Evidence of IL-18 as a Novel Angiogenic Mediator Christy C. Park, Jacques C. M. Morel, M. Asif Amin, Matthew A. Connors, Lisa A. Harlow and Alisa E. Koch This information is current as of September 28, 2021. J Immunol 2001; 167:1644-1653; ; doi: 10.4049/jimmunol.167.3.1644 http://www.jimmunol.org/content/167/3/1644 Downloaded from

References This article cites 50 articles, 13 of which you can access for free at: http://www.jimmunol.org/content/167/3/1644.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on September 28, 2021

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2001 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Evidence of IL-18 as a Novel Angiogenic Mediator1

Christy C. Park,2*† Jacques C. M. Morel,2* M. Asif Amin,* Matthew A. Connors,* Lisa A. Harlow,* and Alisa E. Koch3*†

Angiogenesis, or new blood vessel growth, is a key process in the development of synovial inflammation in rheumatoid arthritis (RA). Integral to this pathologic proliferation are proinflammatory . We hypothesized a role for IL-18 as an angiogenic mediator in RA. We examined the effect of human IL-18 on human microvascular endothelial cell (HMVEC) migration. IL-18 induced HMVEC migration at 1 nM (p < 0.05). RA synovial fluids potently induced endothelial cell migration, but IL-18 im- ␤ ␣ ؎ munodepletion resulted in a 68 5% decrease in HMVEC migration (p < 0.05). IL-18 appears to act on HMVECs via v 3 . To test whether IL-18 induced endothelial cell tube formation in vitro, we quantitated the degree of tube formation on Matrigel matrix. IL-18, 1 or 10 nM, resulted in a 77% or 87% increase in tube formation compared with control (p < 0.05). To determine whether IL-18 may be angiogenic in vivo, we implanted IL-18 in Matrigel plugs in mice, and IL-18 at 1 and 10 nM induced (p < 0.05). The angiogenesis observed appears to be independent of the contribution of local TNF-␣,as Downloaded from evidenced by adding neutralizing anti-TNF-␣ Ab to the Matrigel plugs. In an alternative in vivo model, sponges embedded with IL-18 or control were implanted into mice. IL-18 (10 nM) induced a 4-fold increase in angiogenesis vs the control (p < 0.05). These findings support a novel function for IL-18 as an angiogenic factor in RA and may elucidate a potential therapeutic target for angiogenesis-directed diseases. The Journal of Immunology, 2001, 167: 1644–1653.

ngiogenesis, or new blood vessel formation, is integral IL-18 is a with many proinflammatory functions. IL-18 http://www.jimmunol.org/ to many physiological processes, such as wound healing was initially described as IFN-␥-inducing factor in 1989 (15). A and placental development, but also to disease states Thus, IL-18 stimulates Th1 cytokines by T cells and NK cells and such as tumorigenesis (1), inflammatory conditions (2), and rheu- promotes Th1 cell differentiation and immune responses (16, 17). matoid arthritis (RA)4 (3–5). The process of angiogenesis is con- IL-18 can act in synergy with IL-12 in regulating IFN-␥ (18). trolled by a complex interaction of inhibitors and activators, and an However, IL-18 also up-regulates other cytokines, such as TNF-␣, inappropriate balance leads to the initiation and maintenance of IL-1␤, and IL-8 in non-CD14ϩ PBMCs, resting T cells, and NK angiogenesis. Some of these proinflammatory mediators include cells (19). IL-18 up-regulates GM-CSF in PBMCs (20) and also in

cytokines, , adhesion molecules, growth factors, and synovial membrane cultures, directly affecting to by guest on September 28, 2021 extracellular matrix receptors (6). Specifically, some known an- stimulate cytokine production (21). giogenic stimuli include growth factors (7, 8) such as basic fibro- IL-18 has structural homology with IL-1, shares some common blast (bFGF) or vascular endothelial growth factor signaling pathways (22), and also requires the cleavage at its as- ␣ (VEGF) (9), TNF- (10), C-X-C chemokines (11, 12) such as IL-8 partic acid residue by IL-1-converting enzyme to become the ac- or epithelial -activating peptide (ENA-78), fractalkine tive, mature (23, 24). Thus, IL-1 and IL-18 share many (13), and adhesion molecules such as soluble E- or soluble biologically similar inflammatory functions. Previous work has VCAM-1 (14). implicated IL-18 in RA, as higher levels are present in RA vs osteoarthritic synovial fluid (SF) and sera (21). Also, IL-18 en- hanced erosive, inflammatory arthritis in a murine model of sys- temic arthritis (21). In the study by Leung et al. (25) in which *Department of Medicine, Northwestern University Medical School, Chicago, IL 60611; and †Veteran’s Administration Chicago Health Care System, Lakeside Divi- IL-18 promoted collagen-induced inflammatory arthritis through a sion, Chicago, IL 60611 mechanism shown to be distinct from IL-12, cytokine treatment Received for publication November 22, 2000. Accepted for publication May enhanced synovial hyperplasia, inflammatory infiltrate, and carti- 25, 2001. lage erosion. Interestingly, the IL-18-treated mice produced sig- The costs of publication of this article were defrayed in part by the payment of page ␣ charges. This article must therefore be hereby marked advertisement in accordance nificant amounts of TNF- , and splenic macrophages from mice with 18 U.S.C. Section 1734 solely to indicate this fact. cultured with IL-18 also stimulated high levels of TNF-␣. The 1 This work was supported by National Institutes of Health Grants AR30692, influential role of IL-18 in articular inflammation was confirmed in AI40987, and HL58695. Additional support included the Gallagher Professorship for mice lacking the IL-18 that had reduced incidence and se- Arthritis Research, and funds from the Veteran’s Administration Research Service and Northwestern Memorial Hospital Intramural Grant Program. verity of collagen-induced arthritis, which was reversed by treat- 2 C.C.P. and J.C.M.M. contributed equally to the work reported in this study. ment with rIL-18 (26). One important source of IL-18 is the mac- 3 Address correspondence and reprint requests to Dr. Alisa E. Koch, Gallagher Re- rophage, which is a critical synovial tissue producer of IL-12 (27), search Professor of Medicine, Northwestern University Medical School, 303 East and responsible for the production of many other cytokines (28). Chicago Avenue, Ward 3-315, Chicago, IL 60611. E-mail address: ae- [email protected] However, various other sources of IL-18 have been identified, in- 4 Abbreviations used in this paper: RA, rheumatoid arthritis; aFGF, acidic FGF; cluding Kupffer cells, dendritic cells, , articular chon- bFGF, basic FGF; EBM, endothelial cell basal medium; ENA-78, epithelial neutro- drocytes, osteoblasts, and synovial fibroblasts (18, 21, 29–33). phil-activating peptide; FGF, fibroblast growth factor; HMVEC, human microvascu- In regard to arthritis, if IL-18 functions in an autocrine or para- lar endothelial cell; rh, recombinant human; SF, synovial fluid; VEGF, vascular en- dothelial growth factor. crine fashion, the increased expression of IL-18 in the synovium

