Differences in pericyte contractile function in rat cardiac and skeletal muscle microvasculatures

In order to evaluate contractile activity of pericytes, rat hearts and hindlimbs were perfused separately with angiotensin (1 μg/ml), norepinephrine (1 μg/ml), or vasopressin (10⁻³ U/ml) for 3 min, fixed by perfusion, then processed for electron microscopy and morphometry. An electronic planimeter was employed to quantify configurational alterations (buckling) of endothelium between pericyte processes which provided an index of pericyte contraction (PCI). Inherent in this technique is the assumption that contracting pericytes distort and convolute endothelial cell membranes (increasing the PCI) abutting pericytes, but do not affect endothelial configuration elsewhere around the capillary circumference. PCI values were virtually identical in control hearts and skeletal muscle and in low-flow hindlimb perfusions (101.2 ± 0.3% (SD), 101.2 ± 0.7%, and 102.3 ± 0.8%, respectively). All three vasoactive agents increased perfusion pressure significantly in both hearts and hindlimbs with one exception (norepinephrine produced a slight pressure drop in hearts). While perfusion with vasoactive agents had no effect on the cardiac PCI, skeletal PCI values were markedly elevated for all three vasoactive agents (124.1 ± 8.2, 118.7 ± 4.9, and 120.2 ± 6.8% for angiotensin, norepinephrine, and vasopressin, respectively). Marked buckling of endothelium in apposition with pericytes and the absence of such changes elsewhere in the vessel wall were documented in morphologic studies of drug-perfused skeletal muscle. In both control hearts and hindlimbs and drug-perfused hearts, endothelial configuration in apposition with pericytes did not differ from the rest of the vessel. These observations, together with ultrastructural and morphometric data documenting much more extensive interaction between pericytes and endothelium in skeletal than in cardiac muscle, strongly suggest that pericytes in rat hindlimb skeletal muscle constrict in response to selected vasoactive agents while those in cardiac muscle do not.     

Analysis of relationships between pericytes and gas exchange capillaries in neonatal and mature bovine lungs

Neonatal and mature bovine lungs were examined ultrastructurally to quantitatively assess pericyte envelopment of gas exchange capillaries and proximities of pericyte margins to endothelial cell junctions. Pericytes were observed on 91% of the cross-sectioned capillary profiles examined, with 18 and 26% coverage in neonatal and mature lungs, respectively. Chi-square analysis indicated a random occurrence of endothelial cell junctions under pericytes; however, pericyte processes tended to end near endothelial cell junctions. In the neonatal and mature lungs, 38 and 40%, respectively, of all endothelial junctions were within 0.5 μm of the margins of pericyte processes; such distances covered only 16 and 17% of the circumferences of the capillary profiles examined. Thus, endothelial cell junctions were located near pericyte margins over twice as often as would occur in a random distribution. If pericytes are contractile cells, as recent research indicates, they may function as regulators of lymph formation by influencing the permeability of endothelial cell junctions in the gas exchange capillaries.     

Differential arterial/venous expression of NG2 proteoglycan in perivascular cells along microvessels: identifying a venule-specific phenotype

OBJECTIVE: Similar to other vascular pericyte markers, including smooth muscle (SM) alpha-actin, desmin, and PDGF-beta-receptor, NG2 proteoglycan is not pericyte specific. Therefore, the use of NG2 as a pericyte marker, especially in cell lineage studies, in comparison to other nonspecific pericyte markers requires an understanding of how its expression varies spatially within a microvascular network. The objective of this study was to characterize NG2 expression along vessels within rat microvascular networks and compare this to SM alpha-actin expression. METHODS: Mesenteric tissue, subcutaneous tissue, spinotrapezius muscle, and gracilis muscle were harvested from 250-g, female, Sprague-Dawley rats and stained for NG2 and SM alpha-actin. The distribution of NG2 expression was evaluated in mesenteric networks (n = 28) with complementary observations in subcutaneous tissue and skeletal muscle. RESULTS: Perivascular cells, including mature smooth muscle cells (SMCs), immature SMCs, and pericytes, expressed NG2. Most importantly, NG2 expression was primarily confined to perivascular cells along arterioles and capillaries, and continuous expression was not observed along venules beyond the immediate postcapillary vessels. The differential expression of NG2 along the arteriolar side of microvascular networks was also observed in rat subcutaneous and skeletal muscle. CONCLUSIONS: The results indicate that NG2 is expressed by all perivascular cells along arterioles, and its absence denotes a venule-specific phenotype. These results identify for the first time a marker that differentiates venous smooth muscle and pericytes from other capillary- and arteriole-associated perivascular cells.     

