Fibrocytes induce an angiogenic phenotype in cultured endothelial

Fibrocytes induce an angiogenic phenotype in cultured endothelial cells and promote angiogenesis in vivo INGO HARTLAPP,*,1 RIICHIRO ABE, RUBINA W. SAEED, TINA PENG, WOLFGANG VOELTER,* RICHARD BUCALA, AND CHRISTINE N. METZ2 Laboratory of Vascular Biology, The Picower Institute for Medical Research, Manhasset, New York 11030, USA; and *University of Tu¨bingen, Tu¨bingen, Germany Angiogenesis is an ordered process requiring the inter-play of numerous cellular and humoral factors. Studies over the past 20 years have identified several growth factors, cytokines, and enzymes that promote blood vessel formation. Most have revealed how individual factors promote an angiogenic phenotype in endothelial cells in vitro or contribute to blood vessel formation in vivo. However, the fundamental question that remains unanswered is how the cellular microenvironment contributes to angiogenesis. Fibrocytes are a recently characterized mesenchymal cell type isolated from peripheral blood that rapidly enter subcutaneously implanted wound chambers and sites of tissue injury. Here we describe the induction of an angiogenic phenotype in microvascular endothelial cells in vitro and promotion of angiogenesis in vivo by cultured fibrocytes. Fibrocytes constitutively secrete extracellular matrix-degrading enzymes, primarily matrix metalloproteinase 9, which promotes endothelial cell invasion. In addition, fibrocytes secrete several proangiogenic factors including VEGF, bFGF, IL-8, PDGF, and hematopoietic growth factors that promote endothelial cell migration, proliferation, and/or tube formation. By contrast, they do not produce representative antiangiogenic factors. Finally, both autologous fibrocytes and fibrocyte-conditioned media were found to induce blood vessel formation in vivo using the Matrigel angiogenesis model.—Hartlapp, I., Abe, R., Saeed, R. W., Peng, T., Voelter, W., Bucala, R., Metz, C. N. Fibrocytes induce an angiogenic phenotype in cultured endothelial cells and promote angiogenesis in vivo. FASEB J. 15, 2215–2224 (2001)

ABSTRACT

Key Words: fibroblast 䡠 antigen 䡠 MMP-9 䡠 wound healing 䡠 neovascularization Angiogenesis or neovascularization define the process of forming new capillaries from preexisting blood vessels. Blood vessel formation during normal physiological processes, such as corpus luteum formation and wound healing, is highly regulated by a delicate balance between pro- and antiangiogenic factors. In contrast, uncontrolled blood vessel formation occurs in many pathological conditions where the presence of angiogenic factors exceeds angiogenic inhibitors. These conditions characterized by excessive 0892-6638/01/0015-2215 © FASEB

angiogenesis include tumorigenesis, diabetic retinopathy, rheumatoid arthritis, and psoriasis. Neovascularization is initiated by the activation of quiescent endothelial cells by angiogenic factors produced by tumor or host inflammatory cells. After activation of the endothelial cells, angiogenesis ensues by a succession of overlapping events that includes 1) basement membrane proteolysis, 2) endothelial cell proliferation, 3) migration, and 4) alignment to form cord-like structures, followed by 5) vessel maturation or patency (reviewed in ref 1). Much progress has been made in identifying specific factors that regulate angiogenesis. These molecules include polypeptides, proteins, lipids, nucleotides, and copper. An important challenge is to develop an understanding of how the tissue microenvironment and cell-to-cell interactions result in the production of angiogenic mediators necessary for neovascularization in vivo. Several circulating cell types, including neutrophils, monocytes, macrophages, lymphocytes, and granulocytes, which are recruited to sites of tumor growth, chronic inflammation, or wounds, contribute to the angiogenic process. However, their specific roles in the neovascularization process are only partially understood. In previous studies we have established that a population of connective tissue fibroblast-like cells exist in the circulating peripheral blood (2). These cells, termed ‘fibrocytes’, were identified by their rapid and specific recruitment from the blood to subcutaneously (s.c.) implanted wound chambers in mice, where they proliferate. Fibrocytes purified from peripheral blood and grown in culture express vimentin and collagens I and III. However, these cells are molecularly distinct from fibroblasts exhibiting a unique profile of cell surface markers including CD34, CD45, CD80, CD86, and MHC class II (2, 3). These cells do not express typical surface antigens present on dendritic cells (CD11a, CD25, CD10, CD38), macrophages (CD14, CD16), B cells (CD19), fibroblasts, or endothelial cells (von Willebrand factor) (2). Past studies have revealed 1 Current address: University of Tu¨bingen, Tu¨bingen, Germany. 2 Correspondence: The Picower Institute for Medical Research, 350 Community Dr., Manhasset, NY 11030, USA. E-mail: [email protected]

2215

the unique cytokine and chemokine profile produced by fibrocytes (4) and demonstrated their ability to present antigen in vitro and in vivo (3, 5). More recent studies show the critical interaction between the chemokine receptor CCR7, expressed on fibrocytes, with secondary lymphoid chemokine as a mechanism by which fibrocytes migrate to wound sites (35). Although circulating fibrocytes have been shown to migrate to early wound sites where angiogenesis occurs, the precise contribution of these cells to wound healing is not known. In the present report, we characterize the production of proangiogenic factors by isolated fibrocytes and reveal the potential for fibrocytes to induce an angiogenic phenotype in microvascular endothelial cells in vitro and promote angiogenesis in vivo. Fibrocytes produce and secrete active matrix metalloproteinase 9 (MMP-9; gelatinase B), which is implicated in the proteolysis of the basement membrane early during the invasion stage of angiogenesis. In addition, cultured fibrocytes constitutively secrete several growth factors (i.e., VEGF, bFGF, PDGF, and hematopoietic factors) that induce endothelial cell migration, proliferation, and/or alignment of endothelial cells into tubular-like structures in vitro. Last, we observed that cultured fibrocytes (and fibrocyte-conditioned media) promote angiogenesis in vivo using the Matrigel implant model. Taken together, these results suggest that fibrocytes comprise a population of cells that may play a functional role in angiogenesis.

