Published online before print September 7, 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Laboratories of Molecular Immunology and
Immunobiology, Rega Institute for Medical Research, University of Leuven, Leuven, Belgium; and
Department of Pathology, University Hospital Leuven, Leuven, Belgium
1Correspondence: Laboratory of Molecular Immunology, Rega Institute, University of Leuven, Minderbroedersstraat 10, B-3000 Leuven, Belgium. E-mail: jo.vandamme{at}rega.kuleuven.be
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
but inhibited by IFN-
, which synergized with IL-1β to produce the angiostatic CXCL10/IFN--induced protein-10. These findings indicate that the equilibrium between angiostatic and angiogenic factors during inflammation and tumor progression is rather complex and differs depending on the chemokine, cell type, and stimulus. Selective intervention in the chemokine network may drastically disturb this delicate balance of angiogenesis and tissue repair. Application of angiostatic CXCL4L1/PF-4var without attraction of protumoral phagocytes may be beneficial in cancer therapy.
Key Words: angiogenesis • chemokine • cytokine • monocytes
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The chemokine family can be subdivided into CXC and CC chemokine ligands (CXCL, CCL), depending on the positioning of the conserved cysteines in the aminoterminal part of these small, inducible proteins [4
5
6
]. Most, if not all, chemokines activate leukocytes through binding to G protein-coupled receptors designated CXCR or CCR [7
, 8
]. Some CXC chemokines, such as CXCL8/IL-8, attract neutrophilic granulocytes and exert angiogenic activity, whereas other CXC chemokines, e.g., CXCL10/IFN--induced protein-10 (IP-10), rather chemoattract lymphocytes and inhibit angiogenic chemokines as well as other EC growth factors [9
, 10
]. More recently, additional CXC chemokines were identified and found to have potent, angiogenic effects, e.g., CXCL6/granulocyte chemotactic protein-2 (GCP-2), or rather angiostatic activities, e.g., CXCL4L1/platelet factor-4 variant (PF-4var) [10
11
12
]. CXCL6/GCP-2 is the only chemokine in addition to CXCL8/IL-8, which activates CXCR1 and CXCR2 [13
] and constitutes a major neutrophil chemoattractant in the mouse compensating for the absence of CXCL8/IL-8 in this species [14
]. CXCL4L1/PF-4var was isolated from platelets and is a more potent inhibitor of angiogenesis than CXCL4/PF-4 in vitro and in vivo [11
]. This variant of CXCL4/PF-4 is encoded by a second, nonallelic gene [15
, 16
].
The regulated production of these more recently described chemokines in various body compartments adds to the balance of angiogenesis versus angiostasis, as well as to the composition of the leukocyte influx, and hence, determines the outcome of tumor development and inflammation. However, the precise role of these chemokines still remains elusive, also taking into account that the chemokine network is only apparently redundant [17 ]. Indeed, the conditions and sites of production differ for each chemokine, in addition to the variation in the spectrum of activities and target cells, as chemokine receptors are also expressed in nonhematopoietic tissues. Furthermore, tumor cells expressing chemokine receptors can metastasize to chemokine-secreting organs in response to the chemotactic gradient [18 , 19 ]. For these reasons, a detailed study of the regulated production of angiogenic CXCL6/GCP-2 and angiostatic CXCL4L1/PF-4var was initiated. It was found that in contrast to CXCL4/PF-4, CXCL4L1/PF-4var was also induced in tumor cells, whereas the regulated production of CXCL6/GCP-2 by mesenchymal cells differed significantly from that of CXCL4L1/PF-4var, CXCL8/IL-8, and CXCL10/IP-10.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
were purchased from Peprotech (Rocky Hill, NJ, USA). rhIFN-, rhIL-18, and the plant lectin Con A were obtained from Bioferon (Laupheim, Germany), R&D Systems (Minneapolis, MN, USA), and Calbiochem (La Jolla, CA, USA), respectively. The cytokine inducers, bacterial LPS from Escherichia coli 0111:B4, bacterial peptidoglycan from Staphylococcus aureus, viral dsRNA polyriboinosinic:polyribocytidylic acid, PMA, thrombin, and thrombin receptor activator (TRA; Ser-Phe-Leu-Leu-Arg-Asn-Pro-Asn-Asp-Lys-Tyr-Glu-Pro-Phe), were purchased from Sigma-Aldrich (St. Louis, MO, USA).
Cell cultures and induction experiments
Human mononuclear cells were obtained from buffy coats of healthy donors provided by the Blood Transfusion Center of Leuven (Belgium) and isolated as described previously [20
]. Leukocytes were purified by subsequent sedimentation in hydroxyethyl starch and density gradient centrifugation in lymphoprep. Neutrophils were fractionated from PBMC and erythrocytes by density gradient centrifugation followed by hypotonic shock. The PBMCs (2.5x106 cells/ml) were seeded in 24-well plates (Techno Plastic Products AG, Trasadingen, Switzerland) in DMEM plus 2% (v/v) FCS and washed twice with PBS (Invitrogen, Grand Island, NY, USA) after 2 h. Induction experiments on adherent cells were conducted in RPMI 1640 (Cambrex Bio Science, Verviers, Belgium) with 10% (v/v) FCS for 72 h at 37°C/5% CO2. Macrophages differentiated for 5 days from adherent blood monocytes in RPMI 1640 containing 10% (v/v) FCS were induced in 24-well plates for 48 h at 37°C/5% CO2. Human neutrophilic granulocytes were freshly isolated from peripheral blood obtained from healthy donors (Blood Transfusion Center of Leuven and Laboratory of Experimental Immunology, University of Leuven, Belgium), as described by Gouwy et al. [21
]. These cells were stimulated with various inducers in 24-well plates (4x106 cells/ml) in RPMI 1640 [serum-free or containing 10% (v/v) FCS] at 37°C/5% CO2 for 3 h or 48 h, respectively. Conditioned media of mononuclear cells, macrophages, and neutrophilic granulocytes were collected and centrifuged at 220 g for 10 min at 4°C, and supernatants were stored at –20°C.