Characteristics of lung pericytes in culture including their growth inhibition by endothelial substrate

Pericytes and endothelial cells from the same sample of adult rat lung have been separately established in culture by use of selective growth media. The endothelial cells are positive and the pericytes negative for angiotensin-converting enzyme activity and tissue plasminogen activator. Morphologically in culture, the pericytes are similar to pericytes from bovine retina and other sites and show positive immunofluorescence to both human platelet (non-muscle) myosin and smooth muscle myosin. In this respect they resemble smooth muscle cells grown from the rat main pulmonary artery, but lack the myofilaments and dense bodies characteristic of muscle cells. Lung endothelial cells and fibroblasts are positive only for platelet myosin. Pericytes in culture demonstrate an unusual growth response to endothelial substrate, obtained by removing confluent endothelial monolayers with nonionic detergent or alkali. When plated onto this material at low density, pericyte growth is inhibited. By contrast, the substrate stimulates the growth of endothelial cells and has no effect on smooth muscle cells. Initial attachment of endothelial cells and pericytes to the substrate is similar.     

Lack of type XV collagen causes a skeletal myopathy and cardiovascular defects in mice

Type XV collagen occurs widely in the basement membrane zones of tissues, but its function is unknown. To understand the biological role of this protein, a null mutation in the Col15a1 gene was introduced into the germ line of mice. Despite the complete lack of type XV collagen, the mutant mice developed and reproduced normally, and they were indistinguishable from their wild-type littermates. However, Col15a1-deficient mice showed progressive histological changes characteristic for muscular diseases after 3 months of age, and they were more vulnerable than controls to exercise-induced muscle injury. Despite the antiangiogenic role of type XV collagen-derived endostatin, the development of the vasculature appeared normal in the null mice. Nevertheless, ultrastructural analyses revealed collapsed capillaries and endothelial cell degeneration in the heart and skeletal muscle. Furthermore, perfused hearts showed a diminished inotropic response, and exercise resulted in cardiac injury, changes that mimic early or mild heart disease. Thus, type XV collagen appears to function as a structural component needed to stabilize skeletal muscle cells and microvessels.     

Associations between pericytes and capillary endothelium in the eel rete mirabile

Morphometric analysis of electron micrographs of the eel rete mirabile revealed that pericytes occupied nearly one-third of the cellular volume of this organ with over 75% of the pericyte volume associated with arterially derived capillaries. These pericytes were highly arborized, extending processes which encircled the capillaries and covering, on average, more than 85% the ablumenal surface of arterially derived capillaries. Pericytes and endothelial cells interacted in a manner suggesting that pericyte activity modulates both capillary blood flow and permeability.     

The source of epifluorescence in isolated perfused heart loaded with fura 2-AM or indo 1-AM

Summary The purpose of this study is to clarify the kinds of cells that fluoresce in isolated perfused heart intracoronarily loaded with fura 2-AM or indo 1-AM. The hearts from hamsters perfused by the Langendorff method were loaded with fura 2-AM or indo 1-AM intracoronarily. The epicardium was excited by ultraviolet light at a wavelength of 340 or 380 nm (fura 2) or 360 nm (indo 1). Epifluorescent images from arrested hearts prepared with St Thomas' cardioplegic solution were observed through an inverted microscope and taken by a still camera. Endothelial cells and pericytes of the capillaries showed intense fluorescence. In the intercapillary spaces, a less intense but still apparent fluorescence emitted by myocytes could also be detected. On the other hand, fluorescence from smooth muscle cells and fibroblasts could not be identified. Therefore, epifluorescence in the isolated perfused heart loaded with fura 2-AM or indo 1-AM was mainly generated within endothelial cells, pericytes, and myocytes. To evaluate calcium fluorescence in isolated hearts quantitatively, the fluorescence of endothelial cells and pericytes of the capillaries must be taken into consideration.     

