Granulocyte–macrophage colony-stimulating factor as an autocrine survival-growth factor in human gliomas

Cytokine 57 (2012) 347–359

Contents lists available at SciVerse ScienceDirect

Cytokine journal homepage: www.elsevier.com/locate/issn/10434666

Granulocyte–macrophage colony-stimulating factor as an autocrine survival-growth factor in human gliomas Roberto P. Revoltella a,⇑, Michele Menicagli b, Daniela Campani b a b

Institute for Chemical-Physical Processes (IPCF), National Research Council of Italy (CNR), Pisa, Italy Department of Surgery, University of Pisa, Pisa, Italy

a r t i c l e

i n f o

Article history: Received 8 March 2011 Received in revised form 12 October 2011 Accepted 20 November 2011 Available online 24 December 2011 Keywords: GM-CSF GM-CSF receptor Human glioma Glioblastoma multiforme Tumor progression

a b s t r a c t We studied the expression of Granulocyte–Macrophage Colony-Stimulating Factor (GM-CSF) and its receptors (GM-CSF.R) in 20 human brain gliomas with different tumor gradings and demonstrated constitutive high levels of both mRNA gene expression and protein production exclusively in the highestgrade tumors (WHO, III–IV grade). Five astrocytic cell lines were isolated in vitro from glioma cells, which had selectively adhered to plates pre-coated with rhGM-CSF. These cells were tumorigenic when xenografted to athymic mice, and produced GM-CSF constitutively in culture. Two lines, particularly lines AS1 and PG1, each from a patient with glioblastoma multiforme, constitutively over-expressed both GM-CSF and GM-CSF.R genes and secreted into their culture media biologically active GM-CSF. Different clones of the AS1 line, isolated after subsequent passages in vitro and then transplanted to athymic mice, demonstrated higher tumorigenic capacity with increasing passages in vivo. Cell proliferation was stimulated by rhGM-CSF in late-stage malignant clones, whereas apoptosis occurred at high frequency in the presence of blocking anti-GM-CSF antibodies. In contrast, rhGM-CSF did not induce any apparent effect in early-stage clones expressing neither GM-CSF nor GM-CSF.R. The addition of rhGM-CSF or rhIL-1b, to cultures induced the overproduction of both GM-CSF and its receptors and increased gene activation for several functional proteins (e.g. NGF, VEGF, VEGF.R1, G-CSF, MHC-II), indicating that these cells may undergo dynamic changes in response to environmental stimuli. These findings thus revealed: (1) that the co-expression of both autocrine GM-CSF and GM-CSF.R correlates with the advanced tumor stage; (2) that an important contribution of GM-CSF in malignant glioma cells is the prevention of apoptosis. These results imply that GM-CSF has an effective role in the evolution and pathogenesis of gliomas. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction Granulocyte–Macrophage Colony-Stimulating Factor (GM-CSF) is a cytokine whose best-known functions are to stimulate survival, proliferation and differentiation of hematopoietic myeloid cells. In synergy with other factors, GM-CSF also acts on more primitive hematopoietic multipotent precursor cells as well as on differentiated cells of other lineages [1–4]. GM-CSF binds to a dimeric receptor (GM-CSF.R) [5–8] consisting of a ligand-specific a-chain subunit binding GM-CSF with low affinity, and a b-chain subunit, which is shared between GM-CSF, interleukin-3 (IL-3) and interleukin-5 (IL-5). The b-subunit alone does not bind GM-CSF, but does so only after interacting with the a-subunit/GM-CSF complex, forming a high-affinity receptor triggering a number of cellular responses [9–12]. The availability of the purified recombinant pro⇑ Corresponding author. Address: Institute for Chemical-Physical Processes (IPCF), Laboratory of Regenerative Medicine, National Research Council of Italy (CNR), Via Derna 1, 56126 Pisa, Italy. Tel.: +39 050 2211510, mobile: +39 3341748107; fax: +39 050 2211527. E-mail address: [email protected] (R.P. Revoltella). 1043-4666/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.cyto.2011.11.016

tein has enabled several therapeutic applications of GM-CSF in pathological conditions where the number and/or functional capacity of leukocytes is sub-optimal including in patients undergoing marrow transplantation or patients with malignant tumors, in order to reduce the duration and severity of leukopenia in standard cancer therapy (review in [13]). However, GM-CSF appears to have a much broader spectrum of activities than was originally believed. It plays a role in stimulating survival and growth in a variety of cell types of non-hemopoietic origin [14], including osteoblastic cells [15], fibroblasts [16], endothelial cells [17], dendritic cell precursors [18] and keratinocytes [19,20]. GM-CSF plays an important role in accelerating placenta growth, fetal growth and fetal survival [21]. GM-CSF is an autocrine survival factor for mature and malignant B-lymphocytes [22] and may play a role as a stimulator of tumor progression in human solid tumors of different non-hematopoietic cell lineages including human prostate cancer, bladder carcinoma, melanoma, gastric cancer, and non-small-cell lung carcinoma [23–29]. In some cases, these tumor cells were found to produce GM-CSF and to co-express the appropriate GM-CSF.R, suggesting an effect of GM-CSF on tumor growth, tumor progression and invasion. These findings led to a careful

348

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359

re-evaluation of GM-CSF therapy in patients with malignant tumors. They also highlighted the need to identify other possible effects that autocrine GM-CSF as well as the passive administration of recombinant human rhGM-CSF, may have in these patients. This includes not only on hematopoietic cells, but also on other potential normal and malignant target cells of non-hemopoietic origin bearing functional GM-CSF.R. Malignant glioma is one of the most aggressive and fastestgrowing brain tumors in the adult, with a very short survival time for patients [30,31]. The expression of GM-CSF and its receptor genes within human glioma specimens has been previously reported [32–37]. However, the results published to-date have often been contradictory. For example, glioblastoma cells produced GMCSF in vitro, but not in vivo without expressing its receptor [33]. The autocrine/paracrine growth regulation by combined GM-CSF and granulocyte colony-stimulating factor (G-CSF) in gliomas and astrocytomas has been proposed [34–37], but so far, the demonstration of GM-CSF production has not been substantiated by functional assays, providing unequivocal evidence for GM-CSF of possible autocrine/paracrine loops. The aim of this study was to investigate the expression of GMCSF and its membrane receptors in surgical specimens of 20 unrelated human brain gliomas of different malignant gradings and in cell lines isolated in culture from these tumors, revealing increasing malignancy grades in late-stage passages in vivo by transplantation to athymic mice. We tested whether the survival and growth rate of these glioma cultures correlates with the production of GM-CSF and co-expression of membrane GM-CSF.R. This was done in order to test the hypothesis that the co-expression of both GM-CSF and GM-CSF.R is required by glioma cells for their survival and growth, providing evidence that GM-CSF plays an important regulatory role in the pathogenesis and progression of these tumors. 2. Materials and methods 2.1. Reagents rhGM-CSF (unglycosylated, from Escherichia Coli) was obtained through the courtesy of Dr. Federico Bertolero. Rabbit polyclonal antibodies and mouse hybridoma monoclonal anti-rhGM-CSF antibodies (clone G3.7/1, IgG1k) were described [38,39]. These antibodies (Abs) are specific for GM-CSF and do not decrease the activity of macrophage-colony-stimulating factor (M-CSF), IL-3, Interleukin1b (IL-1b) or G-CSF. They have been used as affinity purified reagents and were determined to be free from detectable endotoxin (<1 ng/ml) by the Limulus amoebocyte lysate assay. These Abs blocked cell survival and proliferation of the GM-CSF/IL-3/erythropoietin-sensitive leukemia TF-l cell line [40], and inhibited 21-day myeloid colony formation from normal human bone marrow mononuclear precursors (BM-MNC) in soft agar assays [3] in a concentration-dependent manner. Equivalent concentrations of nonimmune rabbit or mouse immunoglobulins (IgG) produced no such blocking. In the present study, optimal concentrations were first determined for each set of experiments in which these blocking Abs were used. Mouse hybridoma monoclonal antibodies (mAbs) against the a- and b-chain subunit of GM-CSF.R (clone S50 and clone S16, respectively) were obtained from Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA. Purified rhIL-1b came from Janssen Biochimica (Beerse, Belgium). Tumor Necrosis Factor-a (rhTNFa) was obtained from Hoffmann-La Roche (Basel, Switzerland). 2.2. Cell lines The human neuroblastoma SK-N-SH cell line was a gift of Dr. June Biedler, Sloan-Kettering Institute, New York, USA. The cells

were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (GIBCO Laboratories, Grand Island, NY, USA) supplemented with l0% fetal calf serum (FCS), L-glutamine, penicillin and streptomycin (all from Euroclone, West York, UK) and subcultured as appropriate. Cell aliquots were kept frozen in liquid nitrogen.

2.3. Cell survival and apoptosis A range of culture conditions were employed. Cells were cultured in DMEM supplemented with 1 mg/ml bovine serum albumin (BSA) (Sigma Chemical Company, St. Louis, MO, USA) without or with rhGM-CSF (0.1–10 ng/ml). Cells were subsequently washed, procured, and stained with propidium iodide (PI; 5 lg/ml) or with 40 nM 3,30 -dihexolyloxacarbocyanine iodide (DiOC6; Sigma) according to published methods [22,23,41–43]. Dead cells become permeable to PI, and being fluorescent, they emit bright red light. DiOC6 is a cell permeable green fluorochrome that is taken up by charged but not by depolarised mitochondria which therefore stains live but not dead cells. Apoptosis was specifically detected by doubling staining cells with annexin V-FITC (BD Biosciences PharMingen, San Diego, CA, USA) and PI. Briefly, the cells were washed in PBS and incubated for 15 min in 50 ll of a 1:20 dilution of annexin V-FITC, added to 350 ll PI and analyzed. Annexin V specifically binds to phosphatidylserine, a phospholipid expressed on the surface of apoptotic but not live cells. The results were expressed as percent viable cells.

