Angiogenesis and Blood Vessel Formation

Vasculogenesis is the de novo formation of blood vessels from mesoderm-derived precursor cells called angioblasts, which differentiate into endothelial cells and assemble into primitive tubular networks without any pre-existing vessel as a template. This happens first in the extra-embryonic mesoderm of the yolk sac, where clusters of cells called blood islands form; the peripheral cells become endothelium and the central cells become the first blood cells (hemangioblasts giving rise to both lineages). A parallel process establishes the dorsal aortae and cardinal veins within the embryo proper.

Angiogenesis is the subsequent (and lifelong) process by which new vessels sprout, branch, and remodel from this primary plexus or from any existing vessel. Unlike vasculogenesis, angiogenesis does not require new endothelial cells to differentiate from mesenchyme — it operates on cells that are already endothelial, directing them to migrate, proliferate, and form new lumens. Angiogenesis is responsible for essentially all vascular growth after the initial plexus is laid down: the maturation of the embryonic circulatory tree, the growth of vessels into every developing organ, the cyclical remodeling of the uterine endometrium, wound healing, and — pathologically — tumor vascularization and retinal neovascularization. The key conceptual bridge between these two processes and the rest of developmental biology is that angiogenesis is a paradigm case of a local tissue need converted into a long-range, directional signal. A hypoxic cell cannot itself become a blood vessel; it can only broadcast a message. That message — vascular endothelial growth factor, or VEGF — diffuses outward, forms a concentration gradient, and is read by endothelial cells at the margin of the existing vasculature much the way a morphogen gradient is read in patterning problems like inductive vulval signaling. The endothelial response to that gradient is itself patterned by a lateral-inhibition circuit (Notch/Dll4) formally analogous to the lateral inhibition that sharpens boundaries in neurogenesis and in the vulval precursor cell decisions — a recurring motif in which a graded signal is converted into a sharp, binary cell-fate choice among initially equivalent neighbors.

Step 1 — Hypoxia is sensed by a molecular oxygen switch. Every cell continuously synthesizes the transcription factor subunit HIF-1α (hypoxia-inducible factor 1-alpha). Under normal oxygen tension, prolyl hydroxylase domain enzymes (PHDs) use molecular O₂ as a substrate to hydroxylate HIF-1α at specific proline residues. Hydroxylated HIF-1α is recognized by the von Hippel-Lindau (VHL) protein, ubiquitinated, and degraded by the proteasome within minutes. When oxygen tension falls — as it does in any tissue growing faster than its blood supply, or in an infarcted or wounded region — the PHDs (which have a high Km for O₂) cannot hydroxylate HIF-1α efficiently. HIF-1α escapes degradation, accumulates, translocates to the nucleus, dimerizes with the constitutively expressed HIF-1β, and binds hypoxia-response elements (HREs) in the promoters of dozens of target genes.

Step 2 — HIF-1 transcribes VEGF-A. Among HIF-1's most important targets is the gene encoding VEGF-A. Hypoxic cells — whether embryonic mesenchyme outrunning its vasculature, a healing wound bed, or a solid tumor core — secrete VEGF-A, which diffuses into the surrounding extracellular matrix and forms a gradient with the highest concentration nearest the hypoxic source. VEGF-A is alternately spliced into several isoforms (VEGF₁₂₁, VEGF₁₆₅, VEGF₁₈₉ in humans) that differ in their affinity for heparan sulfate in the matrix — VEGF₁₆₅ is the dominant, moderately matrix-bound isoform that produces a stable local gradient rather than a fully diffusible one.

Step 3 — Endothelial cells read the gradient through VEGFR-2. The gradient is transduced by receptor tyrosine kinases on nearby endothelial cells, chiefly VEGFR-2 (KDR/Flk-1), with VEGFR-1 (Flt-1) acting largely as a decoy/modulatory receptor with much higher ligand affinity but weaker kinase signaling. VEGF binding dimerizes VEGFR-2, triggering autophosphorylation and downstream activation of PLCγ–PKC–MAPK (driving proliferation), PI3K–Akt (driving survival), and Src-family kinases (driving cytoskeletal remodeling and migration). The net cellular outcome is that the endothelial cell nearest the highest VEGF concentration becomes migratory and invasive.