Human dermal microvascular EC (HMVEC) were purchased from Cambrex Bio Science and cultured following the manufacturer’s instructions in endothelial basal medium (EBM)-2, supplemented with the endothelial growth medium (EGM)-2MV Bullet kit (Cambrex Bio Science). After subcultivation in 48-well plates, cells were grown for 5 days to confluency and then stimulated with various cytokine inducers (vide supra) in 0.5 ml culture medium at 37°C/5% CO2 or were left untreated (control). Conditioned media were collected 72 h after induction and stored at –20°C until assay.
Human diploid fibroblasts (E1SM or E6SM strains of embryonic skin and muscle cells) and human MG-63 osteosarcoma cells (ATCC CRL 1427) developed in our laboratory were grown to confluency in Eagle’s minimal essential medium with Earle’s salts (EMEM; Invitrogen), supplemented with 10% (v/v) FCS (PAA Laboratories, Pasching, Germany). Human colorectal adenocarcinoma HT-29 cells (ATCC HTB-38) were grown to confluency in DMEM (Cambrex Bio Science), supplemented with 10% (v/v) FCS. Fibroblasts and HT-29 cells were treated with cytokines and cytokine inducers in 48-well dishes for 48 h at 37°C/5% CO2 in 2% (v/v) FCS.
Western blot analysis
Human natural CXCL4/PF-4 and rCXCL4L1/PF-4var, purified to homogeneity in our laboratory [11
], were loaded at different dilutions (200, 20, or 2 ng/lane) onto a Tris/tricine gel and subjected to SDS-PAGE [22
]. The separating, spacer, and stacking gels contained 13% total concentration of acrylamide and bisacrylamide monomers in w/v percent (T) and 5% amount of bisacrylamide cross-linker relative to T in w/w percent (C), 10% T and 3.3% C, and 5% T and 5% C, respectively. Prestained molecular weight markers (Bio-Rad Laboratories, Hercules, CA) were soybean trypsin inhibitor, lysozyme, and aprotinin (relative molecular weights 29,000, 19,000, and 7000, respectively). Proteins were transferred to a polyvinylidene difluoride (PVDF) membrane (MP Biochemicals, Irvine, CA, USA) by electroblotting. To block nonspecific binding sites, the blot was immersed in 100 mM NaCl, 10 mM Tris-HCl, pH 7.4 (NT buffer) with 3% (w/v) BSA (Sigma-Aldrich) before incubation with polyclonal rabbit anti-human CXCL4/PF-4 antibody (5 µg/ml diluted in NT buffer) or polyclonal rabbit anti-human CXCL4L1/PF-4var peptide antibody (10.8 µg/ml diluted in NT buffer), which were developed in our laboratory. After three flushes in wash buffer [NT buffer with 0.01% Tween-20 (Sigma-Aldrich) and 0.3% (w/v) BSA], the blot was incubated with alkaline-phosphatase-conjugated goat anti-rabbit antibody (Jackson ImmunoResearch Laboratories, West-Grove, PA, USA; 0.2 µg/ml diluted in wash buffer). The blot was washed, and proteins were visualized with nitroblue tetrazolium chloride/5-bromo-4-chloro-3-indolyl phosphate (Merck, Darmstadt, Germany).
In vitro chemotaxis
The chemotactic activity of natural CXCL4/PF-4 [11
] and rCXCL4L1/PF-4var [23
], purified to homogeneity in our laboratory, for PBMC and neutrophils, was performed in the Boyden microchamber (Neuro Probe, Cabin John, MD, USA). Natural human CXCL8/IL-8 and CCL2/MCP-1, purified to homogeneity from monocyte-derived conditioned medium [20
, 24
], were used as positive controls. The chemokine solution, diluted in HBSS (Invitrogen) plus 0.1% (w/v) human serum albumin (HSA; Belgian Red Cross), was added in triplicate to the lower compartment of the chamber and covered with a 5-µm pore polyvinyl pyrrolidone-free polycarbonate membrane (Nuclepore, Corning Costar, Acton, MD, USA) for neutrophils or polyvinyl pyrrolidone polycarbonate membrane (GE Osmonics Labstore, Minnetonka, MN, USA) for PBMC. The upper wells of the chamber, separated from the lower wells by a silicone gasket, were filled with a neutrophil (106 cells/ml) or PBMC suspension (2x106 cells/ml). Neutrophils and PBMCs were allowed to migrate at 37°C/5% CO2 for 45 min and 2 h, respectively. Cells, which migrated through the membrane, were stained with Hemacolor solutions (Merck) and counted microscopically in 10 oil immersion fields (500x magnification). The chemotactic index (CI) was calculated as the number of cells that migrated to the test chemokine, divided by the number of cells that migrated spontaneously to the chemotaxis buffer (1xHBSS+0.1% HSA).