MATERIALS AND METHODS Cells/cell lines Human dermal microvascular endothelial cells (HuMVEC) from adult donors were obtained from Clonetics (Walkersville, MD) and grown in endothelial growth medium-2 MV (complete endothelial cell growth media, or EGM-2; Clonetics). Cells obtained from passages 5–10 were used for in vitro experiments as indicated. These cells stained positive for acetylated low density lipoprotein uptake, von Willebrand’s factor VIII, and CD34 and were negative for alpha smooth muscle actin (data not shown). Human umbilical vein endothelial cells (HuVEC) and NIH/3T3 fibroblast cells were obtained from American Type Culture Collection (ATCC, Manassas, VA). Human dermal fibroblasts were obtained from Clonetics and grown in fibroblast growth media as suggested by the manufacturer. Human fibrocytes were purified from peripheral blood as described (2– 4). Total PBMCs were first isolated from blood by centrifugation over FicollHypaque (Pharmacia, Uppsala, Sweden), following the manufacturer’s protocol, and plated at 1 ⫻ 107 cells/ml in Dulbecco’s minimal essential media (DMEM) supplemented with 20% fetal calf serum (FCS; Hyclone, Logan, UT) and 1% penicillin/streptomycin/glutamine (DMEM 20%FCS, PSQ). After overnight culture, the nonadherent cells were removed by gentle aspiration and the cells were cultured for ⬃10 –12 days. Mouse fibrocytes were purified from the peripheral blood of BALB/c mice (female, retired breeders), as described (3– 4). Mice were killed by CO2 asphyxiation and bled by cardiac puncture. Mouse blood was mixed with PBS (1/5 volume); 4 ml blood was layered over 8 ml Ficoll-Hypaque and centrifuged at 400 g for 30 min at room temperature. The 2216

Vol. 15

October 2001

buffy coat was isolated, washed with PBS and spun at 500 g, then washed with PBS. Mouse fibrocytes were maintained in DMEM supplemented with 10% FCS, 10% mouse sera (Sigma, St. Louis, MO), and 1% P/S/Q for ⬃ 2–3 wk. Typical yields were 0.8 – 4 ⫻ 104 cells/ml of mouse blood. Fibrocytes (mouse and human) were analyzed to be ⬎95% pure by adherent, spindle-shaped morphology as well as by flow cytometry for CD34 and collagen I, as described previously (4). Antibodies, other reagents Neutralizing antibodies to basic fibroblast growth factor (bFGF; AF233), vascular endothelial growth factor (VEGF; AF493), IL-8 (MAB209), MCSF (AF216), and GCSF (NA-214) were purchased from R&D Systems (Minneapolis, MN). The matrix metalloproteinase inhibitor, GM6001 (Ilomastat) was purchased from Chemicon International (Temecula, CA). Preparation of conditioned media Two- to 3-wk-old fibrocytes (80% confluent), mouse NIH/ 3T3 fibroblasts (80% confluent), or human dermal fibroblasts (80% confluent) were washed once with PBS and cultured for 48 h in either OptiMEM (Life Technologies, Grand Island, NY) or DMEM, as indicated. Fibrocyte-conditioned media (FcCM) was centrifuged at 5000 rpm for 10 min, filtered through 0.22 ␮m filters, and stored at ⫺80°C until use. Conditioned media was concentrated as indicated using Microcon 10 MWCO filters (Amicon, Beverly, MA). Zymography gels Gelatin zymography of FcCM (or control media) was performed to visualize the gelatinolytic activity (6). Briefly, 20 ␮l aliquots containing ⬃10 –20 ␮g protein were electrophoresed using 10% SDS-PAGE gels copolymerized with gelatin (BioRad, Hercules, CA) at 60V (4°C). After electrophoresis, the gels were washed with 2.5% Triton X-100 to remove the SDS, then incubated in development buffer (50 mM Tris pH 8.0 alone or 50 mM Tris pH 8.0 containing 10 mM EDTA or 10 mM Ca2⫹) overnight at 37°C. The lysed regions indicating gelatinase activity were visualized as clear bands after Coomassie blue staining/destaining. RT-PCR Total cellular RNA was isolated from human fibrocytes by a modified guanidinium isothiocyanate method (RNAzol, TelTest, Friendswood, TX). The cDNA was prepared from 1.0 ␮g of RNA using 0.25 ng of oligo-(dT)12–18 and MMLV reverse transcriptase following the manufacturer’s protocol (Life Technologies). Two microliter aliquots of cDNA were amplified using Supermix (Life Technologies) by PCR in a Perkin Elmer model 9600 thermal cycler using specific primers and conditions as described previously: aFGF, bFGF, VEGF (7), ␤-actin, GADPH, IL-8, IL-1␤, GM-CSF, interferon ␥ (IFN-␥), tumor necrosis factor (TNF-␣) (8), CTGF (9), HGF (10), IGF-1 (11), platelet-derived growth factor A (PDGF-A) (12), M-CSF (purchased from Clontech, Palo Alto, CA), and MMP-1, -2, -9, -14 (13). Amplified fragments of expected sizes were analyzed using a 2% agarose gel and photographed over UV light. MMP-9 Western blotting Fibrocyte-conditioned media (or control media) was concentrated (fivefold) using a Microcon 10 concentrator and

The FASEB Journal

HARTLAPP ET AL.

electrophoresed (10 ␮g/lane) using a 12% SDS-PAGE gel under both reducing and nonreducing conditions. Proteins were Western blotted using an MMP-9 monoclonal antibody (Calbiochem, Cambridge, MA) according to the manufacturer’s instructions, using detection by enhanced chemiluminescent methods (Amersham, Piscataway, NJ). Under nonreducing conditions, both the latent (92 kDa) and active (83 kDa) forms of MMP-9 are detected, whereas under reducing conditions only the latent form of the protein (92 kDa) is revealed. Endothelial cell proliferation assay HuMVEC (3–5⫻103/well; passages 5–10) were plated in 96-well plates in EGM-2 media. After 24 h, the medium was aspirated, replaced with EGM-2 (positive control), human FcCM (prepared by incubation of fibrocytes with DMEM 0.2% FCS for 24 h), or control media (DMEM 0.2% FCS) containing 0.5 ␮Ci [3H]-thymidine/well, and incubated overnight (n⫽8 per condition). The cells were harvested onto GF-B filter plates (Packard, Downers Grove, IL) and [3H]thymidine incorporation was measured by ␤-liquid scintillation counting (Packard). For endothelial cell proliferation experiments to identify specific mitogenic factors present in FcCM, neutralizing antibodies (or isotype control antibodies) were included in the proliferation assay between 0.1 and 10 ␮g/ml (according to the manufacturer’s directions). The data were reported as the mean % change in proliferation ⫾ sd when compared with control media and the Student’s t test was used to determine significance (P⬍0.05 was considered significant). Similar results were obtained using FcCM preparation obtained from cultured fibrocytes from more than six different donors. Determination of cytokines and chemokines present in fibrocyte-conditioned media ELISA kits for the quantification of the following factors were purchased from R&D Systems: angiogenin, bFGF, EGF, GMCSF, HGF, IL-1␤, IL-8, IL-12, IFN-␣, IFN-␥, VEGF, M-CSF, and substance P. Endothelial cell migration assay Endothelial cell migration was determined using the ChemoTX® disposable chemotaxis 96-well microplates (Neuroprobe, Gaithersburg, MD) according to the manufacturer’s instructions. Human FcCM (prepared in OptiMEM or DMEM and supplemented with 2%FCS), EGM-2, or basal endothelial cell media containing 2% FCS and 5 ng/ml VEGF (as a positive control) or OptiMEM supplemented with 2% FCS (as a negative control) was placed in the bottom chamber (29 ␮l/well; n⫽6 per sample condition). The chemotaxis microplate was then fitted with a filter (8 ␮ pore size) bonded to a metal frame containing 96 – 6.0 mm diameter test sites and 50 ␮l containing 1–2 ⫻ 104 human microvascular endothelial cells (p5–7; resuspended in OptiMEM 2% FCS) was pipetted onto each site. After overnight incubation, cells were wiped from the top filter and the plate was centrifuged at 500 g. The cells that migrated through the filter were fixed and stained using Diff-Quik reagents I and II (Baxter, Morton Grove, IL). Migrated cells were counted by microscopy [four or five random fields (400⫻)/sample]. The data were reported as the mean ⫾ se and the Student’s t test was used to determine significance (P⬍0.05 was considered significant). To identify the specific chemotactic agents present, FcCM (or control media) was preincubated specific neutralizing antibodies at 25 ␮g/ml for 30 min before the addition of the FIBROCYTES INDUCE ANGIOGENESIS

