In vitro self-assembly of human pericyte-supported endothelial microvessels in three-dimensional coculture: a simple model for interrogating endothelial-pericyte interactions.

We describe a method for coculture of macro- or microvascular human endothelial cells (ECs) and pericytes (PCs) within a 3-dimensional (3-D) protein matrix resulting in lumenized EC cords invested by PCs. To prevent apoptotic cell death of ECs in 3-D culture, human umbilical vein or dermal microvascular ECs were transduced to express the antiapoptotic protein Bcl-2. To prevent PC-mediated gel contraction, the collagen-fibronectin gel was polymerized within a polyglycolic acid nonwoven matrix. Over the first 24-48 h, EC-only gels spontaneously formed cords that developed lumens via vacuolization; such vascular networks were maintained for up to 7 days. In EC-PC cocultures, PCs were recruited to the EC networks. PC investment of EC cords both limited the lumen diameter and increased the degree of vascular network arborization. Peg and socket junctions formed between ECs and PCs in this system, but dye transfer, indicative of gap junction formation, was not observed. This simple system can be used to analyze bidirectional signals between ECs and PCs in a 3-D geometry.

Human endothelial cell presentation of antigen and the homing of memory/effector T cells to skin.

Dermal microvascular endothelial cells (ECs) form a continuous lining that normally bars blood-borne T lymphocytes from entering the skin, but as part of the response to foreign antigen, dermal ECs undergo alterations in their surface proteins so as to provide signals to circulating T cells that lead to their activation and recruitment. Several observations suggest that human dermal microvascular ECs may help initiate cutaneous immune reactions by presentation of cognate antigens to circulating T memory cells: (1) antigen-specific inflammatory responses in the skin, as in other organs, involve accumulation of memory and effector T cell populations that are enriched in cells specific for the eliciting antigen; (2) recall responses to intradermal protein antigens in the skin start very rapidly within two hours of challenge; (3) dermal microvascular ECs in humans and other large mammals basally display high levels of class I and class II MHC molecules, the only known purpose of which is to present antigenic peptides to lymphocytes; (4) the lumen of dermal capillaries are narrower than the diameter of circulating T cells, ensuring surface contact; and (5) cultured human ECs effectively present antigens to resting memory T cells isolated from the circulation. Upon contact with activated T cells or their secreted products (cytokines), dermal ECs themselves become activated, increasing their capacity to recruit memory and effector T cell populations in an antigen-independent manner. Specifically, activated ECs express inducible leukocyte adhesion molecules such as E-selectin, ICAM-1, and VCAM-1; and several lines of evidence, including neutralizing antibody experiments and gene knockouts, have supported a role of these molecules in T cell recruitment. Dermal ECs have unique expression patterns of adhesion molecules that can determine the subsets of memory T cells that are recruited into the skin. For example, slow internalization of E-selectin allows more persistent expression of this protein on the surface of dermal ECs, favoring interactions with CLA-1+ T cells. VCAM-1 expression, normally confined to venular EC may extend to capillaries within the dermal papillae and contribute to epidermal inflammation, recruiting alpha4beta7 integrin-expressing T cells that also express the cadherin-binding integrin alphaEbeta7. New models involving transplantation of normal and genetically modified human dermal ECs into immunodeficient mice may be used to further explore these properties.

Mechanism of sustained E-selectin expression in cultured human dermal microvascular endothelial cells.

Persistent E-selectin expression has been proposed to be a unique property of dermal vascular endothelium that directs skin-specific homing of a subpopulation of circulating memory T cells. We compared the kinetics of E-selectin expression on cultured human dermal microvascular endothelial cells (HDMEC) with expression on HUVEC. Following treatment with TNF, E-selectin on HDMEC appears more slowly than on HUVEC (peak values 6-8 vs 4 h, respectively) and is sustained at significantly higher levels after 24 h. E-selectin mRNA, analyzed by S1 nuclease protection, consists of a single predominant transcript that follows a similarly transient time course in both cell types. Cell surface E-selectin is internalized more slowly on HDMEC than on HUVEC (t1/2 = 4.3 vs 1.6 h, respectively) as measured by serial FACS analyses in the presence of the protein synthesis inhibitor cycloheximide. In comparison, intercellular adhesion molecule-1 (ICAM-1) expression is not measurably reduced by either cell type under the same conditions. HDMEC are similar to HUVEC in rates of pinocytosis or receptor-mediated endocytosis. Pulse-chase analysis indicated that the degradative half-life of E-selectin protein is greater in HDMEC than in HUVEC (1.9 vs 1.5 h, respectively). E-selectin internalization in microvascular endothelial cells (EC) from lung and subcutaneous fat is slow, like HDMEC, whereas internalization in large vessel EC from saphenous vein and aorta is rapid, like HUVEC. We conclude that HDMEC sustain higher levels of expression at 24 h by slower internalization and degradation of E-selectin protein and that this may be a general property of microvascular EC.