Step 4 — The tip/stalk decision by Notch–Dll4 lateral inhibition. Not every endothelial cell at the vessel front can become a leading "tip cell" — that would produce a chaotic tangle rather than an organized sprout. The cell that receives the most VEGF signal upregulates Delta-like ligand 4 (Dll4) on its surface. Dll4 activates Notch1 receptors on the immediately adjacent cells. Notch activation in those neighbors releases the Notch intracellular domain (NICD), which suppresses VEGFR-2 and Dll4 expression in the neighbor and upregulates VEGFR-1 and Notch target genes (e.g., HES/HEY family repressors). This is classic lateral inhibition: the cell that "wins" (highest VEGF exposure) actively suppresses its neighbors from adopting the same fate, in exactly the same logic used in Delta-Notch neural precursor selection and analogous to the primary/secondary vulval precursor cell decision. The winning cell becomes the tip cell — highly migratory, extending filopodia studded with VEGFR-2 that sample the gradient, non-proliferative. The suppressed neighbors become stalk cells — proliferative, forming the lumenized trunk of the new vessel, and expressing Notch target genes.

Step 5 — Tip cell migration and lumen formation. The tip cell's filopodia respond to VEGF-A gradients and also to guidance cues shared with the nervous system — netrins, semaphorins (Sema3A repels, acting through Neuropilin-1, a VEGFR-2 co-receptor), Slits, and ephrin-B2/EphB4 (which additionally distinguishes arterial from venous identity). Stalk cells behind the tip proliferate under continued VEGF/Notch signaling and undergo lumen formation via intracellular vacuole fusion and cell-cell junctional remodeling (VE-cadherin re-organization), converting a solid cord of cells into a hollow, perfusable tube.

Step 6 — Anastomosis and pruning. Advancing tip cells from different sprouts meet and fuse (anastomose), guided by tip-cell filopodial contact and macrophages that act as physical bridges at some anastomosis points. Blood flow is established through the new loop. Vessels that fail to carry adequate flow are pruned by endothelial cell migration away from low-flow segments — a flow-dependent remodeling step that converts the initially uniform, redundant capillary plexus into a hierarchically branched, efficient tree matched to perfusion needs (first demonstrated quantitatively in the chick yolk sac and retina).

Step 7 — Vessel maturation and stabilization. New vessels are initially leaky and fragile. Maturation requires recruitment of mural cells — pericytes on capillaries, smooth muscle cells on larger vessels — via endothelial secretion of PDGF-B (which recruits PDGFRβ-expressing mesenchymal cells) and Angiopoietin-1/Tie2 signaling, which stabilizes endothelial-pericyte contacts and suppresses further sprouting. Angiopoietin-2, secreted by activated endothelium, antagonizes Ang-1/Tie2 locally and — together with renewed VEGF — permits vessels to re-enter a plastic, sprouting-competent state, which is exactly how quiescent adult vasculature can be re-activated for wound healing or exploited by tumors.

Sabin and the yolk sac blood islands (early 1900s). Florence Sabin's classic studies of chick and pig embryo yolk sacs, using vital dyes injected into living embryos and careful serial sectioning, showed that the earliest vessels arise in situ from clusters of mesodermal cells ("blood islands") without migrating in from elsewhere, and that the peripheral cells of these islands flatten into endothelium while the central cells become primitive blood cells. This established vasculogenesis as a genuinely local, de novo process rather than an extension of pre-existing vessels — the foundational observation the entire field builds on.

Folkman's angiogenesis hypothesis and the diffusion-limit experiments (1971). Judah Folkman, a surgeon, observed that tumors transplanted into isolated, avascular sites (such as an isolated perfused organ, or later a chamber lacking blood vessels) stopped growing at roughly 1–2 mm³ and never became lethal, whereas the same tumor cells grown in a vascularized site grew to enormous, lethal size. He proposed, in a landmark 1971 New England Journal of Medicine paper, that tumor growth beyond this diffusion-limited size is strictly dependent on the tumor's ability to recruit its own blood supply, and that tumors must secrete a diffusible "tumor angiogenesis factor" (TAF). This was initially received with considerable skepticism, but it directly predicted that blocking angiogenesis should be able to control tumor growth — a hypothesis validated decades later by anti-VEGF drugs. The diffusion-limit number he worked from (a few hundred micrometers to ~1–2 mm, the maximum distance oxygen can diffuse to sustain viable tissue) is still the standard figure cited for why avascular tumor spheroids plateau in size.

