- Email: [email protected]
Glioma apoptosis induced by macrophages involves both death receptor-dependent and independent pathways
Glioma apoptosis induced by macrophages involves both death receptor– dependent and independent pathways GEORGE G. CHEN, ERNEST C. W. CHAK, YING S. CHUN, ISA K. Y. LAM, FANNY L. F. SIN, BILLY C. S. LEUNG, HO K. NG, and WAI S. POON HONG KONG
Apoptosis of glioma may represent a promising intervention for tumor treatment. Macrophages are able to induce apoptosis in a number of tumor cells, including glioma. It is known that apoptosis of cells is executed on either a death receptor– dependent or independent pathway. Whether and how apoptosis of glioma cells induced by activated macrophages is involved in these two pathways simultaneously are not known. Using in vitro and in vivo experimental models, we investigated Bcl-2 system and Fas/FasL channel, representing the death receptor– dependent and independent pathways, respectively, in glioma cells treated with the supernatant from the activated macrophages, which was rich in tumor necrosis factor-␣ and interferon-␥. We found that levels of Fas and FasL were up-regulated both in vitro and in vivo, accompanying an increase in the expression of caspase-8. The number of apoptotic cells was also increased significantly, although the percentage of death cells exceeded the number of tumor cells positive for Fas or FasL. It was also evident that the expression of Bax was increased, whereas the level of Bcl-2 was decreased, in glioma cells treated with the supernatant from the activated macrophages. The alteration of molecules related to both death pathways led to apoptosis of glioma and the inhibition of xenograft glioma growth in mice. Apoptosis of glioma induced by the activated macrophage is executed by way of both death receptor– dependent and independent pathways, and such an apoptosis-induced approach can effectively inhibit the growth of glioma in vivo. (J Lab Clin Med 2003;141:190-9) Abbreviations: DMEM ⫽ Dulbecco’s modified Eagle medium; ELISA ⫽ enzyme-linked immunosorbent assay; FCS ⫽ fetal calf serum; IETD-pNA ⫽ N-acetyl-Ile-Glu-Thr-Asp-p-nitroanilide; IFN ⫽ interferon; LPS ⫽ lipopolisaccharide; PBS ⫽ phosphate-buffered saline solution; pNA ⫽ p-nitroanilide; TNF ⫽ tumor necrosis factor
M
alignant glioma is the most common malignant primary brain tumor in human beings. Despite aggressive therapeutic strategies of surgical debulking, radiation, and chemotherapy, the median survival for patients afflicted with this malignancy remains just 9 months.1,2 Less than 1 year after
chemotherapy or irradiation, tumor recurs in most patients, and because of intrinsic drug resistance or acquired resistance, these patients respond poorly and only briefly to conventional therapy. Increasing evidence suggests that immunologic modulation and genetic approach are two logical choices to improve
From the Departments of Surgery and Anatomical and Cellular Pathology, the Sir Y. K. Pao Center for Cancer at Prince of Wales Hospital, and the Chinese University of Hong Kong. Supported by an RGC direct grant (2040833), Chinese University of Hong Kong. Submitted for publication April 22, 2002; revision submitted September 9, 2002; accepted November 13, 2002.
Reprint requests: George G. Chen, MD, PhD, Department of Surgery, Prince of Wales Hospital, Chinese University of Hong Kong, Shatin, NT, Hong Kong; e-mail: [email protected].