2.4. Tumor cell implantation and growth in athymic mice Exponentially growing cultured glioma cells were washed with phosphate buffered saline, pH 7.2 (PBS) and briefly incubated with trypsin–EDTA. The enzyme was neutralized by the growth medium and suspended cells were washed once. Five million cells per line suspended in 100 ll of PBS in the presence of antibiotics and antimycotics (Collaborative Biomedical Products, Bedford, MA, USA) and 100 ll of Matrigel (Collaborative Biomedical Products), were inoculated sub-coutaneously (s.c.) into the flanks of 2-month-old athymic mice. Two or three animals were injected for each cell line and each animal received two injections with tumor cells, one in each flank with a total of 5  l06 cells injected s.c. per animal. Animals from the same set were caged together. Animals were observed weekly, beginning 7 days after cell injections; tumor size was measured sequentially. Six weeks after inoculation, all animals were sacrificed. At that time, tumors were dissected and weighed individually before being processed for histology, immunohistochemistry and RNA isolation. The tumor volume (V) was calculated by the equation V = 1/2  A  B2 in which A and B are the experimental measurements in mm of the length and width, respectively. The relative tumor volume (RV) was evaluated using the equation RV = Vx/Vo where Vx is the tumor volume at day x and Vo is the tumor volume at day 7. 2.5. Tissue preparation and tumor-derived cell lines Fragments of different primary brain glioma specimens were resected surgically from unrelated patients who had given their informed consent. The specimens were obtained intraoperatively, avoiding necrotic or degenerative tissues. Twenty different gliomas and astrocytomas were analyzed and classified histopathologically as follows, according to the World Health Organization (WHO) grading: four glioma IV (AS1, PG1, TV1, PR3), six astrocytoma III/ glioma IV (AV1, DG1, PL1, VS2, VS4, GS3), two astrocytoma III (VS3, MB1), two astrocytoma II–III (SB2,VC1), three astrocytoma II (CA1, CA2, MC1) and three astrocytoma I (VI2, LA1, BB1).

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359

2.5.1. Primary cultures and established lines Five glioma cell lines (PG1, AS1, TV1, PR3 and PL1) were obtained from the most malignant tumors. Briefly, small (1–3 mm) pieces of the tumors were incubated with 0.025% trypsin (DIFCO, Detroit, Mich., USA) in a Krebs Ringer’s buffer at 37 °C for 20 min under continuous gentle agitation, and then were centrifuged and washed. The cell pellet was resuspended in high-glucose– DMEM, filtered through sterile gauze and finally cultivated at a density of 1.5–2.0  105 cells/cm2 on poly-L-lysine (10 lg/ml)coated 35 mm-diameter culture dishes in DMEM. DMEM contained D-valine (GIBCO) which has been reported to be a selective agent in preventing the growth of fibroblasts in primary cultures without affecting neural cell growth [43,44]. It was then supplemented with 44 mM glucose, 2 mM glutamine, 10% non-inactivated fetal calf serum (FCS) (Hyclone, Logan, UT, USA), and 10 lg/ml gentamycin (Sigma) in an incubator in humidified 5% CO2 in air atmosphere at 37 °C. Twice a week, fresh medium was changed at 1:1 dilution. Primary cultures of neuronal-like cells developed and formed a confluent monolayer after about 1 month. These cells were incubated with 0.25% trypsin and 0.02% EDTA for 10 min at 37 °C and detached using a rubber policeman, dispersed by gentle pipetting and passaged at a split ratio of 1:3 into a new plate. Second passage-cultures were established after 1 month by trypsinization and replating of the cells at 1:5 dilutions. The cells were detached, re-suspended in serum-free DMEM, seeded on 6-well plates precoated with rhGM-CSF (20 ng/ml of PBS for 4 h at 37 °C and then washed), and incubated for about 16 h at 37 °C. Non-adherent cells were removed, and DMEM containing glucose, D-valine, 20% FCS, penicillin and streptomycin was replaced. When adherent cells reached semi-confluence, they were divided into normal 6-well culture plates and subsequently at a 1:3 dilution, with changes of fresh medium twice a week. Cells were characterized at low passages (3rd–4th passages) and at prolonged passages (14th and 26th passages) thereafter. Cultured cells were characterized for their tissue origin by the detection of the following markers, using a panel of monospecific Abs: NGF (rat mAbs, a gift of Dr. L. Aloe, Institute of Neuroscience, CNR, Rome, Italy), NGF receptor (NGF-R p75), vimentin and glial fibrillary acidic protein (GFAP) (mouse mAbs IgG, all three from Santa Cruz Biotechnology), as astrocyte markers; CD14 and CD11c (respectively mouse mAbs clones LeuM3 and LeuM5 from Santa Cruz Biotechnology), as monocytemicroglia markers; 200 KD a-neurofilament (mouse mAbs NG14, from Sigma) as neuron marker; fibronectin (mouse mAbs IgG1 anti-fibronectin, from Santa Cruz Biotechnology) as fibroblast marker; F VIII (mouse mAbs anti- F VIII-B clone MAS352, from Harlan Sera-Lab., Hilcrest, Belton, Loughborough, UK) and PECAM-1 (mouse mAbs IgG clone H300, from Santa Cruz Biotechnology) as endothelial markers; galactose cerebroside and 04-sulfatide antigen (respectively mouse mAbs IgG clone A2B5 and mAbs IgM clone 04, from Boehringer Manneheim, Germany) as oligodendrocyte markers. Additional markers included CD11a integrin aL, the ICAM-1 receptor (mouse mAbs IgG clone 16B8, from Santa Cruz Biotechnology), MHC-I (mouse IgG2 mAbs clone W6/32) and MHC-II (mouse mAbs IgG clone YE2/32I), both from Serotec, and CD58 (LFA3/FL250) rabbit polyclonal Abs from Immunotech SA. Non-relevant Igs (respectively mouse mAbs IgG anti-CD8ct clone D-9, and rabbit polyclonal serum Igs, both from Santa Cruz Biotechnology) were used as control Igs. For intracellular staining, cells were fixed in 4% formalin in PBS, permeabilized with acid alcohol (5% acetic acid, 95% ethanol), exposed first to primary Abs and then to the FITC-labeled secondary Abs. Only stable cell cultures showing a predominant proportion of cells with homogeneous staining for GFAP (>40% positive cells at the 4th cell passage and >90% at the 14th cell passage) almost free of neuron, oligodendrocyte, endothelial, macrophage-microglia cell contamination (the latter determined by staining with lineage specific Abs, see

349

above) were used in this study. Glioma cell cultures (14th–16th cell passage) were transplanted to athymic mice (2–3 month-old athymic mice from Harlan Sprague Dawley, Inc., Minneapolis, MN). When tumors developed, transplants were dissected and immediately processed for histology, immunohistochemistry and RNA isolation, or for cultures. 2.5.2. Line AS1 To investigate further a role of GM-CSF and GM-CSF.R in tumor progression, we used the human glioma cell line AS1, which was originally isolated from a 49-year-old male patient with a malignant glioblastoma multiforme and excessive granulocytosis in the peripheral blood. The original tumor showed cystic and necrotic components with abundant neutrophils. Karyological analysis of tumor cells revealed this karyotype: 49,xy,+1,6q,+7,10,17,+18. Immunohistochemically, a large cell fraction expressed GM-CSF protein as detected using affinity-purified polyclonal rabbit antirhGM-CSF IgG Abs and alkaline-phosphatase-labeled goat anti-rabbit IgG Abs. Fig. 1 Panel A, shows that AS1 cells (>90% GFAP positive at the 14th passage) formed tumors if xenografted to two recipient athymic mice. A tumor mass developed in each of two mice at the sites where 5  106 cells of the original AS1 line had been inoculated; the tumor latency was 56 and 52 days, respectively. Each tumor (with a size of approximately 10  10 mm in diameter) was subsequently surgically removed: one-half was fixed in 10% formalin and processed for histology, proving to be a glioma. The second half of the tumor was transferred onto a culture plate and cut into small pieces, which were incubated in a solution of trypsin–EDTA. Glioma cells were dispersed using a rubber policeman and by pipetting, washed by centrifugation and transferred into a polylysinecoated plastic culture dish in DMEM supplemented with FCS. They were then expanded by serial divisions at sub-confluence. Three new mass cultures were established: AS1A from the tumour of one mouse; GP1A and GP3A from the tumor of the other mouse. After four successive passages in vitro, the cells of each of these three lines were injected s.c. at a single site (5  106 cells/site) into new athymic mice (three mice in each case), generating a tumor at the site of the inoculum with a latency that was in all cases approximately 35 days. Each of these tumors was then removed, and the cells were dispersed and replaced in culture following the same procedure described above. Three new mass cultures, labeled GP1A1, GP3A2 and AS1A2, were isolated, expanded in vitro and subsequently injected to new athymic mice, generating tumors at the sites of the inoculation (tumor latency approximately 20 days with 5  106 cells/site). Two new mass cultures were isolated from the tumor induced by the AS1A2 variant, and glioma cells were further purified by selected GM-CSF affinity binding and cloned by limited cell dilution. Two clones were isolated by limiting cell dilution, named AS1A4 and AS1A3, and were characterized for their GFAP positive immuno-staining (>45% positive). They were determined to be free of endothelial cell contamination by negative staining with endothelial cell-specific antibodies against both Factor VIII (F VIII) and PECAM-1. Both AS1A4 and AS1A3 clones appeared to be highly tumorigenic when injected to nude mice (tumor latency approximately 10 days, with 5  106 cells/site), developing a greater tumor volume than the initial AS1 line and leading to metastasis at several body sites in the recipient animal after 8 weeks. All these variants had human karyotype and maintained their distinct tumorigenic property upon re-injection. Aliquots of all these lines, which excluded Mycoplasma contamination, were frozen and kept in liquid nitrogen. 2.6. Histology Tumor samples were fixed overnight in 10% formalin, dehydrated in 70% ethanol, then processed using standard procedures

350

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359

Abs for 30 min at room temperature, washed with cold DMEM, incubated with fluorescein-isothiocyanate (FITC)-labeled-rabbit anti-mouse IgG.F(ab)2 secondary antibody (Zymed Laboratories, San Francisco, CA, USA), washed and then fixed in 4% formalin in PBS. In some experiments, the cells washed in serum-free medium were detached by treatment with PBS/0.02% EDTA, suspended by gentle pipetting, transferred to a tube, centrifuged at l50g for 10 min, re-suspended in 0.5 ml PBS containing 0.5% formalin, and analyzed with a fluorescence-activated cell sorter (FACS scan, Becton Dickinson, Mountainview, CA, USA); 3000–5000 cells were gated for condition. The background fluorescence was evaluated by staining the cells either with the secondary Abs alone or with non-relevant Abs (anti-CD8a, clone D-9). FITC-labeled cells were also examined using a Polyvar (Reichert-Yung, Wien, Austria) confocal microscope, equipped with interference contrast and fluorescence optics. Their tissue origin was characterized by the immunofluorescent detection of specific cytoplasmic markers.

