Holger Gerhardt : Vascular Biology

The Vascular Biology Laboratory aims to unravel the basic cellular principles and the genetic/molecular control of blood vessel patterning in development and disease. Blood vessels are critical for tissue growth and healthy organ function. Effective blood vessel function requires that the endothelial cells lining the vessel assemble a regular network of interconnected tubes with adequate diameter and branching frequency to allow regulated blood flow. As different organs serve distinct functions, and possess different metabolic requirements, blood vessel patterning bears organ specific characteristics.

We use a cell biology approach in various model systems in vivo and in vitro, in combination with computational modelling to investigate how individual endothelial cells respond to signals from the tissue and communicate with each other in order to orchestrate behaviour leading to functional network formation.Disorganised vascular patterns characteristically associate with pathology. Unlike the regular hierarchical branching pattern found in healthy organs, ischemic, inflammatory and malignant tissues harbour irregular, tortuous, dilated and leaky vessels that are poorly perfused. Understanding the cellular and molecular mechanisms underlying healthy and pathological vessel patterning is of greatest importance for the development of therapy aimed at restoring blood vessel function, or depriving growing tumours of blood supply.

Extracellular matrix functions in vascular patterning

Endothelial cells forming nascent blood vessels engage in coordinated cell behaviour in which individual cells perform distinct functions, such as guidance, migration, proliferation, and lumen formation. The collective actions of the population and the orchestration of the behaviours not only shape the emerging vascular sprout, but ultimately pattern the functional network.

The endothelial cells and their supporting mural cells as well as the surrounding tissue produce extracellular matrix components that influence cell behaviour, mediate physical stability, establish cell polarity and affect cell signalling. Over the past years, we have been interested in understanding when and where endothelial cells express specific matrix components and how endothelial as well as tissue matrix components influence cell behaviour during vascular patterning.

Current concepts state that endothelial cells must actively degrade their basement membrane to break out of the stable vessel and form new sprouts. In addition, the leading endothelial cells are assumed to degrade tissue matrix by proteases in order to ‘tunnel’ through the tissue. Indeed, in 3D collagen matrices, embedded endothelial cells utilise matrix metalloproteases to invade the gel. How important such a mechanism may be in the tissue environment is less clear.

Astrocytic matrix-growth factor interactions drive tip cell migration

In the post-natal mouse retina, we observed that astrocytes serving as migration template for the endothelium deposit fibrillar fibronectin (FN) transiently ahead of the advancing endothelium. The leading endothelial tip cells express various integrins on their filopodia, including the FN receptor a5b1 integrin, suggesting that FN might serve as migration template for haptotactic guidance of endothelial tip cells (Stenzel et al., 2011; Development. 138: 4451-4463).

Using conditional knockout strategy, we studied the function of astrocytic FN. Deletion of FN led to a mild migration delay (~10%), with little overall effects on vascular patterning. Surprisingly, deletion of just one allele also led to a migration delay of ~5%, suggesting that the FN dose directly gauges migration speed. To investigate whether this effect was integrin mediated, we studied mice carrying an FN RGD to RGE mutation, inactivating the RGD motif-dependent binding to a5b1 integrin. Compound mutants in which astrocytes only express FN RGE displayed no additional migration defects, indicating that the RGD motif is not essential for tip cell migration. Selective endothelial deletion of a5 integrin also had no effect on tip cell migration, illustrating that FN mediated tip cell migration is likely integrin independent.

Given that FN binds VEGF-A and VEGF-A gradients drive directed tip cell migration, we investigated whether interference with FN binding VEGF-A affects migration. Indeed, intraocular injection of inhibitory peptides stalled the migratory front. Both FN deletion and inhibitory peptides impaired Akt but not Erk signalling, consistent with the selective effect on endothelial migration, but not proliferation.

Fibronectin however is not the only matrix component with strong VEGF binding properties produced by astrocytes. Heparan-sulfate proteoglycans (HS) are key surface moieties that through the negative charge of their highly sulfated saccharide side chains bind growth factors and morphogens. Staining for HS revealed a strikingly similar pattern to fibronectin on astrocytes. Deletion of the key glycosyltransferase EXT1 in astrocytes also led to reduced vascular migration. Deletion of both EXT1 and FN caused a dramatic decrease, suggesting that the combined binding capacity of HS and FN establishes a cell surface density gradient of VEGF-A stimulating directed migration. Whilst integrin interactions appeared dispensable for migration, they do affect other aspects of tip cell morphology. The filopodia of endothelial cells normally track along astrocytic processes; an association that lost precision in the absence of a5 integrin.

Endothelial matrix regulates stalk cell formation through Dll4-Notch signalling

In a parallel study, we investigated the expression and function of endothelial specific matrix molecules that are deposited as the vessels form (Stenzel et al., 2011; EMBO Rep. 12: 1135-1143). Laminins are a major component of basement membranes, and the endothelium expresses two distinct heterotrimeric proteins, laminin 411 and 511. Studying mice and embryonic stem cell sprouting assays deficient in lama4, the gene coding for the alpha4 chain combining to form the mature 411 protein, we observed dramatic hypersprouting with excessive tip cell formation at the expense of stalk cell specification. Dll4 production and Notch activation were significantly reduced, suggesting a previously unappreciated role for laminin 411 in Dll4/Notch signalling. Indeed cultured endothelial cells responded with strong Dll4 induction when seeded on laminin 411, but not on laminin 511.

Based on a series of gain- and loss-of function studies, and investigation of interaction with the VEGF and integrin signalling pathway, as well as mosaic loss-of function analysis, we conclude that laminin411 produced in the nascent vessels promotes Dll4 production through integrin b1 mediated signalling, in concert with VEGF-A signalling. The precise transcriptional control is currently under investigation. The integration of integrin and Notch signalling in the endothelium provides a molecular mechanism for the stabilising effect of the endothelial basement membrane on nascent blood vessels.

These studies provide insights into how distinct matrix molecules in the tissue environment and around the vessels can shape the response of individual endothelial cells and their coordination during vessel morphogenesis. Given the wealth of matrix modifying activities of proteases during inflammation and tumour angiogenesis, vascular patterning defects might at least in part be caused by deficiencies in matrix-VEGF retention and guidance as well as laminin/Dll4 mediated tip/stalk cell specification.