EC migration
HMVEC were cultured in 24-well plates in EBM-2 with EGM-2MV Bullet kit. When HMVEC were grown to confluency, the culture medium was replaced with 0.5 ml/well EBM-2 with 40 µg/ml mitomycin-C (Acros Organics) without EGM-2MV Bullet kit. After 30 min at 37°C/5% CO2, a plastic pipette tip was used to draw a linear wound in the cell monolayer of each well. The monolayers were washed twice with PBS to remove the cells, which were detached from the monolayer, and chemokines were added in 0.5 ml EBM-2 with EGM-2MV Bullet kit. After 24 h, HMVEC cultures were fixed and stained with Hemacolor solutions. In every 24-well plate, three control cultures were included, to which no chemokine was added. The difference of the width of the individual wounds before and after treatment was scored under a microscope and set at zero for control cultures. A scar, which was broader compared with the control scars (inhibition of HMVEC migration), received an inhibition score. Optical scores ranged from zero (narrow) to two (broad). All samples were tested in duplicate in each 24-well plate and were scored double-blind and independently by two investigators.
Zymography analysis
Gelatinase activity in cell culture fluids was detected by gelatin zymography analysis [25
]. The zymography gels were analyzed for 91 kDa matrix metalloproteinase 9 (MMP-9) by scanning densitometry, and MMP-9 levels were expressed as arbitrary scanning units per equivalent culture supernatant volume.
Immunoassays
Sandwich ELISAs for human CXCL6/GCP-2 [26
], CXCL8/IL-8 [26
], CXCL10/IP-10 [27
], CXCL4/PF-4, and CXCL4L1/PF-4var using antibodies purified by protein A/G chromatography were developed in our laboratory. The human CXCL4L1/PF-4var ELISA was specific in that CXCL4/PF-4 was not detectable, whereas the CXCL4/PF-4 ELISA measured CXCL4/PF-4 and CXCL4L1/PF-4var. Microtiter plates (Costar, Corning, NY, USA) were coated overnight at 4°C with a polyclonal rabbit anti-human CXCL4/PF-4 (1.9 µg/ml) or polyclonal rabbit anti-human CXCL4L1/PF-4var peptide (6.5 µg/ml) antibody in PBS, respectively. Remaining binding sites were blocked for 1 h at 37°C with PBS containing 0.1% (w/v) casein (Sigma-Aldrich) and 0.05% (v/v) Tween-20 (PBS/Tween/casein). Natural CXCL4/PF-4 and natural or rCXCL4L1/PF-4var, purified to homogeneity in our laboratory, served as standards. Standard and samples were diluted in PBS/Tween/casein buffer and incubated for 1 h at 37°C. Subsequently, monoclonal mouse anti-human CXCL4/PF-4 (R&D Systems, Abingdon, UK), at 500 ng/ml in PBS/Tween/casein, was added as a secondary antibody and incubated for 1 h at 37°C. After washing with PBS/Tween, the detection was performed with peroxidase-conjugated goat anti-mouse antibody (0.16 µg/ml, Jackson ImmunoResearch Laboratories). Peroxidase activity was quantified by measuring the conversion of 3,3',5,5'-tetramethylbenzidine (Sigma-Aldrich) at 450 nm.
Both ELISAs did not show cross-reactivity with any other chemokine (e.g., CXCL6/GCP-2, CXCL8/IL-8, CXCL10/IP-10) or any used chemokine inducer (vide supra).
Immunocytochemistry
MG-63 cells were grown in 16-well glass plates (Lab-Tek, Nunc, Roskilde, Denmark) to confluency in EMEM plus 10% (v/v) FCS, induced for 24 h at 37°C/5% CO2, fixed with aceton for 10 min, and stored at –70°C. Fresh PBMC were seeded in 16-well glass plates in DMEM plus 2% (v/v) FCS, washed twice with PBS after 2 h, and induced in RPMI 1640 with 10% (v/v) FCS for 24 h at 37°C/5% CO2. Cells were then incubated with 1 µg/ml brefeldin-A from Penicillium brefeldianum (Sigma-Aldrich) for 4 h at 37°C and fixed with 4% paraformaldehyde (Merck) in PBS overnight at 4°C. After fixation, cells were washed three times with PBS and permeabilized for 5 min at room temperature with 0.2% Triton X-100 (Sigma-Aldrich) in PBS. Nonspecific binding sites were blocked for 20 min at room temperature with 5% normal goat serum (NGS; Santa Cruz Biotechnology, Santa Cruz, CA, USA) in PBS. Cells were then incubated with the primary antibody (diluted in 1% NGS in PBS) for 1 h at room temperature. The different primary antibodies included polyclonal rabbit anti-CXCL4/PF-4, polyclonal rabbit anti-CXCL4L1/PF-4var peptide (both at 15 µg/ml), and a rabbit IgG antibody (100 µg/ml, Sigma-Aldrich), all purified by chromatography. Cells were washed three times and incubated for 1 h at room temperature with the secondary antibody, goat anti-rabbit Alexa Fluor 647 at 4 µg/ml (Molecular Probes, Invitrogen) in 1% NGS in PBS. After three washes with PBS, cells were stained with 1.1 µM 4',6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich) and washed three times with PBS. Coverslips were mounted with Prolong Gold Antifade medium (Molecular Probes, Invitrogen). Fluorescent microscopic analysis was done with a Carl Zeiss Axiovert 200 M inverted microscope (Zeiss, Göttingen, Germany), using an EC Plan-Neofluar 40x/1.30 oil objective. Pictures were taken with an AxioCam MRm camera and processed with AxioVision Release 4.6 software (Zeiss). Identical exposure conditions for control, anti-CXCL4/PF-4, and anti-CXCL4L1/PF-4var antibodies were respected.