Copyright © 2001 by The American Association of Immunologists 0022-1767/01/$02.00 The Journal of Immunology 1645 may play a critical role in the development of synovial inflamma- Bioassay for HMVEC migration activity tion, synovial hyperplasia, and articular degradation, to which an- Subconfluent HMVECs (passage 3–12) were fed the night before the assay, giogenesis may contribute. Given the importance of angiogenesis to optimize the cellular conditions, with endothelial cell basal medium-2 in the pathophysiology of RA, we hypothesized a role for IL-18 as (EBM-2; Clonetics) and 0.1% FBS. HMVECs (3.75 ϫ 104 cells/25 ␮l an angiogenic mediator. Supportive of this function is the finding EBM and 0.1% FBS) were placed in the bottom wells of a 48-well Boyden chamber (NeuroProbe, Cabin John, MD) with gelatin-coated that IL-18 has been shown to stimulate production of angiogenic ␮ ␣ polycarbonate membranes (8 m pore size; Nucleopore, Pleasant, CA) TNF- (19). The establishment of the role of IL-18 in inflamma- (14). The chamber was inverted and incubated in a humidified incubator tion and RA led to the question of its angiogenic potential. with 5% CO2/95% air at 37°C for 2 h, allowing endothelial cell attachment In this study, we examined the capacity of IL-18 to mediate to the membrane. The chamber was reinverted, and the test substances, angiogenesis in in vitro and in vivo models. We further compared PBS, positive control bFGF (60 nM) in PBS, or RA SF were added to the wells in the top of the chamber, and the chamber was further incubated for the angiogenic activity of IL-18 with that of known mediators, 2 h at 37°C. IL-18 (R&D Systems) was used in varying concentrations such as bFGF. One potential mechanism of the angiogenic activity (11.25 ng/ml or 0.625 nM to 180 ng/ml or 10 nM). PBS served as the ␣ ␤ of IL-18 that we examined is through the involvement of v 3 negative control throughout these experiments, and testing PBS with 2% ␣ ␤ BSA, the diluent used to reconstitute the lyophilized cytokines, resulted in integrin (34). This role of v 3 was examined in the IL-18-induced migration of endothelial cells. IL-18 induced tube formation in similar baseline levels of HMVEC migration. Additionally, this assay was also performed with a representative RA SF sample after incubation for 1 h Matrigel matrix in vitro. To examine the role of IL-18 in angio- at 37°C with combinations of neutralizing Ab (25 ␮g/ml) to human IL-18 genesis in vivo, we also implanted IL-18 in Matrigel plugs in mice, and ENA-78 or IL-8 before adding samples to the bottom of the wells. and found a significant increase in blood vessel formation over Comparison was made to control nonspecific Abs, goat IgG for anti-IL-18 control, as measured by hemoglobin concentration. This effect was control, and mouse IgG for anti-IL-8 and anti-ENA-78. A confirmatory Downloaded from ␣ assay was performed with rhIL-18 and the addition of neutralizing anti- not decreased by implanting neutralizing anti-TNF- Ab in the human IL-18 Ab or nonspecific isotype-matched control Ab (25 ␮g/ml of Matrigel plug. Finally, in an alternative model of angiogenesis, a each). A comparative dose response to the cytokines IL-18, IL-8, and wound model, we implanted sponge discs embedded ENA-78 (PeproTech) was also examined. The membranes were then re- with IL-18 into mice and showed a 4-fold increase in angiogenesis moved, fixed in methanol for 1 min, and stained with Diff-Quik. Readings compared with controls. These findings support a novel role for represent the number of cells migrating through the membrane (the sum of three high power ϫ40 fields/well, averaged for each quadruplicate well).

IL-18 as an angiogenic mediator in RA and may elucidate a po- http://www.jimmunol.org/ tential therapeutic target for angiogenesis-directed disease. Checkerboard analysis Checkerboard analyses were performed as previously described (14, 35) Materials and Methods using the Boyden chamber. Briefly, increasing concentrations of IL-18 (0, Reagents 1, 10, and 100 nM) were added to the cell suspensions in the bottom wells in addition to the top wells. The effect of an alteration or decrease in the Recombinant human (rh) IL-1␤ and TNF-␣, with sp. act. of 2 ϫ 107 and resultant gradient was examined. The assay was performed three times and 1.3 ϫ 107 U/mg, respectively, were gifts from Upjohn-Pharmacia analyzed in a similar manner as above. ␣ (Kalamazoo, MI). Some TNF- was from R&D Systems (Minneapolis, ␣ ␤ MN). rhIL-18 was purchased from R&D Systems, with sp. act. measured The role of v 3 integrin in IL-18-mediated migration

␥ by guest on September 28, 2021 by ability to induce IFN- produced by KG-1 cells in the presence of 10 To determine whether IL-18 induces endothelial migration via ␣ ␤ inte- ␣ ␥ v 3 ng/ml rhTNF- (100 ng/ml rhIL-18 induces 0.5–2 ng/ml IFN- produced grin, we employed the modified Boyden chamber migration assay. by 105/ml KG-1 cells), and was used for the majority of the assays. An ␣ ␤ HMVECs were preincubated for1hat37°C with Ab to integrin v 3 (25 additional source of rhIL-18 was PeproTech (Rocky Hill, NJ), with IL-18 ␮g/ml), in an attempt to inhibit its function in the migration process, or ␥ sp. act. determined by the dose-dependent stimulation of IFN- production nonspecific isotype-matched control Ab. The cell suspensions were then by human PBMC costimulated with human IL-12 (ED50 of 5 ng/ml); and ␣ ␤ incubated again with Ab to v 3 or nonspecific Ab and used in the cell the experiments in which this IL-18 was used included the checkerboard migration assay, as described above. The test substances included IL-18 assay, the chemotaxis assays comparing the activity of IL-18 with IL-8 or (10 nM), bFGF (60 nM) as positive control, and PBS as negative control. epithelial neutrophil-activating peptide-78 (ENA-78), and the Matrigel in This assay was performed three times, with additional positive control ␣ vivo assays with IL-18 and anti-TNF- Ab. PeproTech was also the source TNF-␣ (15 ng/ml) or negative control VEGF (10 ng/ml). for the rhIL-8 and ENA-78, with biological activities determined by sig- nificant ability to chemoattract human peripheral blood using a IL-18 induced HMVEC proliferation concentration of 10–100 ng/ml and 5–10 ng/ml, respectively. Neutralizing monoclonal mouse anti-human IL-8, mouse anti-human ENA-78, and HMVEC proliferation in response to rhIL-18 was determined using a polyclonal goat anti-human IL-18 Ab were obtained from R&D Systems. modified method from previously described procedures (14, 36). Briefly, Neutralizing mouse anti-TNF-␣ was obtained from R&D Systems. Mono- HMVECs fed 1 day before use (with a fresh media exchange to obtain ␣ ␤ optimal growth conditions) were trypsinized and resuspended at 5 ϫ 104 clonal mouse anti-human integrin v 3 (vitronectin receptor) was obtained from BioWhittaker Clonetics (Temecula, CA). Nonimmune rabbit serum cells/ml in EBM (Clonetics) ϩ 2% FBS ϩ gentamicin. An equivalent 3 (R&D Systems) and mouse serum IgG (PharMingen, San Diego, CA) were number of cells (2.5 ϫ 10 )in50␮l cell suspension was plated noncon- negative controls. Nonspecific goat IgG control from R&D Systems was fluently onto each well of a 96-well plate and allowed to adhere for4hat also used. Recombinant bovine acidic FGF (aFGF) and recombinant hu- 37°C. Serial IL-18 dilutions were prepared in this media at 2ϫ concentra- tion, of which 50 ␮l was added to the 50 ␮l cell suspension already in the man bFGF with sp. act. defined as ED50 of 0.1–0.3 ng/ml and 0.1–0.25 ng/ml were purchased from R&D Systems. Recombinant VEGF was ob- wells, resulting in 100 ␮l1ϫ dilutions. In a similar manner, serial dilutions tained from Endogen (Woburn, MA). Growth factor-reduced Matrigel Ma- of bFGF were also prepared for a positive control; media alone were added trix, derived from Engelbreth-Holm-Swarm mouse tumor, was purchased for the negative control. All dilutions were performed in quadruplicate from Becton Dickinson (Bedford, MA). Human microvascular endothelial wells. The plate was incubated for up to 72 h at 37°C. The Cell Titer 96 cells (HMVECs) were obtained from Clonetics (San Diego, CA). Diff-Quik Aqueous nonradioactive cell proliferation assay kit (Promega, Madison, was purchased from Baxter (North Chicago, IL). A Kontes’ homogenizer WI), which contains a tetrazolium compound and phenazine methosulfate was obtained from Kontes Glass (Vineland, NJ); tetramethylbenzidine and solution, was diluted according to the recommended protocol and added to methemoglobin were obtained from Sigma (St. Louis, MO); and ethylene each well (20 ␮l/well). The plates were incubated at 37°C for 2–12 h until vinyl acetate copolymer was obtained from Elvax (Dupont, color development occurred. Absorbance was read at 490 nm on an ELISA Wilmington, DE). plate reader with Microplate Manager (Bio-Rad, Richmond, CA). The ab- sorbance readings were performed when the maximal response (plateau Patient population y-axis value) by bFGF concentrations represented at least a doubling from baseline control (no stimulation), and the absorbance at 490 nm (y-axis) vs SFs were isolated from patients with RA during therapeutic arthrocentesis. concentration of stimulant (x-axis) was plotted and examined. Proliferation RA SFs were used for HMVEC migration assays. All specimens were curves for HMVECs were determined, and measurements were obtained obtained with Institutional Review Board approval. and compared in the log proliferation phase, which was 3 days (t ϭ 72 h) 1646 EVIDENCE OF IL-18 AS A NOVEL ANGIOGENIC MEDIATOR