Immunofluorescent imaging of capillaries and pericytes in human skeletal muscle and retina

A new technique for visualizing small vessels and pericytes in unsectioned tissues has been developed based on immunofluorescent staining of vascular basement membranes. Following brief fixation, tissue from human skeletal muscle and retina is gently teased apart with fine forceps and then incubated sequentially with: (1) rabibit antibody directed against purified human glomerular basement membrane and (2) goat anti-rabbit immunoglobulin tagged with fluorescein isothiocyanate. The tissue is then examined by fluorescence microscopy using incident light. In skeletal muscle, branching vascular networks and loops are seen overlying lightly stained muscle fibers; pericyte cell bodies and their processes are visualized within the vascular basement membrane. Pericyte processes demonstrate considerable variation in shape and frequently form bridges between neighboring capillaries. Pericytes in retinal vascular networks, obtained without trypsin digestion, are more uniform in shape and in closer apposition to vessels than those in skeletal muscle. Pericyte processes and branching patterns are not evident in retinal vessels even at high magnification. The technique described provides a simple method for light microscopic visualization of the microvasculature with preservation of three dimensional relationships that otherwise could only be appreciated by tedious reconstruction of tissue sections.     

Cerebral cortical capillary basement membrane thickening in galactosaemic rats

Summary Wistar-Kyoto rats fed a diet containing 30% by weight galactose for 15–21 months developed significant thickening of the endothelial basement membranes of capillaries from the frontal cortex of the cerebrum, by comparison with cerebral capillary basement membranes from animals on a standard diet (p<0.001), or animals receiving a diet containing 30% galactose together with 250 mg/kg diet of the aldose reductase inhibitor, Sorbinil (0.001<p<0.01). The effect was similar to that which we have reported previously in the retinal capillaries of these animals. Spontaneously hypertensive rats on the high-galactose diet showed modest cerebral capillary basement membrane thickening (0.02<p<0.05) only for one of the measurement protocols utilised, and the process was not prevented by Sorbinil. Biochemical assays of retina, cerebral cortex, and blood serum from Wistar-Kyoto and spontaneously hypertensive rats maintained on the Sorbinil regimen showed that the drug did cross the blood-retinal and blood-brain barriers. Similar to our previous study on the retinal capillaries, we observed no degeneration of pericyte or endothelial cell cytoplasm, and no alteration in the pericyte/endothelial cell nuclear ratio in the cerebral capillaries of galactosaemic animals, by comparison with controls. Based on immunocytochemical studies in the human retina, it has been claimed that aldose reductase is present in capillary pericytes but absent in the endothelial cells. However, we observe a considerably smaller pericyte/ endothelial cell nuclear ratio in the capillaries of the cerebral cortex of the rat, by comparison with those of the retina. Also, pericyte coverage of the cerebral cortical capillaries is much less than that of the retinal capillaries of these animals. Therefore, it appears that the biochemical process(es) responsible for basement membrane thickening are unlikely to reside within the pericytes.     

Epidermal growth factor receptor at endothelial cell and pericyte interdigitation in human granulation tissue.

Angiogenic immature capillaries in human granulation tissue possess many cytoplasmic interdigitations between endothelial cells and pericytes (CIDEP). Epidermal growth factor (EGF) is a potent mitogenic polypeptide which accelerates angiogenesis in vivo and in vitro. We have recently demonstrated in immature capillaries that EGF is present at the CIDEP and proposed that the CIDEP may be involved in a signaling pathway for EGF (Wakui et al., 1990c. Microvasc. Res. 40, 285-291). This study follows that previous report. In the present study, I have investigated the ultrastructural localization of the EGF receptor (EGF-r) at CIDEP. Immunoreactivity for the EGF-r in immature capillaries was located strictly at the CIDEP in large numbers, at the coated pits and vesicles of endothelial cells, and at some lysosome-like structures of the endothelial cells and the pericytes. On the other hand, immunoreactivity for EGF-r was absent in mature capillaries. At the CIDEP, EGF-r immunoreactivity was present at the cell membrane and a few cytoplasmic elements at the tip of pericyte cytoplasmic projections, but it was completely absent at the corresponding endothelial membrane indentation. The present results support our proposed hypothesis that the CIDEP in immature capillaries act as a pathway for EGF transportation from the endothelial cell to the pericyte by a receptor-mediated process.     

On pericytes, particularly their existence on lung capillaries

The existence of pericytes on the capillaries of lung alveoli has long been questioned. A systematic electron microscopic study has shown them to exist in human lungs, as well as in lungs of dogs, guinea pigs, and rats, whereas they could not be found in the lungs of the smallest mammal, the Etruscan shrew. The study of systemic capillaries (myocardium, pancreas) shows the pericyte to be characterised by a close association with the capillary basement membrane within which it is ensheathed, by the formation of branched cytoplasmic processes which approach the endothelial cell more and more the finer they become, by the presence of a web of cytoplasmic filaments along the membrane close to the endothelium, and by the exclusive or predominant occurrence of pinocytosis along the outer cell membrane; lysosomal elements are a rare occurrence. These features are comparable to those characterising smooth muscle cell processes in venules. Several cells and numerous cell processes showing all these characteristics have been found associated with lung capillaries. It is concluded that pulmonary alveolar capillaries are also provided with pericytes, but that these appear to be rarer than on systemic capillaries and that they are less densely branched. Their frequency furthermore appears to be proportional to body and lung size.     