endothelial cells. The experiment was repeated twice with similar results. In addition, the endothelial cell migration assays in response to FcCM were assessed using Costar transwell chambers (8 ␮M; Costar Corp., Cambridge, MA) under similar conditions and according to the manufacturer’s directions. Similar results were observed with four donors. Endothelial cell morphogenesis assay Approximately 2 ⫻ 104 human microvascular endothelial cells (passages 4 –7) were resuspended in 0.4 ml EGM-2, human FcCM (prepared in Opti-MEM or DMEM) supplemented with 2% FCS or control media (OptiMEM 2% FCS or DMEM 2% FCS) and plated onto 4-well chamber slides (Nunc, Naperville, IL) precoated with Matrigel (0.3 ml/well; Collaborative Biomedical Products, Bedford, MA), as described previously (14, 15). After 3.5 h at 37°C/5%CO2, the media was aspirated, the cells were fixed and stained with Diff-Quik reagents, and the slides were examined for endothelial cells alignment at 100⫻ using an Olympus microscope (Model BX60). Images were obtained using MetaMorph Software (West Chester, PA). Similar results were observed with four donors. In vivo angiogenesis assay In vivo angiogenesis was assayed by the growth of blood vessels from the s.c. tissue into a Matrigel implant (16, 17) containing cultured fibrocytes or concentrated FcCM. Animal protocols were approved by the North Shore University Hospital Institutional Animal Care and Use Committee. Mice (BALB/c, retired female breeders, n⫽5 per group) were injected along the abdominal midline into the s.c. tissue with 0.4 ml Matrigel containing heparin (60 U/ml) (Life Technologies) mixed with fibrocytes (3–5⫻105/mouse), NIH/3T3 fibroblast cells (as a negative control), FcCM (4 ␮l; 20-fold concentrated), control medium (4 ␮l; 20-fold concentrated), or NIH/3T3 fibroblast cell-conditioned media (4 ml; 20-fold concentrated). As positive and negative controls for angiogenesis, mice were injected with 0.4 ml Matrigel containing heparin (60 U/ml) and bovine aFGF (1.25 ng/ml; R&D, Abingdon, Oxon, UK) (positive control) or heparin alone (60 U/ml) (negative control, Life Technologies). The implanted gels were harvested 6 –7 days after injection. A small piece of the gel implant (100 ␮g) was saved for quantitative hemoglobin (Hb) analysis using the Drabkin’s kit (Sigma). The data are reported as the mean concentration of Hb (g/dl) ⫾ sd and the Student’s t test (two-sample, equal variance) was used to determine significance. The remainder of the Matrigel implant was fixed with 10% buffered formalin, embedded in paraffin, sectioned at 3 ␮m, and stained with Masson’s Trichrome. To further assess vascularization, equivalent sections were stained using an anti-CD31 mAb (PharMingen, San Diego, CA) and analyzed as described previously (18). Matrigel angiogenesis was photographed at 100⫻ using an Olympus BX60 microscope and MetaMorph imaging software; representative sections are shown. The experiment was repeated once with similar results.

RESULTS Fibrocytes produce and secrete active MMP-9: RTPCR, zymography, and Western blotting analysis of MMP-9 Previous studies have shown that metalloproteinasemediated proteolysis plays an important role in early 2217

tissue repair events (19). In mouse wounds, MMP-9 is induced within 24 h after injury (20) and localized mainly to tissue fibroblast-like cells and the endothelium (21). Based on these observations, we investigated whether fibrocytes express matrix metalloproteinases. RT-PCR analysis of human fibrocytes revealed the presence of MMP-9 (gelatinase B) mRNA (Fig. 1A) but not MMP-1, MMP-2 (gelatinase A), or MMP-14 (data not shown). Fibrocytes did not produce MMP-3 (active or latent stromelysin 1), as determined using a FITClabeled MMP-3 substrate (Chemicon International) or MMP-10 (active or latent stromelysin 2), as determined by Western blotting using an MMP-10 specific monoclonal antibody (Chemicon International) (data not shown). To demonstrate the presence of gelatinase proteins, conditioned media obtained from unstimulated human fibrocytes was analyzed by SDS-PAGE

gelatin zymography. As shown in Fig. 1B (left panel), gelatinolytic bands with an apparent molecular mass of 92 and 83 kDa were revealed, suggesting the presence of both latent and active forms of MMP-9, respectively. Incubation of the gelatin zymogram in EDTA prevented the appearance of these bands (Fig. 1B, middle panel) whereas incubation with calcium enhanced proteolysis of the gelatin, further suggesting the presence of MMP-9 (Fig. 1B, right panel). In contrast, control media lacked gelatinase activity. For comparison, conditioned media from mouse fibrocytes also contained MMP-9 activity, whereas human monocytes/macrophages (prepared in the same manner) contained ⬃10to 20-fold less MMP-9 activity (data not shown). The presence of MMP-9 was further confirmed by Western blotting using an MMP-9 mAb (Fig. 1C). Under reducing conditions, this antibody reveals the latent form of MMP-9 (lane 4); under nonreducing conditions, this antibody detects the latent (92 kDa) and active (83 kDa) forms (lane 2). Control media lacks MMP-9 reactivity (lanes 1 and 3). Fibrocytes secrete factors that promote endothelial cell proliferation