190
Copyright © 2003 by Mosby, Inc. All rights reserved. 0022-2143/2003 $30.00 ⫹ 0 doi:10.1067/mlc.2003.22
J Lab Clin Med Volume 141, Number 3
the current medications or to provide alternative therapy. Destruction of tumor cells by macrophages represents an important defensive system in human beings. Compared with systemic neoplasms, glioma has a high macrophage content,3 suggesting that a special channel exists to recruit and keep macrophages at the site of brain tumors. This makes macrophage-based immunotherapy of glioma particularly attractive. Macrophages have been recognized as potent tumor-cell killers for many years. As early as 1978, experiments by Reinhart et al4 showed that macrophages were more cytotoxic to tumor cells than to monocytes or lymphocytes and that they did not attack nonneoplastic cells such as fibroblasts and lymphocytes. Macrophage-tumoricidal function is a multiple-step process involving the activation of macrophages by a variety of stimulating agents. The cytotoxicity of macrophages to tumor cells can be accomplished through a direct contact or indirect killing or both. It is believed that the effectiveness of the tumoricidal pathway depends on the type of target and the channel of macrophage activation.5 Apoptosis is a normal, active, genetically controlled process of cell elimination. The induction of apoptosis has been shown to be an important determinant of the response of tumors to therapy. It can be induced by several different stimuli, including ligation of death receptors, cellular stresses, DNA damage, growth-factor withdrawal, and pathogen assault. Increasing evidence has accumulated to indicate that apoptosis occurs in glioma cells when these cells are treated with a variety of agents, including nitric oxide, ceramide, cycloheximide, cisplatin, proteasome inhibitors and interleukin-1.6-11 It is now known that whatever the stimuli are, they must go through death receptor– dependent or independent pathways.12,13 However, in no report have the questions of whether and how both pathways function simultaneously in apoptosis of glioma cells induced by macrophages. In this study we demonstrate that the apoptosis of glioma induced by activated macrophages involves both death receptor– dependent and independent pathways. METHODS Chemicals. Cell-culture medium, DMEM, and FCS were
purchased from Life Technologies (Grand Island, NY). LPS, collagenase, pronase, DNase, and mouse anti–rabbit IgG with fluorescein conjugate were from Sigma Chemical Co (St Louis, Mo). Anti–TNF-␣ (clone TN3-19.12) and anti–IFN-␥ (clone H22) were obtained from BD Pharmingen (San Diego, Calif). Percoll solution was from Amersham Pharmacia Biotech (Buckinghamshire, England). Mouse anti–rat ED1 antibody was obtained from Serotec Ltd (Oxford, UK), and rabbit anti–rat Fas, FasL, Bax, Bcl-2, Bcl-xL, and caspase-8 were
Chen et al
191
from Santa Cruz Biotechnology (Santa Cruz, Calif). Fluorescein-conjugated annexin V and propidium iodine were from Molecular Probes (Eugene, Ore). ABC reagent was from Vector Laboratories (Burlingame, Calif). The sources for various assay kits used are indicated elsewhere in this section. Animals. BALB/c nude mice (6-7 weeks) were supplied by the Laboratory Animal Services Center of the Chinese University of Hong Kong. The animals were housed under specific pathogen–free conditions in air-controlled rooms, which were specifically designed for maintenance of nude mice. Mice were fed with chow and sterile water ad libitum. The care and use of the animals was in compliance with institutional guidelines. Glioma cell line and culture. The F98 glioma cell line was derived from an undifferentiated neoplasm transplacentally induced by N-ethyl-N-nitosourea in an inbred CD Fischer rat and has been propagated in vitro and in vivo since 1971. Its structure and growth characteristics closely resemble those of human glioblastoma multiforme, forming a progressively growing tumor with islands of tumor cells at