Immunohistochemistry
Immunohistochemical stainings were performed on tissue sections from patients with a high-grade leiomyosarcoma with extraosseous bone formation and a liposarcoma. Semiserial sections were cut from paraffin blocks containing formalin-fixed biopsy fragments. The sections were stained using an immunoperoxidase technique. Following inhibition of endogenous peroxidase, the tissue sections were incubated with the primary antibody (dilution 1/200) for 30 min at room temperature. After rinsing the sections in PBS, the sections were incubated for 30 min with the biotinylated secondary antibody and subsequently incubated with the avidin-biotinylated peroxidase complex (EnVision+, Dako, Carpinteria, CA, USA). The reaction product was visualized by incubation with 3,3'-diaminobenzidine tetrahydrochloride (Sigma-Aldrich) and hydrogen peroxide, resulting in brown, immunoreactive sites. The slides were faintly counterstained with Harris hematoxylin. Finally, the sections were rinsed with distilled water and coverslipped with glycerol. Negative controls consisted of similar sections, processed without the primary antibody.
Statistical analysis
Data were analyzed using Statistica (StatSoft Inc., Tulsa, OK, USA). All results were evaluated by the Mann-Whitney U-test. In view of the differences in induction capacity in function of the experiment, ELISA data were, for each cell type, expressed as percentage of the control for the statistical analysis.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
1.6) on monocytes, in that up to 2.5 µg/ml was required to obtain a chemotactic response. For comparison, the CC chemokine CCL2/MCP-1 was effective on monocytes, as a CI of 12 was observed at only 10 ng/ml. Similarly, both PF-4 forms were weakly active for neutrophils (indices of 1.3) at high concentrations (2.5 µg/ml), whereas CXCL8/IL-8 was clearly active at 1 ng/ml. In sharp contrast to its weak phagocyte chemotactic activity, CXCL4L1/PF-4var is a potent inhibitor of HMVEC migration, as evidenced in an EC migration assay. Indeed, at concentrations as low as 10 ng/ml, CXCL4L1/PF-4var inhibited HMVEC migration significantly, whereas for CXCL4/PF-4, a higher (20-fold) dose was required (Fig. 1C)
. It must be concluded that the chemokine CXCL4L1/PF-4var is a potent angiostatic factor, rather than an inflammatory mediator responsible for phagocyte mobilization.
 View larger version (24K): [in a new window] |
Figure 1. Effect of CXCL4L1/PF-4var and CXCL4/PF-4 on the migration of blood phagocytes and EC. Human natural CXCL4/PF-4 and rCXCL4L1/PF-4var (2500, 500, 100, and 20 ng/ml) were tested for their ability to induce migration of monocytes (A) and neutrophils (B) in the Boyden microchamber assay. CCL2/MCP-1 and CXCL8/IL-8 (10 or 1 ng/ml) were used as positive controls for monocyte and neutrophil chemotaxis, respectively. Results represent the mean CI (CI±SEM) of four independent experiments conducted in triplicate wells. Confluent monolayers of HMVEC were wounded, and repair was monitored microscopically after a 24-h treatment with natural CXCL4/PF-4 (1000, 200, and 40 ng/ml) or CXCL4L1/PF-4var (50, 10, and 2 ng/ml; C). Zero scores correspond to the change in width for untreated, control EC cultures. A wound, which retained a broader width compared with the control wounds, received an inhibition score, ranging from zero to two. Results represent average inhibition scores ± SEM from three to four independent experiments, each tested in duplicate wells in one 24-well plate. Assays were scored double-blind by two independent investigators. The Mann-Whitney U-test was used for statistical analysis (*, P<0.05; **, P<0.01).
|
 View larger version (21K): [in a new window] |
Figure 2. Specificity of anti-CXCL4L1/PF-4var antibody in ELISA and in Western blot. (A) For the development of a specific and sensitive CXCL4L1/PF-4var ELISA, a rabbit anti-human CXCL4L1/PF-4var peptide antibody (Ab) was coated at different dilutions (1/300, 1/1000, and 1/3000) as a primary (coating) antibody. A commercial monoclonal mouse anti-human CXCL4/PF-4 was added as a secondary antibody at 500 ng/ml. The x-axis represents the concentration of CXCL4/PF-4 or CXCL4L1/PF-4var proteins (ng/ml); the y-axis shows the mean absorption value measured at 450 nm. With the rabbit anti-human CXCL4L1/PF-4var peptide antibody at a dilution of 1/300, as low as 0.3 ng/ml CXCL4L1/PF-4var could be detected, whereas CXCL4/PF-4 was still not detectable at 100 ng/ml. (B and C) Western blot with human natural CXCL4/PF-4 and rCXCL4L1/PF-4var loaded at different dilutions: 200 ng CXCL4/PF-4 (lane 1), 200 ng CXCL4L1/PF-4var (lane 2), 20 ng CXCL4/PF-4 (lane 3), 20 ng CXCL4L1/PF-4var (lane 4), and 2 ng CXCL4L1/PF-4var (lane 5). After loading on a Tris/tricine gel and SDS-PAGE, chemokines were electroblotted to a PVDF membrane and incubated with a polyclonal rabbit anti-human CXCL4/PF-4 antibody (5 µg/ml, B) or polyclonal rabbit anti-human CXCL4L1/PF-4var peptide antibody (10.8 µg/ml, C). Prestained molecular weight markers represent soybean trypsin inhibitor (29 kDa), lysozyme (19 kDa), and aprotinin (7 kDa).
|
1%) of this PF-4 immunoreactivity was a result of CXCL4L1/PF-4var (Fig. 3B)
. Indeed, as the CXCL4L1/PF-4var ELISA is recognizing CXCL4/PF-4 at least 300-fold less efficient, the detected CXCL4L1/PF-4var cannot be a result of cross-reactivity with CXCL4/PF-4 (Fig. 2)
. Angiogenic CXCL8/IL-8 was also inducible in monocytes (data not shown) by the same concentrations of thrombin, IL-1β, Con A, PMA, and LPS. In contrast, none of these stimuli, when tested at various concentrations, could induce detectable CXCL6/GCP-2 (<0.2 ng/ml) production (data not shown). Together, these results demonstrate that some angiogenic (CXCL8/IL-8) and angiostatic (CXCL4/PF-4) chemokines are induced simultaneously in mononuclear cells, whereas other angiogenic chemokines (CXCL6/GCP-2) are not.