Endosialin/TEM 1/CD248 is a pericyte marker of embryonic and tumor neovascularization

The formation of functional, mature blood vessels depends on the interaction between endothelial cells and pericytes. Commonality exists in the processes involved in vasculature development between tissues whether healthy or diseased. Endosialin/TEM 1 is a cell membrane protein that is expressed in blood vessels during embryogenesis and tumorigenesis but not in normal mature vessels. Antibodies developed to human endosialin were used to investigate endosialin expression and function in human prenatal brain pericytes and pericytes residing in tumors. Anti-endosialin was capable of preventing pericyte tube formation in culture and inhibited migration. Brain pericytes in culture had higher levels of endosialin/TEM 1 than TEMs-2, -3, -4, -5, -7, and -8. Immunocytochemistry revealed that endosialin was present in the cytoplasmic body and in the elongated extensions essential to pericyte function. Transgenic mice engineered to express human endosialin bred on an immunocompromised background allowed the growth of human tumor xenografts. In human colon carcinoma Colo205 and HT29 xenografts grown in human endosialin-transgenic mice, endosialin expression was largely confined to NG2-expressing perivascular cells and not CD31-positive endothelial cells. Similar methods applied to human ovarian and colon tumors confirmed endosialin expression by pericytes. The data indicate that endosialin is strongly expressed by pericytes during periods of active angiogenesis during embryonic and tumor development. Anti-endosialin antibodies may have value in identifying vasculature in malignant tissues. With the appropriate agent, targeting endosialin may interfere with blood vessel growth during tumor development.     

Endosialin/TEM 1/CD248 is a pericyte marker of embryonic and tumor neovascularization

The formation of functional, mature blood vessels depends on the interaction between endothelial cells and pericytes. Commonality exists in the processes involved in vasculature development between tissues whether healthy or diseased. Endosialin/TEM 1 is a cell membrane protein that is expressed in blood vessels during embryogenesis and tumorigenesis but not in normal mature vessels. Antibodies developed to human endosialin were used to investigate endosialin expression and function in human prenatal brain pericytes and pericytes residing in tumors. Anti-endosialin was capable of preventing pericyte tube formation in culture and inhibited migration. Brain pericytes in culture had higher levels of endosialin/TEM 1 than TEMs-2, -3, -4, -5, -7, and -8. Immunocytochemistry revealed that endosialin was present in the cytoplasmic body and in the elongated extensions essential to pericyte function. Transgenic mice engineered to express human endosialin bred on an immunocompromised background allowed the growth of human tumor xenografts. In human colon carcinoma Colo205 and HT29 xenografts grown in human endosialin-transgenic mice, endosialin expression was largely confined to NG2-expressing perivascular cells and not CD31-positive endothelial cells. Similar methods applied to human ovarian and colon tumors confirmed endosialin expression by pericytes. The data indicate that endosialin is strongly expressed by pericytes during periods of active angiogenesis during embryonic and tumor development. Anti-endosialin antibodies may have value in identifying vasculature in malignant tissues. With the appropriate agent, targeting endosialin may interfere with blood vessel growth during tumor development.     

HMG-CoA reductase inhibitors induce apoptosis in pericytes

Pericytes, which surround endothelial cells in precapillary arterioles, capillaries, and postcapillary venules, are important for the development, maturation, and maintenance of the vascular system. Pericytes are also pluripotent cells that can differentiate into a variety of mesenchymal cells including smooth muscle cells and osteoblasts. Possibly because of their vasculature regulating activities and ability to differentiate in situ, pericytes are implicated in several diseases with vascular complications, including diabetic retinopathy, as well as Reynaud's Syndrome, central nervous system dementias, and vascular calcification among others. Statin drugs, which block the conversion of HMG-CoA to mevalonate in the cholesterol synthesis pathway, are known to have apoptotic and growth inhibitory effects on cells in vitro and complex pleiotropic effects on cells and tissues in vivo. Recently, evidence has emerged that statin drug use in human patients results in a significant 20% reduction in cancer incidence. It is not known whether these results are due to direct statin action on normal tissue, growth inhibitory/pro-apoptotic effects on tumor cells, and/or effects on angiogenesis. Because of the role of pericytes in angiogenesis and the effects of statins on cancer incidence, we tested the direct effects of statins on pericytes. Specifically, we demonstrate that 3 statins, simvastatin, lovastatin, and mevastatin induce dose-dependent apoptosis in the TR-PCT1 pericyte cell line, that simvastatin (empirically shown to be the most potent of the 3 statins) induces similar levels of apoptosis in freshly isolated pericytes, and that simvastatin-induced apoptosis in pericytes is cholesterol, caspase-3, and caspase-7 mediated.     