varying distances from the centrally growing mass and being weakly immunogenic.14,15 F98 cells were grown in DMEM with 10% FCS at 37°C in 5% CO2/95% air and saturated humidity. Isolation and culture of macrophages. It is extremely difficult to isolate macrophages from the central nervous system. However, it is relatively convenient to obtain macrophages from other sources. One rich pool of macrophages is liver; 30% of the hepatic nonparenchymal cell population is Kupffer cells. Kupffer cells constitute 80% to 90% of the body total macrophage mass while retaining the property of macrophages.16,17 Therefore macrophages from liver were used in this experiment. Kupffer cells were isolated from rats (250-320 g); the procedure for the isolation has been described in detail elsewhere.18,19 We identified and assessed the purity of macrophages using ED1 antibody staining, which recognized a single-chain glycoprotein expressed by resident and recruited rat hepatic macrophages.20 The purity of Kupffer cells was greater than 95%. Macrophage-conditional supernatant. Macrophages were activated by means of incubation with 100 ng/mL LPS. The culture was maintained in DMEM with 10% FCS at 37°C for 24 hours. At the end of the culture period, supernatants were collected, aliquoted, and stored at 0°C for future experiments. This supernatant, macrophage-conditional supernatant, was called test supernatant. The supernatant from the macrophage culture without LPS stimulation was used as a control and called control supernatant. Establishment of tumor-bearing mice and observation of tumor growth. Groups of 25 BALB/c nude mice were used
to develop tumor-bearing mice. Mice were divided into 5 groups, each comprising 5 mice (Fig 1). Mice were subcutaneously injected with 1 ⫻ 106 malignant glioma cells, which had not been treated (glioma only) or treated with control supernatant, test supernatant, anti–TNF-␣, or anti–IFN-␥. Tumor size was recorded on the sixth day of injection and was thereafter checked every 6 days. Experimental and control animals were killed after 36 days of tumor inoculation. The
192
Chen et al
J Lab Clin Med March 2003
Fig 1. Expression of Fas and FasL on glioma cells. Tumor cells were cultured in the presence of test supernatant (10%) for 2 days, after which the cells were stained with anti-Fas or anti-FasL antibody and the percentage of positive cells was determined on flow cytometry. In neutralization experiments, test supernatant was incubated with anti–TNF-␣ (2.5 g/mL) or anti–IFN-␥ (0.25 g/mL) for 2 hours before it was applied to tumor cells. We set up a control using an isotype antibody (IgG) in each experiment. *P ⬍ .05, n ⫽ 3, paired Student t test, vs tumor cells treated with control supernatant.
tumors were resected and the tissues immediately processed for immunohistochemical staining and protein isolation. Fas and FasL detection. The expression of Fas and FasL proteins was detected by means of indirect immunofluorescence labeling and flow-cytometric analysis. Cells resuspended in 100 L of PBS with 0.1% sodium azide were incubated with 1000 ng of anti-Fas or anti-FasL antibody at 4°C for 30 minutes. Cells were washed and recovered by means of centrifugation at 1000g for 5 minutes at 4°C before being resuspended in 20 L of mouse anti–rabbit IgG with fluorescein conjugate (20 g/mL); cells were then incubated in dark for 30 minutes at 4°C. Finally, cells were analyzed with a Becton Dickinson FACScan machine (BD Biosciences, San Jose, Calif) for cell-associated fluorescence. Caspase-8 activity assay. We measured the activity of caspase-8 using a caspase-8 assay kit from Clontech (Palo Alto, Calif). The assay is based on spectrophotometric detection of the chromophore pNA after cleavage from the labeled
substrate IETD-pNA.21 The pNA light emission can be quantified using a spectrophotometer at 405 nm. Comparison of the absorbance of pNA from an apoptotic sample with an uninduced control allows determination of the fold increase in caspase-8 activity. The result was expressed as nmol/50 g protein/hr. Immunohistochemical examination for Fas, FasL, Bax, Bcl-l, Bcl-xL, and Caspase-8. Tissues were subjected to