2.8. Bioassays for GM-CSF 2.8.1. Human marrow in soft agar clonogenic cultures Normal human marrow aspirates were obtained with informed consent. Non-adherent light-density (30 cells/colony) were read after 21 days.

Fig. 1. The AS1 line and its clones. Panel A. The protocol adopted for these experiments and the cumulative results obtained are shown. Three phenotypically stable culture variants (GP1A, GP3A, AS1A) were reconstituted from one of the early tumors which developed in nude mice inoculated with AS1 cells (a total of 5xl06 cells in the two flanks per mouse; tumor latency, approximately 50 days); these cultures formed local tumors (latency approximately 35 days, always with 5  l06 cells injected s.c. per mouse). As a consequence of serial in vitro passages, three new malignant clonal variants were then isolated by limiting cell dilution in culture (GP1A1, GP3A2 and AS1A2) which generated tumors with a latency of approximately 20 days when injected into animals. Two high-grade malignant clones were then obtained from AS1A2 cells, by limiting cell dilution in culture (AS1A4 and AS1A3). Fast-growing tumors formed in two nude mice (latency 10 days, with 5  l06 cells injected s.c. per animal), developing a greater tumor volume than the initial AS1A2 line and leading to metastasis at several body sites in the recipient animal after 4 weeks. Panel B (a) Positive expression of GM-CSF (arrows) in a glioma which developed after xenograftingAS1 cells in a nude mouse. (b–f) Immunofluorescence evaluation of AS1 cells (14th passage) after incubation with purified antiGFAP Abs (b), anti-GM-CSF Abs (c), anti-GM-CSF.Ra Abs (d), anti-NGF-R p75 (e) and anti-MHC-I Abs (f).

and embedded in paraffin. Sections (<10 lm) were cut, mounted on slides, and stained with hematoxylin and eosin (H&E) or further analyzed by immunohistochemistry.

2.7. Immunohistochemistry The paraffin-embedded sections were deparaffinized in xylene, rehydrated in a graded series of ethanol, and rinsed in distilled water. GM-CSF, GM-CSF.R and GFAP production was evaluated after incubation first with purified monospecific primary Abs and then with alkaline phosphatase-labelled secondary Abs. Tumor cells were grown on poly-L-lysine (Sigma)-coated 12-mm round glass coverslips maintained in 24-multi-well plates and characterized phenotypically, by indirect immunofluorescence, for expression of the following surface markers: MHC-I, MHC-II, ICAM-1, CD11a, NGF Receptor (NGF.R p75). Coverslips were washed in cold serum-free DMEM between steps. Cells were exposed to the primary

2.8.2. Cultured glioma cell proliferation assay Cell proliferation was assessed by [3H]Thymidine incorporation. Briefly, 6000–10,000 cells per well were plated in quadruplicates in 96-well plates in DMEM supplemented with 10% FCS and antibiotics in 96-well plates. After 20 h, medium was shifted to DMEM with 0.5% FCS in the absence or presence (0.1 to 20 ng/well) of rhGM-CSF. Every day for 6 days of culture, cells were counted. Additionally, in parallel experiments 1 lCi of [3H]Thymidine was added to each well and the cells were harvested after 20 h, after extraction with 1 N NaOH, using a cell harvester. The radioactivity incorporated into DNA was quantified by liquid scintillation.

2.9. Conditioned medium and cell product preparation Conditioned medium and cell lysates were prepared using the following protocol [45]. Briefly, for every cell line, six-well plates were seeded with 6000–25,000 cells/well on day 0 in 2 ml medium containing 10% FCS. Each day, starting the following morning (day 1) for a total of 6 days, the culture supernatant from six replicas was harvested and the cells were washed twice with serum-free medium and pre-conditioned for 3 h in 2 ml medium containing 2% FCS for 24 h at 37 °C. They were then collected in separate tubes containing 1 ll of 100 mM phenylmethylsulfonyl-fluoride (PMSF). The medium was then centrifuged and the supernatant (conditioned medium) was collected and stored at 20 °C until further analysis. The cells were washed briefly with cold PBS, detached by mechanical scraping, counted, and then lysed with a lysis buffer (150 mM NaCl, 0.5% NP-40, 5 mM EDTA, 20 mM Tris pH 7.6). To 1 ml of lysis buffer, 1 ll of leupeptin (10 mg/ml) and 10 ll of PMSF (100 mM) were added. The lysates were collected, passed at least four times through a 23 gauge needle with a syringe, and incubated on ice for 1 h. Tubes were then centrifuged at 4 °C for 20 min. The supernatant was collected, divided into aliquots, and stored until further analysis.

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359

2.10. Detection of GM-CSF by solid phase enzyme-linked immunosorbent assay (ELISA) To quantify endogenous or secreted GM-CSF, a sandwich technique using a capture/detection antibody pair was used. Affinity purified capture mAb (clone G7.A/1, 2 lg/ml in 0.1 M NaHCO3, pH 8.6) [38,39] was adsorbed onto a 96-well microplate (Falcon Plastic, Oxnard, CA, USA) by overnight incubation at 4 °C while shaken. After washing three times with PBS-0.05% Tween, the plate was blocked with 0.5% bovine gelatin for 4 h at room temperature. rhGM-CSF (5.0 lg/ml) or test samples were added after washing, and the plate was incubated for l6 h while shaken gently at room temperature. Bound rhGM-CSF was detected using alkaline phosphatase-conjugated rabbit anti-rhGM-CSF secondary Abs and 4-nitro-phenyl-phosphate (4-NPP) (Merck, Darmstadt, Germany) as substrate. The enzyme reaction was evaluated by absorbance at a wavelength of 405 nm. In competition binding assay, the free GM-CSF concentration was determined by extrapolation from a standard curve from known amounts of the cytokine. 2.11. Receptor binding assays For binding, 7  106 cells were suspended in each well of a 6-well plate on day 0 in medium containing 0.2% bovine serum albumin (BSA) and increasing concentrations of 13II-labeled rhGM-CSF (range from 5 pmol/L to 10 nmol/L) with or without excess unlabeled rhGM-CSF. Cells were incubated overnight at 4 °C, washed in PBS, detached by mechanical scraping, centrifuged for 5 min in the cold through a cushion of FCS, and the cell pellets were washed with cold PBS. Bound rhGM-CSF was quantified by c-spectrometry. The number of binding sites per cell and their affinity were evaluated by Scatchard analysis. Because of differences in seeding efficiency among experiments and a variation in the proliferation rates among cell lines, the number of receptors was evaluated by final unit cell numbers (receptors/106 cells). 2.12. RNA-isolation, Northern blot analysis and RT-PCR The total cellular RNA was extracted as described [45–47] from glioma cells, tissue-derived cells and the following control cells: the SK-N-SH neuroblastoma cell line (negative control for GM-CSF) and human tumor prostate tissue (positive control for GM-CSF). The integrity and amount of each RNA sample were determined by staining the gel and visualizing it under UV light. For Northern blot analysis, RNA was size-fractionated in acid formaldehyde agarose gel and transferred onto Hybond C extra membrane (Amersham, Bioscences, Milano, Italy). For hybridization, the probes were labeled by the hexanucleotide primer method with 32P-dCTP (Amersham). The poly(A)+ RNA from tissues and cells for transcripts homologous to GM-CSF, G-CSF, VEGF, and VEGF.R cDNA probes were isolated after two cycles of oligo(dt)cellulose chromatography. This RNA was denatured, fractionated by electrophoresis in 1% (w/v) agarose gel containing formamide, blotted (3 lg/lane) after appropriate calibration with a b-actin gene probe and hybridized to the desired 32P-radiolabeled probe after recovering from the previous probe. The same filter was tested sequentially with each probe to ensure that observed hybridization represented true variations among samples rather than variations in sample preparation. Radiolabelled probes had a specific activity of about 3  108 dpm/ lg DNA. For RT-PCR analysis, total cellular RNA was extracted using RNA zol (Cinna/Biotec, Houston, TX, USA) from frozen tissue specimens and cell lines. The cDNA synthesis was performed by reverse transcription at 42 °C for 45 min in a volume of 10 lg total RNA with 0.5 lg oligo(dT)primer, 1 mM dNTPs (20 deoxynucleotyde-5triphosphate), 5 ll of 10 RT buffer (100 mM Tris–HCl, pH 8.8, 500 mM KCl, and 1% Triton X-100), 5 mM MgCl2, 20 U RNase, and

