Vascular endothelial cells: Targets for studying the activity of hair follicle cell-produced VEGF

Fetal bovine aortic endothelial cells (FBAEC) were exposed to purified fractions of conditioned medium from cultures of hair dermal papilla cells (DPC) to determine the existence of any vascular endothelial growth factor (VEGF)-like paracrine activity of the latter. Such fractions were tested for stimulation of growth and migration of cultured FBAEC. In addition, VEGF secretion by DPC was measured by radioassay of VEGF receptors using FBAEC as target cells. The results showed that stimulation of FBAEC proliferation and migration following exposure to purified conditioned medium was dose-dependent. Radioreceptor assays of recombinant VEGF and purified DPC-conditioned medium showed competitive VEGF binding in FBAEC.Abbreviations CM conditioned medium - DMEM Dulbecco's modified eagle's medium - DPC dermal papilla cells - EDTA ethylenediaminetetra-acetic acid - FBAEC fetal bovine aortic endothelial cells - FCS fetal calf serum - VEGF vascular endothelial growth factor     

Human placental multipotent mesenchymal stromal cells modulate placenta angiogenesis through Slit2-Robo signaling

abstract

The objective of this study was to investigate whether human placental multipotent mesenchymal stromal cell (hPMSC)-derived Slit2 and endothelial cell Roundabout (Robo) receptors are involved in placental angiogenesis. The hPMSC-conditioned medium and human umbilical vein endothelial cells were studied for Slit2 and Robo receptor expression by immunoassay and RT-PCR. The effect of the conditioned medium of hPMSCs with or without Slit2 depletion on endothelial cells was investigated by in vitro angiogenesis using growth factor-reduced Matrigel. hPMSCs express Slit2 and both Robo1 and Robo4 are present in human umbilical vein endothelial cells. Human umbilical vein endothelial cells do not express Robo2 and Robo3. The hPMSC-conditioned medium and Slit2 recombinant protein significantly inhibit the endothelial cell migration, but not by the hPMSC-conditioned medium with Slit2 depletion. The hPMSC-conditioned medium and Slit2 significantly enhance endothelial tube formation with increased cumulated tube length, polygonal network number and vessel branching point number compared to endothelial cells alone. The tube formation is inhibited by the depletion of Slit2 from the conditioned medium, or following the expression of Robo1, Robo4, and both receptor knockdown using small interfering RNA. Furthermore, co-immunoprecipitation reveals Slit2 binds to Robo1 and Robo4. Robo1 interacts and forms a heterodimeric complex with Robo4. These results suggest the implication of both Robo receptors with Slit2 signaling, which is involved in endothelial cell angiogenesis. Slit2 in the conditioned medium of hPMSCs has functional effect on endothelial cells and may play a role in placental angiogenesis.     

Molecular cloning and characterization of prolactin-like protein C complementary deoxyribonucleic acid.

In this report, we describe the isolation and characterization of a full length cDNA clone for rat prolactin-like protein C (PLP-C) and describe the expression of PLP-C mRNA in the developing rat placenta. Nucleotide sequence analysis of the PLP-C cDNA clone predicted a mature protein of 238 amino acids, including a 30-amino acid signal sequence. The predicted PLP-C amino acid sequence contains seven cysteine residues, three tryptophan residues, and two putative N-linked glycosylation sites. Six of the cysteine residues in PLP-C are located in positions homologous to the cysteines of pituitary prolactin (PRL). Additional sequence similarities with pituitary PRL and other members of the rat placental PRL family are evident. The PLP-C gene was localized to rat chromosome 17. Northern blot analysis showed that the PLP-C cDNA clone specifically hybridized to a 1.0-kilobase mRNA. PLP-C mRNA was first detectable between days 13 and 14 of gestation, peaked by day 18 of gestation, and remained elevated until term. In situ hybridization analysis indicated that PLP-C mRNA was specifically expressed by spongiotrophoblast cells and some trophoblast giant cells in the junctional zone region of rat chorioallantoic placenta.     

Glycosylation of vascular endothelial growth factor is not required for its mitogenic activity.

