Quantitation and physiological characterization of angiogenic vessels in mice: effect of basic fibroblast growth factor, vascular endothelial growth factor/vascular permeability factor, and host microenvironment.

A prerequisite for the development of novel angiogenic and anti-angiogenic agents is the availability of routine in vivo assays that permit 1) repeated, long-term quantitation of angiogenesis and 2) physiological characterization of angiogenic vessels. We report here the development of such an assay in mice. Using this assay, we tested the hypothesis that the physiological properties of angiogenic vessels governed by the microenvironment and vessel origin rather than the initial angiogenic stimulus. Gels containing basic fibroblast growth factor (bFGF) or vascular endothelial growth (VEGF) were implanted in transparent windows in the dorsal skin or cranium of mice. Vessels could be continuously and non-invasively monitored and easily quantified for more than 5 weeks after gel implantation. Newly formed vessels were first visible on day 4 in the cranial window and day 10 in the dorsal skinfold chamber, respectively. The number of vessels was dependent on the dose of bFGF and VEGF. At 3000 ng/ml, bFGF- and VEGF-induced blood vessels had similar diameters, red blood cell velocities, and microvascular permeability to albumin. However, red blood cell velocities and microvascular permeability to albumin were higher in the cranial window than in the dorsal skinfold chamber. Leukocyte-endothelial interaction was nearly zero in both sites. Thus, newly grown microvessels resembled vessels of granulation and neoplastic tissue in many aspects. Their physiological properties were mainly determined by the microenvironment, whereas the initial angiogenic response was stimulated by growth factors.

Tumor necrosis factor alpha-induced leukocyte adhesion in normal and tumor vessels: effect of tumor type, transplantation site, and host strain.

Tumor necrosis factor alpha (TNF-alpha) can lead to tumor regression when injected locally or when used in an isolated limb perfusion, and it can enhance the tumoricidal effect of various therapies. TNF-alpha can also up-regulate adhesion molecules, and thus, facilitate the binding of leukocytes to normal vessels. The present study was designed to investigate the extent to which the host leukocytes roll and adhere to vessels of different tumors (MCaIV, a murine mammary adenocarcinoma; HGL21, a human malignant astrocytoma) at a given site or to the same tumor at different sites (dorsal skin and cranium), in different mouse strains [C3H and severe combined immunodeficient (SCID)], both with and without TNF-alpha-activation. There was no significant difference in hemodynamic parameters such as RBC velocity, diameter, or shear rate between PBS-treated control groups and corresponding TNF-alpha-treated groups. Under PBS control conditions, the leukocyte rolling count in MCaIV tumor vessels in the dorsal chamber in C3H and SCID mice and in the cranial window in C3H mice was significantly lower than that in normal vessels (P < 0.05), but stable cell adhesion was similar between normal and tumor vessels. TNF-alpha led to an increase (P < 0.05) in leukocyte-endothelial interaction in vessels in the following cases: normal tissue regardless of sites and strains, MCaIV tumor in the cranial window in C3H mice, and HGL21 tumor in the cranial window in SCID mice. However, the increase in rolling and adhesion in the MCaIV tumor in response to TNF-alpha was significantly lower than in the corresponding normal vessels (P < 0.05) in the dorsal chamber in C3H and SCID mice and in the cranial window in C3H mice. The HGL21 tumor in the cranial window in SCID mice showed leukocyte rolling and adhesion comparable to that in normal pial vessels. These findings suggest that (a) in general, basal leukocyte rolling is lower in tumor vessels than in normal vessels; (b) leukocyte rolling and adhesion in tumors can be enhanced by TNF-alpha-mediated activation; and (c) the TNF-alpha response is dependent on tumor type, transplantation site, and host strain. These results have significant implications in the gene therapy of cancer using TNF-alpha-gene-transfected cancer cells or lymphocytes.

Interaction of activated natural killer cells with normal and tumor vessels in cranial windows in mice.

A mammary carcinoma, MCa IV, was grown in syngeneic C3H mice in a cranial window preparation which permitted the in vivo observation of the growth and microcirculation of the tumors. Fluorescently labeled activated natural killer (A-NK) cells were injected into the external carotid artery and their interactions with normal and tumor vessels were quantified by video microscopy. Cells which entered the tumor vessels adhered heterogeneously to these vessels, regardless of vessel size or blood flow rates and bound with an efficiency ranging from 0 to 82% of the incoming cell flux. Normal brain tissue showed significantly fewer binding cells per microscopic field (9 +/- 5 vs 85 +/- 27 cells/1.3 mm2) and the few cells which were retained by the normal tissue were highly deformed, suggesting mechanical rather than adhesive entrapment. These studies indicate that A-NK cells bind in high numbers to segments of the vessels of mammary tumors growing in an intracranial site when administered through an arterial route; however, some tumor vessels may escape recognition by these cells. These findings suggest that A-NK cells may be used as carriers of genes for anti-cancer agents.

Vascular permeability and microcirculation of gliomas and mammary carcinomas transplanted in rat and mouse cranial windows.

Many brain tumors are highly resistant to chemotherapy, presumably due to the presence of a tight blood-tumor barrier. For a better understanding of the regulation of this barrier by the brain environment, a new intravital microscopy model was established by transplanting tumor tissue into cranial windows in both rats and mice. The model was characterized by RBC velocities, vessel diameters, and vascular permeabilities of various tumors: R3230AC (a rat mammary adenocarcinoma), MCaIV (a mouse mammary adenocarcinoma), and U87 and HGL21 (human malignant astrocytomas). Our results showed that tumor blood flow in cranial windows was one to three orders of magnitude lower than the blood flow in pial vessels and similar to that in dorsal skin-fold chambers observed in previous studies. The mean vessel diameter ranged from 6.8 +/- 1.3 microns for HGL21 to 30.4 +/- 8.5 microns for MCaIV. At least one order of magnitude difference in vascular permeability to albumin was observed between tumor lines: 0.11 +/- 0.05 x 10(-7) cm/s for HGL21 versus 3.8 +/- 1.2 x 10(-7) cm/s for U87. The low vascular permeability of HGL21, which was also confirmed by both sodium fluorescein and Lissamine green injections, suggests that not all tumors are leaky to tracer molecules and that the blood-tumor barrier of this tumor still possesses some characteristics of blood-brain barrier as observed in other intracranial tumors. The model presented here will allow us to manipulate the vascular permeability in brain tumors and thus may provide new information on the regulation of the blood-tumor barrier and new strategies for improving drug delivery in brain tumors.