Purification and identification of VEGF (Ferrara and colleagues, 1989). Napoleone Ferrara and Judah Folkman's TAF hypothesis converged when Ferrara's group at Genentech purified a protein from bovine pituitary follicular cells that specifically stimulated vascular endothelial cell proliferation but not other cell types, naming it vascular endothelial growth factor. Cloning of its cDNA confirmed it as a secreted, disulfide-linked homodimeric glycoprotein distinct from other known growth factors, and subsequent knockout studies in mice showed that even loss of a single VEGF-A allele (VEGF+/–) is embryonic lethal around E11–E12 due to severe defects in vascular development — an extraordinary haploinsufficiency phenotype demonstrating just how precisely VEGF dosage must be tuned, a result reported independently by two groups (Carmeliet et al. and Ferrara et al., both Nature, 1996).

Gerhardt and colleagues establish the tip/stalk cell model (2003). Holger Gerhardt, working with Christer Betsholm's group, used the developing mouse retina — an accessible, flat, post-natally vascularizing tissue perfect for live and fixed imaging — to show that endothelial cells at the vascular front extend long, dynamic filopodia specifically toward regions of higher VEGF-A (visualized via astrocyte-derived VEGF-A gradients preceding the vascular front), and that these "tip cells" behave migratorily while trailing "stalk cells" proliferate. Genetic and pharmacological VEGFR-2 blockade collapsed filopodial extension, directly showing VEGF-A gradient-sensing by tip cell filopodia drives directional sprouting — the paper that gave the tip/stalk model its modern molecular definition.

Notch/Dll4 haploinsufficiency phenotype (2007). Multiple groups (Hellström et al.; Suchting et al.; Lobov et al., all in 2007) showed that mice with only one functional copy of Dll4 — again a haploinsufficient, dosage-sensitive phenotype echoing the VEGF-A story — develop hypervascularized, hyperbranched, non-productive vessel networks, because without adequate Dll4-Notch lateral inhibition too many endothelial cells adopt the tip-cell fate simultaneously. Pharmacological Notch blockade produced the same "too many tip cells, too little perfusion" phenotype, directly demonstrating that Notch signaling limits tip-cell number and that this limitation is required for functional (not just abundant) vascularization — a counterintuitive finding since Dll4/Notch blockade paradoxically increases vessel density while decreasing tumor perfusion and had to be pursued as an oncology strategy with caution.

Example 1 — Reading a VEGF/HIF dose-response and predicting a phenotype. A researcher engineers three mouse embryonic stem cell lines: (A) wild-type, (B) heterozygous VEGF-A knockout (VEGF+/–), (C) homozygous VEGF-A knockout (VEGF–/–), and implants each into an in vitro embryoid body assay that measures blood island and primitive vascular plexus formation at day 8. Reasoning: Because Carmeliet et al. (1996) and Ferrara et al. (1996) both independently showed that even a 50% reduction in VEGF-A dosage (the heterozygote) is embryonic lethal around E11–E12 with disorganized, deficient vasculature, this is a case where gene dosage — not simple presence/absence — determines phenotype. Predict:

The heterozygote is not simply "mildly less vascularized" — it is lethal, illustrating that VEGF-A operates in an unusually narrow, tightly tuned dosage window compared to most developmental genes, where heterozygous loss is typically silent due to buffering. This dosage-sensitivity is precisely why anti-VEGF drugs (which partially neutralize, not eliminate, VEGF) can meaningfully suppress pathological angiogenesis without needing complete blockade.

Example 2 — Calculating oxygen diffusion limits and predicting tumor behavior. A spherical tumor spheroid growing in culture, with no blood supply, is measured to develop a necrotic core once its radius exceeds approximately 200 μm — consistent with the classical estimate that oxygen can diffuse effectively only about 100–200 μm from a capillary before falling below the level needed to sustain aerobic metabolism. Given: a spheroid of radius r has surface area growing as r² but volume (and hence oxygen demand) growing as r³. Once r exceeds the diffusion limit d (~150–200 μm), the innermost cells (at distance > d from the surface) become hypoxic.