for the conditions described. Serial dilutions of IL-18, ranging from 1 fM was added to each sample, the plate was allowed to develop at room tem- to 100 nM, or bFGF (positive control), ranging from 10 fM to 100 nM, in perature for 15–20 min with gentle shaking, and the reaction was termi- ␮ ␮ EBM and 2% FBS in a 100- l vol were assayed. nated with 150 l2NH2SO4 after 3–5 min. Absorbance was read with an ELISA plate reader at 450 nm. To calculate hemoglobin concentrations, the Matrigel in vitro HMVEC tube formation assay for in vitro values were normalized by the weight of the plugs. angiogenesis Masson’s trichrome staining of Matrigel plugs Matrigel was used to examine HMVEC tube formation in response to IL-18. Matrigel was plated in eight-well chamber slides after thawing on To examine the Matrigel plugs histologically, some of the plugs were par- ice, and the Matrigel was allowed to polymerize at 37°C for 30–60 min. affin embedded. Sections (5 ␮m) were deparaffinized and stained using HMVECs were removed from culture, trypsinized, and resuspended at 4 ϫ Masson’s trichrome straining. In brief, sections were hydrated in distilled 104 cells/ml in Medium 199 (Invitrogen, Carlsbad, CA) containing 2% water, dipped in Bouin’s solution for1hat56°C, washed in water, dipped FBS and 200 ␮g/ml endothelial cell growth supplement (Becton Dickin- in Weigert’s iron solution for 7 min, washed, and dipped in Biebrich scar- son, Bedford, MA). Four hundred microliters of cell suspension were let-acid fuchsin solution for 2 min. Sections were then rinsed, incubated in added to each chamber, followed by 1, 5, or 10 nM IL-18, 50 nM PMA, or phosphomolybdic-phosphotungstic acid solution for 10 min, dipped in an- vehicle control (PBS for IL-18, and DMSO for PMA) placed directly into iline blue solution for 5 min, rinsed, dipped in glacial acetic acid solution the suspension in the chambers. The chamber slides were then incubated for 3–5 min, and dehydrated in two changes of 95% alcohol, 100% alcohol, for 16–18 h at 37°C in 5% CO2 humidified atmosphere. Culture media and xylene. The slides were mounted with Cytoseal 60 (Stephens Scien- were aspirated off the Matrigel surface, and the cells were fixed with meth- tific, Kalamazoo, MI). anol and stained with Diff-Quick Solution II. Each chamber was photo- ϫ graphed using Polaroid Microcam camera at 22 magnification. The num- Sponge granuloma in vivo angiogenesis assay ber of tubes formed was quantitated blindly, as previously described (36).

Briefly, a connecting branch between two discrete endothelial cells was The animal model of inflammatory angiogenesis using a sponge granuloma Downloaded from counted as one tube and required a consistent intensity, thickness, and has been described by Fajardo and colleagues (41). C57BL/6 mice were minimum length (Ͼ1mmona4ϫ enlarged copy of the photomicrograph) divided into several groups: negative control PBS; positive control, aFGF; to be counted. This tube analysis was determined from one focal plane and test substance, IL-18. One-centimeter sponge discs cut from sheets of generated from a photomicrograph of the center of the well, focusing on the 2-mm-thick polyvinyl alcohol foam sponges obtained from sterile packs surface of the Matrigel, which was also the most prominent portion of the (M-pact, Eudora, KS) were prepared, and a 2-mm hole was cut into the disc Matrigel (Ͻ0.5 cm deep after pipetting 400 ␮l into the chamber). Each center to serve as a depot for control PBS or test substances, and then concentration of the test substance and vehicle control was performed in closed back with the cut plug. After adding the stimulant to the center hole, triplicate within the same chamber, while PMA and its vehicle DMSO were the sponge discs were coated with an inert slow release ethylene vinyl http://www.jimmunol.org/ tested in the chamber’s remaining two wells. This assay was performed acetate copolymer, Elvax (DuPont Packages), and both disc surfaces were four times for each IL-18 concentration. sealed with Millipore filters (0.45 ␮m) using Millipore glue number 1 (Millipore, Bedford, MA). The animals were anesthetized with pentobar- Matrigel plug angiogenesis in vivo assay bital 40 mg/kg of body weight, the hair on their left back area was shaved and sprayed with 70% ethanol, and the disc inserted into the s.c. layer at a To examine the effect of IL-18 on angiogenesis in vivo, we used a Matrigel site 2 cm distant from the incision, which was then sutured to prevent disc plug assay (37, 38). C57BL/6 mice were anesthetized by Metofane inha- dispersion. After 9–12 days, the animals were sacrificed, and the sponge lation. Each mouse was shaved on its ventral aspect and given an s.c. discs were harvested. The sponges were then analyzed by hemoglobin injection of sterile Matrigel (500 ␮l/injection) with a 27-gauge needle. quantitation. Matrigel plus PBS served as the negative control, Matrigel containing bo- vine aFGF (1 ng/ml) served as the positive control, and Matrigel with IL-18 by guest on September 28, 2021 (10 nM) was the test substance. After 7–10 days, the mice were sacrificed Hemoglobin quantitation of sponge by by Metofane inhalation. The Matrigel plugs were then carefully dissected tetramethylbenzidine out, with removal of any surrounding connective tissue, and then analyzed by hemoglobin measurement or by histology. The Drabkin’s reagent To quantitate angiogenesis in the sponge model, the sponges were placed method of determining hemoglobin content was employed. For histological in 2 ml double-distilled water in 24-well plates. The sponges were cut into analysis, we also embedded some of these plugs in paraffin for tissue sec- small pieces with scissors and then homogenized with a Kontes’ hom- tioning and staining. genizer. The samples were then spun at 8000 rpm for 6 min, and the ␮ Additionally, we examined the effect of IL-18 in the presence of neu- resultant supernatants were filtered through a 0.22- m filter. The hemo- tralizing anti-TNF-␣ Ab in the Matigel plug. Although the aim of this globin percentage was determined by tetramethylbenzidine (39, 40). The experiment was not to block endogenous TNF-␣ completely with the dose protocol as described for the hemoglobin concentration determination in ␮ employed, the effect of IL-18 while inhibiting the local activity of TNF-␣ the Matrigel plug assay above was then followed, in this case using 50 l on angiogenesis at the site of the Matrigel was assessed. Matrigel plugs of the filtered supernatant. To calculate hemoglobin concentrations, the containing 10 nM IL-18 (PeproTech) and 25 ␮g/ml anti-TNF-␣ Ab or values were normalized by the weight of the sponges. isotype-matched control IgG were injected into mice (n ϭ 18 per group). Matrigel with PBS also served as a negative control. Statistical analysis Hemoglobin determination of angiogenesis in Matrigel plugs Statistical analysis was performed using the nonparametric Wilcoxon rank- sum test, with statistical significance defined by p Ͻ 0.05. Values were The Matrigel plugs dissected from the mice were carefully stripped of any expressed as means Ϯ SEM. remaining peritoneum. The plugs were weighed by placing them into pre-

weighed 1-ml tube of ddH20 and homogenized for 5–10 min on ice. The samples were spun at 10,000 rpm on a microcentrifuge for 6 min, and the Results supernatants collected for hemoglobin measurement. Fifty microliters of IL-18 induces HMVEC migration supernatant were mixed with 950 ␮l Drabkin’s reagent and allowed to sit at room temperature for 15–30 min, and then 100 ␮l of this mixture was To determine the effect of rhIL-18 on HMVEC migration, we placed in a 96-well plate. Absorbance was read with a Microplate Manager tested varying concentrations of IL-18 in a modified Boyden ELISA reader at 540 nm. Hemoglobin concentration was determined by chamber assay. A representative assay of three is shown (Fig. 1). comparison with a standard curve in g/dl. These values were normalized by IL-18 induced HMVEC migration in a concentration-dependent dividing the hemoglobin percentage by the plug weight. Hemoglobin con- Ͻ centration is a reflection of the number of blood vessels in the plugs. manner, reaching significance at 1 nM ( p 0.05). Cell migration For the Matrigel in vivo experiments comparing plugs containing IL-18 measured by the number of migrating cells/well was determined and anti-TNF-␣ or IL-18 and control Ab, the hemoglobin concentrations by three blinded readings per each quadruplicate well. The mean were determined by the tetramethylbenzidine method (39, 40). In brief, cell migration induced by 1 and 10 nM IL-18 was 18 Ϯ 0.8 and methemoglobin (17.5 mg) in 25 ml double-distilled water was brought to Ϯ Ϯ room temperature, and glycerol 30% was added. Serial dilutions were pre- 20 0.4 cells/well, respectively, while 100 nM IL-18 (29 3.3 pared to generate a standard curve. Fifty microliters of supernatant or stan- cells/well) exceeded cell migration by control stimulant 60 nM dard were added to a 96-well plate in duplicate, 50 ␮l tetramethylbenzidine bFGF (24.7 Ϯ 1 cells/well). The Journal of Immunology 1647