351

12.5 U avian Moloney virus (AMV)-RT. All reagents for cDNA synthesis were from Promega (Promega, Madison, WI, USA). Five microliters of cDNA were amplified in the presence of 25 pmol of each sense and antisense primer, 10 PCR buffer (0.5 M KCl, 0.1 M Tris–HCl, pH 8.8, 0.025 M MgCl2, 0.002 M each dNTP, and 2 mg/ml BSA), and 2.5 U of AmpliTaq DNA polymerase (Perkin–Elmer/Cetus, Monza, Italy). The mixture was overlaid with mineral oil and amplified in a Thermal Cycler (Cetus Corp., Emeryville, CA, USA) with PCR cycle conditions specific for GM-CSF and its receptor chain subunits. Fifteen microliters of PCR product was electrophoresed in a 2% agarose gel in Tris/boric acid/EDTA buffer. Gels were stained with ethidium bromide and photographed. The oligonucleotide primers that were used are the following: GM-CSF: sense: 50 -TGGCCTGCAGCATCTCTGCA-30 ; antisense: 50 ACACGTTGGGTCTGATAGTG-30 . GM-CSF.R b: sense: 50 -CTACAAGCCCAGCCCAGATGC-30 ; antisense: 50 -ACCCGTAGATGCCACAGAAGC-30 ; GM-CSF.R a: sense: 50 -AGCCCGAGCAAAACACA-30 ; antisense: 50 -CCATGCCATTCCTACACCCT-30 G-CSF: sense: 50 -ACTTTGCCACCACCATCTGG-30 ; antisense: 50 GATTCCCCAGCAAATTCGTG-30 b-Actin: sense: 50 -GAAGTGTGACGTGGACATC-30 ; antisense: 50 ACTCGTCATACTCCTGCTTG-30 VEG.F.R1: sense: 50 -TGCTTGAAACCGTAGCTGG-30 ; antisense: 50 GGTGCCAGAACCACTTGATT-30 VEGF: sense: 50 -ATGGCAGAAGGAGGAGGGCAGAAT-30 ; antisense: 50 -TTGGTGAGGTTTGATCCGCATAAT-30 3. Results 3.1. GM-CSF production in gliomas 3.1.1. Histology and immunohistochemistry Tumors were stained by H&E and analyzed by immunohistochemical staining conducted on consecutive sections after incubation with both anti-rhGM-CSF Abs and anti-GM-CSF.R (a- or b-chain) Abs. The results of three separate experiments are reported in Table 1. Tumor cells from the highest-grade malignant gliomas WHO grade IV (AS1, PG1, TV1, PR3), and to a lesser extent from astrocytoma grade III/glioma grade IV (AV1, DG1, PL1, VS2, VS4, GS3), consistently stained positively after incubation with anti-rhGMCSF Abs (see for example Fig. 1 Panel B). Instead, tumour cells isolated with a lower malignant grade (VS3, MB1, SB2, VC1, CA1, CA2, MC1, VI2, LA1 and BB1) compared to the sensitivity of the immunostaining method, did not stain with these Abs or revealed only sporadically isolated clusters of cells, which reacted positively with these Abs. Tumor sections with cells that were positively immunostained for the anti-GM-CSF Abs often revealed the presence of cells positive for the Abs anti-GM-CSF.R (a- or b-chain receptor subunit) as determined by immunohistochemical analysis conducted on consecutive sections. Sections from a prostate tumour which stained intensively with these Abs were used as a positive tissue control, while sections of a normal human skin explant and cells from the human neuroblastoma SK-N-SH line grown in monolayer, which never reacted positively with these same antibodies, were both used as negative controls in these assays. 3.1.2. Gene expression As shown in Fig. 2 and summarized in Table 1, by using the Northern blot analysis, a high level of positive GM-CSF and GM-CSF.R (a-chain) mRNA expression was obtained in the highest clinical grade malignant tumors, whereas only weak positivity of gene expression was revealed in some lower malignant grade gliomas (for example SB2, MB1, CA1) or no detectable levels (for example CA2, MC1, VI2, LA1 and BB1. These results correlated closely with

352

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359

Table 1 Expression of GM-CSF and its receptor (GM-CSF.R) in human gliomas evaluated by immunohistochemical staining and Northern blot analysis. Glioma

WHO grades

AS1 PG1 TV1 PR3 AV1 DG1 PL1 VS2 VS3 VS4 GS3 MB1 SB2 VC1 CA1 CA2 MC1 BB1 VI2 LA1

Glio IV Glio IV Glio IV Glio IV Astro III/Glio Astro III/Glio Astro III/Glio Astro III/Glio Astro III Astro III/Glio Astro III/Glio Astro III Astro II–III Astro II–III Astro II Astro II Astro II Astro I/II Astro I Astro I

GM-CSFa

IV IV IV IV IV IV

GM-CSF.Ra

mRNA

Protein

mRNA

Protein

++ +++ ++ ++ + + +/++ + ± ± +  ±/+ ±/+ ± ±    

++ ++ ++ ++  + + + nd nd nd  ±       

+/++ +/++ +/++ ++ ±/+ + + + + ±/+ + ±/+ ± ± +     

++ ++ ±/+ +   + + ± nd +         

Values shown are cumulative results of three separate experiments, using an arbitrary scale of evaluation expression:  , negative; ± to +++, positive based on an arbitrary scale of positive expression; nd, not done. a BB1 benign astrocytoma or SK-N-SH neuroblastoma cells were used as negative control and human prostate tumor as positive control for GM-CSF and GM-CSF.R production.

constitutively expressing GM-CSF and GM-CSF.R, a preliminary selection was made by sorting glioma cells (over 80% GFAP positive) (Fig. 1 Panel B), based on their capacity to adhere strongly to GM-CSF-pre-coated culture plates and to generate by division enlarging colonies. Once the cells had reached sub-confluence they could be subsequently divided and continued to expand thereafter. Continuos cultures were obtained from only 13 of 20 tumors. 3.2.2. Immunophenotyping A large proportion of the cells of these primary mass cultures produced GM-CSF constitutively, as revealed by immunostaining (Fig. 1 Panel B). The presence of GM-CSF was clearly discernible in the cytoplasm, with staining at the cell membrane for GM-CSF.R. Specificity of the Abs against GM-CSF was confirmed by competition with the soluble recombinant protein. Positive staining with anti-GM-CSF.R Abs was also confirmed in a large cell fraction in early primary cultures, but the number of positive cells and the intensity of their immunostaining for GM-CSF.R significantly decreased with successive cell passages (>80% at passage 10th, <30% at passage 14th). A large cell fraction always expressed NGF.R and MHC-I constitutively. 3.2.3. Northern blot analysis Fig. 2 shows the results of an experiment of gene expression analysis by Northern blot assays of these 13 primary glioma cultures performed with mRNA extracted from pooled cells from their 12th to the 14th passage. GM-CSF and its receptor mRNA (both chain subunits) were not always equally co-expressed in the same cell cultures. In lines AS1, PG1, PR3, PL1 and VS2 mRNA expression of both GM-CSF and GM-CSF.R (both chain subunits) were always found co-expressed. However, in lines AV1, MB1 and SB2, we consistently found that the cells could constitutively express mRNA for one or both GM-CSF.R chain subunits but did not detect levels of GM-CSF. GM-CSF and GM-CSF.R were never detected by Northern analysis in the five low-grade astrocytoma-derived cultures CA2, MC1,VI2, LA1 and BB1 and in the human neuroblastoma SK-N-SH line, confirming similar results obtained by immuno-histochemical staining (Table 1). 3.3. Establishment of five lines

Fig. 2. Northern blot analysis of GM-CSF, GM-CSF.Ra and GM-CSF.Rb in purified glioma cells isolated from primary mass cultures. Co-expression of both GM-CSF and receptor mRNA were always found in high-grade malignant tumor cells. On the contrary, low levels of one or both gene expression was usually shown in lowergrade tumors. The human SK-N-SH line and human prostate mRNA were the negative and positive control lanes. Hybridization of the same blots with a b-actin cDNA probe was performed to ascertain for the amount and quality of RNA loaded in each lane.

the levels of positive antigen production measured by the immunohistochemical analysis of the same cell lines shown in the Table. 3.2. Glioma-derived primary mass cultures 3.2.1. Adopted procedure To examine whether there was any possible association between constitutive production of GM-CSF and GM-CSF.R with tumorigenicity and high cell-proliferation rate in gliomas, we attempted to isolate in vitro primary glioma mass cultures from all 20 tumors. In order to increase the initial fraction of neural cells

3.3.1. Cell composition Continuous lines that could be maintained in culture for >10 cell passages (approximately 30 population doublings), containing high proportion of cells that expressed glial-specific GFAP markers constitutively and were free of endothelial cell-specific markers (e.g. F VIII and PECAM-1), occurred only in five of these primary 13 cultures (i.e. PG1, AS1, TV1, PR3 and PL1). These five lines derived from the most highly malignant tumors with high constitutive expression of the GM-CSF and GM-CSF receptor genes. Fig. 3A shows the characteristic growth curves of lines PG1, AS1, TV1, PL1 and PR3 (all tested between the 8th to 10th passage). Line PG1 showed the greatest proliferation, compared to lines AS1 and TV1 which exhibited apparently similar doubling times, and PL1 or VS2 lines which divided more slowly. All these five lines maintained contact inhibition of cell growth. Subconfluent cultures were analyzed after 12 successive cell passages of the five established glioma cell lines: the cells of these cultures were singlestained with specific markers by immunocytochemistry. As an example, we report the composition of the glial AS1 line tested at different passages (passages 4th, 14th, 26th), stained positively for vimentin and GFAP, NGF and NGF.R. Interestingly, we found that the cells of the early passages (14th passages) expressed GFAP (75–95% positive by immunostaining). This was much less NGF.R (about 10–35% of the cells) and than NGF (about 5–10%) and even less than vimentin (1–2%). With successive passages of the cells in

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359

353

Fig. 3. Growth and GM-CSF production of primary glioma cultures. (A) Growth curves of five glioma lines (PG1, AS1, TV1, PL1, and PR3) tested between the 8th and 10th passage. Six thousands cells were seeded in culture and then counted daily. Data are shown as average values of cells from quadruplicates. Standard deviations are not shown because always <5%. (B). Kinetics of GM-CSF production by five glioma lines (AS1, PG1, TV1, PL1 and PR3). Cells (6000 cells seeded per well) were cultured for 6 days before harvesting. GM-CSF was assayed in the culture supernatant conditioned media by ELISA, as described in Section 2. Results are average daily GM-CSF (pg/106/ml) from quadruplicates. (C). Competition ELISA. A fixed amount of partially purified GM-CSF released from 6-day AS1A3 cells into the culture supernatant conditioned medium was first adsorbed onto the well of a 96-well plate which was then saturated with gelatin. A fixed amount (5 lg) of purified mouse anti-GM-CSF clone G3.7/1 Abs was then incubated in the absence (100% binding capacity) or in the presence of varying concentrations of purified rabbit polyclonal anti-GM-CSF Abs or normal isotypic IgG as negative control. Mouse antibody binding was then evaluated using AP-labeled anti-mouse IgG secondary Abs. (D). BM-MNC clonogenic assay. Seven-day old supernatants of four glioma lines (AS1, PG1, PL1 and MC1) were tested for the capacity to stimulate myeloid colony formation (CGU-GM) from 2  104 BM-MNC in soft agar cultures. Culture supernatants were tested in the absence or presence of purified blocking anti-GM-CSF polyclonal Abs (2 lg and 20 lg respectively) or normal isotypic IgG (20 lg) as negative control. Results are number of myeloid colonies per culture dish. Standard deviations are not shown because they were always <5%.