immunohistochemical processing and embedded in paraffin. Formalin fixation before embedding was of less than 30 hours’ duration throughout, which is important for preserving the antigenic determinants analyzed in this study. Tissues were sectioned at a thickness of 4 m. Immunostaining was performed on paraffin sections in accordance with the instructions of the ABC kit from Vector Laboratories. In brief, tissue sections were deparaffinized and rehydrated through three changes of xylene and graded alcohol. Tissue sections were then boiled in citrate-based antigen-unmasking solution for 1
J Lab Clin Med Volume 141, Number 3
Chen et al
193
Table I. Scoring standard for immunohistochemical staining Grade
Scoring by tissue
0 1 2
None Slight Moderate
3
Strong
4
Extremely strong
Scoring by cell
None Part of the cytoplasm Whole cytoplasm without reticular deposits Reticular deposits in less than half of cytoplasm Reticular deposits in more than half of cytoplasm
minute and cooled in Milli-Q water (Millipore, Billerica, Mass), the endogenous peroxidase activity in tissue sections was quenched with 3% hydrogen peroxide solution for 5 minutes. Normal blocking serum (1.5%) supplemented with avidin solution (avidin/biotin blocking kit; Vector Laboratories) was used to block tissue sections for 30 minutes. Afterward, preparations were incubated with a primary antibody overnight at 4°C. The primary antibody was prepared in 1.5% normal blocking serum supplemented with biotin solution from the Vector Laboratories avidin/biotin blocking kit and used at a working dilution of 1:200. After tissue sections were washed with PBS, a biotinylate-labeled secondary antibody, IgG, was applied for 30 minutes. Tissue sections were then washed with PBS, ABC reagent (avidin/biotin kit; Vector Laboratories) conjugated with horseradish peroxidase was applied for 30 minutes. Bid antigen staining was visualized with the use of NovaRED or DAB substrate (Vector Laboratories). We terminated the reaction by rinsing the sections in tap water. All incubations were performed in a humidified environment at room temperature, except where indicated. Finally, sections were counterstained with Vector Gill hematoxylene. After dehydration through graded alcohol and being cleared with xylene, they were mounted with DPX permanent mountant (Vector Laboratories). Negative controls were prepared by replacing the primary antibody with PBS. Immunoreactivity for antigens in the tissues was examined with the use of the Eclipse TE300 microscope (Nikon, Tokyo, Japan). The image was analyzed with the MetaMorph Imaging System (Universal Imaging Corp, Downington, Pa) and scored in accordance with the standard described in Table I. Western blotting for Fas, FasL, Bax, Bcl-l, Bcl-xL, and caspase-8. Tissue samples were homogenized with ice-cold
PBS and then subjected to lysis in a solution containing 8 mol/L urea, 0.1 mol/L Na2H2PO4, and 0.01 mol/L Tris-HCl. Supernatants were obtained after centrifugation at 10,000g. After boiling, proteins were separated on 10% sodium dodecyl sulfate–polyacrylamide gels. Proteins were then electrophoretically transferred from the gel onto nitrocellulose membranes, and the membranes were blocked for 1 hour in PBSTween buffer containing 5% dry milk powder (fat-free) at room temperature. The membranes were then incubated with a primary antibody for 1 hour. After washing, the membranes were incubated with a secondary antibody, IgG-HRP. Finally they were treated with the reagents in the chemilumines-
Fig 2. Activity of caspase-8. Glioma cells were cultured in the presence of test supernatant (10%) for 2 days, after which the cell lysate was isolated for the measurement of caspase-8 activity. *P ⬍ .05, n ⫽ 6, paired Student t test, vs tumor cells treated with control supernatant.