Transcapillary permeability and subendothelial distribution of endothelial and amniotic fluid insulin-like growth factor binding proteins in the rat heart.

Insulin-like growth factor (IGF) binding proteins (IGFBP) were purified from conditioned media of cultured bovine endothelial cells (ECBP) and from human amniotic fluid (IGFBP-1), and then labeled by radioiodination. 125I-ECBP and 125I-IGFBP-1 were perfused through isolated beating rat hearts for 1 and 5 min, and the hearts fixed and analyzed for 125I-BP content and distribution. One to 4% of the perfused 125I-ECBP and 125I-IGFBP-1 crossed the capillary boundary. The ECBPs predominantly localized as intact 125I-BP in connective tissue elements of the heart with less 125I-BP in cardiac muscle. The ratio of 125I-ECBP in connective tissue: muscle (normalized to percent vol of these compartments) was greater than or equal to 10:1. In contrast, the IGFBP-1 had a greater affinity for cardiac muscle with ratios of 125I-IGFBP-1 in connective tissue:muscle of approximately 1:2. When 125I-IGF-I, in the absence of any BPs, was perfused through the hearts approximately 3-5% left the microcirculation and was found in subendothelial tissues. 125I-IGF-I localized primarily to cardiac muscle with a distribution of connective tissue:cardiac muscle of about 1:3. The findings in the isolated perfused heart were confirmed in intact animals. After 125I-IGFBP-1 was injected into anesthetized rats and allowed to circulate for 5 min, substantial radioactivity was associated with the heart. As in the isolated heart, the IGFBP-1 preferentially localized to cardiac muscle with a connective tissue:cardiac muscle ratio of 1:3. We conclude that IGFBPs produced by endothelial cells and the IGFBP-1 contained in amniotic fluid can cross the capillary boundaries of the rat heart, and that the ECBPs preferentially localize in connective tissue elements of the myocardium, whereas IGFBP-1 predominantly localizes in cardiac muscle.     

Increased expression of biglycan mRNA in pressure-overloaded rat heart.

Biglycan mRNA expression in rat myocardium after abdominal aortic banding with renal ischemia was examined. The Northern blot analysis demonstrated that expression of biglycan mRNA in the pressure-overloaded hearts on days 2, 7, 14 and 28 was 2.88 +/- 0.89, 2.32 +/- 0.49, 2.17 +/- 0.57 and 1.81 +/- 0.46-fold higher, respectively, than that in the sham-operated hearts. In situ hybridization showed an increased density of biglycan mRNA signal-positive cells in the pressure-overloaded hearts. The cells with positive signals were spindle-shaped mesenchymal cells in the myocardial interstitium. A marked increase in biglycan mRNA signal expression was also observed in endothelial cells and smooth muscle cells of the thickened myocardial capillary wall. These results demonstrated an increase in biglycan mRNA in the pressure-overloaded heart in mesenchymal cells in the myocardial interstitium, and in endothelial and smooth muscle cells of the capillaries, indicating that biglycan contributes to the ventricular and vascular remodeling in response to pressure overload.     