FIGURE 1. IL-18 induces HMVEC migration. An assay with increas- ing concentrations of IL-18-inducing HMVEC migration, which is signif- p Ͻ ,ء) icant at a concentration of 1 nM and higher compared with control 0.05). The mean number of migrating cells per well Ϯ SEM per replicate well is shown compared with positive control stimulant bFGF (60 nM). This assay is representative of three similar assays. Downloaded from Neutralization of IL-18 inhibits HMVEC migration To confirm the specificity of rhIL-18-induced HMVEC migration, we preincubated the IL-18 stimulant with either neutralizing goat anti-human IL-18 IgG Ab or nonspecific isotype-matched control IgG Ab, at a concentration of 25 ␮g/ml for 1 h before and during FIGURE 2. The specificity of IL-18-induced HMVEC migration. A,

the2hofthecell migration assay. The response to the stimulant http://www.jimmunol.org/ with or without Ab was determined by the number of migrating Anti-IL-18 Ab inhibits IL-18-induced HMVEC migration. Inhibition of cells/well based on three blinded readings per each quadruplicate IL-18-induced HMVEC migration was demonstrated with preincubation of well (Fig. 2A). Neutralization of IL-18 resulted in inhibition of cell HMVECs with neutralizing anti-IL-18 Ab vs nonspecific control goat IgG p Ͻ 0.05). Results represent the mean number of cells per well Ϯ ,ء) Ab migration to basal values (9 Ϯ 1.32 cells/well) similar to that seen Ϯ SEM migrating, and this assay is representative of three similar assays. upon incubation of HMVECs with control PBS (7 0.71 cells/ Positive control stimulation with bFGF vs PBS is also shown. B, IL-18 is well). Results are representative of three similar assays. In addi- primarily chemotactic for HMVECs. A checkerboard chemotaxis assay tion, 10 nM IL-18 alone stimulated HMVEC migration (27.25 Ϯ performed with increasing concentrations of IL-18 in the upper and lower 1.44 cells/well) comparable to that by 60 nM bFGF (28.75 Ϯ 0.65 wells (containing HMVECs) of the chamber is shown, with results ex- cells/well). pressed as mean number of cells/well Ϯ SEM migrating across the mem- by guest on September 28, 2021 brane filter. In this assay, bFGF induced 24.9 Ϯ 2.2 cells/well and PBS IL-18 is primarily chemotactic for HMVECs induced 12 Ϯ 5 cells/well, and this assay is representative of three similar The checkerboard analysis was performed with the chemotaxis assays. assay using increasing concentrations of IL-18 in the cell suspen- sions at the bottom of the chamber as well as in the top of the chamber. A representative assay of three is shown (Fig. 2B). The concentrations were chosen for their known induction of SF alone (40.6 Ϯ 1.4 cells/well) ( p Ͻ 0.05). These findings sug- HMVEC migration, starting at 1 nM, although the effect of 100 nM gest a prominent role for IL-18 in RA SF-induced HMVEC may not have been greatly dampened by the addition of 1 or 10 nM migration. on the opposide side of the membrane. There was evidence of Contribution of IL-18, IL-8, and ENA-78 to RA SF-induced some HMVEC migration induced by the presence of IL-18 within HMVEC migration the cell suspension, which exceeded that by PBS, reflecting a che- mokinetic contribution. However, the presence of equivalent con- To compare the relative contribution to RA SF-induced HMVEC centrations of IL-18 on either side of the membrane reduced the migration, we preincubated four different RA SF samples with resultant migration to levels closer to PBS control. These data neutralizing Abs to IL-18, IL-8, or ENA-78 (Fig. 4). Immu- suggest a primarily chemotactic effect of IL-18 on HMVEC nodepletion of IL-18 resulted in a similar percentage of HMVEC migration. migration inhibition (68 Ϯ 5%) as immunodepletion of IL-8 (64 Ϯ 7%). However, immunodepletion of IL-18 resulted in a 1.6-fold IL-18 contributes to RA SF-induced HMVEC migration greater percent inhibition than immunodepletion of ENA-78 (43 Ϯ To determine the relative contribution of IL-18 to RA SF-induced 6%). Values shown are means generated from four different RA SF HMVEC migration, we preincubated RA SF samples with neu- patient samples and are significant compared with isotype-matched p Ͻ 0.05). Inhibition was not further enhanced with ,ء) tralizing goat anti-human IL-18 Ab or nonspecific isotype-matched control Ab control goat IgG Ab, at a concentration of 25 ␮g/ml for 1 h before combinations of neutralizing Abs to IL-18 and either IL-8 or and during2hoftheassay. We found significant inhibition of RA ENA-78 (62 Ϯ 3% and 70 Ϯ 3%, respectively). SF-induced HMVEC migration with anti-IL-18 Ab, compared with nonspecific Ab (Fig. 3). The inhibition of HMVEC migration Comparison of IL-18-induced HMVEC migration with after neutralization of the RA SF samples (n ϭ 4) with the anti- -induced migration by IL-8 and ENA-78 IL-18 Ab (13.5 Ϯ 2.9 cells/well) was 68 Ϯ 5% compared with The neutralization of IL-18 present in RA SF reduced SF-induced sham neutralization with the control IgG Ab (41.5 Ϯ 4.3 cells/ HMVEC migration to a similar or greater degree as neutralization well). HMVEC migration with control IgG was equivalent to RA of IL-8 or ENA-78, respectively. We compared the bioactivities of 1648 EVIDENCE OF IL-18 AS A NOVEL ANGIOGENIC MEDIATOR

FIGURE 3. Inhibition of RA SF induced HMVEC migration by IL-18 FIGURE 5. IL-18 induces HMVEC migration comparable with the che- immunodepletion. RA SF-induced HMVEC migration was inhibited by mokines IL-8 and ENA-78. The relative bioactivities of rhIL-18, IL-8, and neutralization of RA SF IL-18 with anti-human IL-18 Ab before perform- ENA-78 were compared in a dose-response chemotaxis assay using similar p Ͻ 0.05). Comparatively, there was no effect with non- increasing concentrations (0.001–100 nM). The chemotactic effect of IL-8 ,ء) ing the assay specific Ab. Results represent the mean number of cells/well Ϯ SEM mi- became significant at doses lower than shown (0.001–0.01 nM) and was grating, and shown is a composite of the assays (n ϭ 4 patients examined). greater at 0.1 nM compared with that of IL-18 and ENA-78, both of which are effective at higher doses (0.1–1 nM), although to a less degree than IL-8 Downloaded from at the highest dose (100 nM). Results represent the mean number of mi- these recombinant human cytokines in the HMVEC migration as- grating cells per well Ϯ SEM, and shown is a representative assay of three say. The dose responses were generated from three different ex- similar assays. periments, and a representative assay is shown (Fig. 5). IL-18 and ENA-78 induced HMVEC migration to a similar degree at the lower concentrations 0.1–10 nM. IL-18 exceeded the effect of ␣

on HMVECs. TNF- is angiogenic in part via its action on integrin http://www.jimmunol.org/ ␣ ␤ ␣ ␤ ENA-78 at higher concentrations (10–100 nM). In comparison, v 3 (34). Thus, we examined whether blocking v 3 would in- IL-8 showed greater effect on HMVEC migration at the lower and hibit IL-18-induced angiogenesis in the HMVEC migration assay. ␣ ␤ higher doses compared with IL-18 and ENA-78, which had more HMVECs were first incubated with anti- v 3 Ab or nonspecific comparable bioactivities. Overall, assaying a wide range of con- control IgG at 25 ␮g/ml for 1 h before and during the chemotaxis centrations, it appears that IL-8 is a more potent angiogenic agent assay. IL-18 (10 nM) was then used to stimulate HMVEC migra- than IL-18, which is of greater potency than ENA-78. tion, and comparison was made with the control stimulants bFGF and TNF-␣. To illustrate a contrasting mechanism, VEGF was also ␣ ␤ IL-18 induces HMVEC migration via v 3 integrin used as another stimulant, which would not be expected to be

␣ ␤ by guest on September 28, 2021 To determine the mechanism by which IL-18 mediates angiogen- inhibited by blocking v 3, since the angiogenic ability of VEGF ␣ ␤ esis, we hypothesized that IL-18 may act through present is thought to occur via the integrin v 5 (42). Blocking HMVEC ␣ ␤ v 3 resulted in significant inhibition of IL-18-induced cell mi- gration comparable with that seen with control (Fig. 6). The same effect was seen with 25 or 10 ␮g/ml blocking Ab concentration. ␣ ␤ Significant inhibition of cell migration by blocking HMVEC v 3 before stimulation with bFGF and TNF-␣ was also demonstrated, while no change in cell migration occurred after stimulation with VEGF. Shown are the representative data from three separate assays.