Figure 1. Expression of MMP-9 mRNA and protein by fibrocytes. A) RT-PCR analysis of MMP-1, -2, -9, and -14 mRNA expression by cultured human fibrocytes; B) gelatin zymography analysis of human FcCM or control media (fivefold concentrated). Unreduced samples were fractionated on a 12% SDS/PAGE gel containing gelatin. Zymograms were developed in either Tris buffer alone (untreated), 10 mM EDTA (to inhibit MMP activity), or 10 mM Ca2⫹ (to activate MMP activity). Lane A, OptiMEM media alone; lane B, FcCM-donor 1; lane C, FcCM-donor 2; lane D, endothelial cell-conditioned media (as a control). White bands with an apparent molecular mass of 92 and 83 kDa correspond to the latent and activated forms of MMP-9, respectively. C) Western blotting of FcCM (fivefold concentrated) confirms the presence of latent (92 kDa) and active (83 kDa) MMP-9. Lanes 1, OptiMEM under nonreducing conditions; lanes 2, FcCM under nonreducing conditions; lanes 3, OptiMEM under reducing conditions; lanes 4, FcCM under reducing conditions. 2218

Vol. 15

October 2001

Endothelial cell proliferation is a critical step in angiogenesis. Previous studies revealed that mouse fibrocytes secrete bFGF, a potent angiogenic factor (4). Therefore, we tested whether fibrocyte-conditioned media induced endothelial cell proliferation in vitro. Fibrocyte-conditioned media promoted microvascular endothelial cell proliferation by 35– 40%, as determined by incorporation of [3H]-thymidine (Fig. 2A). This increase in endothelial cell proliferation was confirmed by counting the cells after incubation with EGM-2 media, FcCM, or control media using a Coulter counter, as well as by a nonradioactive method for measuring cell proliferation using sulforhodamine (22)(data not shown). Similar proliferation results were obtained using HuVEC (data not shown). A striking observation was that serum-free human FcCM was as mitogenic for the endothelial cells as the commercially available EGM-2 containing FCS, VEGF, bFGF, aFGF, hydrocortisone, heparin, etc. The mitogenic effect of FcCM (and EGM-2) on endothelial cell proliferation was completely lost by boiling FcCM for 10 min before the assay, suggesting the presence of labile growth factors (Fig. 2A). Several preparations of FcCM from multiple donors exhibited similar mitogenic properties (data not shown). When we examined the effect of mouse FCCM on human endothelial cell proliferation (due to a lack of mouse primary endothelial cells), we found it was almost half as active as human FcCM despite species differences (data not shown). In addition, screening conditioned media (serum-free) obtained from several primary cell cultures, including human dermal fibroblasts and numerous tumor cell lines for endothelial cell mitogenic factors, revealed that these media were significantly less active than

The FASEB Journal

HARTLAPP ET AL.

Figure 2. Fibrocyte-conditioned media promotes proliferation of endothelial cells. A) Proliferation of endothelial cells in response to FcCM, EGM-2 media (complete endothelial cell growth media), boiled FcCM, or boiled EGM-2 media and untreated control media was measured by 3H-thymidine incorporation assays as described previously. Data are represented as % change in proliferation (⫾sd) compared with proliferation in response to control media (n⫽8 per sample condition). The Student’s t test was used to determine significance, *P ⬍ 0.01, **P ⬍ 0.001. B) FcCM-mediated endothelial cell proliferation was assessed by 3 H-thymidine incorporation in the presence or absence of isotype control antibodies, specific neutralizing antibodies (10 ␮g/ ml), or all neutralizing antibodies (1 ␮g/ml and 10 ␮g/ml) and compared with endothelial cell proliferation in response to control media alone. Data are represented as the mean % change in proliferation (⫾sd) compared with control media. The Student’s t test was used to determine significance, *P ⬍ 0.05.

human FcCM (⬍20% change in proliferation; data not shown). To identify the potential growth factors produced by fibrocytes mitogenic for endothelial cells, the fibrocytes and fibrocyte-conditioned media were screened for several known endothelial cell mitogenic factors by RT-PCR (mRNA) and ELISA (protein), respectively. As shown in Table 1, human fibrocytes express and constitutively secrete several factors known to induce endothelial cell proliferation, including bFGF, GM-CSF, IGF-I, IL-1␤, IL-8, PDGF, and VEGF. We also characterized the expression of proangiogenic factors by mouse fibrocytes by ELISA and found that they secrete VEGF, GM-CSF, IL-1␤ (data not shown; bFGF and IL-8 were not tested). Because angiogenesis is the result of a balance of both pro- and antiangiogenic factors, we also screened fibrocytes and FcCM for representative wellknown inhibitors of endothelial cell proliferation (Table 2). Fibrocytes do not express IFN-␣ or IFN-␥, both of which have been shown to block the angiogenic effects of bFGF, PDGF, and MMP-9 (23–27). Nor do they express IL-12 or TSP-1, a potent macrophagederived antiangiogenic agent. To identify the specific growth factors present in the human FcCM mitogenic for endothelial cells, we performed proliferation assays in the presence of neutralizing antibodies. Endothelial cell proliferation induced by FcCM was reduced by ⬎35% in the presence of anti-bFGF antibodies (10 ␮g/ml), ⬎30% in the presence of VEGF antibodies (10 ␮g/ml), and only 15% with G-CSF antibodies (10 ␮g/ml) (see Fig. 2B). Inclusion of all inhibitory antibodies (each at 10 ␮g/ml) reduced FcCM-mediated endothelial cell proliferation to that of control media containing 2% serum, but no additional growth factors. In contrast, endothelial cell

TABLE 1. Analysis of fibrocytes for pro-angiogenic factorsa Factor

aFGF angiogenenin bFGF CTGF EGF GM-CSF HGF IGF-1 IL-1␤ IL-8 M-CSF MMP-9 PD-ECGF PDGF-A PIGF Substance P VEGF

Function on endothelial cells

RT-PCR

Protein (conc.)

