The aim of the present study was to evaluate whether primary macrovascular endothelial cells isolated from human umbilical cords prove useful as a cell culture model to study tight junction assembly. The decision to choose HUVEC for these assays was based on the fact that umbilical cords are easy to obtain and the subsequent isolation of the endothelial cells out of the vein is well described and does not require specific technical skills [21, 23]. We also decided to concentrate on human cells as a broad variety of appropriate antibodies directed against human tight and adherens junction-associated proteins are available.
Although endothelial cells apparently form dense monolayer in vitro microscopic analysis revealed intercellular gaps and clefts as well as discontinuous junction strands  demonstrating that single monocultures of endothelial cells are not sufficient to induce proper cellular junction formation. Therefore, many approaches were undertaken to overcome these limitations by co-cultivating endothelial cells with "helper cells" like astrocytes or pericytes or with their appropriate conditioned cell culture medium. Under these conditions cerebral microvascular endothelial cells succeeded in building blood-brain barrier like monolayer with enhanced transelectrical resistance and decreased paracellular permeability [15, 25–27].
Another approach takes advantage of the junction-protecting effect of the second messenger cAMP. Beneath its ability to modulate PKA activity cAMP directly influences Epac, a guanine nucleotide exchange factor for Rap1 that modulates actin reorganization and distribution of adherens and tight junction-associated proteins [28, 29].
We here focused on the cAMP-approach and described the effects of different cAMP analogues on formation of cellular junctions and changes in paracellular permeability in HUVEC. Although the effect of cAMP on assembly of tight junctions in endothelial cells is well known a characterization of the cellular and molecular events that are modulated by different cAMP derivates is necessary as endothelial cells from macro- or microvascular beds may differ in their response to cAMP. Moreover, evidence exists that different chemically modified cAMP analogues modulate specific cellular responses. Recently, a study by Sand and colleagues demonstrated that 8-CPT-conjugated but not bromine-conjugated cAMP analogues act as competitive thromboxane receptor antagonists .
We here demonstrated that, although all cAMP derivates improved formation of continuous junction strands in HUVEC, they differ regarding their cell compatibility and kinetics of junction assembly. cAMP, 8-Br-cAMP and its sodium salt 8-Br-cAMP/Na induced formation of continuous tight junction strands as soon as 24 h after addition. Nevertheless, longer incubation periods led to a disassembly of the junction strands suggesting that these compounds proved useful for solely short-termed experiments. Moreover, they did not show any significant effect on improving paracellular permeability of HUVEC monolayer and, more importantly, cAMP and 8-Br-cAMP seemed to induce extensive vacuolization when exposed for longer time periods. pCPT-cAMP, on the other hand, displayed its best effects on HUVEC when added for longer time periods (≥72h) and generated the most robust phenotype as shown by calcium switch experiments. Moreover, pCPT-cAMP is the only compound that actually induced an improvement in barrier properties. Nevertheless, all four compounds reduced thrombin-induced increases in paracellular permeability to a certain degree and partly diminished breakdown of junctions in response to calcium depletion. These results suggest that cAMP and its derivates exhibit rather protective properties towards barrier breakdown than improving existing barrier properties.
For us, it was interesting to note that the compounds tested here had different effects on transcript expression of tight junction-associated proteins in HUVEC. All four cAMP derivates induced enhanced synthesis of CLDN5 mRNA after 24 h but only in pCPT-cAMP and 8-Br-cAMP/Na-treated cells mRNA level of CLDN5 remained elevated in the next 48 h. CLDN5 is one of the main endothelial tight junction proteins and its expression correlates with formation of tight junctions and endothelial barrier properties [31–33]. On the other hand, substances known to disrupt endothelial barrier properties down-regulate CLDN5 transcript amounts . Concordantly, the expression profiles of CLDN5 mRNA correlated quite well with the assembly of proper tight junction strands in HUVEC treated with the different cAMP derivates. Expression of OCLN mRNA followed a slightly different pattern. Although pCPT-cAMP and 8-Br-cAMP/Na showed the greatest effect on induction of OCLN after 24 h at the end of the experiment mRNA level did not differ compared to untreated cells. Surprisingly, the remaining two cAMP derivates did not induce any significant changes in OCLN expression. We therefore performed western blots for CLDN5 and OCLN. Whereas the protein and mRNA data for CLDN were consistent expression of OCLN was increased after stimulation with all four cAMP derivates. This discrepancy might be explained by the fact that changes in mRNA expression do not necessarily induce altered protein synthesis and that in case of doubt the protein analysis provided the most reliable data.
It is well known that cAMP exerts both PKA-dependent and -independent effects. In HUVEC cAMP-induced expression of OCLN and CLDN5 is controlled by PKA whereas membrane translocation and formation of continuous strands seemed to be independent of PKA at least for pCPT-cAMP, cAMP and 8-Br-cAMP. Interestingly, these findings are in opposite to a study by Ishizaki et al. who showed that induction of CLDN5 by pCPT-cAMP in brain microvascular endothelial cells is independent of PKA . This discrepancy may be explained by the different sources of the endothelial cells; we used human macrovascular cells whereas the group of Ishizaki worked on microvascular cells isolated from porcine brain capillaries.
The intention of this study was to evaluate whether HUVEC proved to be suitable for studying endothelial tight junction formation. These primary endothelial cells express most if not all of the known endothelial adherens and tight junction-associated proteins and form proper junction strands when cultured under appropriate conditions. Nevertheless, they definitely miss some of the typical vascular endothelial characteristics like increased transelectrical resistance or highly impermeable paracellular barriers found in other endothelial cell culture models established to study e.g. blood-brain barrier properties in vitro [35, 36]. Certainly, these different properties of the cells are due to the specific vascular beds they derived from. Whereas most of the culture models used for permeability and drug transport studies utilizes cerebral microvascular endothelial cells we worked with macrovascular venous endothelial cells from the umbilical cord. Accordingly, HUVEC are definitely not an adequate model to study blood-brain barrier-related topics. But on the other hand, umbilical cords are easy to obtain, and isolation of HUVEC does not require highly specific technical skills and is not as time-consuming as, for example, isolation of microvascular endothelial cells. Therefore, they represent a suitable culture model for studying formation or disassembly of endothelial intercellular junctions and the signaling pathways that are linked to this process.