We have stably expressed the cDNA encoding the 165 amino-acid long form of human vascular endothelial growth factor (VEGF) in BHK-21 cells. VEGF was partially purified from the conditioned medium of transfected cells using heparin-sepharose affinity chromatography. The partially purified VEGF was mitogenic for various types of endothelial cells and inhibited the binding of pure [125I]VEGF to its receptors. Western blot analysis, using anti-VEGF antibodies, revealed a 47 kDa VEGF homodimer in the partially purified VEGF fraction. Preincubation of the transfected cells with the N-glycosylation inhibitor tunicamycin resulted in the conversion of the 47 kDa VEGF homodimer into a smaller, deglycosylated form of 42 kDa. Partially purified preparations of the deglycosylated VEGF displayed a mitogenic activity that was similar to that of the glycosylated form and efficiently inhibited the binding of native [125I]VEGF to the VEGF receptors of bovine aortic arch derived endothelial cells.     

Comparative in vitro effects of guinea pig VIP and common VIP on liver and lung membranes from guinea pig and rat and on human lymphoblastic SUP-T1 membranes

Guinea pig VIP differs from VIP of several mammals by its amino acids in positions 5, 9, 19 and 26. We tested a) its ability to occupy VIP receptors in liver and lung membranes of rat and guinea pig and in the human lymphoblastic SUP-T1 cell line and b) the ensuing adenylate cyclase stimulation. In liver and lung membranes from rat, guinea pig VIP was less potent than common VIP to occupy high and low affinity VIP receptors. In rat liver both VIP activated adenylate cyclase mostly through high affinity receptors. In rat lung, guinea pig VIP activated the enzyme mostly through high affinity receptors and was less efficient than common VIP acting through both classes of receptors. In guinea pig liver and lung membranes, binding inhibition curves were steeper than with rat preparations and adenylate cyclase appeared to be mostly activated through high affinity VIP receptors in liver and through both classes of receptors in lung. On human lymphoblastic SUP-T1 membranes both VIP were equally potent and efficient to inhibit tracer binding and activate adenylate cyclase.     

Expression, purification and characterization of recombinant human insulin-like growth factor I in yeast

Insulin-like growth factor I (IGF-I) is a 70 amino acid (aa) protein that is structurally similar and functionally related to insulin. We have inserted a synthetic gene coding for human IGF-I into a Saccharomyces cerevisiae expression vector utilizing the MF alpha 1 promoter and pre-pro leader peptide. This vector directs the expression and secretion of native, biologically active growth factor. Cleavage of the pre-pro alpha factor leader sequence in vivo results in the secretion of a 70-aa recombinant IGF-I molecule with the native N-terminal glycine residue. Human IGF-I purified from yeast culture supernatant is equipotent to serum-derived IGF-I in inhibiting [125I]IGF-I binding to type-I IGF receptors and crude human serum-binding proteins. Recombinant IGF-I is also equipotent to human IGF-I in the stimulation of DNA synthesis in rat aortic smooth-muscle cells. In contrast, yeast recombinant IGF-I is less potent than serum-derived IGF-I in binding to type-2 IGF receptors. The ability to produce native, biologically active IGF-I in yeast will allow the elucidation of binding domains through the expression and characterization of specific structural analogs.     

Characterization of a Mr 20,000 basic fibroblast growth factor-like protein secreted by normal and transformed fetal bovine aortic endothelial cells

A mitogenic and plasminogen activator (PA)-inducing activity for endothelial cells has been identified in serum-free culture medium of normal AG 7680 and transformed tumorigenic GM 7373 fetal bovine aortic endothelial (FBAE) cells. The activity binds to heparin-Sepharose and it is quenched by polyclonal anti-human placental basic fibroblast growth factor (bFGF) antibodies. In the serum-free conditioned medium of FBAE cells, the anti-bFGF antiserum recognizes an immunorective Mr 20,000 molecule which co-purifies with the mitogenic and PA-inducing activity on a heparin-Sepharose column. The partially purified Mr 20,000 bFGF-like molecule competes with the typical Mr 18,000 125I-bFGF form for the binding to high-affinity bFGF receptors in intact GM 7373 cells. Immunoprecipitation of biosynthetically labeled GM 7373 cells with anti-bFGF antiserum confirms the presence of a Mr 20,000 bFGF-like molecule in the conditioned medium of these cells and identifies the typical Mr 16,000 and Mr 18,000 bFGF forms and two high-molecular-weight immunoreactive Mr 22,000 and Mr 25,000 bFGF forms in their cell extract. Immunoreactive Mr 20,000 bFGF is detectable also in the conditioned medium of transformed nontumorigenic FBAE GM 7372 cells and of adult bovine aortic endothelial cells, but not in the culture medium of nonendothelial cell types, including rat and mouse fibroblasts, human hepatoma, and human endometrial adenocarcinoma cells. The results indicate that bovine endothelial cells secrete a Mr 20,000 bFGF-like molecule which shares several biological, biochemical, and immunological characteristics with the typical cell-associated Mr 18,000 bFGF.     

