLVAP is a novel therapeutic target for the treatment of HCC (27).

Spinal cord injury (SCI) is a potentially life-threatening condition (44,45). Following SCI, both ECs and neurons are rapidly lost (45,46). Expeditious and extensive cell death leads to inflammation followed by adaptive angiogenic responses at and around the epicenters of primary injury (45,46). In contusive SCI mice, PLVAP was detected in several microvascular beds at 12 h post-SCI, with the level of PLVAP peaking at around 3–7 days post-injury (25). The majority of newly generated vessels at the injury sites were PLVAP-positive on day 3 post-injury, thus suggesting that PLVAP is expressed in neovasculature post-SCI (25). Furthermore, microvessels expressing PLVAP appeared to be spatially correlated with tissues containing actively extravasated neutrophils (25). On day 3 post-SCI, the expression of intracellular junction component zonula occludens-1 (ZO-1) and occludin was negligible in PLVAP-positive microvessels, which was indicative of a significant disruption of neurovascular integrity (25). Therefore, the upregulation of PLVAP may lead to secondary injury post-SCI via the induction of inflammation and deterioration of neurovascular function (25).

Normal mature glomerular endothelium contains fenestrae without FDs for trafficking of macromolecules (26,47) (Fig. 3). One of the histopathological features of TG is the upregulation of PLVAP expression and the increased numbers of caveolae in glomerular ECs (26). Transcytotic pathways that deliver albumin and immunoglobulins are largely dependent on vesicles originating from caveolae (20). As glomerular ECs in TG patients typically demonstrate reduced fenestration under the transmission electron microscope (26), Yamamoto et al (26) hypothesized that the elevation of PLVAP expression and caveolae formation may be a compensatory mechanism to increase the permeability of ECs to macromolecules. TG with double contouring of the glomerular capillaries is a characteristic manifestation of chronic rejection in an allograft recipient (48). The double contouring in TG is caused by endothelial injury and subsequent formation of a basement membrane-like structure beneath the endothelium; hence, the upregulation of PLVAP expression in glomerular capillaries may reflect vascular remodeling post-injury (26,47,48).

In ischemic brain disease, damage to the blood-brain barrier results in an increased permeability of the microvasculature and fluid accumulation in the extracellular space, leading to brain ischemia, hypoxia and eventually mortality (43). PLVAP was suspected to be induced during hypoxia, as PLVAP upregulation was detected in the brain tissues of patients with acute ischemia (17). In C57/B6 mouse models with focal cerebral ischemia, negligible PLVAP staining was observed at 48 h following disease onset, and marked upregulation was detected 5 days later (43). Furthermore, all PLVAP-positive cells were shown to be located around the area of ischemic damage (43). Within the tissues, ~17% of CD31-positive vessels were expressing PLVAP by 48 h (43). As a result of PLVAP upregulation, no statistical difference was observed between the number of PLVAP-positive vessels and CD31 expressing vessels on day 7 following disease onset (43).

Norrie disease, which is characterized by abnormal angiogenesis and exudative vitreoretinopathy, is caused by mutations in the Norrie disease pseudoglioma (NDP) gene (49–51). One-week-old NDP-null mice had reduced retinal capillarization (52), and increased endothelial fenestration and disrupted vascular integrity were observed in the retinas of NDP-null mice a week later (21). Unlike the wild-type mice, ectopic expression of PLVAP protein was demonstrated in the developing retinal vasculature of NDP-null mice, providing evidence for a potential role of PLVAP in the pathogenesis of Norrie disease (21).

Loss of blood-retinal barrier (BRB) integrity is an important feature underlying the pathogenesis of diabetic macular edema (DME) (22). Increased retinal VEGF levels are associated with BRB breakdown in diabetic rodents, primates and humans (22,23,34). VEGF is considered to induce transcellular transport via caveolae (22). In previous studies, exposure to exogenous VEGF led to an increase in the number of pinocytotic vesicles at the lumenal side of retinal capillaries, and to increased levels of PLVAP in caveolae (23,53). To mimic the in vivo pathophysiology of DME, bovine retinal ECs (BRECs) were stimulated with VEGF, which led to an increase in the mRNA expression levels of caveolin-1 and PLVAP (22). In addition, a cell-based enzyme-linked immunosorbent assay detected a significant increase in the PLVAP content of BRECs at 72 h following VEGF treatment (23).

Mating the diabetic Ins2^Akita (Akita) mice with the human VEGF photoreceptor-overexpressing trVEGF029 (Kimba) mice resulted in the Akimba (Ins2^AkitaVEGF⁺/⁻) mouse model that displayed retinal neovascularization and hyperglycemia, which are characteristics of advanced clinical DR (24). BRB loss, which was characterized by fluorescein leakage, was observed in both Kimba and Akimba mice (24). Among these mice, fluorescein leakage was associated with focal angiogenesis and PLVAP gene expression, suggesting a vital role for PLVAP in regulating the permeability of the BRB (24).