Angiogenesis in the corpus luteum

The ovarian corpus luteum plays a critical role in reproduction because it is the primary source of circulating progesterone. After ovulation, as the corpus luteum forms from the wall of the ruptured follicle, it grows and vascularizes extremely rapidly. In fact, the rates of tissue growth and angiogenesis in the corpus luteum rival those of even the fastest growing tumors. Thus, the corpus luteum provides an outstanding model for studying the factors that regulate the angiogenic process, which is critical for normal tissue growth, development, and function. In agreement with data from other tissues, vascular endothelial growth factors (VEGF) seem to be a major angiogenic factor responsible for vascularization of the developing corpus luteum. Recent data suggest that luteal expression of VEGF occurs primarily in specific perivascular cells, including arteriolar smooth muscle and capillary pericytes, and is regulated primarily by oxygen levels. In addition, soon after ovulation, pericytes derived from the thecal compartment appear to be the first vascular cells to invade the developing luteal parenchyma. The granulosa-derived cells produce a factor that stimulates pericyte migration. Moreover, nitric oxide (NO), which is a potent vasodilator and can stimulate VEGF production and angiogenesis, is expressed in endothelial cells of luteal arterioles and capillaries, often in association with expression of VEGF by luteal perivascular cells. Thus, we have proposed a model for the initial process of luteal vascularization in which hypoxia plays a major role. In this model, which we believe will apply to other tissues as well, a paracrine loop exists between the vascular endothelial cells, which produce NO, and the peri-endothelial cells (vascular smooth muscle and pericytes), which produce VEGF, to ensure coordinate regulation of luteal vasodilation and angiogenesis.     

NG2 proteoglycan promotes tumor vascularization via integrin-dependent effects on pericyte function

The NG2 proteoglycan stimulates the proliferation and migration of various immature cell types, including pericytes. However, the role of NG2 in mediating pericyte/endothelial cell interaction has been less clear. In this study, we show that pericyte-specific NG2 ablation causes several structural deficits in blood vessels in intracranial B16F10 melanomas, including decreased pericyte ensheathment of endothelial cells, diminished formation of endothelial junctions, and reduced assembly of the vascular basal lamina. These deficits result in decreased tumor vessel patency, increased vessel leakiness, and increased intratumoral hypoxia. NG2-dependent mechanisms of pericyte interaction with endothelial cells are further explored in pericyte/endothelial cell co-cultures. siRNA-mediated NG2 knockdown in pericytes leads to reduced formation of pericyte/endothelial networks, reduced formation of ZO-1 positive endothelial cell junctions, and increased permeability of endothelial cell monolayers. We also show that NG2 knockdown results in loss of β1 integrin activation in endothelial cells, revealing a mechanism for NG2-dependent cross talk between pericytes and endothelial cells.     

INCREASED EXPRESSION OF BIGLYCAN mRNA IN PRESSURE-OVERLOADED RAT HEART

Biglycan mRNA expression in rat myocardium after abdominal aortic banding with renal ischemia was examined. The Northern blot analysis demonstrated that expression of biglycan mRNA in the pressure-overloaded hearts on days 2, 7, 14 and 28 was 2.88 ± 0.89, 2.32 ± 0.49, 2.17 ± 0.57 and 1.81 ± 0.46-fold higher, respectively, than that in the sham-operated hearts. In situ hybridization showed an increased density of biglycan mRNA signal-positive cells in the pressure-overloaded hearts. The cells with positive signals were spindle-shaped mesenchymal cells in the myocardial interstitium. A marked increase in biglycan mRNA signal expression was also observed in endothelial cells and smooth muscle cells of the thickened myocardial capillary wall. These results demonstrated an increase in biglycan mRNA in the pressure-overloaded heart in mesenchymal cells in the myocardial interstitium, and in endothelial and smooth muscle cells of the capillaries, indicating that biglycan contributes to the ventricular and vascular remodeling in response to pressure overload.     

Early Contribution of Pericytes to Angiogenic Sprouting and Tube Formation

Immunostaining with endothelial and pericyte markers was used to evaluate the cellular composition of angiogenic sprouts in several types of tumors and in the developing retina. Confocal microscopy revealed that, in addition to conventional endothelial tubes heavily invested by pericytes, all tissues contained small populations of endothelium-free pericyte tubes in which nerve/glial antigen 2 (NG2) positive, platelet-derived growth factor beta (PDGF beta ) receptor-positive perivascular cells formed the lumen of the microvessel. Perfusion of tumor-bearing mice with FITC-dextran, followed by immunohistochemical staining of tumor vasculature, demonstrated direct apposition of pericytes to FITC-dextran in the lumen, confirming functional connection of the pericyte tube to the circulation. Transplantation of prostate and mammary tumor fragments into NG2-null mice led to the formation of tumor microvasculature that was invariably NG2-negative, demonstrating that pericytes associated with tumor microvessels are derived from the host rather than from the conversion of tumor cells to a pericyte phenotype. The existence of pericyte tubes reflects the early participation of pericytes in the process of angiogenic sprouting. The ability to study these precocious contributions of pericytes to neovascularization depends heavily on the use of NG2 and PDGF beta -receptor as reliable early markers for activated pericytes.