CONSTITUTIVELY MIGRATING MALIGNANT RAT LIVER EPITHELIAL CELLS PRODUCE A MIGRATION-STIMULATING ACTIVITY FOR EPITHELIAL CELLS (eMSA)

Constitutively migrating malignant rat liver epithelial cells obtained by Ha ras transformation exhibit a fibroblastoid phenotypein vitro. The cells deposit the anti-adhesive extracellular matrix (ECM) protein tenascin into their ECM migration tracks. The serum-free medium conditioned by these constitutively migrating cells contains an epithelial migration-stimulating activity (eMSA) that is neither cell-type-, nor species-specific. This eMSA fractionates in the range of 30 to 50kDa and binds to Mono-Q, Mono-S, and with low affinity to heparin—Sepharose. The conditioned medium also induces the expression of the serine proteinase inhibitor PAI-1. Both migration and expression of PAI-1 are inhibited by cyclic AMP, as previously shown for the migration of the non-transformed liver epithelial cells induced by several growth factors that act through tyrosine kinase receptors. These results suggest that the eMSA might act through signal transduction pathways similar to those of the growth factors previously studied. It is postulated that the eMSA, through both autocrine and paracrine mechanisms, is at least partially responsible for the malignant phenotype of the transformants.     

Effects of dopamine and melatonin on the regulation of the PIT-1 isotype, placental growth hormone and lactogen gene expressions in the rat placenta

Rat placenta produces several members of the placental prolactin-growth hormone (PRL-GH), including placental lactogen (PL) and placental prolactin like protein (PLP), during pregnancy. It is important to study placental local regulators that control the expression of PRL-GH genes. We have previously reported that dopamine (DA) can regulate Pit-1 and PL-II gene expressions. In this study we aimed to investigate the local expression of melatonin receptor 1a (Mel1a) and the effects of DA and melatonin on the expressions of PL-Iv, PL-II, PLP-C genes and Pit-1 gene that are involved in the expression of PRL-GH genes in the rat pituitary and placenta. According to the Northern blot analysis, DA receptor 2 (D2) was expressed in the rat placenta. We also report on the local expression of Mel1a in the rat placenta for the first time. Injected DA agonist, bromocriptine (in vivo) decreased PL-Iv, PLP-C and Pit-1 mRNA levels in the rat placenta. The melatonin agonist, chloromelatonin in culture media also decreased the levels of PL-Iv, PL-II and PLP-C mRNA. However, melatonin does not affect the Pit-1 mRNA level. These data suggest that D2 and Mel1a may control the expression of PRL-GH genes in the rat placenta and its response to the extracellular changes of DA and melatonin secreted from the maternal organ. However, Pit-1 may not be involved in the Mel1a induced inhibition of PRL-GH gene expressions in the rat placenta.     