Effect of IL-18 on endothelial cell proliferation The effect of rhIL-18 on endothelial cell proliferation was assessed using varying concentrations of IL-18 (10Ϫ6 to 100 nM) to stim- ulate HMVEC growth. Control stimulation with bFGF (10Ϫ5 to 10 nM) demonstrated an expected 2-fold increase in HMVEC pro- liferation. However, IL-18 did not directly stimulate or inhibit HMVEC proliferation on repeated assays (n ϭ 4), but behaved similarly to PBS control. The time point of 72 h at which cell growth was measured was selected when control bFGF-induced FIGURE 4. Comparison of RA SF IL-18-induced HMVEC migration proliferation was optimal. Additionally, proliferation curves for with the contribution of other cytokines. We examined four different RA HMVECs were determined for bFGF, media alone, and IL-18; and SF samples and immunodepleted IL-18 to compare the relative contribu- measurements were compared in the log proliferation phase (Fig. 7). tion to RA SF-induced HMVEC migration with that of the chemokines IL-8 and ENA-78. RA SF samples were preincubated with neutralizing IL-18 induces angiogenesis in Matrigel in vitro Abs to IL-18, IL-8, or ENA-78 and compared with nonspecific Ab controls Endothelial cell tube formation in Matrigel is one measure of an- (25 ␮g/ml). Immunodepletion of IL-18 resulted in similar percent inhibi- tion of cell migration as immunodepletion of IL-8 and a 1.6-fold greater giogenesis in vitro. The role of IL-18 in angiogenesis was assessed percent inhibition of cell migration than immunodepletion of ENA-78. The by determining the induction of HMVEC tube formation on Ma- effect of combining neutralizing Abs is also shown. Results represent the trigel matrix plated onto eight-well chamber slides. Tube-like mean number of migrating cells/well Ϯ SEM, and shown is a composite of structures formed after a 16-h incubation in the presence of IL-18, four assays (n, number of RA SF patient samples). whereas grossly fewer tubes formed with PBS control (Fig. 8). The Journal of Immunology 1649

␣ ␤ FIGURE 6. Anti- v 3 Ab inhibits IL-18-induced HMVEC migration. Inhibition of IL-18-induced HMVEC migration was ␣ ␤ demonstrated by preincubation of HMVECs with anti- v 3 Ab ␣ ␤ vs nonspecific control Ab. The effect of blocking v 3 integrin before stimulation with IL-18 or the positive controls bFGF and TNF-␣ is compared with nonspecific blockade with goat IgG. Results represent the mean number of cells/well Ϯ SEM mi- grating, and shown is a representative assay (n ϭ 3 assays). There was significant suppression of IL-18-induced HMVEC ␣ ␤ migration after v 3 blockade, as was seen with bFGF and TNF-␣. VEGF served as a negative control stimulant. Ab con- centrations were 25 ␮g/ml.

Shown are representative wells (magnified ϫ88): A shows tube for each test group. The hemoglobin content of the 10 nM and 1 formation with IL-18 (1 nM), compared with its PBS control (B), nM IL-18-containing plugs was 52-fold and 64-fold higher, re- and C shows tube formation with IL-18 (10 nM), compared with spectively, than that for PBS control (Fig. 9B): 1.65 Ϯ 0.45 g/dl/ Downloaded from its PBS control (D). IL-18 at 1 and 10 nM induced significant (mg) for 10 nM IL-18, 2.05 Ϯ 1.3 g/dl/(mg) for 1 nM IL-18, and HMVEC tube formation with an increase of 77% and 87% com- 0.032 Ϯ 0.02 g/dl/(mg) for PBS ( p Ͻ 0.05). IL-18-stimulated pared with PBS control ( p Ͻ 0.05). Results represent the average angiogenesis is comparable with positive control aFGF in the Ma- of three wells per four similar assays at two different IL-18 con- trigel plugs. centrations, in which the number of tubes formed per each well

were blindly counted from photomicrographs taken by Polaroid at http://www.jimmunol.org/ ϫ22 magnification of the center of the well focused on the Ma- trigel surface. Shown are the data for 1 nM IL-18 stimulation (Fig. 8E). Comparison with the stimulant PMA and its vehicle control DMSO is also shown.

IL-18 induces angiogenesis in Matrigel in vivo The angiogenic role of IL-18 in vivo was assessed by examining the effect of IL-18 on blood vessel growth in the Matrigel plugs in mice. The Matrigel plugs containing IL-18 (10 nM) induced sig- by guest on September 28, 2021 nificantly greater angiogenesis than control PBS. Histology illus- trates the difference qualitatively in the representative photomicro- graphs of Masson trichrome-stained sections (ϫ50 magnification) (Fig. 9A). Marked new blood vessel growth can be seen in the IL-18-containing plugs (left panel) compared with the control (right panel). Five Matrigel plugs were embedded with stimulus

FIGURE 7. IL-18 does not affect HMVEC proliferation. Varying con- centrations of rhIL-18 (10Ϫ6–100 nM) were used to assess the effect on FIGURE 8. IL-18 induces endothelial cell tube formation on Matrigel in HMVEC proliferation, which was compared with that of control bFGF vitro. Photomicrographs (taken at ϫ22, enlarged to ϫ88 magnification) of (10Ϫ5–10 nM) and nonstimulated PBS controls. Cell growth was measured representative wells show the increased IL-18-induced HMVEC tube for- as absorbance readings at 72 h, during the optimal phase of growth. An mation in comparison with PBS control: A, IL-18 (1 nM) with B, PBS expected 2-fold increase was demonstrated after 100 pM and 1 nM bFGF control; and C, IL-18 (10 nM) with D, PBS control. Shown are results stimulation, with significant difference from PBS controls seen also at 10 representative of three experiments. E, HMVECs form significantly greater p Ͻ 0.05). However, the effect of IL-18 was similar to number of tubes in Matrigel matrix in response to 1 nM IL-18 vs PBS ,ء) nM bFGF p Ͻ 0.05). Positive control stimulant is PMA; vehicle control is ,ء) nonstimulated PBS controls on repeated assays, and shown is a composite control of the data (n ϭ 3). DMSO; and the control for IL-18 is PBS. 1650 EVIDENCE OF IL-18 AS A NOVEL ANGIOGENIC MEDIATOR

FIGURE 10. IL-18-induced angiogenesis in the Matrigel plug in vivo is not inhibited by the addition of anti-TNF-␣ Ab. Blood vessel formation in Matrigel plugs containing IL-18 (10 nM) and neutralizing anti-TNF-␣ Ab (25 ␮g/ml) or control IgG (25 ␮g/ml) was determined after 7-day incuba- tion. The two test groups had statistically similar results, although both

were assoicated with 2–3 times greater hemoglobin contents than the PBS Downloaded from p Ͻ 0.05). The values represent the concentration of hemoglobin ,ء) group (grams per deciliter)/Matrigel plug weight (milligrams) Ϯ SEM, with n ϭ number Matrigel plugs.

g/dl/(mg) for IL-18, 4.45 Ϯ 1.81 g/dl/(mg) for aFGF, and 1.13 Ϯ

0.29 g/dl/(mg) for PBS ( p Ͻ 0.05). http://www.jimmunol.org/

Discussion FIGURE 9. IL-18 induces angiogenesis in the Matrigel plug assay in In this study, we examined the potential role of IL-18 in mediating vivo. A, A representative assay shows Masson trichrome staining of blood angiogenesis through its action in both in vitro and in vivo bioas- vessels in a paraffin section of the Matrigel plug (ϫ50). IL-18-induced says. We found evidence that IL-18 possesses angiogenic proper- blood vessel formation is compared with PBS control. Blood vessels are ties, a novel function for IL-18. While IL-18 has been implicated present in the IL-18-treated plug (arrows). B, The IL-18 (1 and 10 nM) in RA pathogenesis, IL-18’s complex array of functions and mech- induced angiogenesis in the Matrigel plug to a significantly greater degree anisms by which it promotes inflammation has not been fully elu- p Ͻ 0.05). The values represent the by guest on September 28, 2021 ,ء) compared with PBS vehicle control concentration of hemoglobin (grams per deciliter)/Matrigel plug weight cidated. IL-18 is up-regulated in the SF and serum from RA pa- (milligrams) Ϯ SEM, with n ϭ number Matrigel plugs. tients compared with osteoarthritis patients (21). The use of IL-18 has also been shown to facilitate the development of erosive, in- flammatory arthritis when administered in murine type II collagen- IL-18 induces angiogenesis in Matrigel in vivo independently of induced arthritis (25). IL-18 promotes articular Th1 responses and TNF-␣ is thus important in the pathophysiology of RA (21). Integral to the inflammation in RA is the process of angiogen- To examine the angiogenic effect of IL-18 in the presence of neu- esis. We found that IL-18 induced endothelial migration in the ␣ tralizing anti-TNF- Ab, Matrigel plugs containing 10 nM IL-18 HMVEC migration assay, which is one aspect of angiogenesis in ␮ ␣ and 25 g/ml anti-TNF- Ab or isotype-matched control IgG were vitro. IL-18 induced HMVEC migration in the nanomolar range in injected into mice (n ϭ 18 per group). After 7 days incubation, the mean normalized hemoglobin contents were determined, and there was no significant difference between the two groups: 2.67 Ϯ 0.62 g/dl/(mg) for the IL-18 plus IgG-containing plugs vs 2 Ϯ 0.31 g/dl/(mg) for the IL-18 plus anti-TNF-␣-containing plugs (Fig. 10). The hemoglobin content of the 10 nM IL-18 plus IgG-con- taining plugs was 3.7 times greater than that of PBS control.