 View larger version (28K): [in a new window] |
Figure 3. Production of CXCL4/PF-4 and CXCL4L1/PF-4var by human monocytes stimulated with various chemokine inducers. Human mononuclear cells were stimulated for 72 h with thrombin, TRA, IL-1β, Con A, PMA, LPS, or dsRNA or were left untreated (Co). Conditions for induction were as described in Materials and Methods. CXCL4/PF-4 (A) and CXCL4L1/PF-4var (B) protein concentrations were determined by ELISA and are expressed as the mean ± SEM from five independent experiments. A significant increased induction of chemokine in treated compared with untreated cultures was evidenced by the Mann-Whitney U-test (*, P<0.05; **, P<0.01).
|
0.032 ng/ml, respectively) upon induction with various stimuli for 2 days. For comparison, these macrophage cultures were as capable as monocytes to produce more than 50 ng/ml CXCL8/IL-8 after stimulation with the inflammatory mediators (Fig. 4B)
. Freshly isolated peripheral blood neutrophils released low amounts (1 ng/ml) of CXCL4/PF-4 immunoreactivity spontaneously. However, this constitutive production could not be enhanced by any of the inflammatory stimuli (IL-1β, LPS, or PMA) tested (Fig. 5A
). In contrast, such neutrophils clearly released MMP-9/gelatinase B upon activation for 3 h with LPS or PMA (Fig. 5B)
, whereas CXCL4L1/PF-4var again remained below the detection limit at any time-point.
 View larger version (32K): [in a new window] |
Figure 4. Induction of angiostatic CXCL4/PF-4 versus angiogenic CXCL8/IL-8 in macrophages, which when differentiated for 5 days from adherent blood monocytes, were stimulated for 48 h with thrombin, IL-1β, Con A, PMA, LPS, or dsRNA or were left untreated (Co), as described in Materials and Methods. Secreted CXCL4/PF-4 (A) and CXCL8/IL-8 (B) protein concentrations were measured by specific ELISAs. Results represent the mean ± SEM of three to five independent experiments. Significant chemokine induction (above constitutive expression) was demonstrated by the Mann-Whitney U-test (*, P<0.05; **, P<0.01).
|
 View larger version (22K): [in a new window] |
Figure 5. Production of CXCL4/PF-4 and MMP-9/gelatinase B by human neutrophilic granulocytes. Human neutrophils, freshly isolated from peripheral blood, were induced for 3 h (MMP-9 release) or 20 h (chemokine production) with various doses of IL-1β, LPS, or PMA or were left untreated (Co). Secreted CXCL4/PF-4 protein concentration (A) was measured by specific ELISA, and MMP-9/gelatinase-B activity (B) was determined by zymography, as described in Materials and Methods. Results represent the mean ± SEM of two to four independent experiments.
|
(data not shown), a known inducer of angiostatic CXCL10/IP-10. However, MG-63 monolayers stimulated with single proinflammatory cytokines (IL-1β, IL-17A, TNF-) produced PF-4 immunoreactivity dose-dependently, as determined with the specific CXCL4L1/PF-4var ELISA (Fig. 6A
). Optimal induction was obtained at 1 ng/ml IL-1β, 100 ng/ml IL-17A, and 10 ng/ml TNF-. Analysis of the secreted PF-4 immunoreactivity demonstrated that in contrast to leukocytes, solely, this more angiostatic CXCL4L1/PF-4 variant was produced in MG-63 cells (Fig. 6A)
. Indeed, no higher PF-4 levels were detected with the ELISA, measuring both PF-4 forms (data not shown). For comparison, the angiogenic CXCL6/GCP-2 revealed a similar induction pattern in that TNF-, IL-1β, and IL-17A, but not IL-18, also induced CXCL6/GCP-2 dose-dependently in these mesenchymal tumor cells (Fig. 6B)
. As the optimal cytokine doses were similar for induction of both chemokines, it is clear that under inflammatory conditions, these osteosarcoma cells produce angiogenic and angiostatic chemokines. Furthermore, CXCL10/IP-10 was up-regulated equally well in MG-63 cells by TNF-, IL-1β, and IFN- (data not shown). In contrast, human colorectal adenocarcinoma HT-29 cells only produced low amounts (1 ng/ml) of CXCL6/GCP-2 in response to IL-1β, whereas no detectable PF-4 immunoreactivity was induced in these cells (data not shown).