Endothelial proteoglycans inhibit bFGF binding and mitogenesis

Basic fibroblast growth factor (bFGF) is a known mitogen for vascular smooth muscle cells and has been implicated as having a role in a number of proliferative vascular disorders. Binding of bFGF to heparin or heparan sulfate has been demonstrated to both stimulate and inhibit growth factor activity. The activity, towards bFGF, of heparan sulfate proteoglycans present within the vascular system is likely related to the chemical characteristics of the glycosaminoglycan as well as the structure and pericellular location of the intact proteoglycans. We have previously shown that endothelial conditioned medium inhibits both bFGF binding to vascular smooth muscle cells and bFGF stimulated cell proliferation in vitro. In the present study, we have isolated proteoglycans from endothelial cell conditioned medium and demonstrated that they are responsible for the bFGF inhibitory activity. We further separated endothelial secreted proteoglycans into two fractions, PG-A and PG-B. The larger sized fraction (PG-A) had greater inhibitory activity than did PG-B for both bFGF binding and bFGF stimulation of vascular smooth muscle cell proliferation. The increased relative activity of PG-A was attributed, in part, to larger heparan sulfate chains which were more potent inhibitors of bFGF binding than the smaller heparan sulfate chains on PG-B. Both proteoglycan fractions contained perlecan-like core proteins; however, PG-A contained an additional core protein (approximately 190 kDa) that was not observed in PG-B. Both proteoglycan fractions bound bFGF directly, and PG-A bound a significantly greater relative amount of bFGF than did PG-B. Thus the ability of endothelial heparan sulfate proteoglycans to bind bFGF and prevent its association with vascular smooth muscle cells appears essential for inhibition of bFGF-induced mitogenesis. The production of potent bFGF inhibitory heparan sulfate proteoglycans by endothelial cells might contribute to the maintenance of vascular homeostasis. J. Cell. Physiol. 172:209–220, 1997. © 1997 Wiley-Liss, Inc.     

Sustained neurotensin exposure promotes cell surface recruitment of NTS2 receptors

In this study, we investigated whether persistent agonist stimulation of NTS2 receptors gives rise to down-regulation, in light of reports that their activation induced long-lasting effects. To address this issue, we incubated COS-7 cells expressing the rat NTS2 with neurotensin (NT) for up to 24 h and measured resultant cell surface [125I]-NT binding. We found that NTS2-expressing cells retained the same surface receptor density despite efficient internalization mechanisms. This preservation was neither due to NTS2 neosynthesis nor recycling since it was not blocked by cycloheximide or monensin. However, it appeared to involve translocation of spare receptors from internal stores, as NT induced NTS2 migration from trans-Golgi network to endosome-like structures. This stimulation-induced regulation of cell surface NTS2 receptors was even more striking in rat spinal cord neurons. Taken together, these results suggest that sustained NTS2 activation promotes recruitment of intracellular receptors to the cell surface, thereby preventing functional desensitization.     

Bacterial fimbriae stimulate proinflammatory activation in the endothelium through distinct TLRs

The major and minor fimbriae proteins produced by the human pathogen Porphyromonas gingivalis are required for invasion of human aortic endothelial cells and for the stimulation of potent inflammatory responses. In this study, we report that native forms of both the major and minor fimbriae proteins bind to and signal through TLR2 for this response. Major and minor fimbriae bound to a human TLR2:Fc chimeric protein with an observed K(d) of 28.9 nM and 61.7 nM, respectively. Direct binding of the major and minor fimbriae to a human chimeric CD14-Fc protein also established specific binding of the major and minor fimbriae to CD14 with classic saturation kinetics. Using a P. gingivalis major and minor fimbriae mutant, we confirmed that TLR2 binding in whole cells is dependent on the expression of the major and minor fimbriae. Although we did not observe binding with the major or minor fimbriae to the TLR4-Fc chimeric protein, signaling through TLR4 for both proteins was demonstrated in human embryonic kidney 293 cells transfected with TLR4 and only in the presence MD-2. Transient transfection of dominant-negative forms of TLR2 or TLR4 reduced IL-8 production by human aortic endothelial cells following stimulation with major or minor fimbriae. The ability of two well-defined microbe-associated molecular patterns to select for innate immune recognition receptors based on accessory proteins may provide a novel way for a pathogen to sense and signal in appropriate host environments.     

Growth hormone action in rat insulinoma cells expressing truncated growth hormone receptors

Transfection of the insulin-producing rat islet tumor cell line RIN-5AH with a full length cDNA of the rat hepatic growth hormone (GH) receptor (GH-R1-638) augments the GH-responsive insulin synthesis in these cells. Using this functional system we analyzed the effect of COOH-terminal truncation of the GH receptor. Two mutated cDNAs encoding truncated GH receptors, GH-R1-294 and GH-R1-454, respectively, were generated by site-directed mutagenesis and transfected into the RIN cells. Both receptor mutants were expressed on the cell surface and displayed normal GH binding affinity. Whereas GH-R1-638 had a molecular mass of about 110 kDa, GH-R1-294 and GH-R1-454 showed molecular masses of 49 and 80 kDa, respectively. Cells expressing GH-R1-454 internalized GH to a similar extent as cells transfected with the full length receptor and the parent cell line, but GH-R1-294-expressing cells showed a markedly reduced capability of GH internalization. In contrast to cells transfected with GH-R1-638, none of the cell lines expressing truncated GH receptors exhibited any increase of the GH-stimulated insulin production. We conclude that domains within the COOH-terminal half of the cytoplasmic part of the GH receptor are required for transduction of the signal for GH-stimulated insulin synthesis, whereas cytoplasmic domains proximal to the transmembrane region are involved in receptor-mediated GH internalization.     

