J.A. Firth

Endothelial barriers: from hypothetical pores to membrane proteins

The anatomical counterpart of the physiologically defined small pore system of capillary endothelia has proved difficult to establish. In non‐brain continuous capillaries, the contributions of caveolar and transmembrane pathways are likely to be small and paracellular clefts are probably the dominant routes. Analogy with epithelial paracellular pathways suggests that tight junctions may be the most restrictive elements. However, structural features of tight junction‐based models are incompatible with physiological data; it is more likely that the tight junction acts as a shutter limiting the available cleft area. Proposed molecular sieves elsewhere in the paracellular pathway include the glycocalyx and the cadherin‐based complexes of the adherens junctions. The molecular architecture of tight junctions and adherens junctions is moderately well defined in terms of molecular species, and there are differences at both sites between the endothelial and epithelial spectra of protein expression. However, definition of the size‐restricting pore remains elusive and may require structural biology approaches to the spatial arrangements and interactions of the membrane molecular complexes surrounding the endothelial paracellular clefts.

Distinct Patterns of Microvascular Endothelial Cell Morphology Are Determined by Extracellular Matrix Composition

Endothelial interactions with the extracellular matrix (ECM) play important roles in angiogenesis but whether specific ECM signals can determine specific cellular morphologies is unclear. The authors compared in vitro ECM-induced morphological responses of the phenotypically distinct human placental microvascular endothelial cells (HPMECs) with large vessel endothelial cells (HUVECs). HPMECs showed distinct patterns of reorganization in response to collagen-I or collagen-IV (monolayer disruption, sprouting, migration) and Matrigel or laminin-A (intussusception, cord formation, tubulogenesis), and an intermediate response to fibrin; whereas HUVECs responded similarly to collagen-1 and Matrigel (elongation, lattice formation, vacuolation) and showed little response to fibrin. Although the extent of collagen and Matrigel responses of HPMECs were increased by serum, acidic or basic fibroblast growth factor (aFGF, bFGF), or vascular endothelial growth factor (VEGF), and varied with matrix protein concentration, the basic patterns were matrix specific, and were independent of fibronectin. The collagen responses correlated with disruption of adherens and tight junctions and the formation of filopodial protrusions. Matrigel responses were associated with up-regulated junctional localization of VE-cadherin, and tubulogenesis developed mainly through paracellular remodeling rather than intracellular vacuolation. Overall, these findings suggest that distinct ECM interactions stimulate specific morphological responses. These signals may regulate morphological behaviour in the angiogenesis cycle, switching endothelial cells between migratory and vasculogenic phenotypes.

Not trophoblast alone: A review of the contribution of the fetal microvasculature to transplacental exchange

The fetal microcirculation of the term human placenta offers an interesting microvascular model. A perfused placenta can be used for integrated studies of vascular permeability-structure relationships. The organization of the paracellular pathway in human placental microvessels closely resembles not only that of the guinea-pig placenta, but also that seen in typical continuous non-cerebral capillaries such as those of the myocardium. This uniformity of organization has allowed the development of a model of the organization of endothelial junctional complexes that allows testable predictions about the relationship between junctional organization and microvascular permeability. The key features of this model are: (1) molecular size restriction may be determined by a fibre matrix based on cadherin arrays in the zonula adhaerens. (2) The zonula occludens (tight junction) is discontinuous and so cannot act as a molecular sieve for solutes. It may serve as a shutter that limits the proportion of the paracellular cleft available for permeation. The main implication for placental function is that the human placental microcirculation is relatively tight and is an important restriction to diffusive permeation of the maternal-fetal barrier by large molecules.

Characterization of cationic amino acid transporters and expression of endothelial nitric oxide synthase in human placental microvascular endothelial cells

We investigated the expression and activity of arginine transporters and endothelial nitric oxide synthase (eNOS) in human placental microvascular endothelial cells (HPMEC). Using RT‐PCR amplification products for eNOS, CAT1, CAT2A, CAT2B, CAT4, 4F2hc (CD98), rBAT and the light chains y+LAT1, y+LAT2, and b0+AT1 were detected in HPMEC, but not B0+. Immunohistochemistry and Western blotting confirmed the presence of 4F2hc and CAT1 protein in HPMEC. 4F2hc‐light chain dimers were indicated by a shift in molecular mass detected under nonreducing conditions. l‐Arginine transport into HPMEC was independent of Na+ or Cl− and was inhibited by the neutral amino acid glutamine, but not by cystine. The Ki for glutamine inhibition was greater in the absence of Na+. Kinetic analysis supported a two‐transporter model attributed to system y+L and system y+. Expression of eNOS in HPMEC was detectable by immunohistochemistry and ELISA but not by Western blotting. Activity of eNOS in HPMEC, measured over 48 h, either as the basal production of nitric oxide (NO) or as the accumulation of intracellular cGMP was not detectable. We conclude that HPMEC transport cationic amino acids by systems y+ and y+L and that basal eNOS expression and activity in these cells is low.