 View larger version (27K): [in a new window] |
Figure 6. Induction of angiostatic CXCL4L1/PF-4var and angiogenic CXCL6/GCP-2 in mesenchymal tumor cells by cytokines. Confluent monolayers of MG-63 cells were stimulated for 48 h at 37°C/5% CO2 with different doses of IL-1β, IL-17A, IL-18, or TNF- or were left untreated (Co), as described in Materials and Methods. Secreted CXCL4L1/PF-4var (A) and CXCL6/GCP-2 (B) protein concentrations were measured by specific ELISAs. Results represent the mean ± SEM of eight independent experiments. Significant chemokine induction (above constitutive expression) was demonstrated by the Mann-Whitney U-test (*, P<0.05; **, P<0.01; ***, P<0.001).
|
 View larger version (24K): [in a new window] |
Figure 7. Kinetics of CXCL4L1/PF-4var and CXCL6/GCP-2 production in human osteosarcoma cells. MG-63 osteosarcoma cells grown to confluency were stimulated with 100 ng/ml IL-1β or were left untreated (control). Samples taken at different time-points were analyzed for the presence of CXCL4L1/PF-4var (A) or CXCL6/GCP-2 (B) immunoreactivity by specific ELISAs. Results represent the mean ± SEM of four independent experiments. Significant induction by IL-1β above constitutive expression was evidenced by the Mann-Whitney U-test (*, P<0.05).
|
) or in monocytes (IL-1β, LPS) was capable of stimulating detectable (<0.8 ng/ml) CXCL4L1/PF-4var production in microvascular EC and various strains (E1SM, E6SM) of diploid fibroblasts (Fig. 8A
, and data not shown). For comparison, IL-1β, TNF-, or IL-17A dose-dependently induced CXCL6/GCP-2 production in fibroblasts at even higher levels than in MG-63 cells (Fig. 8B)
. Furthermore, in microvascular EC, CXCL6/GCP-2 but not CXCL4L1/PF-4var was detectable after induction by LPS, IL-1β, and TNF- (Figs. 9
10
11
, and data not shown). Thus, the balance between angiogenic CXCL6/GCP-2 and angiostatic CXCL4L1/PF-4var in normal connective and vascular tissue might be different from that in tumoral tissue.
 View larger version (19K): [in a new window] |
Figure 8. Induction of CXCL6/GCP-2 but not CXCL4L1/PF-4var in fibroblasts by cytokines. Confluent monolayers of human diploid fibroblasts (E1SM) were induced for 48 h with various doses of IL-1β, IL-17A, IL-18, TNF-, or LPS or were left untreated (Co), as described in Materials and Methods. Results represent the mean ± SEM of CXCL4L1/PF-4var (A) and CXCL6/GCP-2 (B) protein concentration measured in three to five independent experiments. Significant chemokine induction (above constitutive expression) was demonstrated by the Mann-Whitney U-test (*, P<0.05; **, P<0.01).
|
 View larger version (35K): [in a new window] |
Figure 9. IL-1β synergizes with TNF- to produce CXCL6/GCP-2 and CXCL8/IL-8 in microvascular EC. Confluent monolayers of HMVEC were incubated for 72 h with different doses of IL-1β in combination with TNF-, as described in Materials and Methods. Results (five to eight independent experiments) represent the mean CXCL6/GCP-2 (A) or CXCL8/IL-8 (B) concentration measured by ELISA and were analyzed by the Mann-Whitney U-test (*, P<0.05; **, P<0.01; ***, P<0.001). Synergistic induction corresponds to an increase above the sum of the induction capacity by the individual cytokines. For clarity, error bars were not depicted in the figure.
|
 View larger version (27K): [in a new window] |
Figure 10. IFN- inhibits the induction of CXCL6/GCP-2 by IL-1β in EC, whereas it synergizes with IL-1β to induce CXCL10/IP-10 production. Confluent monolayers of HMVEC were incubated for 72 h with different doses of IL-1β in combination with IFN-. Results represent the mean (three to eight independent experiments) CXCL6/GCP-2 (A), CXCL8/IL-8 (B), or CXCL10/IP-10 (C) concentration measured by ELISA. Data were analyzed by the Mann-Whitney U-test (*, P<0.05; **, P<0.01). Synergistic induction of CXCL10/IP-10 corresponds to an increase above the sum of the induction capacity by the individual cytokines. For clarity, error bars were not depicted in the figure.
|
 View larger version (28K): [in a new window] |
Figure 11. IFN- inhibits the induction of angiogenic CXCL6/GCP-2 by LPS in EC, whereas it synergizes with LPS to induce angiostatic CXCL10/IP-10 and has no effect on CXCL8/IL-8 production. Confluent monolayers of HMVEC were incubated for 72 h with different doses of LPS in combination with IFN-. Results represent the mean (three to eight independent experiments) CXCL6/GCP-2 (A), CXCL8/IL-8 (B), and CXCL10/IP-10 (C) production measured by ELISA and were analyzed by the Mann-Whitney U-test (*, P<0.05; **, P<0.01; ***, P<0.001). For clarity, error bars were not depicted in the figure.
|
synergized in a dose-dependent manner to induce CXCL6/GCP-2 in EC. A maximal production (47.5±16.4 ng/ml) of CXCL6/GCP-2 was obtained with 10 ng/ml IL-1β plus 10 ng/ml TNF-, corresponding to a 2.5-fold increase above the sum of the induction capacity by the individual cytokines (Fig. 9A)
. However, under these optimal circumstances, no induction of the angiostatic chemokines CXCL10/IP-10 or CXCL4L1/PF-4var was detected (data not shown), whereas for CXCL8/IL-8, a significant (but more restricted), synergistic induction (at 1 and 0.1 ng/ml IL-1β in combination with 1 ng/ml TNF-) was noticed (Fig. 9B)
.
Next, the effect of IFN- on IL-1β-induced CXCL6/GCP-2 production by EC was investigated. In contrast to TNF-, IFN- inhibited CXCL6/GCP-2 production induced by IL-1β dose-dependently, with a fivefold reduction at 20 ng/ml IFN- in combination with 10 ng/ml IL-1β (Fig. 10A)
. However, for CXCL8/IL-8 induction by IL-1β, no inhibitory effect of IFN- was observed, whereas for CXCL10/IP-10, production synergy (up to threefold above cumulative effect) between these two cytokines was clearly evidenced (Fig. 10B
and 10C)
.