culture we found the gradual phenotypic changes. After 26 passages, the line was stabilized, in a population containing approximately 10% of the cells positive for GFAP, about 2% positive for NGF.R, but virtually every cell expressing vimentin (>90% positive). All cultures were free from endothelial cell contaminations as determined by the lack of staining with endothelial-specific Abs against F VIII-B and PECAM-1. In addition, the cells expressed histocompatibility antigens (MHC-I constitutively and MHC-II mainly after induction with rhIL.1b), adhesion proteins (ICAM-1, LFA3) and members of the integrin family. Cells did not stain for neurofilament (200 KDa) (neuron marker), O4-sulfatide or galactocerebroside (oligodendrocyte markers), CD14 and CD11c (monocyte–macrophage and microglia markers). From these findings, we concluded that all of these five cell lines constitutively expressed a typical pattern of glial-specific antigens. With prolonged passages in culture (26th passages), the cells gradually lost constitutive GFAP expression, while a large proportion of the cells continued to stain positively for vimentin, NGF, NGF.R and MHC-I. However, upon induction with rhIL-1 (4 ng/ml), GFAP expression could be rapidly induced in a predominant (>50%) cell fraction,

suggesting different stages of differentiation in the cell population (data not shown). Additional analysis on the PG1 and As1 lines showed that cultures of advanced passages (i.e. after the 14th passage) normally failed to be immunoreactive with anti-MHC-II Abs. On the other hand, the cells could be positively induced to modulate the expression of these membrane surface molecules after adding rhIL-1b (4 ng/ml) to their growth medium (data not shown). Immunophenotype analysis was performed in the AS1 cell line (passages 14th and 26th) by comparing the wild-type cell population and a sorted cell population, based on high affinity binding to culture plates pre-coated with purified anti-NGF.R or anti-GMCSF.R antibodies. No major difference in growth curve, life span or marker antigen expression was observed by comparing the initial and the sorted cells, confirming a homogeneous population (data not shown). 3.3.2. Production of GM-CSF and GM-CSF.R 3.3.2.1. Immunological staining. Based on immunoperoxidase staining, more than 60% of the cultured cells (14th passage) of the five glioma lines PG1, AS1, TV1, PL1 and VS2 were constitutively

354

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359

producing GM-CSF. A largely scattered fraction of these cells (range 15–45%) also co-expressed the GM-CSF specific receptor. Expression of GM-CSF and GM-CSF.R was always negative in MC1, BB1 and SK-N-SH cells (data not shown). 3.3.2.2. Elisa. Subsequently, we used ELISA to examine GM-CSF levels in soluble cell lysates and culture conditioned media prepared from these five lines. We found that unless cells were detached from the culture plates by scraping with a rubber policeman, most of the GM-CSF was kept within the cells, with the exception of AS1 and PG1 cell lines which consistently released GM-CSF into their conditioned media, starting from the 4th to 5th day after the moment of seeding. Fig. 3B shows the results of a representative experiment of daily GM-CSF production by the five cell lines after seeding 6000 cells in each well of a six-well plate on day 0. By ELISA, we measured GM-CSF levels in conditioned media of cells that had been pre-conditioned and then conditioned as described in Section 2. The results of a representative experiment of antiGM-CSF antibody binding inhibition performed with AS1A3 cells (6-day culture) using conditioned media of cells that had been pre-conditioned and then conditioned are shown in Fig. 3C. By ELISA we also measured endogenous GM-CSF extracted in cell lysates of 6-day old cultures from all cell lines. GM-CSF was detected in cell lysates at high levels in AS1 and PG1 cells (19,852 ± 760 pg and 15,465 ± 610 pg per l06 cells, respectively), at moderate levels in TV-1 and PL-1 cells (5680 ± 810 pg and 3487 ± 410 pg per l06 cells, respectively) and at a low level in PR3 cells (748 ± 42 pg per l06 cells). Cell lysates from the two CA1 line and BB1 primary cultures did not produce detectable GM-CSF (<40 pg/ml per l06 cells). Inhibition of binding experiments with Abs anti-GM-CSF confirmed the specificity of these reactions.

3.4.2. GM-CSF production in clonal variants The presence of GM-CSF was analyzed by competition ELISA in both cell lysates and culture conditioned media of all the AS1 variant mass cultures and clones described above. Fig. 4A shows that all these glioma cell variants produced GM-CSF, clearly detected in cell lysates particularly from the highly-tumorigenic AS1A2, AS1A3 and AS1A4 lines. On the contrary, low amounts of GM-CSF were detected in cell lysate of exponentially growing cells from the lowtumorigenic AS1A, GP1A, GP3A and GP1A1 variants. The former also released large amounts of GM-CSF into their culture conditioned media, as determined by competition ELISA, while the latter did not release the growth factor into their culture-conditioned media until 6 days after the moment of seeding, probably when confluent cells began to die (not shown). Therefore, the level of GM-CSF measured in the cell lysates of these lines on the sixth day was considered the total accumulation for 6 days. 3.4.3. rhGM-CSF added to culture medium selectively stimulates growth of individual clones derived from the AS1 line As shown in Fig. 4B the addition of rhGM-CSF to cultures selectively stimulated proliferation of clones AS1A3 and AS1A4, constitutively normally and had high GM-CSF.R mRNA expression (both a- and b-chain tested), but failed to stimulate proliferation of clones GP3A, which does not normally co-express the high affinity GM-CSF receptors. The specific stimulatory activity of rhGM-CSF was inhibited by anti-GM-CSF antibodies in the two responder

3.3.2.3. Bioassays. To demonstrate that functionally active GM-CSF was produced by the positive lines, GM-CSF activity in culture cell lysates was tested for its ability to support growth and myeloid differentiation of 21-day-old colonies from human non-adherent BMMNC in soft agar assays. Soluble lysates of 7-day-old glioma cultures were concentrated by lyophilization, reconstituted (10x concentrated the original volume) and then tested in the absence or presence of purified polyclonal rabbit anti-rhGM-CSF Abs (2 and 20 lg) and normal rabbit IgG (2 and 20 lg) as controls, respectively. In a representative experiment, the number of 21-day myeloid colonies (CFU-GM)/2  105 BM-MNC) (Fig. 3D) clearly demonstrated that cell lysates from positive AS1 and PG1 lines contained GM-CSF clonogenic activity, which was specifically inhibited by anti-GM-CSF antibodies. On the contrary, lysates from PL1 and MC1 cells as well as control neuroblastoma SK-N-SH cells (the latter not shown) failed to stimulate colony formation in soft agar cultures. This was consistent with the findings of the ELISA assays reported above. 3.4. The AS1 cell line as a model for studying correlation between GMCSF production GM-CSF receptor co-expression, tumor growth and progression 3.4.1. The AS1 line We wanted to determine whether differences in tumorigenicity and tumor cell proliferation in vivo might correspond to a different expression and productivity of GM-CSF and GM-CSF receptors by the cell lines in vitro. We developed a model by injecting s.c. AS1 cells into athymic mice. The cells isolated from tumors were expanded in culture and new clonal variants with increased tumorigenicity were produced after subsequent in vivo and in vitro cell passages, as described in Section 2 and shown in Fig. 1 Panel A.

Fig. 4. GM-CSF release in cell extracts of AS1 cells and other clones and lines, measured by ELISA in 7-day culture cell lysates. (A) Higher amounts of this cytokine were particularly detected from the highly tumorigenic AS1A2, AS1A3 and AS1A4 clones, compared to the low-tumorigenic GP1A, GP3A, AS1A and GP1A1 cultures. Results are pg GM-CSF/106 cells/ml. (B) rhGM-CSF selectively stimulated proliferation of AS1-derived clones in a dose-dependent fashion. Values represent the mean of dpm from triplicates after subtracting the dpm measured in the absence of added rhGM-CSF. Control values (no rhGM-CSF) were: 7350 ± 717 dpm for AS1A3; 4103 ± 318 dpm for AS1A4; 5320 ± 712 dpm for GP3A cells. (C) Cultures were also incubated with anti-GM-CSF purified polyclonal Abs (0.5 lg/ml and 2 lg/ml respectively) or normal IgG (2 lg/ml) as negative controls. Results are shown as relative proliferation values (1 = absence of IgG).