Permeability of the fetal villous microvasculature in the isolated perfused term human placenta

1. Capillary permeability‐surface area (PS) products for the low molecular weight radioactive tracers, 22Na, 51Cr‐EDTA (relative molecular mass 357) and 57Co‐cyanocobalamin (relative molecular mass 1353) were measured in the fetal circulation of isolated dually perfused lobules of normal term human placentae using the single circulation, multiple‐tracer dilution technique. 2. In lobules perfused with M199 medium, containing dextran and 5 g l‐1 bovine albumin, the extractions of all three tracers decreased as the flow was increased over the range of 2‐8 ml min‐1, and PS products for 51Cr‐EDTA and 57Co‐cyanocobalamin, but not for 22Na, reached constant values at flows above 0.1 ml min‐1 g‐1. 3. Flow‐independent PS products in the presence of albumin were 0.025 +/‐ 0.002 ml min‐1 g‐1 (mean +/‐ S.E.M., n = 25) for 57Co‐cyanocobalamin and 0.057 +/‐ 0.003 ml min‐1 g‐1 (n = 25) for 51Cr‐EDTA. The ratio of PS values (51Cr‐EDTA/57Co‐cyanocobalamin) was 2.28, while the ratio of the corresponding free diffusion coefficients was 1.79, indicating substantial restriction to the diffusion of the 57Co‐cyanocobalamin. 4. In another series of lobules perfused in the absence of albumin, extraction values for all three test tracers were constant over the same flow range. Values at high flow rates were therefore about twice those measured in the presence of albumin, and PS products for all three tracers failed to reach diffusion‐limited values. 6. Lobules perfused with and without albumin were fixed using a glutaraldehyde fixative containing 1% Alcian Blue dye. An ultrastructural examination of the endothelium showed no significant changes in cell or cleft morphology, or in the glycocalyx, in the absence of albumin which might account for the observed permeability change. 7. These data are the first physiological measurements specifically characterizing fetal microvascular permeability in the human placenta. The results suggest that permeability resembles that found in skeletal muscle and, as such, the endothelium presents a significant barrier to the diffusion of large solutes. The observed ‘protein effect’ indicates that albumin can interact with elements of the solute pathway to increase its restrictiveness.

Isolation of endothelial cells from human term placental villi using immunomagnetic beads

Chorionic villi excised from freshly delivered human term placentae and small endothelial cell aggregates were released from them by the sequential use of collagenase and trypsin. The endothelial cells were further isolated by rosetting with magnetic polystyrene beads which were coated with QB END/40, the endothelial-specific monoclonal antibody (mAb) to thrombomodulin. Cell rosettes were plated on gelatin coated Petri dishes. The cells initially grew as discrete colonies but reached confluence within 7 days. The monolayers were sub-cultured five times, and grew to confluence each time. All the cells were immunoreactive to the endothelial markers von Willebrand factor, QB-End/40 and Ulex europaeus-1 lectin. They did not show immunoreactivity to trophoblast markers (mAbs ED341 and ED235). The isolated cells could also incorporate acetylated low-density lipoprotein. Most of the cells possessed an elongated morphology, though some were slightly spread and polygonal in shape. The cell monolayers did not resemble the typical cobblestone appearance of endothelial cells isolated from large vessels. Ultrastructurally, most of the cells resembled placental microvascular cells in shape and frequency of caveolae; undifferentiated cell-cell contacts and extracellular matrix material was observed. Human placental microvascular endothelial cells may offer an in vitro model which complements the use of the perfused term placental lobule in studies of microvascular permeability.

Endothelial phenotype of the materno–fetal barrier

The terminal villous capillaries of the human placenta are the sites of maximal materno‐fetal exchange. These capillaries are highly angiogenic and are formed in the last trimester of pregnancy by looped angiogenesis from existing microvessels in the intermediate villi. They represent an ‘activated’ phenotype by possessing adherens junctions which contain VE‐cadherin (– and (– catenin but lack plakoglobin, the component of well differentiated junctions. The tight junctions present here show negative immunoreactivity to occludin and claudin‐1. The conduit vessels further upstream in the stem villi have the full complement of adherens and tight junctional molecules. VEGF and angiopoietin‐2 is present in terminal villi whilst angiopoietin‐1 appears to be a predominant feature of stem villi. The phenotype of junctions present in nascent vessels of the first trimester placentae was similar to the angiogenic terminal villous capillaries. The predominant growth factor here was VEGF, localised to all villi. Isolation of endothelial cells from last trimester placenta microvessels (HPMEC) demonstrated that these cells are capable of differential expression and dynamic regulation of cell‐cell junctions. Cells grown in the presence of ECGS showed a reduction in the level and junctional localisation of plakoglobin and occludin, with associated increased permeability to macromolecules. These adhesion molecules were up‐regulated and targeted back to cell‐cell contact regions by growing HPMEC in the presence of cAMP enhancing agents. This resulted in a 1.7 times increase in transendothelial resistance whilst proliferation of HPMEC cells was inhibited by elevated cAMP. HUVEC cells isolated from the umbilical cords of these placentae contain the full repertoire of AJ molecules in normal culture conditions with no change in composition or resistance under elevated cAMP. The differential expression of junctional adhesion molecules along the placental vascular tree and their in vitro plasticity suggests there is a physiological relevance of junctional regulation in the placenta. The activated junctional phenotype of the newly formed terminal microvessels allows them to participate in increased growth, proliferation and solute exchange in the last trimester when fetal demand is at its greatest. Furthemore, this necessary junctional immaturity of new/angiogenic human vessels appear to be at the expense of barrier function.