Finally, the effect of IFN- on LPS-induced chemokine production by EC was verified. Similar to the induction by IL-1β, IFN- (0.2–20 ng/ml) inhibited (about fivefold) the CXCL6/GCP-2 production by LPS (0.5–50 µg/ml) dose-dependently, without having an effect on the CXCL8/IL-8 production (Fig. 11A
and 11B)
. Again, in sharp contrast with these angiogenic chemokines, the production of angiostatic CXCL10/IP-10 induced by IFN- was enhanced synergistically (about threefold) in the presence of LPS (Fig. 11C)
. However, the production of CXCL4L1/PF-4var was not enhanced by IL-1β or LPS in combination with IFN- (data not shown).
Taken together, these data demonstrate that the induction of various angiogenic CXCR2 agonists (CXCL8/IL-8 and CXCL6/GCP-2) as well as angiostatic chemokines (CXCL4L1/PF-4var and CXCL10/IP-10) is regulated differently, depending on the cell type and inflammatory stimuli applied. Hence, it can be deduced that these angiogenic and angiostatic chemokines may antagonize or cooperate with each other, depending on the type of inflammation and the disease state.
Immunocytochemical and immunohistochemical detection of CXCL4L1/PF-4var in cultured cells and tumor tissue
To confirm the in vitro expression of CXCL4L1/PF-4var detected by ELISA, immunocytochemical and immunohistochemical staining was performed on normal and tumoral mesenchymal cells and tissues. In adherent mononuclear leukocytes and MG-63 osteosarcoma cells, expression of CXCL4L1/PF-4var immunoreactivity (as observed by ELISA) was confirmed by cytoplasmic staining (Fig. 12E
and F) using the specific CXCL4L1/PF-4var antibody (Fig. 2C)
. Irrelevant control Ig failed to demonstrate such staining, confirming the specificity of the antibody (Fig. 12A
and 12B)
. Staining of the monocyte cultures with the nondiscriminatory CXCL4/PF-4 antibody provided a more pronounced signal (Fig. 12C)
, corresponding with the higher CXCL4/PF-4 levels detected by ELISA. In contrast, the staining intensity of MG-63 cells by the aspecific CXCL4/PF-4 antibody was similar to that with anti-CXCL4L1/PF-4var antibody (Fig. 12D
and 12F)
, which is also in agreement with the ELISA data showing that only CXCL4L1/PF-4var is produced. Finally, the relevance of CXCL4L1/PF-4var secretion in tumoral tissue was evidenced by immunohistochemical staining of human leiomyosarcoma and liposarcoma tissue sections. Immunoperoxidase reaction demonstrates granular, brown cytoplasmic staining in the spindle-shaped tumor cells (Fig. 12H
, and data not shown).
 View larger version (40K): [in a new window] |
Figure 12. Immunocytochemical and immunohistochemical staining of CXCL4L1/PF-4var. Immunofluorescent staining of thrombin (20 U/ml)-stimulated monocytes (A, C, and E) and IL-1β (10 ng/ml)-stimulated MG-63 osteosarcoma cells (B, D, and F). After fixation, cells were stained with a rabbit IgG antibody (Control, A and B), a polyclonal rabbit anti-CXCL4/PF-4 antibody (C and D), or polyclonal rabbit anti-CXCL4L1/PF-4var peptide (E and F) antibody. The red color was obtained with Alexa Fluor 647 fluorescent-labeled, anti-rabbit antibody, and nuclei were stained blue with DAPI. Identical exposure conditions for control, anti-CXCL4/PF-4, and anti-CXCL4L1/PF-4var antibodies were respected. The immunohistochemical microphotographs (lower part, G and H) show a leiomyosarcoma composed of spindle-shaped cells with moderate variability in cell size and shape. Several cells show granular, brown cytoplasmic staining with the antibody directed against CXCL4L1/PF-4var (H) but not if stained without the primary antibody being added (G). The original scale bars correspond to 20 µm.
|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, IL-1β, IL-17, and IL-18 [38
39
40
]. In this study, we have investigated two more chemokines identified recently, i.e., CXCL4L1/PF-4var and CXCL6/GCP-2, which possess angiostatic and angiogenic activity, respectively [11
, 12
]. Although, CXCL4/PF-4 has been reported to act on monocytes [41
, 42
] and neutrophils [37
, 43
], we observed that CXCL4L1/PF-4var is a weak chemoattractant for protumoral phagocytes, i.e., neutrophils and monocytes, whereas it is a potent inhibitor of EC chemotaxis in vitro (Fig. 1)
and prevents new blood vessel formation in vivo [11
]. In contrast, CXCL6/GCP-2 is chemotactic for neutrophils and exerts angiogenic activity in vitro and in vivo [10
, 30
]. To better predict the outcome of the balance between angiogenic and angiostatic factors in various tissues, the regulated production of CXCL4L1/PF-4var and CXCL6/GCP-2 in different cell types was compared in detail. To that goal, a sensitive and specific immunotest for CXCL4L1/PF-4var, not detecting CXCL4/PF-4, was developed (Fig. 2)
.