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359

AS1A4 and AS1A3 clones, but cell proliferation was not affected in the cytokine-insensitive GP3A clone (Fig. 4C). 3.4.4. Regulation of different gene expression and protein production by exogenous cytokines These preliminary findings led us to investigate further the possibility of regulating GM-CSF and GM-CSF.R and other gene expression on cultured AS1 cells and its clonal variants in vitro. Three different series of experiments were performed. 3.4.4.1. Northern blot analysis. In one series of experiments, a Northern blot analysis was performed in order to examine the effects of stimulating AS1A3 cells (a pool of cells from the 10th to the 12th passage) with rhGM-CSF (50 ng/ml), on the expression of the GM-CSF gene as well as the GM-CSF.Ra, within 3 h from stimulation. These experiments were performed in order to verify a potential stimulatory role of this growth factor via autocrine/paracrine loops during tumor development. As shown in Fig. 5A, we found a significant progressive increase of mRNA GM-CSF and GM-CSF.R expression in response to rhGM-CSF as a function of time. We then further investigated the effect of stimulating AS1A3 cells with rhGM-CSF (50 ng/ml) on the production of G-CSF and the concomitant production of other functional pro-

355

teins, such as VEGF and VEGF.R1, potentially stimulating angiogenic regulation and tumor growth in vivo. These three proteins are constitutively poorly expressed in AS1A3 cells (not shown). We found that the addition of rhGM-CSF to the culture medium up-regulated cell mRNA co-expression of the G-CSF gene in AS1A3 clone, and also G-CSF, VEGF and its specific membrane surface receptor (VEGF.R1) mRNA expression. 3.4.4.2. Elisa. In another series of experiments, we tested AS1A3 cells (6th passage) for NGF production, by competition ELISA. As shown in Fig. 5B, the cells grown on 6-well plates for 5 days until they had reached confluence, constitutively released low levels of NGF into their conditioned medium (approximately 22 pg/ml). However, rhIL-1b added to the growth medium (4 ng/ml) strongly stimulated NGF release (1023 ± 125 pg/ml). TNFa (10 ng/ml) added alone or in combination with IL-1b did not influence NGF release (44 ± 5 pg/ ml and 1086 ± 58 pg/ml, respectively). In the same experiments, we found that both rhIL-1b and rhTNFa strongly stimulated increased NGF.R and MHC-II membrane surface expression, as quantitatively measured by flow cytometric analysis (data not shown). 3.4.4.3. Binding assays. In a third series of experiments, AS1A3 cells (8th passage) and three other variants, respectively AS1A, AS1A2

Fig. 5. Regulation of different gene expression and protein production by exogenous cytokines in AS1A3 cells. (A) Northern blot analysis of GM-CSF mRNA, GM-CSF.R,a G-CSF, VEGF and VEGF.R1 in AS1A3 cells (10th-12th passage) within 3 h from stimulation in culture with rhGM-CSF (50 ng/ml). This cytokine induced a significant concomitant increased expression of all these five genes. The amount and quality of RNA extracted from cells were ascertained by hybridization of the same blots with a b-globin cDNA probe. RNA extracted from SK-N-SH (negative control) and human prostate (positive control) were also analyzed. (B) Production of NGF in response to rhIL-1b (4 ng/ml) after 5 days in culture. TNFa (10 ng/ml) added alone or in combination with rhIL-1b did not influence NGF production. Each bar represents the mean (±SE) of three independent experiments. (C) Effects on GM-CSF.Ra expression induced by rhIL-1b (4 ng/ml) and by rhEGF (50 ng/ml) added to cultures, measured by RIA with 131I-labeled rhGM-CSF, as described in Section 2. Results are shown as binding sites/cell.

356

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359

and AS1A4, expression of GM-CSF.R was evaluated by binding assays. Cells (7  106) were incubated with radiolabeled GM-CSF at concentrations that ranged from 5 pmol/L to 10 nmol/L. 131 I-GM-CSF binding was dose dependent and saturable at approximately at 10 nmol/L. Scatchard analysis of data obtained from these binding experiments revealed an average of approximately 500 sites per cell, although two classes of binding sites were detected (about 160 ± 18 sites of Kd 40 pmol/L, and approximately 320 ± 27 sites with a Kd of 2.8 nmol/L). Cells were stimulated in culture with EGF (10 ng/ml) or rhIL-1b (4 ng/ml). The latter, but not the former, stimulated enhanced membrane binding sites expression (range 13,000–19,000 total binding sites with a significantly increased proportion of high-affinity sites, about 70% with a Kd of 46 pmol/L) as compared to untreated cells. Fig. 5C shows the results of a representative experiment. Taken together, the data from these three series of representative experiments confirmed that AS1A3 cells were able to undergo significant dynamic changes in response to different environmental signals. 3.4.5. Anti-apoptotic effects of GM-CSF Given our demonstration that AS1A3 cells express both GM-CSF and its receptors constitutively, and that exogenous rhGM-CSF was capable of stimulating their proliferation in culture, we next considered the possibility that GM-CSF might effect survival and an anti-apoptotic function in these cells. From the evidence reported above of the effects of the exogenous rhGM-CSF added to the culture medium of clones AS1A3, we employed blocking polyclonal anti-GM-CSF antibodies in an attempt to neutralize endogenously-produced GM-CSF, using the same clone AS1A3 and also clones AS1A4, GP3A and GP1A1, whose production of GM-CSF and its receptor significantly differ from one another (see above). The effect of anti-GM-CSF antibodies on cell viability was evaluated by FACS analysis as an increase of PI staining and a reduction of DiOC6 staining, a method that measures apoptosis via collapse of mitochondria membrane potential. Apoptosis was tested also by an increase of annexin V-FITC staining. The IgG isotypic control was used at the same concentration. As shown in Fig. 6, we found that both our blocking monoclonal and polyclonal antibodies on the two GM-CSF-sensitive AS1A3 and AS1A4 clones specifically produced a marked reduction in their survival, but did not affect the survival of GP3A or GP1A1 clones that did not express GMCSF.R (Fig. 6A). To provide further evidence that GM-CSF could act as an autocrine rescue factor for AS1A3 cells, we employed a range of cell densities. Blocking GM-CSF by polyclonal antibodies induced cell killing by apoptosis over a wide range of cell densities (Fig. 6B). The results shown are for 48 h but comparable data were obtained for 24 h or 72 h and later time points (not shown). These results strongly suggest that a likely important contribution of GM-CSF is to provide protection from apoptosis in AS1A3 cells. 3.4.6. Correlation between high tumorigenicity in mice and levels of GM-CSF/GM-CSF.R mRNA expression in tumors The results reported above induced us to measure the tumours which developed in nude mice inoculated with glioma cultures and the AS1 cell variants described above, in an attempt to correlate high tumorigenicity with RT-PCR analysis of the tumors, testing polyadenylated mRNA for GM-CSF and for GM-CSF.R (both a- and b-chains). Our data shown in Fig. 7 indicate a positive direct correlation between the developing relative tumor volume (Fig. 7A) and the high GM-CSF and GM-CSF.R mRNA co-expression measured in the same tumors, for the lines and clones tested in these studies (Fig. 7B). These results provide further evidence indicating that constitutive high GM-CSF and GM-CSF.R protein production and mRNA

Fig. 6. Effect of blocking anti-rhGM-CSF Abs on viability of glioma cells. (A) Glioma cells from four different lines (AS1A3, AS1A4, GP3A and GP1A1) were cultured at 2.5  105 cells/ml. Viability was measured by PI exclusion and DiOC6 staining at 24 h. Purified anti-rhGM-CSF neutralizing monoclonal Abs (clone G3.7/1) and polyclonal Abs were used at 20 lg/ml, a concentration that had previously shown to produce maximal cell killing. The normal IgG isotypic control was also used at the same concentration. Both Abs but not the IgG control induced a marked reduction in the survival of AS1A3 and As1A4 cells, but had no effect on GP3A and GP1A1 cells. (B) Effect of purified anti-rhGM-CSF polyclonal Abs on the survival of AS1A3 cells cultured at different cell densities. Cells were cultured for 3 days in the presence of blocking Abs or normal isotypic IgG control. Cell death by apoptosis was measured as marked increase of PI staining, reduction in DiOC6 staining and increased surface staining with annexin V-FITC. In both panels, the data summarize the results of all and are expressed as percent viable cells.

expression levels correlate positively with cell malignancy in the gliomas, and are important in the induction and development of tumor progression.

4. Discussion We investigated the expression of both GM-CSF and GM-CSF.R in brain gliomas of different malignant grades isolated from different unrelated patients. We found that these genes were overexpressed in the most malignant tumors, whereas benign tumors failed to express either GM-CSF or its receptors or produced these proteins at significantly lower levels. Subsequently we isolated primary astrocytic cell lines from the surgically removed tumors and found that the in vitro growth of the tumor cells was in a strong correlation with the production of GM-CSF and the expression of its receptor. Five cell lines (PG1, AS1, TV1, PR3, PL1) derived from the mostly-malignant tumors, were isolated after selection by adherence to plates coated with rhGM-CSF. They contained almost exclusively cells expressing glial specific markers and produced cell-associated GM-CSF that could be measured in their cell lysates. The two cell lines PG1 and AS1 (both isolated from a malignant glioblastoma multiforme), appeared to produce more GM-CSF than did the other lines and they secreted this protein into their culture-conditioned media. When these cultured cells were injected to athymic mice, they developed malignant gliomas whose growth rate closely correlated with their constitutive high production of GM-CSF and its receptor. These results therefore demonstrate that the baseline GM-CSF and GM-CSF.R production in these highly

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359

Fig. 7. GM-CSF and its receptor (both a- and b-chains) gene expression in xenografts of different gliomas. Glioma cells from tumor-derived primary mass cultures as well as from established lines co-expressing high level of both GM-CSF and its appropriate receptor genes developed bigger tumors in nude mice (A) than those not expressing these genes (B). Two or three mice were inoculated s.c. in the two flanks with a total of 5  106 cells per animal. The tumor volume was measured after 6 weeks. Values are shown as relative tumor volume evaluated by the equation RV = VX/V0 where VX is the tumor volume at day 42 and V0 is the palpable mass at the site of the inoculum detected at day 7.

tumorigenic cells is influenced by this constitutive activation (presumably by intrinsic oncogenic events). We also found that, the gene expression for both the growth factor and its receptor as well as other growth factors produced from the over-producing glioma cells isolated from the tumors, could be significantly increased after stimulation of the cultures with exogenous rhIL-1, and rhGM-CSF itself, revealing sensitivity to environmental stimulation. It is known that GM-CSF receptor internalizes after ligand binding [12,33,46]. Since we showed that malignant glioma cells produce more GM-CSF than benign glioma cells, it seems likely that the difference of membrane receptors observed in highly malignant cell lines as compared to benign glioma cell lines that express neither GM-CSF nor its receptor after immunostimulation, might be linked to the higher rate of GM-CSF secretion. It had been shown by others that by blocking endogenously produced GM-CSF with neutralizing anti-GM-CSF antibodies added to the culture medium, different types of GM-CSF-sensitive cells die via apoptosis [23–25]. The work presented here fully supports these previous studies, revealing that a likely important contribution of GM-CSF is to provide protection from apoptosis in GM-CSFsensitive glioma cells in an autocrine/paracrine fashion. Additionally, we confirmed that GM-CSF plays an important role in high grade glioma promotion and progression .When we tested variant clones of the AS1 line, with low- or highly-tumorigenic capacity, we showed that GM-CSF had the potential to contribute to the malignant behavior of the tumor cells in vitro either in the presence or absence of external stimulation. This suggests that in vivo not only environmental stimulation but also disease-related intrinsic activation, may offer signals for GM-CSF production and autocrine/paracrine stimulation in highly malignant gliomas. Our study also demonstrates that besides its influence on the survival and proliferation of the tumor cells