Proliferation, migration Vessel maturation (in vivo), migration Proliferation, migration, tube formation Proliferation, migration, tube formation Proliferation Proliferation, migration, tube formation Proliferation, migration, tube formation Proliferation 1 VEGF and VEGF-R expression by ECs Proliferation (high dose), migration (in vivo) Migration Degrade basement membrane Proliferation, migration Proliferation, migration Proliferation Proliferation, angiogenesis (in vivo) Proliferation, migration, tube formation

⫺ nd ⫹ ⫺ nd ⫹ ⫺ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ ⫹ ⫺ nd ⫹

⫺ ⫹ 100–200 pg/ml ⫹ 100–300 pg/ml nd ⫺ ⫹ 200–500 pg/ml ⫺ nd Not detectedb ⫹ 50–100 ng/mlb ⫹ 30–50 ng/mlb ⫹ no ELISA done nd Not detectedb nd ⫺ ⫹ 25–200 pg/ml

a aFGF, acidic FGF; bFGF, basic FGF; CTGF, connective tissue growth factor; EGF, epidermal growth factor; GM-CSF, granulocytemacrophage colony stimulating factor; HGF, hepatocyte growth factor; IGF-1, insulin growth factor 1; M-CSF, macrophage colony stimulating factor; MMP-9, matrix metalloproteinase 9; PD-ECGF, platelet-derived endothelial cell growth factor; PDGF, platelet-derived growth factor; PIGF, placental growth factor; VEGF, vascular endothelial cell growth factor; nd ⫽ not done. b Reported in ref 4.

FIBROCYTES INDUCE ANGIOGENESIS

2219

TABLE 2. Analysis of fibrocytes for antiangiogenic factorsa

a

Factor

Function on endothelial cells

RT-PCR

Protein

IFN-␣ IFN-␤ TSP-1 TNF-␣ IL-12

Blocks the effect of angiogenic factors Blocks the effect of angiogenic factors Inhibits EC proliferation and migration Inhibits EC proliferation—in vitro Induces IFN␥ which 2VEGF & bFGF by ECs

⫺ ⫺ ⫺ ⫺ nd

⫺ ⫺ nd ⫺ ⫺

IFN-␣, interferon-␣; IFN-␤, interferon-␤; TSP-1, thrombospondin-1; nd ⫽ not done.

proliferation induced by FcCM was not significantly inhibited by inclusion of isotype control antibodies, antiangiogenin antibodies, anti-IL-8 antibodies, anti-MCSF antibodies, or the addition of a MMP inhibitor, GM6001 or Ilomastat, which inhibits the enzymatic activity of MMP-9, as well as MMP-1, -2, -3, and -8 (data not shown). The effect of commercially available media (EGM-2) on endothelial cell proliferation was significantly inhibited only by the inclusion of anti-bFGF (40%) or anti-VEGF antibodies (30%) when present at 10 ␮g/ml (data not shown). Fibrocytes secrete factors that promote endothelial cell migration The identification of potent endothelial cell chemotactic factors and active MMP-9 in the conditioned media of fibrocytes prompted us to investigate whether the FcCM could also induce endothelial cell migration. As shown in Fig. 3, we observed a two- to threefold increase in endothelial cell chemotaxis with FcCM prepared using Opti-MEM vs. control media alone (Opti-MEM). Similarly, endothelial cells migrated four-

Figure 3. Fibrocytes release endothelial cell chemoattractant agents. Endothelial cells (1–2⫻104 cells) were resuspended in OptiMEM 2% FCS and pipetted onto ChemoTX® disposable 96-well plates (8 ␮M) containing either control media ⫹ VEGF (5 ng/ml), control media alone, FcCM alone, FcCM ⫹ isotype control antibody (25 ␮g/ml), or FcCM ⫹ anti-MCSF antibody (25 ␮g/ml) in the bottom chambers. Chemotaxis assays (n⫽6 per sample) were performed as described in Materials and Methods. The number of cells that migrated was counted by microscopy and the data are presented as the average number of migrated cells per high-power field (HPF, 400⫻). The data are presented as the mean ⫾ se and the Student’s t test was used to determine significance, *P ⬍ 0.01. 2220

Vol. 15

October 2001

fold more in response to VEGF (5 ng/ml). The addition of neutralizing antibody to M-CSF (25 ␮g/ml) inhibited FcCM-mediated chemotaxis whereas antiIL-8, anti-VEGF, and anti-bFGF had no significant effect on endothelial cell migration (data not shown). The ability FcCM to promote endothelial cell migration was confirmed using 8 ␮ Costar transwell chambers. VEGF induced the migration of 65 ⫾ 6% of the endothelial cells to the bottom chamber, whereas FcCM promoted the migration of 46 ⫾ 4% endothelial cells compared with 12 ⫾ 5% migrated cells in the control media group. A similar induction of endothelial cell migration was observed using DMEM-based fibrocyte-conditioned media compared with DMEM alone (data not shown). Fibrocytes secrete factors that enhance endothelial cell differentiation in vitro Because fibrocyte-conditioned media was chemotactic for endothelial cells, we explored whether it could support the alignment of endothelial cells or the formation of lumen-like structures. As shown in Fig. 4, microvascular endothelial cells when plated on Matrigel and incubated with FcCM aligned to form lumenlike structures and anastomosing tubes with multicentric junctions (Fig. 4B) vs. nonconditioned control media (Fig. 4C). Similar endothelial cell morphogenesis occurred in response to incubation with DMEMbased fibrocyte-conditioned media vs. DMEM alone (data not shown). This capillary-like network of endothelial cells induced by FcCM appears to be weaker than that induced by EGM-2, commercially prepared endothelial cell media (Fig. 4A). By contrast, no endothelial cell morphogenesis occurred after the incubation of endothelial cells plated on Matrigel with human fibroblast-conditioned media (data not shown). Fibrocytes secrete factors that are angiogenic in vivo Based on our observations that fibrocytes constitutively secrete factors that induce endothelial cell proliferation and migration and support tube formation, we next examined whether cultured fibrocytes or FcCM could promote angiogenesis in vivo. We used an experimental model of blood vessel formation in vivo. When combined with Matrigel containing heparin, mouse FcCM (Fig. 5C) and mouse fibrocytes (Fig. 5E) both induce neovascularization when implanted s.c. in mice. Similar neovascularization of Matrigel implants was

The FASEB Journal

HARTLAPP ET AL.

not shown), nonconditioned control media (OptiMEM or DMEM), or mouse NIH/3T3 cells (Fig. 5G).