First, it was verified whether CXCL4L1/PF-4var is produced by hematopoietic cells other than thrombocytes. Schaffner et al. [44
] already described production of PF-4 immunoreactivity by monocytes, but the detection system used did not discriminate between CXCL4/PF-4 and CXCL4L1/PF-4var. Unlike the angiostatic CXCL10/IP-10 induced by IFN- [27
], both PF-4 forms were rather up-regulated in monocytes by thrombin, the cytokine IL-1β, and the chemokine inducers LPS and PMA, the latter not being reported before as an inducer of CXCL4/PF-4 (Fig. 3)
. However, monocytes did not produce angiogenic CXCL6/GCP-2 in response to any of these inducing substances, in contrast to CXCL8/IL-8, which has been identified as a monocyte-derived chemokine [24
], and was coinduced selectively with the PF-4 forms. Thus, in monocytes, the production of the angiostatic CXCR3 ligands (e.g., CXCL10/IP-10) and PF-4 forms is inducer-dependent and hence, complementary. As CXCL10/IP-10 and CXCL4L1/PF-4var strongly inhibit EC chemotaxis induced by CXCL8/IL-8, coinduction of PF-4 forms in monocytes may directly counteract the monokine CXCL8/IL-8 [45
, 46
] during inflammation and tumorigenesis. Macrophages responded to the same stimuli as monocytes to produce CXCL4/PF-4 but at a much lower concentration (Fig. 4)
, whereas CXCL4L1/PF-4var induction remained undetectable. Neutrophils, which are also reportedly a cellular source of chemokines [28
], secrete low amounts of CXCL4/PF-4 spontaneously but not CXCL4L1/PF-4var (Fig. 5)
, which could not be enhanced by inflammatory stimuli active on neutrophils to induce MMP-9 [25
].
In normal mesenchymal cells such as fibroblasts and EC, the induction of PF-4 forms by cytokines and cytokine inducers was found to be marginal or absent. However, CXCL6/GCP-2 was induced in these cell types by the TLR agonist LPS and inflammatory cytokines, such as TNF-, IL-1β, and IL-17 (Figs. 8
9
10
11)
. CXCL8/IL-8 and the angiostatic CXCR3 ligands CXCL9/monokine induced by IFN-, CXCL10/IP-10, and CXCL11/IFN-inducible T cell chemoattractant are also inducible in fibroblasts and EC by these cytokines and cytokine inducers [30
, 45
, 47
, 48
]. This indicates that these CXCR3 ligands and not CXCL4L1/PF-4var may counteract the angiogenic CXCR2 ligands such as CXCL6/GCP-2 produced by mesenchymal tissue.
The complexity of the regulated chemokine production in mesenchymal cells, such as EC, is increased further, as cytokines and cytokine inducers act in concert. Indeed, IL-1β and TNF- enhanced CXCL6/GCP-2 and IL-8/CXCL8 production synergistically in microvascular EC, whereas no effect on the CXCL4L1/PF-4var expression was observed (Fig. 9)
. In contrast, IFN- inhibited CXCL6/GCP-2 production in microvascular EC induced by IL-1β or LPS, but CXCL10/IP-10 expression was enhanced synergistically under these conditions (Figs. 10
and 11)
[49
]. This indicates that during a strong, inflammatory response, a better antitumoral environment might be created by enhanced angiostasis, paralleled by an increased lymphocyte influx (augmentation in CXCL10/IP-10 vs. reduction in CXCL6/GCP-2).
Finally, in cancer, the tumor cells themselves also determine the chemokine profile. Indeed, in response to the inflammatory cytokines IL-1β, TNF-, and IL-17 but not IL-18, osteosarcoma cells coproduced chemokines abundantly, including angiogenic CXCL6/GCP-2, as well as angiostatic CXCL4L1/PF-4var (Fig. 6)
. The fact that mesenchymal tumor cells induced by these cytokines selectively produced CXCL4L1/PF-4var and not CXCL4/PF-4 has not been reported before. Nevertheless, Lasagni et al. [50
] have demonstrated recently spontaneous CXCL4L1/PF-4var mRNA expression in normal, smooth muscle cells. It remains difficult to understand why, under inflammatory conditions, tumor cells (Fig. 7)
produce different chemokines simultaneously, which can counteract each other. It was therefore interesting to determine chemokine expression in vivo. In sarcoma tissue, weak to strong expression of CXCL4L1/PF-4var was detected in the tumor cells by immunohistochemistry (Fig. 12)
, whereas normal stromal tissue and blood vessels remained negative. Immunohistochemical detection of CXCL6/GCP-2 in gastrointestinal carcinoma and stromal tumors demonstrated that this chemokine was rather expressed by EC at the junction of tumoral and nontumoral tissue, i.e., at sites of neovascularization [47
]. The fact that sarcomas express multiple chemokines, such as CXCL8/IL-8 and CCL2/MCP-1, which are potent chemoattractants for neutrophils and monocytes, respectively, explains the presence of tumor-associated leukocytes beneficial for tumor growth [32
, 51
]. The production of CXC chemokines (CXCL10/IP-10) by normal and tumor cells contributes to the angiostatic counterbalance and the attraction of tumoricidal lymphocytes and NK cells.
Taken together, this study demonstrates that the role of chemokines in tumor development is extremely complex in view of the angiogenic versus angiostatic activities. The fact that these chemokines are regulated differently in normal versus malignant tissue, depending also on the cell type and inflammatory stimulus, only adds to this complexity. As a consequence, it is extremely difficult to predict the outcome of such chemokine balance during tumor progression, but once the triggers and players are known, it might be possible to exploit this information in a beneficial way.
Received April 4, 2007; revised July 4, 2007; accepted July 10, 2007.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
B activation Int. J. Cancer 75,900-907[CrossRef][Medline]: comparison with interleukin-8/CXCL8 Lab. Invest. 83,23-34[Medline] and provide a mechanism for enhanced synovial chemokine levels in septic arthritis Eur. J. Immunol. 33,3146-3153[CrossRef][Medline] production by T cells Nature 378,88-91[CrossRef][Medline]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