357

themselves, GM-CSF also plays a role as a stimulator of the production of other powerful pleiotropic growth factors and their receptors (i.e. NGF, G-CSF, VEGF). These receptors are directly derived from glioma cells as well as from endothelial and stromal cells, involved in the formation of new blood vessels [45,47,48,] and have potential functional consequences on tumor angiogenesis and tumor development. A lot of biochemical and genetic evidences suggest that multiple and complex signaling pathways are initiated and transmitted by a cascade of protein/protein interactions and signal transduction pathways activated by GM-CSF.R in response to GM-CSF stimulation. Despite the lack of tyrosine kinase domains within the cytoplasmic regions of the a and b subunits of its receptor, GM-CSF stimulation induces a rapid phosphorylation of various cellular proteins including its receptor subunits [48–52]. These findings suggest that an autocrine/paracrine loop is indeed activated by GM-CSF in the primary tumour of the patients, contributing to regulation of glioma growth and development, and this correlates with advanced tumor stage, excluding an adaptive response by the cells binding on plastic or to the manipulation done in cell culture experiments. Further studies are in progress in an attempt to elucidate the potential molecular mechanisms and the differential signaling networks involved in gliomagenesis. Our present findings thus demonstrate that GM-CSF, may act as a broad regulator of a spectrum of functions in different cell types bearing the appropriate functional receptors, serving as a pleiotropic molecular link controlling the environmental homeostasis, as previously suggested by others [33–37]. GM-CSF and G-CSF produced by malignant glioma cells may account for the marked leukocytosis observed in the peripheral blood as well as the massive infiltration of activated leukocytes (neutrophilic granulocytes, macrophages and lymphocytes) occurring without evidence of infection. The latter is often seen as a paraneoplastic sign in several malignant solid brain tumors, particularly in malignant glioblastoma multiforme specimens [53,54]. This paraneoplastic leukocytosis, with activated leukocytes producing likely abundant immunologically active cytokines, favors the malignant transformation of a benign glioma, stimulating tumour growth and its rapid progression, which is a characteristic behavior of this disease. Additionally, GM-CSF specifically binds to components of the extracellular matrix: the abnormally local high production of GM-CSF by glioma cells might serve as a mechanism for concentrating the growth factor and allowing high cell levels to be achieved in particular sites where it is produced. This constitutes a mechanism contributing to tumor progression and may be related to the frequent presence of multiple, widespread inflammatory foci and cystic and necrotic components common in this malignant disease. As a result, the differential diagnosis of primary brain tumour versus brain abscess is difficult to determine, particularly on the basis of radiological imaging and presenting clinical manifestations [53–55]. Despite the important progresses in improved neurosurgical management combined to radiological treatment and chemo -therapy, only modest improvement has been achieved in the survival of patients with malignant glioma. Current opinion is that more effective tumor-targeted therapies, and a more rational, biological-base approach, are required in order to induce regression and eradication of established intracranial malignant gliomas [56–66]. Recent research has actively focused on immunotherapeutical approaches for malignant gliomas reporting in some cases progress and potentials using rhGM-CSF (for example: immunotherapy with novel monoclonal antibodies or using engineered biomaterial-based vaccines to increase dendritic-cell antigen presentation and the number of activated tumoricidal T-cells; vaccination against growth factor producing tumor cells in combination with appropriate cytokines; intracellular effector molecules downstream of these factors; induced cytotoxicity by mesenchymal stem cells), showing some promising results [67–73]. However,

358

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359

the demonstration that the production of GM-CSF is an essential element of the tumorigenic process in gliomas of high grading strengthens the case for careful evaluation of the clinical application of this growth factor in patients with malignant gliomas. Acknowledgments We thank Mrs Eliana Vigneti for her generous and valuable collaboration. Ms. Julie Peters and Ms Laura Fatuzzo post-edited the English style. Supported in part by C.U.C.C.S. University of Pisa. References [1] Wong GG, Witek JS, Temple PA, et al. Human GM-CSF: molecular cloning of the complementary DNA and purification of the natural and recombinant proteins. Science 1985;228:810–5. [2] Lopez AF, Williamson DJ, Gamble JR, et al. Recombinant human granulocyte– macrophage colony-stimulating factor stimulates in vitro mature human neutrophil and eosinophil function, surface receptor expression, and survival. J Clin Invest 1986;78:1220–8. [3] Metcalf D, Nicola NA. Biological actions of the colony-stimulating factors in vitro, in: The Hematopoietic Colony-Stimulating Factors. Cambridge, UK: Cambridge University Press; 1995. [4] Metcalf D, Nicola NA. Clinical use of the colony-stimulating factors. In: The Hematopoietic Colony-Stimulating Factors. Cambridge, UK: Cambridge University Press; 1995. [5] Gasson JC, Kaufman SE, Weisbart RH, Tomonaga M, Golde DW. High-affinity binding of granulocyte–macrophage colony-stimulating factor to normal and leukemic human myeloid cells. Proc Natl Acad Sci USA 1986;83:669–73. [6] Gearing DP, King JA, Gough NM, Nicola NA. Expression cloning of a receptor for human granulocyte–macrophage colony-stimulating factor. EMBO J 1989;8:3667–76. [7] Hayashida K, Kitamura T, Gorman DM, Arai K, Yokota T, Miyajima A. Molecular cloning of a second subunit of the receptor for human granulocyte– macrophage colony-stimulating factor (GM-CSF): reconstitution of a high affinity GM-CSF receptor. Proc Natl Acad Sci USA 1990;87:9655–9. [8] Cannistra SA, Koenigsmann M, DiCarlo J, Groshek P, Griffin JD. Differentiationassociated expression of two functionally distinct classes of granulocyte– macrophage colony-stimulating factor receptors by human myeloid cells. J Biol Chem 1990;265:12656–63. [9] Isfort RJ, Ihle JN. Multiple hematopoietic growth factors signal through tyrosine phosphorylation. Growth Factors 1990;2:213–20. [10] Kanakura Y, Druker B, Cannistra SA, Furukawa Y, Torimoto Y, Griffin JD. Signal transduction of the human granulocyte–macrophage colony-stimulating factor and interleukin-3 receptors involves tyrosine phosphorylation of a common set of cytoplasmic proteins. Blood 1990;76:706–15. [11] Okuda K, Sanghera JS, Pelech SL, Kanakura Y, Hallek M, Griffin JD, et al. Granulocyte–macrophage colony-stimulating factor and steel factor induce rapid tyrosine phosphorylation of p42 and p44 MAP kinase. Blood 1992;79:2880–7. [12] Kastelein RA, Shanafelt AB. GM-CSF receptor: interactions and activation. Oncogene 1993;8:231–6. [13] American Society of Clinical Oncology. American Society of Clinical Oncology: recommendations for the use of the hematopoietic colony stimulating factors: evidence based clinical practice guidelines. J Clin Oncol 1994;12:2471–508. [14] Dedhar S, Gaboury L, Galloway P, Eaves C. Human granulocyte–macrophage colony-stimulating factor is a growth factor active on a variety of cell types of nonhemopoietic origin. Proc Natl Acad Sci USA 1988;85:9253–7. [15] Modrowski D, Lomri A, Mariel PJ. Endogenous GM-CSF is involved as an autocrine growth factor for human osteoblastic cells. J Cell Physiol 1997;170:35–46. [16] Hamilton JA, Piccoli DS, Cebon J, et al. Blood 1992;79:1413–9. [17] Bussolino F, Wang JM, Defilippi P, Turrini F, Sanavio F, Edgel CJ, et al. Nature 1989;337:471–3. [18] Inaba K, Steinman RM, Pack MW, Aya H, Inaba M, Sudo T, et al. J Exp Med 1992;175:1157–67. [19] Stoof TJ, Boorsma DM, Nickoloff BJ. Keratinocytes and immunological cytokines. In: I. Leigh, Watt F, editors. The Keratinocyte Handbook. Cambridge. University Press; 1994. p. 363–99. [20] Breuhahn K, Mann A, Muller G, et al. Epidermal overexpression of granulocyte–macrophage colony-stimulating factor induces both keratinocyte proliferation and apoptosis. Cell Growth Differ 2000;11:111–21. [21] Wegmann TG, Athanassakis I, Guilbert L, et al. The role of M-CSF and GM-CSF in fastening placental growth, fetal growth, and fetal survival. Transplant Proc 1989;21:566–71. [22] Harris RJ, Pettitt AR, Schmutz G, et al. Granulocyte–macrophage colonystimulating factor as an autocrine survival factor for mature and malignant Blymphocyte. J Immunol 2000;164:3887–93. [23] Lang SH, Miller WR, Duncan W, Habib FK. Production and response of human prostate cancer cell lines to granulocyte macrophage-colony stimulating factor. Int J Cancer 1994;59:235–41.