DISCUSSION

Figure 4. Fibrocytes secrete factors that support endothelial cell tube formation. Endothelial cells (2⫻104 cells) were resuspended in 0.4 ml EGM-2 (A), FcCM supplemented with 2% FCS (B), or control media containing 2% FCS (C) and plated onto Matrigel coated wells as described in Materials and Methods. After 3.5 h, the media was gently aspirated and the cells were fixed and stained. The slides were examined for endothelial cell tube formation by microscopy (100⫻). Representative figures are shown.

observed using FcCM prepared with Opti-MEM and DMEM (data not shown). Human FcCM also induced blood vessel formation within the Matrigel plug; however, the response was less than that of mouse FcCM, probably due to species differences (data not shown). On the other hand, the addition of control media (Fig. 5D) or NIH/3T3 fibroblast cells (Fig. 5F) did not promote blood vessel formation within Matrigel implants. Vessel formation within the Matrigel plugs can be analyzed by measuring the Hb content using Drabkin’s reagent. Hb content present within the Matrigel plugs containing FcCM or mouse fibrocytes is significantly greater than that implanted with conditioned media obtained from mouse NIH/3T3 fibroblasts (data FIBROCYTES INDUCE ANGIOGENESIS

Normal wound repair requires angiogenesis that facilitates the removal of debris and prepares the wound bed for the development of a critical framework of granulation tissue necessary for wound closure. These newly formed blood vessels represent over 50% of the granulation tissue mass found in early wounds (reviewed in ref 28). Although several mediators of wound healing have been identified, few studies have focused on the role of specific cell types that mediate blood vessel formation during early tissue repair. Fibrocytes are a novel peripheral blood cell type that display a distinct cell surface phenotype and rapidly appear in wound sites (2). More recent studies have demonstrated the interaction between CCR7 (expressed on fibrocytes) and SCL necessary for the trafficking of fibrocytes to the wound site (36). Based on their presence in early wounds and their cytokine, chemokine, and growth factor profile (2, 4), we have examined the role of isolated fibrocytes in angiogenesis, a critical step in the wound healing process. Fibrocytes isolated from peripheral blood and cultured in vitro are able to promote all aspects of neovascularization including basement membrane dissolution, endothelial cell proliferation, migration, and tube formation in vitro and induce angiogenesis in vivo. Angiogenesis within a wound bed occurs by formation of blood vessels from the existing vasculature made up of quiescent endothelial cells and a few pericytes. These cellular structures are surrounded by a stable basement membrane that physically separates the capillary from surrounding stromal tissue. In response to injury or other stimuli, angiogenesis ensues when endothelial cells dissociate from neighboring cells and migrate into the wounded tissue area to form new vessels. Therefore, proteolysis of the basement membrane is a critical component during the very early stages of angiogenesis. Several degradative enzymes have been implicated in this process. One family of enzymes, collectively known as matrix metalloproteinases, has been shown to be important mediators of angiogenesis and tissue remodelling. MMPs are secreted as latent proenzymes that are subsequently cleaved and activated extracellularly. MMP activity is also regulated by tissue inhibitors of matrix metalloproteinases or TIMPs. Although MMPs are not expressed by normal human skin, MMP-2 (gelatinase B) and MMP-9 (gelatinase A) are strongly induced within 24 h after wounding (19, 21). In addition, MMP-9 has been shown to be the main MMP found in wound fluid, with peak activity between 2 and 4 days postwounding (21). Consistent with these observations, we have found that fibrocytes, which home to cutaneous wound sites within 1– 4 days (2, 36), constitutively express MMP-9 mRNA (Fig. 1A) and secrete high levels of active MMP-9 (Fig. 2221

Figure 5. Fibrocytes and FcCM induce angiogenesis within Matrigel implants in vivo. Matrigel was injected (s.c.) into mice (n⫽5 per group) containing A) heparin ⫹ aFGF, B) heparin alone, C) heparin ⫹ FcCM, D) heparin ⫹ control media, E) mouse fibrocytes, or F) NIH/ 3T3 fibroblast cells. After 6 days, the implants were removed, processed, and examined for blood vessel formation after Masson’s Trichrome staining. Stained sections were photographed at 100⫻. All panels (A–F ) are in the same orientation. Epidermis (Ep); Matrigel (Ma); skeletal muscle (SM); vessels (V). F) ‘⬎’ denotes growth of NIH/3T3 cells within the Matrigel implant. G) Quantification of angiogenesis within Matrigel plugs, as measured by hemoglobin (Hb) concentration (g/dl). The data are expressed as the mean ⫾ sd and significance was determined by the Student’s t test, *P ⬍ 0.05.

1B, C). MMP-9 dissolves several extracellular matrix proteins including gelatin, collagens I, IV, V, XIV, as well as aggrecan, elastin, entactin, and vitronectin (29). Although MMP-9 has not been shown to play a key role in basement membrane/collagen degradation in vivo, it has been proposed to initiate and promote angiogenesis. Therefore, fibrocytes might be an important source of active MMP-9 involved in angiogenesis in the very early phases of wound healing. After the degradation of the basement membrane during angiogenesis, endothelial cells invade the stromal tissue and migrate to form a leading edge. We report that cultured fibrocytes secrete chemoattractants for endothelial cells (Fig. 3). Surprisingly, only the addition of M-CSF antibodies inhibited FcCM-mediated endothelial cell migration. The levels of bFGF and VEGF present in FcCM (100 –300 pg/ml) are below the optimal level for inducing endothelial cell migration. Although M-CSF is not considered one of the prominent chemoattractant agents for endothelial cells, studies by Sigounas and co-workers have demonstrated the role of M-CSF and other hemopoietic growth factors on endothelial cell migration (30). The biological role of hemopoietic growth factors and expression of their respective receptors on endothelial cells have not been examined thoroughly. However, endothelial cells do express several surface antigens in common with hemopoietic cells and respond to numerous hematopoietic growth factors including G-CSF, GM-CSF (31), IL-3 (32), and M-CSF (30). During neovascularization, as endothelial cells mi2222

Vol. 15

October 2001

grate away from preexisting blood vessels they begin to proliferate at the leading edge. We report that serumfree conditioned media collected from unstimulated cultured fibrocytes (FcCM) supports endothelial proliferation comparable to commercially available endothelial cell growth media containing serum and several mitogenic factors (Fig. 2A). Analysis of fibrocytes (mRNA) and fibrocyte culture supernatants (protein) revealed the presence of numerous proangiogenic factors including bFGF, VEGF, GM-CSF, IL-1, IL-8, and M-CSF (Table 1). Based on these findings, we performed in vitro endothelial cell proliferation assays using specific neutralizing antibodies and found that bFGF, G-CSF, and VEGF were the most active mitogenic factors present in the FcCM. IL-8, a mitogenic factor for endothelial cells, is present at very high concentrations (⬎50 ng/ml) in the FcCM. However, neutralizing antibodies to IL-8 did not inhibit FcCM-mediated proliferation. These data are consistent with previous studies showing that IL-8 is a potent angiogenic factor only in vivo, because cultured endothelial cells do not express the IL-8 receptor (33). Serum-free FcCM induces lumen formation by endothelial cells when plated on Matrigel. Fibrocytes produce bFGF, the main factor found to mediate endothelial cell morphogenesis on Matrigel (34), as well as VEGF and IL-8, minor morphogenic mediators (34). Because angiogenesis is the result of a balance between pro- and antiangiogenic factors, we also examined fibrocytes for antiangiogenic factors. We found that fibrocytes did not produce any of the antiangiogenic factors analyzed, including TSP-1,

The FASEB Journal

HARTLAPP ET AL.