The death domain of tumor necrosis factor receptor 1 is necessary but not sufficient for Golgi retention of the receptor and mediates receptor desensitization.

TNF signals are mediated through two different receptors, TNFR1 and TNFR2. In endothelial cells, TNFR1 is predominantly localized in the Golgi apparatus and TNFR2 on the plasma membrane. To investigate structural features responsible for the disparate localization, endothelial cells were transfected with epitope-tagged or green fluorescent protein-fused wild type and mutant receptor molecules. Wild type receptors recapitulated the distribution of endogenous receptors. Deletions of the entire TNFR1 intracellular domain or of the C-terminal death domain (TNFR1(-DD)) allowed expression of the receptor on the plasma membrane. However, addition of the death domain to the C-terminus of TNFR2 (TNFR2(+DD)) did not lead to Golgi-retention of this chimeric receptor. Overexpressed TNFR1, TNFR2, and TNFR2(+DD) increased basal expression of a cotransfected NF-kappaB-dependent promotor-reporter gene. Overexpressed TNFR1(-DD) did not activate NF-kappaB but acted as a ligand-specific dominant negative inhibitor of TNF actions. Unexpectedly, TNF responses were also inhibited by overexpressed TNFR1 and TNFR2(+DD), but not TNFR2. We conclude that the death domain of TNFR1 is required for retention of TNFR1 in the Golgi apparatus but is not sufficient to direct Golgi retention of a TNFR2(+DD) chimera, and that overexpressed receptors that contain the death domain (TNFR1 and TNFR2(+DD)) spontaneously activate NF-kappaB while inhibiting TNF responses.

Cytoprotection of human umbilical vein endothelial cells against apoptosis and CTL-mediated lysis provided by caspase-resistant Bcl-2 without alterations in growth or activation responses.

Graft endothelial cells are primary targets of host CTL-mediated injury in acute allograft rejection. As an in vitro trial of gene therapy to reduce CTL-mediated endothelial injury, we stably transduced early passage HUVEC with a caspase-resistant mutant form (D34A) of the anti-apoptotic gene Bcl-2. Bcl-2 transductants were compared with HUVEC transduced in parallel with an enhanced green fluorescent protein (EGFP) gene. Both transduced HUVEC have equivalent growth rates in complete medium and both show contact inhibition of growth. However, compared with EGFP-transduced HUVEC, the Bcl-2-transduced cells are resistant to the apoptotic effects of serum and growth factor withdrawal and are also resistant to the induction of apoptosis by staurosporine or by ceramide, with or without TNF. Transduced Bcl-2 did not reduce TNF-mediated NF-kappaB activation or constitutive expression of class I MHC molecules. HUVEC expressing D34A Bcl-2 were significantly more resistant to lysis by either class I-restricted alloreactive or PHA-redirected CTL than were HUVEC expressing EGFP. We conclude that transduction of graft endothelial cells with D34A Bcl-2 is a possible approach for reducing allograft rejection.

In vivo formation of complex microvessels lined by human endothelial cells in an immunodeficient mouse.

We have identified conditions for forming cultured human umbilical vein endothelial cells (HUVEC) into tubes within a three-dimensional gel that on implantation into immunoincompetent mice undergo remodeling into complex microvessels lined by human endothelium. HUVEC suspended in mixed collagen/fibronectin gels organize into cords with early lumena by 24 h and then apoptose. Twenty-hour constructs, s.c. implanted in immunodeficient mice, display HUVEC-lined thin-walled microvessels within the gel 31 days after implantation. Retroviral-mediated overexpression of a caspase-resistant Bcl-2 protein delays HUVEC apoptosis in vitro for over 7 days. Bcl-2-transduced HUVEC produce an increased density of HUVEC-lined perfused microvessels in vivo compared with untransduced or control-transduced HUVEC. Remarkably, Bcl-2- but not control-transduced HUVEC recruit an ingrowth of perivascular smooth-muscle alpha-actin-expressing mouse cells at 31 days, which organize by 60 days into HUVEC-lined multilayered structures resembling true microvessels. This system provides an in vivo model for dissecting mechanisms of microvascular remodeling by using genetically modified endothelium. Incorporation of such human endothelial-lined microvessels into engineered synthetic skin may improve graft viability, especially in recipients with impaired angiogenesis.