[24] Tachibana M, Miyakawa A, Tazaki H, et al. Autocrine growth of transitional cell carcinoma of the bladder induced by granulocyte-colony stimulating factor. Cancer Res 1995;55:3438–43. [25] Baldwin GC, Golde DW, Widhopf GF, Economou J, Gasson JC. Identification and characterization of a low-affinity granulocyte–macrophage colony-stimulating factor receptor on primary and cultured human melanoma cells. Blood 1991;78:609–15. [26] Baldwin GC, Gasson JC, Kaufman SE, et al. Nonhematopoietic tumor cells express functional GM-CSF receptors. Blood 1989;73:1033–7. [27] Baldwin GC. The biology of granulocyte–macrophage colony stimulating factor: effects on hematopoietic and non-hematopoietic cells. Dev Biol 1992;151:352–67. [28] Mueller MM, Fusenig NE. Constitutive expression of G-CSF and GM-CSF in human skin carcinoma cells with functional consequence for tumor progression. Int J Cancer 1999;83:780–9. [29] Miyagawa K, Chiba S, Shibuya K, et al. Frequent expression of receptors for granulocyte–macrophage colony-stimulating factor on human nonhematopoietic tumor cell lines. J Cell Physiol 1990;143:483–7. [30] Kleihues P, Burger PC, Scheithauer BW. The new WHO classification of brain tumors. Brain Pathol 1993;3:255–68. [31] Salcman M. High-grade glioma. In: Vecht J, editor. Handbook of Clinical Neurology. Neuro-oncology, vol. 24. Amsterdam Elsevier; 1997. p. 95–7. [32] Frei K, Piani D, Malipiero UV, Van Meir E, De Tribolet N, Fontana A. Granulocyte–macrophage colony-stimulating factor (GM-CSF) production by glioblastoma cells. J Immunol 1992;148:3140–6. [33] Sawamura MJ, Tada Y, Sakuma S, Sudo M, Aida T, Abe H. Human glioblastoma cells produce granulocyte–macrophage colony-stimulating factor in vitro, but not in vivo without expressing its receptor. Neurol Med Chem 1993;33:603–9. [34] Yamanaka R, Tanaka R, Saitoh T, Okoshi S. Cytokine gene expression on glioma cell lines and specimens. J Neurooncol 1994;21:243–7. [35] Nitta T, Sato K, Allegretta M, et al. Expression of granulocyte colony stimulating factor and granulocyte-macrophage colony stimulating factor genes in human astrocytoma cell lines and in glioma specimens. Brain Res 1992;571:19–25. [36] Tada M, Diserens AC, Desbaillets I, De Tribolet N. Analysis of cytokine receptor messenger RNA expression in human glioblastoma cells and normal astrocytes by reverse-transcription polymerase chain reaction. J Neurosurg 1994;80:1063–73. [37] Mueller MM, Herold-Mende CC, Riede D, Lange M, Steiner HH, Fusenig NE. Autocrine growth regulation by granulocyte colony-stimulating factor and granulocyte macrophage colony-stimulating factor in human gliomas with tumor progression. Am J Pathol 1999;155:1557–67. [38] Beffy P, Rovero P, Di Bartolo V, Laricchia Robbio L, Danè A, Pegoraro S, et al. An immunodominant epitope in a functional domain near the N-terminus of human granulocyte–macrophage colony-stimulating factor identified by cross-reaction of synthetic peptides with neutralizing anti-protein and antipeptide antibodies. Hybridoma 1994;3:457–68. [39] Laricchia-Robbio L, Liedberg B, Platou-Vikinge T, Rovero P, Beffy P, Revoltella PR. Mapping of monoclonal antibody- and receptor-binding domains on human granulocyte–macrophage colony-stimulating factor (rhGM-CSF) using a surface plasmon resonance-based biosensor. Hybridoma 1996;15:343–50. [40] Kitamura T, Tange T, Terasawa T, et al. IL-3 or erythropoietin. J Cell Physiol 1989;140:323–34. [41] Facchinetti A, Tessarollo L, Mazzocchi M, Kingston RT, Collavo D, Biasi G. An improved method for detection of DNA fragmentation. J Immunol Methods 1991;136:125–31. [42] Corcione A, Corrias MV, Daniele S, Zupo S, Spriano M, Pistoia V, et al. Blood 1996;87:2861–9. [43] Omerod MG. The study of apoptotic cells by cytometry. Leukemia 1998;12:1013–7. [44] Gilbert E, Migeon BR. D-Valine as a selective agent for normal human and rodent epithelial cells in culture. Cell 1975;5:11–7. [45] Dao Ke L, Shi YX, Im SA, Chen X, Yung WK. The relevance of cell proliferation, vascular endothelial growth factor, and basic fibroblast growth factor production to angiogenesis and tumorigenicity in human glioma cell lines. Clin Cancer. Res 2000;6:2562–72. [46] Chomczynski P, Sacchi N. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 1987;162:156–9. [47] Cannistra SA, Groshek P, Grlick R, Muiller J, Griffin JD. Regulation of surface expression of the GM-CSF receptors in normal human myeloid cells. Proc Natl Acad Sci USA 1990;87:93–7. [48] Herold-Mende C, Steiner HH, Andl T, Riede D, Buttler D, Reisser C, et al. Lab Invest 1999;79:1573–82. [49] Lilly MB, Zemskova M, Frankel AE, Salo J, Kraft AS. Distinct domains of the human granulocyte–macrophage colony-stimulating factor receptor alpha subunit mediate activation of Jak/Stat signaling and differentiation. Blood 2001;97:1662–70. [50] Sakurai Y, Ken-ichi A, Watanabe S. In vitro analysis of STAT5 activation by granulocyte–macrophage colony-stimulating factor. Genes to Cells 2000;5:937–47. [51] Kapoor GS, O’Rourke DM. Receptor tyrosine kinase signalling in gliomagenesis: pathobiology and therapeutic approaches. Cancer Biol Ther 2003;2:330–42. [52] Jensen RL. Growth factor-mediated angiogenesis in the malignant progression of glial tumors: a review. Surg Neurol 1998;49:1306–9.

R.P. Revoltella et al. / Cytokine 57 (2012) 347–359 [53] Hintzen RQ, Voormolen J, Sonneveld P, vanDuinen SG. Glioblastoma causing granulocytosis by secretion of granulocyte colony stimulating factor. Neurology 2000;54:259–61. [54] Salcman M. High-grade glioma. In: Vecht J, editor. Handbook of Clinical Neurology. Neuro-oncology, 24. Amsterdam Elsevier; 1997. p. 95–7. [55] Luminiczky K, Desaknai S, Mangel L, Szende B, Hamada H, Hidvegi EJ, et al. Local tumor irradiation augments the antitumor effect of cytokine-producing autologous cancer cell vaccines in a murine glioma model. Cancer Gene Ther 2002;9:44–52. [56] Fenstermaker RA, Ciesleski MJ. Immunotherapeutic strategies for malignant glioma. Cancer Control 2004;11(3):181–91. [57] Nieder CA, Grosu AL, Molls M. A comparison of treatment results for recurrent malignant gliomas. Cancer Treat Rev 2000;26:397–406. [58] Wheeler CJ, Das A, Liu G, Yu JS, Black L. Clinical responsiveness of glioblastoma multiforme to chemotherapy after vaccination. Clin Cancer Res 2004;10(16):5316–26. [59] Yu JS, Liu G, Ying H, Yong WH, Black KL, Wheeler CJ. Vaccination with tumor lysate-pulsed dendritic cells elicits antigen-specific cytotoxic T-cells in patients with malignant glioma. Cancer Res 2004;64:4973–9. [60] Rosenberg SA, Yang JC, Restivo NP. Cancer immunotherapy: moving beyond current vaccines. Nat Med 2004;10:909–15. [61] Ciesielski MJ, Latif Kazim A, Berth RF, Fanstermaker RA. Cellular antitumor immune response to a branched lysine multiple antigenic peptide containing epitopes of a common tumor-specific antigen in a rat glioma model. Cancer Immunol Immunother 2005;54:107–19. [62] Fine HA. Promising new therapies for malignant gliomas. Cancer J 2007;13:349–54. [63] Kang SG, Jeun SS, Lim JY, Kim SM, Yang YS, Oh WI, et al. Cytotoxicity of human umbilical cord-blood-derived mesenchymal stem cells against human malignant glioma cells. Childs Nerv Syst 2008;24:293–302.

359

[64] Mercer RW, Tyler MA, Ulasov IV, Lesniak MS. Targeted therapies for malignant gliomas: progress and potential. BioDrugs 2009;23:25–35. [65] Okada H, Koahanbash G, Zhu X, et al. Immunotherapeutic approaches for glioma. Crit Rev Immunol 2009;29:1–42. [66] Vauleon E, Avril T, Collet B, Mosser J, Quillien V. Overview of cellular immunotherapy for patients with glioblastoma. Clin Dev Immunol 2010:18 (Article ID 689171). [67] Herrlinger U, Kramm CM, Johnston KM, Louis DN, Finkelstein D, Reznikoff G, et al. Vaccination for experimental gliomas using GM-CSF – Transduced glioma cells. Cancer Gene Ther 1997;4(6):345–52. [68] Herrlinger U, Aulwurm S, Strik H, Weit S, Naumann U, Weller M. MIP-1 antagonizes the effect of GM-CSF-enhanced subcoutaneous vaccine in a mouse glioma model. J Neuro Oncol 2004;66:147–54. [69] Jean WC, Spellman SR, Wallenfriendman MA, et al. Effects of combined granulocyte–macrophage colony-stimulating factor (GM-CSF), interleukin-2, and interleukin-12 based immunotherapy against intracranial glioma in the rat. J Neuro Oncol 2004;66:39–49. [70] Tseng SH, Chen Y, Chang CY, et al. Clin Cancer Res 2005;11:1639–49. [71] Parmiani G, Castelli C, Pilla L, Santinami M, Colombo MP, Rivoltini L. Opposite immune functions of GM-CSF administered as vaccine adjuvant in cancer patients. Ann Oncol 2007;18(2):226–32. [72] Smith KE, Fritzell S, Badn W, Ebastal S, Janelidre S, Visse E. Cure of established GL261 mouse gliomas after combined immunotherapy with GM-CSF and IFNc is mediated by both CD8+ and CD4+ cells. Int J Cancer 2009;124:630–7. [73] Clavreul A, Piard N, Tanguy SY, Gamlin E, Rouselet MC, Leynia P, et al. Autologous tumor cell vaccination plus infusion of GM-CSF by a programmable pump in the treatment of recurrent malignant glioma. J Clin Neurosci 2010;17:842–8.