IFN-␣, IFN-␥, TNF-␣, and IL-12. Finally, we observed that FcCM could support endothelial cell tube formation in vitro. We speculate that the production of collagens I, III, VEGF, and angiogenin by cultured fibrocytes mediates this effect. Angiogenesis is a multistep process requiring the coordinated activity of many different cell types. Several circulating cells have been postulated to be proangiogenic. Many studies have revealed the role of activated macrophages in inflammatory and tumor-associated angiogenesis (reviewed in ref 35). Activated macrophages are able to promote all phases of angiogenesis in vitro, but attempts to identify pro- vs. nonangiogenic macrophage subtypes have been unsuccessful. Macrophages secrete a tissue-type plasminogen activator and urokinase-type plasminogen activator that degrade the basement membrane. They secrete MMP-9 in the latent form whereas fibrocytes secrete significantly more latent and active MMP-9 (data not shown). Similar to fibrocytes, macrophages secrete many proangiogenic factors (bFGF, TGF␣, GM-CSF, IL-8, VEGF, substance P, IGF-1, and PDGF). However, macrophages also secrete several antiangiogenic agents not produced by fibrocytes, including IFN-␣, IFN-␥, TSP-1, TNF-␣, and IL-6. Fibrocytes produce factors that support every aspect of angiogenesis: proteolysis of the basement membrane, endothelial cell migration, proliferation, and vessel formation in vitro. Therefore, we evaluated whether isolated fibrocytes or fibrocyte-conditioned media would promote angiogenesis in vivo. FcCM and fibrocytes produced a brisk neovasculararization response when we used the Matrigel implant experimental model (Fig. 5). The levels of many angiogenic factors produced by fibrocytes are relatively low. However, because angiogenesis in vivo is a complex process requiring multiple factors, it is difficult to assess the contribution of the individual factors produced by fibrocytes to the overall process of angiogenesis. Many of these factors act additively or synergistically to promote neovascularization, further complicating analysis of the contribution of specific factors to blood vessel formation. Because of the appearance of fibrocytes within the wound and their proposed role in tissue repair, it would be of interest to examine the role of fibrocytes in wound healing in normal and diabetic animals where wound healing is compromised, specifically at the step of blood vessel formation. We propose that fibrocytes play a role in blood vessel formation during the early stages of wound healing based on their early presence within injured tissue and the ability of cultured fibrocytes to promote an angiogenic phenotype in cultured endothelial cells and support neovascularization in vivo. Further characterization of cultured fibrocytes with angiogenic potential provides a unique approach to examine the direct role of fibrocytes in neovascularization during wound healing. Future studies will examine the role of fibrocytes FIBROCYTES INDUCE ANGIOGENESIS

on angiogenesis during disease processes where aberrant blood vessel formation exacerbates the disease pathogenesis, such as tumor growth, rheumatoid arthritis, and psoriasis. We thank Kirk Manogue for critically reading this manuscript. This work was supported in part by The Picower Institute for Medical Research (C.N.M.) and The Scleroderma Foundation (R.B.).

REFERENCES 1. 2. 3.

4.

5.

6. 7. 8.

9.

10. 11.

12.

13.

14. 15. 16.

17.

Harris, S.R., and Thorgeirsson, U. P. (1998) Tumor angiogenesis: biology and therapeutic prospects. In Vivo 12, 563–570 Bucala, R., Spiegel, L. A., Chesney, J. Hogan, M., and Cerami, A. (1994) Circulating fibrocytes define a new leukocyte subpopulation that mediates tissue repair. Mol. Med. 1, 71– 81 Chesney, J., Bacher, M., Bender, A., and Bucala, R. (1997) The peripheral blood fibrocyte is a potent antigen-presenting cell capable of priming naive T cells in situ. Proc. Natl. Acad. Sci. USA 94, 6307– 6312 Chesney, J., Metz, C., Stavitsky, A. B., Bacher, M., and Bucala, R. (1998) Regulated production of type I collagen and inflammatory cytokines by peripheral blood fibrocytes. J. Immunol. 160, 419 – 425 Grab, D. J., Lanners, H., Martin, L. N., Chesney, J., Cai, C., Adkisson, H. D., and Bucala, R. (1999) Interaction of Borrelia burgdorferi with peripheral blood fibrocytes, antigen-presenting cells with the potential for connective tissue targeting. Mol. Med. 5, 46 –54 Kleiner, D. E., and Stetler-Stevenson, W. G. (1994) Quantitative zymography: detection of picogram quantities of gelatinases. Anal. Biochem. 218, 325–329 Brown, K. J., Maynes, S. F., Bezos, A., Maguire, D. J., Ford, M. D., and Parish, C. R. (1996) A novel in vitro assay for human angiogenesis. Lab. Invest. 75, 539 –555 Nilsen, E. M., Johansen, F. E., Jahnsen, F. L., Lundin, K. E., Scholz, T., Brandtzaeg, P., and Haraldsen, G. (1998) Cytokine profiles of cultured microvascular endothelial cells from the human intestine. Gut 42, 635– 642 Nakanishi, T., Nishida, T., Shimo, T., Kobayashi, K., Kubo, T., Tamatani, T., Tezuka, K., and Takigawa, M. (2000) Effects of CTGF/Hcs24, a product of a hypertrophic chondrocyte-specific gene, on the proliferation and differentiation of chondrocytes in culture. Endocrinology 141, 264 –273 Goke, M., Kanai, M., and Podolsky, D. K. (1998) Intestinal fibroblasts regulate intestinal epithelial cell proliferation via hepatocyte growth factor. Am. J. Physiol. 274, G809 –G818 Gebauer, G., Jager, W., and Lang, N. (1998) mRNA expression of components of the insulin-like growth factor system in breast cancer cell lines, tissues, and metastatic breast cancer cells. Anticancer Res. 18, 1191–1195 Eggert, A., Ikegaki, N., Kwiatkowski, J., Zhao, H., Brodeur, G. M., and Himelstein, B. P. (2000) High-level expression of angiogenic factors is associated with advanced tumor stage in human neuroblastomas. Clin. Cancer Res. 6, 1900 –1908 Giambernardi, T. A., Grant, G. M., Taylor, G. P., Hay, R. J., Maher, V. M., McCormick, J. J., and Klebe, R. J. (1998) Overview of matrix metalloproteinase expression in cultured human cells. Matrix Biol. 16, 483– 496 Baatout, S. (1997) Endothelial differentiation using Matrigel (review). Anticancer Res. 17, 451– 455 Baatout, S., and Cheta, N. (1996) Matrigel: a useful tool to study endothelial differentiation. Rom. J. Int. Med. 34, 263–269 Passaniti, A., Taylor, R. M., Pili, R., Guo,Y., Long, P. V., Haney, J. A., Pauly, R. R., Grant, D. S., and Martin, G. R. (1992) A simple, quantitative method for assessing angiogenesis and antiangiogenic agents using reconstituted basement membrane, heparin, and fibroblast growth factor. Lab. Invest. 67, 519 –528 Ito, Y., Iwamoto, Y., Tanaka, K., Okuyama, K., Sugioka, Y. (1996) A quantitative assay using basement membrane extracts to study tumor angiogenesis in vivo. Int. J. Cancer 67, 148 –152