IL-18 induces angiogenesis in the sponge granuloma model in vivo To examine whether IL-18 induced inflammatory angiogenesis in vivo, the sponge granuloma model was used. In this model, an- giogenesis was measured as increased blood vessel growth in discs made from polyvinyl alcohol foam sponges containing either IL-18 or control stimulant aFGF and implanted into mice. After 11 days, there was a significant increase in blood vessel growth with FIGURE 11. IL-18 induces angiogenesis in the mouse sponge disc as- IL-18 stimulation vs PBS control (Fig. 11). A 5-fold increase in say in vivo. IL-18 (5 nM) induced angiogenesis in the mouse sponge gran- angiogenesis with IL-18 (10 nM) was demonstrated by greater uloma model, as measured by hemoglobin concentration, and there was a p Ͻ 0.05). aFGF served as the ,ء) hemoglobin concentrations measured with tetramethylbenzidine, significant increase over control PBS and the hemoglobin value induced by IL-18 was even greater than positive control. Results represent the mean Ϯ SEM. n, Number of sponge that for aFGF, which resulted in a 4-fold increase: 5.77 Ϯ 1.21 discs. The Journal of Immunology 1651 a concentration-dependent manner, between 1 and 100 nM, with To elucidate a mechanism by which IL-18 mediates angio- ␣ ␤ 100 nM IL-18 achieving a comparable degree of cell migration to genesis, we examined the potential role of the integrin v 3 in 60 nM bFGF, a potent chemotactic stimulus. We were able to HMVEC migration. Endothelial cell invasion, migration, and pro- verify this finding by neutralizing the rhIL-18 with specific anti- liferation are regulated in part by the integrin family of cell adhe- ␣ ␤ IL-18 Ab, thereby blocking its chemotactic function compared sion molecules, and we have shown that v 3 is up-regulated on with blocking with control-nonspecific IgG. Checkerboard analy- RA compared with osteoarthritic or normal synovial blood vessels sis demonstrated a primarily chemotactic property of IL-18, whose (48). We therefore hypothesized that like bFGF and TNF-␣ (34), effect is dependent on a concentration gradient rather than random IL-18 may also act via this integrin in angiogenesis. After blocking ␣ ␤ ␣ ␤ chemokinesis. With the knowledge that RA SF is highly chemo- v 3 on HMVECs with specific anti- v 3 Ab, the degree of IL- tactic and that IL-18 is elevated in RA SF, we examined the con- 18-induced cell migration was inhibited by 34% compared with ␣ ␤ tribution of RA SF IL-18 to endothelial cell migration. Four dif- blocking with isotype-matched control Ab. v 3 antagonists have ferent RA SFs immunodepleted of IL-18 resulted in a 68 Ϯ 5% successfully modulated angiogenesis in a rabbit model of synovitis decrease in endothelial migration. This finding supports a potent (49). Targeting IL-18, which appears to function in part through role of IL-18 in endothelial cell migration in RA. this integrin, could be an alternative therapeutic approach for RA. To compare the contribution of RA SF IL-18 with other medi- Endothelial cell proliferation is another aspect of angiogenesis. ators of RA SF that also induce endothelial cell migration, we We did not demonstrate direct inhibition or enhancement of basal further immunodepleted RA SFs of IL-18 alone and in combina- endothelial cell (HMVEC) proliferation, which is in contrast to tion with the chemokines IL-8 or ENA-78 (12, 43). IL-18 appeared previous reports of the antimitogenic activity of IL-18. IL-18 at to account for a significant portion of chemotactic activity for 1–10 nM concentration has been reported to inhibit FGF-2-in- Downloaded from HMVECs in the SFs comparable with that by IL-8. The contribu- duced capillary endothelial cell proliferation (50). However, the tion by IL-18 was 1.6-fold greater than that by ENA-78. Combin- effect of IL-18 on basal proliferation, as examined in our study, ing immunodepletions of IL-18 and IL-8 or ENA-78 did not aug- was not explored. This group also found that murine, not human, ment suppression of HMVEC migration. The lack of further IL-18 suppressed FGF-2-induced corneal neovascularization in suppression suggests that inhibiting the effect of one of these cy- mice, although the IL-18 was administered by i.p. injections, and tokines may be blocking common pathways involved in HMVEC it is unclear whether local rather than systemic treatment would http://www.jimmunol.org/ have induced divergent results. Interestingly, they also found that migration. IL-18 accounted for approximately two-thirds of the IL-18 suppressed angiogenesis in the chick chorioallantoic mem- migrating activity of HMVECs in RA SFs. These results indicate brane model. Our finding that IL-18 did not induce HMVEC pro- that IL-18 is one of the main mediators of RA SF-induced HM- liferation is consistent with results obtained for other angiogenic VEC migration. mediators, such as soluble VCAM-1 or soluble E-selectin, which A direct comparison of the HMVEC-migrating activity of the rh also do not induce endothelial proliferation (14). IL-18 with that of the chemokines rhIL-8 and rhENA-78 in the Consistent with the chemotactic properties of IL-18, we also HMVEC migration assay revealed that IL-8 was a more potent showed IL-18 induction of HMVEC tube formation in Matrigel