Endothelial thromboxane receptors: biochemical characterization and functional implications

We have identified thromboxane specific receptors in membrane preparations of bovine pulmonary artery endothelial cells using a potent thromboxane specific antagonist, [125I]-PTA-OH in a binding assay. The binding was specific and saturable. Neither thromboxane B2, prostaglandin D2 nor prostaglandin F2 alpha displaced the ligand (0.1 nM) at concentrations up to 10 microM. However, binding was displaced by IPTA-OH greater than SQ29548 greater than U46619. In addition, we observed that thromboxane mimetic U46619 significantly lowered the basal production of prostacyclin and also markedly suppressed bradykinin-stimulated prostacyclin released by endothelial cells. We propose that an important biological effect of thromboxane on vascular endothelial cells may be the suppression of prostacyclin production.     

Stimulation of motility in cultured bovine capillary endothelial cells by angiogenic preparations

Several angiogenic preparations that have been shown to stimulate plasminogen activator (PA) and collagenase production by cultured bovine capillary endothelial (BCE) cells were tested for their ability to stimulate BCE cell motility in the phagokinetic track assay. Bovine retinal extract, medium conditioned by 3T3-F442A differentiated mouse adipocytes, SK HEP-1 human hepatoma cell lysate, mouse sarcoma 180 cell lysate, and medium conditioned by mouse sarcoma 180 cells stimulated motility 68.7%, 48.5%, 140.9%, 56.5%, and 102.1%, respectively, relative to untreated cells. The motility-stimulating activity of these preparations was dose dependent and linear over the 16-h assay period. Several hormones and growth factors were tested for BCE cell motility-stimulating activity, including insulin, vasopressin, fibroblast growth factor, and a partially purified preparation of sarcoma growth factor, and were found to be ineffective. 12-0-tetradecanoyl-phorbol-acetate (TPA), a potent stimulator of both PA and collagenase activities in BCE cells, also did not stimulate motility, indicating that protease production is not sufficient to stimulate BCE cell motility in this assay. Neither SK HEP-1 hepatoma cell lysate nor TPA was effective in stimulating motility in bovine aortic endothelial (BAE) cells. The inability of SK HEP-1 hepatoma cell lysate to stimulate movement in BAE cells is consistent with the observation that angiogenesis occurs by sprouting of capillaries, not large vessels.     

Conformational restrictions of the sheep testicular receptor discriminates pituitary lutropin and placental gonadotropins

A membrane preparation from the testis of maturing Dorset-Leicester-Suffolk sheep, capable of discriminating pituitary LH (lutropin) from placental gonadotropins human choriogonadotropin (hCG) and equine choriogonadotropin is described. Maximum binding of 125I-oLH (ovine lutropin) to the testicular receptors occurred at 4 degrees C in a rapid manner, attaining equilibrium in 12-16 h. Under such optimal conditions, only unlabeled ovine LH or the structurally identical bovine LH effectively competed for receptor occupation. Other highly purified pituitary LH preparations from rat and human pituitaries were weakly (4-10%) active in displacement assays. Purified hCG or equine choriogonadotropin, which were highly potent in rat testicular LH receptor assays, could not compete with 125I-oLH for binding to the sheep LH receptor at 4 degrees C. Thus, the sheep testicular LH receptor was highly specific in recognizing pituitary LH conformation. The presence of an ovine/bovine LH alpha- or beta-subunit in recombinants with hCG subunit counterparts was required to generate an effective conformation capable of receptor recognition. Chemically deglycosylated hCG, containing 75% less carbohydrate and which showed greater binding to other LH receptors, failed to recognize sheep LH receptor, suggesting that excess carbohydrate in hCG was not a factor in hindering binding of the native placental hormone. Scatchard analysis using 125I-hCG/125I-oLH revealed that there were separate sites with similar affinities but vastly different capacities. The hCG binding sites, which could also be effectively occupied by oLH, were less than 10% of oLH binding sites. Thus, the Dorset-Leicester-Suffolk sheep testicular receptor provides an important and unique in vitro test system to distinguish pituitary LH from placental LH-like hormones. We infer that temperature-dependent conformational restrictions of the sheep testicular LH receptor are involved in recognizing differences in these highly similar and structurally homologous hormones.     