2223

18.

19. 20. 21.

22.

23.

24.

25.

26.

2224

Chesney, J., Metz, C., Bacher, M., Peng, T., Meinhardt, A., and Bucala, R. (1999) An essential role for macrophage migration inhibitory factor (MIF) in angiogenesis and the growth of a murine lymphoma. Mol. Med. 5, 181–191 Agren, M. S. (1994) Gelatinase activity during wound healing. Br. J. Dermatol. 131, 634 – 640 Madlener, M. (1998) Differential expression of matrix metalloproteinases and their physiological inhibitors in acute murine skin wounds. Arch. Dermatol. Res. 290, S24 –S29 Salo, T., Makela, M., Kylmaniemi, M., Autio-Harmainen, H., and Larjava, H. (1994) Expression of matrix metalloproteinase-2 and -9 during early human wound healing. Lab. Invest. 70, 176 –182 Rubinstein, L. V., Shoemaker, R. H., Paull, K. D., Simon, R. M., Tosini, S., Skehan, P., Scudiero, D. A., Monks, A., and Boyd, M. R. (1990) Comparison of in vitro anticancer-drug-screening data generated with a tetrazolium assay versus a protein assay against a diverse panel of human tumor cell lines. J. Natl. Cancer Inst. 82, 1113–1118 Heyns, A. D., Eldor, A., Vlodavsky, I., Kaiser, N., Fridman, R., and Panet, A. (1985) The antiproliferative effect of interferon and the mitogenic activity of growth factors are independent cell cycle events. Studies with vascular smooth muscle cells and endothelial cells. Exp. Cell Res. 161, 297–306 Sato, N., Nariuchi, H., Tsuruoka, N., Nishihara, T., Beitz, J. G., Calabresi, P., and Frackelton, A. R. J. (1990) Actions of TNF and IFN-gamma on angiogenesis in vitro. J. Invest. Dermatol. 95, 85S– 89S Dinney, C. P., Bielenberg, D. R., Perrotte, P., Reich, R., Eve, B. Y., Bucana, C. D., and Fidler, I. J. (1998) Inhibition of basic fibroblast growth factor expression, angiogenesis, and growth of human bladder carcinoma in mice by systemic interferon-alpha administration. Cancer Res. 58, 808 – 814 Riedel, F., Gotte, K., Bergler, W., Rojas, W., and Hormann, K. (2000) Expression of basic fibroblast growth factor protein and its down-regulation by interferons in head and neck cancer. Head Neck 22, 183–189

Vol. 15

October 2001

27.

28. 29. 30. 31.

32.

33.

34. 35. 36.

Slaton, J. W., Perrotte, P., Inoue, K., Dinney, C. P., and Fidler, I. J. (1999) Interferon-alpha-mediated down-regulation of angiogenesis-related genes and therapy of bladder cancer are dependent on optimization of biological dose and schedule. Clin. Cancer Res. 5, 2726 –2734 Davidson, J. (1992) Wound repair. In Inflammation: Basic Principles and Clinical Correlates (Gallin, J., and Goldstein, I., eds) pp. 809 – 819, Raven Press, New York Kahari, V. M., and Saarialho-Kere, U. (1997) Matrix metalloproteinases in skin. Exp. Dermatol. 6, 199 –213 Sigounas, G., Steiner, M., and Anagnostou, A. (1997) Synergism of hemopoietic growth factors on endothelial cell proliferation. Angiology 48, 141–147 Bussolino, F., Ziche, M., Wang, J. M., Alessi, D., Morbidelli, L., Cremona, O., Bosia, A., Marchisio, P. C., and Mantovani, A. (1991) In vitro and in vivo activation of endothelial cells by colony-stimulating factors. J. Clin. Invest. 87, 986 –995 Dentelli, P., Del Sorbo, L., Rosso, A., Molinar, A., Garbarino, G., Camussi, G., Pegoraro, L., and Brizzi, M. F. (1999) Human IL-3 stimulates endothelial cell motility and promotes in vivo new vessel formation. J. Immunol. 163, 2151–2159 Petzelbauer, P., Watson, C. A., Pfau, S. E., and Pober, J. S. (1995) IL-8 and angiogenesis: evidence that human endothelial cells lack receptors and do not respond to IL-8 in vitro. Cytokine 7, 267–272 Kumar, R., Yoneda, J., Bucana, C. D., and Fidler, I. J. (1998) Regulation of distinct steps of angiogenesis by different angiogenic molecules. Int. J. Oncol. 12, 749 –757 Sunderkotter, C., Steinbrink, K., Goebeler, M., Bhardwaj, R., and Sorg, C. (1994) Macrophages and angiogenesis. J. Leukoc. Biol. 55, 410 – 422 Abe, R. Donnelly, S. C., Peng, T., Bucala, R., Metz, C. N. (2001) Peripheral blood fibrocytes: Differentiation pathway and migration to wound sites. J. Immunol. 166, 7556 –7562

The FASEB Journal

Received for publication January 31, 2001. Revised for publication June 15, 2001.

HARTLAPP ET AL.