• At r = 100 μm: entire spheroid is within diffusion distance of the surface → fully oxygenated, no HIF-1α stabilization, minimal VEGF secretion, purely diffusion-limited (Folkman's original "avascular, dormant" tumor state).
• At r = 500 μm: a hypoxic core forms from roughly the center out to ~300 μm radius (500 − 200 = 300 μm inside the diffusion-competent shell) — HIF-1α accumulates there, VEGF-A is secreted, and if the tumor is in vivo (not in isolated culture) this VEGF gradient will recruit host vessels, converting the tumor from the "avascular" to the "angiogenic" phenotype Folkman described.
• Clinical implication: this is the biological rationale for anti-angiogenic therapy — a tumor that cannot recruit new vessels is mathematically constrained to remain below roughly a few mm³, explaining Folkman's original isolated-organ observations and motivating VEGF-pathway blockade as a strategy to keep micrometastases dormant.

Judah Folkman trained as a surgeon, not a molecular biologist, and his path to the angiogenesis hypothesis began with a frustratingly practical problem: while developing an implantable, slow-release hemoglobin substitute for the U.S. Navy in the 1960s, he noticed that tumor cells cultured alongside blood vessels grew explosively, while the same cells isolated from a vascular supply stalled at a tiny, stable size. Rather than treating this as a technical nuisance, Folkman built his entire subsequent career on the observation, publishing the hypothesis that tumor growth is fundamentally angiogenesis-dependent in 1971 — a claim so far ahead of the available molecular evidence that it was met with open skepticism from much of the cancer research establishment for over a decade, since no angiogenic factor had yet been isolated and some experiments failed to reproduce his tumor-factor extracts. Folkman persisted, building the assays (the corneal micropocket assay, the chick chorioallantoic membrane assay) that later let Ferrara's group and others actually purify VEGF and its family of receptors, and he lived to see his hypothesis validated by the 2004 FDA approval of bevacizumab (Avastin), the first anti-angiogenic cancer drug, built directly on the pathway his hypothesis predicted must exist.

Oncology. Anti-VEGF monoclonal antibodies (bevacizumab) and VEGFR tyrosine kinase inhibitors (sunitinib, sorafenib, regorafenib) are standard components of treatment regimens for colorectal, renal, ovarian, and several other cancers, typically combined with chemotherapy because anti-angiogenic monotherapy alone rarely cures disease — consistent with the vessel-normalization model rather than simple starvation. Resistance to anti-VEGF therapy, driven by tumor cells upregulating alternative angiogenic factors (FGF, PDGF, Ang-2), is an active area of drug combination research.

Ophthalmology. Intravitreal anti-VEGF injections (ranibizumab, aflibercept, bevacizumab off-label) have become first-line treatment for neovascular ("wet") age-related macular degeneration and diabetic macular edema, transforming these diseases from near-inevitable blindness to manageable chronic conditions — arguably the single most successful clinical translation of developmental angiogenesis biology to date.

Wound healing and regenerative medicine. Controlled, localized VEGF delivery (via engineered scaffolds or gene therapy) is being tested to accelerate healing of chronic diabetic ulcers and ischemic tissue, and pro-angiogenic strategies are central to engineering vascularized tissue constructs and organoids, since the stem cell niche of most organs depends critically on adjacent vasculature for oxygen, nutrients, and niche signals.

Cardiovascular disease. Therapeutic angiogenesis — delivering VEGF or related factors to stimulate collateral vessel growth around blocked coronary or peripheral arteries — has been trialed as an alternative or adjunct to surgical revascularization, though clinical efficacy has been more modest than initially hoped, partly because uncontrolled VEGF delivery reproduces the leaky, disorganized vessel phenotype seen in tumors rather than mature, functional collaterals.

Reproductive and developmental biology. Cyclical angiogenesis in the endometrium underlies normal menstrual physiology and placental development; dysregulated angiogenic balance (excess soluble Flt-1, a VEGF decoy receptor) is now understood to be central to the pathophysiology of pre-eclampsia. The same VEGF-HIF-Notch logic that builds embryonic vasculature will resurface, in a different guise, when germ line and gonadal development are considered — ovarian follicles and testes are themselves intensely angiogenesis-dependent organs, and the corpus luteum is one of the most rapidly vascularizing tissues in adult physiology.

Ask the tutor: How does the VEGF/Notch tip-stalk selection mechanism compare mechanistically to lateral inhibition in vulval precursor cell fate determination?