agent than IL-18, which was more potent than or comparable with by guest on September 28, 2021 matrix in vitro. The degree of tube formation by IL-18 at 1 nM was ENA-78. comparable with the stimulant PMA (50 mM), a potent inducer of IL-18 is also ascribed many inflammatory functions that would HMVEC differentiation and tube formation (51). In the Matrigel in be consistent with a role in angiogenesis. For instance, IL-18 stim- vivo mouse angiogenesis model, IL-18 also stimulated actual ulates the production of other proangiogenic cytokines, such as blood vessel formation, which was visibly in excess of control. ␣ ␤ TNF- , IL-1 and IL-8, from human PBMCs, the responding cells The potency of IL-18 relative to other angiogenic stimulants in this being predominantly resting T cells and NK cells. IL-18’s stimu- assay can be assessed, since 1 nM (or 18 ng/ml) IL-18 produced lation of the chemokine IL-8 has also been shown in the macro- results similar to that induced by 1 ng/ml aFGF, which is clearly phage cell line derived from histiocytic lymphoma cell, U937, and one of the strongest angiogenic stimuli. By comparison, an excess the myelomonocytic leukemia cell line, KG-1 (44, 45), to produce of 100-fold more bFGF would be required to produce the same IL-8 and other C-X-C chemokines. Besides its direct angiogenic effect in this assay as 1 ng/ml aFGF (0.86 g/dl vs 1.30 g/dl hemo- role, other potential mechanisms of the angiogenic function of globin), and 10-fold more TNF-␣ (10 ng/ml hemoglobin) would be IL-18 may be through IL-8 and other angiogenic cytokines that it expected to produce twice the effect as 1 ng/ml aFGF (2.3 g/dl vs up-regulates. Indeed, IL-18 exerts an inflammatory cytokine cas- 1.3 g/dl hemoglobin) (37). cade in the mixed PBMC population by stimulating the constitu- The Matrigel in vivo assay also allowed for another examination tive expression of IL-18R on NK cells and lymphocytes, which of a potential mechanism behind the angiogenic activity of IL-18. leads to TNF-␣ production, ultimately stimulating The angiogenic effect of IL-18 in the presence of anti-TNF-␣ Ab IL-1␤ and IL-8 production (19). In turn, treatment with IL-1␤ and was evaluated by adding IL-18 and neutralizing anti-TNF-␣ Ab to TNF-␣ of RA synovial fibroblast cultures, not constitutively ex- the Matrigel before injection. Although the endothelial cells were pressing IL-18, induced IL-18 gene and protein production, sug- the target of the effect of IL-18 in cell migration and tube forma- gesting a feedback interaction to promote Th1 cytokine and cell tion observed in the earlier assays, the new blood vessel formation development in RA (21). in the Matrigel plug could involve cellular targets other than en- IL-18 promotes inflammation, not only via other cytokine ac- dothelial cells, at least indirectly. In our hands, blocking the local tions as above, but also through the promotion of other inflamma- TNF-␣ did not significantly alter the effect of IL-18 on Matrigel tory mediators, some of which are themselves implicated in an- plug hemoglobin content. However, it is not clear whether sys- giogenesis. IL-18 stimulates the expression of the soluble adhesion temic inhibition of TNF-␣ would have diminished the effect of molecule ICAM-1 on monocytic cell lines (46), and we have found IL-18. In addition, the downstream effect of IL-18 on PBMCs, that IL-18 up-regulates the expression of VCAM-1 on RA synovial lymphocytes, and other cells via up-regulation of other cytokines fibroblasts (47). Soluble VCAM-1 is potently angiogenic (14), and cannot be overlooked. The various mediators known to orchestrate it may be that IL-18 also acts to stimulate RA fibroblast VCAM-1 the angiogenic process may function in part through common path- expression, hence amplifying the angiogenic response. ways. Although redundancy in the system exists and serves to 1652 EVIDENCE OF IL-18 AS A NOVEL ANGIOGENIC MEDIATOR maintain the angiogenic state characteristic of certain physiologic 11. Strieter, R. M., S. L. Kunkel, H. J. Showell, D. G. Remick, S. H. Phan, or pathologic conditions, evidence from clinical trials clearly dem- P. A. Ward, and R. M. Marks. 1989. Endothelial cell of a neu- trophil chemotactic factor by TNF-␣, LPS, and IL-1␤. Science 243:1467. onstrates that blocking one key element through treatment target- 12. Koch, A. E., P. J. Polverini, S. L. Kunkel, L. A. Harlow, L. A. DiPietro, ing TNF-␣ alone reduced RA disease activity (52). Further studies V. M. Elner, S. G. Elner, and R. M. Strieter. 1992. -8 as a macrophage- examining potential mechanisms of angiogenesis, including com- derived mediator of angiogenesis. Science 258:1798. 13. Volin, M. V., J. M. Woods, A. M. Amin, M. A. Connors, L. A. Harlow, and mon signaling pathways, cell adhesion, and up-regulation of cy- A. E. Koch. 1999. Fracktalkine: a novel angiogenic chemokine in rheumatoid tokines, integrins, and other mediators, may shed light on the sig- arthritis. Arthritis Rheum. 42:S250. nificance of the current findings. 14. Koch, A. E., M. M. Halloran, C. J. Haskell, M. R. Shah, and P. J. Polverini. 1995. Angiogenesis mediated by soluble forms of E-selectin and vascular cell adhesion The sponge angiogenesis system allows testing of the effect of molecule-1. Nature 376:517. various antagonists and agonists on the proliferation of blood ves- 15. Okamura, H., H. Tsutsi, T. Komatsu, M. Yutsudo, A. Hakura, T. Tanimoto, sels and stroma, provides histological and quantitative informa- K. Torigoe, T. Okura, Y. Nukada, K. Hattori, et al. 1995. Cloning of a new cytokine that induces IFN-␥ production by T cells. Nature 378:88. tion, and is easily reproducible. This system applied to mice was 16. Ushio, S., M. Namba, T. Okura, K. Hattori, Y. Nukada, K. Akita, F. Tanabe, suitable to test the independent effect of IL-18 on angiogenesis, K. Konishi, M. Micallef, M. Fujii, et al. 1996. Cloning of the cDNA for human ␥ and specifically, new blood vessel formation, which this assay IFN- -inducing factor, expression in Escherichia coli, and studies on the biologic activities of the protein. J. Immunol. 156:4274. measures by virtue of the constant sample area of the disc and the 17. Dinarello, C. A. 1999. IL-18: A Th1-inducing, proinflammatory cytokine and planar growth of the vessels (41). IL-18 at 10 nM strongly induced new member of the IL-1 family. J. Allergy Clin. Immunol. 103:11. angiogenesis, as measured by a 5-fold increase in the amount of 18. Kohno, K., and M. Kurimoto. 1998. , a cytokine which resembles IL-1 structurally and IL-12 functionally but exerts its effect independently of hemoglobin over control, similar to the 4-fold increase seen with both. Clin. Immunol. Immunopathol. 86:11. the positive control stimulant aFGF (1 nM). This finding not only 19. Puren, A. J., G. Fantuzzi, Y. Gu, M. S. Su, and C. A. Dinarello. 1998. Interleu- Downloaded from confirms our results from the Matrigel plug in vivo assay, but it kin-18 (IFN␥-inducing factor) induces IL-8 and IL-1␤ via TNF␣ production from non-CD14ϩ human blood mononuclear cells. J. Clin. Invest. 101:711. establishes IL-18 as a direct angiogenic stimulus. 20. Dinarello, C. A., D. Novick, A. J. Puren, G. Fantuzzi, L. Shapiro, H. Muhl, In summary, we found that IL-18 induces endothelial cell che- D. Y. Yoon, L. L. Reznikov, S. H. Kim, and M. Rubinstein. 1998. Overview of motaxis, and that IL-18 accounts for a significant portion of the interleukin-18: more than an -␥ inducing factor. J. Leukocyte Biol. 63: 658. vascular migration characteristic of RA SF. This contribution is 21. Gracie, J. A., R. J. Forsey, W. L. Chan, A. Gilmour, B. P. Leung, M. R. Greer,