An antibody that blocks insulin-like growth factor (IGF) binding to the type II IGF receptor is neither an agonist nor an inhibitor of IGF-stimulated biologic responses in L6 myoblasts

To better define the biologic function of the type II insulin-like growth factor (IGF) receptor, we raised a blocking antiserum in a rabbit by immunizing with highly purified rat type II IGF receptor. On immunoblots of crude type II receptor preparations, only bands corresponding to the type II IGF receptor were seen with IgG 3637, indicating that the antiserum was specific for the type II receptor. Competitive binding and chemical cross-linking experiments showed that IgG 3637 blocked binding of 125I-IGF-II to the rat type II IGF receptor, but did not block binding of 125I-IGF-I to the type I IGF receptor, nor did IgG 3637 block binding of 125I-insulin to the insulin receptor. In addition, IgG 3637 did not inhibit the binding of 125I-IGF-II to partially purified 150- and 40-kDa IGF carrier proteins from adult and fetal rat serum. L6 myoblasts have both type I and type II IGF receptors. IGF-I was more potent than IGF-II in stimulating N-methyl-alpha-[14C]aminoisobutyric acid uptake, 2-[3H]deoxyglucose uptake, and [3H]leucine incorporation into cellular proteins. IgG 3637 did not stimulate either 2-[3H]deoxyglucose uptake, N-methyl-alpha-[14C]aminoisobutyric acid uptake, or [3H]leucine incorporation into protein when tested alone. Furthermore, IgG 3637 at concentrations sufficient to block type II receptors under conditions of the uptake and incorporation experiments did not cause a shift to the right of the dose-response curve for stimulation of these biologic functions by IGF-II. We conclude that the type II IGF receptor does not mediate IGF stimulation of N-methyl-alpha-[14C]aminoisobutyric acid and 2-[3H]deoxyglucose uptake and protein synthesis in L6 myoblasts; presumably, the type I receptor mediates these biologic responses. The anti-type II receptor antibody inhibited IGF-II degradation in the media by greater than 90%, suggesting that the major degradative pathway for IGF-II in L6 myoblasts utilizes the type II IGF receptor.     

Calcitonin gene-related peptide (CALCA) is a proangiogenic growth factor in the human placental development

Recent studies have shown that homozygous knockout of gene for calcitonin gene-related peptide (CALCA) receptor component, calcitonin receptor-like receptor (CALCRL), led to extreme hydrops fetalis and embryonic death, underlining the critical role of CALCA in embryonic development and fetal growth. The present study was designed to determine the cellular localization of CALCA and its receptor components, CALCRL and receptor activity modifying protein 1 (RAMP1), at the human implantation site during early pregnancy; to assess whether CALCA regulates in vitro angiogenesis of human endothelial cells; and to examine whether CALCA can improve angiogenic imbalance in preeclamptic placental explants. Our studies demonstrated that both protein and mRNA for CALCA were expressed by the villous and extravillous trophoblasts and decidual cells in the first-trimester villous tissues. CALCA receptor components, CALCRL and RAMP1, were expressed by both villous and extravillous trophoblast cells, as well as vascular endothelial cells. CALCA induced both endothelial proliferation and migration in a dose- and time-dependent manner, and it promoted capillarylike tube formation of human umbilical vein endothelial cells (HUVECs) on Matrigel. CALCA-induced angiogenesis of human endothelial cells was completely blocked by CALCA antagonist CALCA(8-37). Further, conditioned medium from preeclamptic placental explants significantly inhibited HUVEC capillarylike tube formation compared with gestational age-matched controls, and conditioned medium from preeclamptic placental explants incubated with CALCA significantly improved capillarylike tube formation. We conclude that CALCA induces in vitro angiogenesis by stimulating endothelial cell proliferation, migration, and capillarylike tube formation; thus, CALCA at the human implantation site may constitute a potential autocrine or paracrine mechanism that could modify placental angiogenesis and neovascularization.