independent of that made by the chemokines IL-8 and ENA-78 K. Kennedy, R. Carter, X. Q. Wei, D. Xu, et al. 1999. A proinflammatory role for http://www.jimmunol.org/ also found in RA SF. One mechanism for IL-18-induced HMVEC IL-18 in rheumatoid arthritis. J. Clin. Invest. 104:1393. ␣ ␤ 22. Matsumoto, S., K. Tsuji-Takayama, Y. Aizawa, K. Koide, M. Takeuchi, T. Ohta, migration is through the integrin v 3, placing IL-18 in company and M. Kurimoto. 1997. Interleukin-18 activates NF-␬B in murine T helper type with TNF-␣ and bFGF, which also act in part through this specific 1 cells. Biochem. Biophys. Res. Commun. 234:454. integrin. In our hands, IL-18 did not affect basal endothelial cell 23. Gu, Y., K. Kuida, H. Tsutsui, G. Ku, K. Hsiao, M. A. Fleming, N. Hayashi, K. Higashino, H. Okamura, K. Nakanishi, et al. 1997. Activation of interferon-␥ proliferation. We demonstrated the ability of IL-18 to induce tube inducing factor mediated by interleukin-1␤ converting enzyme. Science 275:206. formation in Matrigel in vitro as well as significant blood vessel 24. Ghayur, T., S. Banerjee, M. Hugunin, D. Butler, L. Herzog, A. Carter, L. Quintal, L. Sekut, R. Talanian, M. Paskind, et al. 1997. -1 processes IFN-␥-in- formation in the Matrigel plug in vivo. An attempt to block the ␥ ␣ ducing factor and regulates LPS-induced IFN- production. Nature 386:619. effect of TNF- present at the angiogenesis site with neutralizing 25. Leung, B. P., I. B. McInnes, E. Esfandiari, X. Q. Wei, and F. Y. Liew. 2000. Abs in the Matrigel plug did not significantly diminish the angio- Combined effects of IL-12 and IL-18 on the induction of collagen-induced ar- by guest on September 28, 2021 genic effect of IL-18 within the plug. Finally, IL-18 induced neo- thritis. J. Immunol. 164:6495. 26. Wei, X., B. P. Leung, H. M. Arthur, I. B. McInnes, and F. Y. Liew. 2001. vascularization in the sponge disc angiogenesis model in mice. Reduced incidence and severity of collagen-induced arthritis in mice lacking These findings support the notion of IL-18 as a potent angiogenic IL-18. J. Immunol. 166:517. stimulus, with the capacity to mediate vascular migration and new 27. Morita, Y., M. Yamamura, K. Nishida, S. Harada, H. Okamoto, H. Inoue, Y. Ohmoto, R. L. Modlin, and H. Makino. 1998. Expression of interleukin-12 in blood vessel formation in the development of inflammatory and synovial tissue from patients with rheumatoid arthritis. Arthritis Rheum. 41:306. angiogenesis-driven disease, such as RA. Ultimately, this evidence 28. Firestein, G. S. 1991. The immunopathogenesis of rheumatoid arthritis. Curr. for a novel angiogenic function of IL-18 may lend itself to poten- Opin. Rheumatol. 3:398. 29. Stoll, S., H. Jonuleit, E. Schmidtt, G. Muller, H. Yamauchi, M. Kurimoto, tial therapeutic targets in the treatment of RA. J. Knop, and A. H. Enk. 1998. Production of functional IL-18 by different sub- types of murine human dendritic cells (DC): DC-derived IL-18 enhances IL-12 dependent Th1 development. Eur. J. Immunol. 28:3231. Acknowledgments 30. Stoll, S., G. Muller, M. Kurimoto, J. Saloga, T. Tanimoto, H. Yamauchi, We thank Eli Shamiyeh for statistical assistance in verifying all data H. Okamura, J. Knop, and A. H. Enk. 1997. Production of IL-18 (IFN-␥-inducing calculations. factor) messenger RNA and functional protein by murine keratinocytes. J. Im- munol. 159:298. 31. Olee, T., S. Hashimoto, J. Quach, and M. Lotz. 1999. IL-18 is produced by References articular chondrocytes and induces proinflammatory and catabolic responses. 1. Gimbrone, M. A., Jr., S. B. Leapman, R. S. Cotran, and J. Folkman. 1973. Tumor J. Immunol. 162:1096. angiogenesis: iris neovascularization at a distance from experimental intraocular 32. Udagawa, N., N. J. Horwood, J. Elliott, A. Mackay, J. Owens, H. Okamura, tumors. J. Natl. Cancer Inst. 50:219. M. Kurimoto, T. J. Chambers, T. J. Martin, and M. T. Gillespie. 1997. Interleu- 2. Winkler, J. D., and M. P. Seed. 1997. Angiogenesis in inflammatory disease. kin-18 (interferon-␥-inducing factor) is produced by osteoblasts and acts via Inflamm. Res. 46:157. granulocyte/macrophage colony-stimulating factor and not via interferon-␥ to 3. Szekanecz, Z., G. Szegedi, and A. E. Koch. 1998. Angiogenesis in rheumatoid inhibit osteoclast formation. J. Exp. Med. 185:1005. arthritis: pathogenic and clinical significance. J. Invest. Med. 46:27. 33. Naik, S. M., G. Cannon, G. J. Burbach, S. R. Singh, R. A. Swerlick, J. N. Wilcox, 4. Colville-Nash, P. R., and D. L. Scott. 1992. Angiogenesis and rheumatoid ar- J. C. Ansel, and S. W. Caughman. 1999. Human keratinocytes constitutively thritis: pathogenic and therapeutic implications. Ann. Rheum. Dis. 51:919. express interleukin-18 and secrete biologically active interleukin-18 after treat- 5. Paleolog, E. M., and J. M. Moitla. 1998. Angiogenesis in arthritis: role in disease ment with pro-inflammatory mediators and dinitrochlorobenzene. J. Invest. Der- pathogenesis and as a potential therapeutic target. Angiogen 2:295. matol. 113:766. 6. Folkman, J. 1992. Angiogenesis: retrospect and outlook. EXS 61:4. 34. Brooks, P. C., R. A. Clark, and D. A. Cheresh. 1994. Requirement of vascular ␣ ␤ 7. Klagsbrun, M., and P. A. D’Amore. 1991. Regulators of angiogenesis. Annu. Rev. integrin v 3 for angiogenesis. Science 264:569. Physiol. 53:217. 35. Keller, H. U., P. C. Wilkinson, M. Abercrombie, E. L. Becker, J. G. Hirsch, 8. Sunderkotter, C., M. Goebeler, K. Schulze-Osthoff, R. Bhardwaj, and C. Sorg. M. E. Miller, W. Scottramsey, and S. H. Zigmond. 1977. A proposal for the 1991. Macrophage-derived angiogenesis factors. Pharmacol. Ther. 51:195. definition of terms related to locomotion of leukocytes and other cells. Clin. Exp. 9. Ferrara, N. 1995. The role of vascular endothelial growth factor in pathological Immunol. 27:377. angiogenesis. Breast Cancer Res. Treat. 36:127. 36. Volin, M. V., L. A. Harlow, J. M. Woods, P. L. Campbell, M. A. Amin, 10. Leibovich, S. J., P. J. Polverini, H. M. Shepard, D. M. Wiseman, V. Shively, and M. Tokuhira, and A. E. Koch. 1999. Treatment with sulfasalazine or sulfapyri- N. Nuseir. 1987. Macrophage-induced angiogenesis is mediated by tumor necro- dine, but not 5-aminosalicyclic acid, inhibits basic fibroblast growth factor-in- sis factor-␣. Nature 329:630. duced endothelial cell chemotaxis. Arthritis Rheum. 42:1927. The Journal of Immunology 1653

37. Passaniti, A., R. M. Taylor, R. Pili, Y. Guo, P. V. Long, J. A. Haney, R. R. Pauly, 45. Shapiro, L., A. J. Puren, H. A. Barton, D. Novick, R. L. Peskind, R. Shenkar, D. S. Grant, and G. R. Martin. 1992. A simple, quantitative method for assessing Y. Gu, M. S. Su, and C. A. Dinarello. 1998. Interleukin 18 stimulates HIV type angiogenesis and antiangiogenic agents using reconstituted basement membrane, 1 in monocytic cells. Proc. Natl. Acad. Sci. USA 95:12550. heparin, and fibroblast growth factor. Lab. Invest. 67:519. 46. Kohka, H., T. Yoshino, H. Iwagaki, I. Sakuma, T. Tanimoto, Y. Matsuo, 38. Johns, A., A. D. Freay, W. Fraser, K. S. Korach, and G. M. Rubanyi. 1996. M. Kurimoto, K. Orita, T. Akagi, and N. Tanaka. 1998. Interleukin-18/interferon- Disruption of estrogen receptor gene prevents 17 ␤ estradiol-induced angiogen- ␥-inducing factor, a novel cytokine, up-regulates ICAM-1 (CD54) expression in esis in transgenic mice. Endocrinology 137:4511. KG-1 cells. J. Leukocyte Biol. 64:519. 39. Lijana, R. C., and M. C. Williams. 1979. Tetramethylbenzidine: a substitute for 47. Morel, J. C. M., C. C. Park, P. Kumar, and A. E. Koch. 2001. Interleukin-18 benzidine in hemoglobin analysis. J. Lab. Clin. Med. 94:266. induces rheumatoid arthritis synovial fibroblast CXC chemokine production ␬ 40. Liem, H. H., F. Cardenas, M. Tavassoli, M. B. Poh-Fitzpatrick, and through NF B activation. Lab. Invest. In press. U. Muller-Eberhard. 1979. Quantitative determination of hemoglobin and cyto- 48. Johnson, B. A., G. K. Haines, L. A. Harlow, and A. E. Koch. 1993. Adhesion chemical staining for peroxidase using 3,3Ј,5,5Ј-tetramethylbenzidine dihydro- molecule expression in human synovial tissue. Arthritis Rheum. 36:137. chloride, a safe substitute for benzidine. Anal. Biochem. 98:388. 49. Storgard, C. M., D. G. Stupack, A. Jonczyk, S. L. Goodman, R. I. Fox, and D. A. Cheresh. 1999. Decreased angiogenesis and arthritic disease in rabbits 41. Fajardo, L. F., J. Kowalski, H. H. Kwan, S. D. Prionas, and A. C. Allison. 1988. ␣ ␤ treated with an v 3 antagonist. J. Clin. Invest. 103:47. The disc angiogenesis system. Lab. Invest. 58:718. 50. Cao, R., J. Farnebo, M. Kurimoto, and Y. Cao. 1999. Interleukin-18 acts as an 42. Friedlander, M., P. C. Brooks, R. W. Shaffer, C. M. Kincaid, J. A. Varner, and angiogenesis and tumor suppressor. FASEB J. 13:2195. ␣ D. A. Cheresh. 1995. Definition of two angiogenic pathways by distinct v in- 51. Kinsella, J. L., D. S. Grant, B. S. Weeks, and H. K. Kleinman. 1992. Protein tegrins. Science 270:1500. kinase C regulates endothelial cell tube formation on basement membrane matrix, 43. Koch, A. E., and R. M. Strieter, eds. 1996. Chemokines in Disease. R. G. Landes, Matrigel. Exp. Cell Res. 199:56. Austin. 52. Moreland, L. W., M. H. Schiff, S. W. Baumgartner, E. A. Tindall, 44. Ohtsuki, T., M. J. Micallef, K. Kohno, T. Tanimoto, M. Ikeda, and M. Kurimoto. R. M. Fleischmann, K. J. Bulpitt, A. L. Weaver, E. C. Keystone, D. E. Furst, 1997. Interleukin 18 enhances expression and induces apoptosis in P. J. Mease, et al. 1999. Etanercept therapy in rheumatoid arthritis: a randomized, Fas-expressing human myelomonocytic KG-1 cells. Anticancer Res. 17:3253. controlled trial. Ann. Intern. Med. 130:478. Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021