cence-detection kit (ECL system; Amersham Pharmacia Biotech, Piscataway, NJ) in accordance with the manufacturer’s instructions. The densities of the protein bands were determined and analyzed with the Quantity One program (version 4.2; Bio-Rad Laboratories, Hercules, Calif). Anti-actin antibody (Santa Cruz Biotechnology, Santa Cruz, Calif) was used to detect actin, which was used as a control for equal loading. Apoptosis detection. Apoptosis was detected with the use of both flow cytometry and in situ detection. For flow cytometry, cells were washed in cold PBS and incubated with fluorescein-conjugated annexin V and propidium iodine. Flow cytometry of annexin V and propidium iodine-stained cells was performed on a Becton Dickinson FACScan machine; results were analyzed with CellQuest software (BD Biosciences). For in situ detection of apoptotic cells, the DeadEnd apoptosis detection kit (TUNEL assay) from Promega (Madison, Wis) was used to conduct in situ labeling of the 3⬘-end of the DNA fragments generated by apoptosisassociated endonucleases. In brief, the sections, after being dewaxed in xylene and rehydrated in ethanol, were incubated with 20 g/mL proteinase K at room temperature for 15 minutes. The slides were then incubated with a terminal transferase enzyme and biotinylated nucleotide mix in 37°C for 60 minutes to allow the end-labeling reaction to occur. We blocked the endogenous peroxidase activity by treating the slides with 0.3% hydrogen peroxide in PBS, pH 7.2. Horseradish-peroxidase–labeled streptavidin solution was then applied to the slides, which were incubated for 30 minutes at room temperature. After incubation, the color was developed with the peroxidase substrate, hydrogen peroxide, and the stable chromogen diaminobenzidine. The slides were then mounted and examined under a light microscope. Cells were defined as apoptotic if the whole nuclear area of the cell labeled positively. Apoptotic bodies were defined as small
194
J Lab Clin Med March 2003
Chen et al
Fig 3. Expression of death receptor– dependent and independent proteins in tumor tissues. The expression of Fas, FasL, caspase-8, Bax, Bcl-2, and Bcl-xL proteins in tumor tissues was examined with the use of immunohistochemical staining (original magnification ⫻400). A and B, Fas. C and D, FasL. E and F, caspase-8. G and H: Bax. I and J, Bcl-2. K and L, Bcl-xL. A, C, E, G, I, and K represent tissue samples obtained from tumors treated with control supernatant; B, D, F, H, J, and L represent tissue samples obtained from tumors treated with test supernatant. Bar ⫽ 18 mol/L.
positively labeled globular bodies in the cytoplasm of the cells, which were detected either singly or in groups. To estimate the apoptotic index that represented the percentage of apoptotic cells in a given area, we counted apoptotic cells and bodies in 10 high-power fields, then divided this figure by the number of cells in the same high-power fields. TNF-␣ and IFN-␥ determination. The levels of TNF-␣ and IFN- released from cultured cells into the medium were determined with the use of ELISA kits from Chemicon (Temecula, Calif) in accordance with the manufacturer’s instructions. In brief, samples were added to the assay plate, which was precoated with an antibody to capture TNF-␣ or IFN-␥. The biotinylated TNF-␣ or IFN-␥ conjugate and sample/ standard compete for TNF-␣ or IFN-␥, and thus the amount of TNF-␣ or IFN-␥ in sample/standard could be determined. The assay was visualized with the use of a streptavidin– alkaline phosphatase conjugate and an ensuring chromagenic substrate reaction.
Statistical analysis. All values are expressed as mean ⫾ SD. We analyzed statistical comparisons with the Student t or Mann-Whitney U test, using InStat software (GraphPad Software, San Diego, Calif). A P value of less than .05 was taken as statistically significant.
RESULTS Expression of death receptor– dependent proteins. Fas was the best-characterized death receptor. We first examined the expression of Fas and its ligand FasL in glioma cells. The glioma cells were stimulated with test supernatant from the activated macrophages. Fas and FasL were recognized by their antibodies, and the positive signals were detected on flow cytometry. We found that concentrations of both Fas and FasL were much higher in glioma cells with test supernatant than
J Lab Clin Med Volume 141, Number 3
Chen et al
195
Fig 3. Continued
in those without (Fig 1). The effect of test supernatant on Fas expression was significantly blocked by the application of a TNF-␣– or IFN-␥–neutralizing antibody (Fig 1). However, the high level of FasL expression could be inhibited only by the TNF-␣–neutralizing antibody, not by the IFN-␥ neutralizing antibody. The activity of caspase-8, which is an upstream caspase and a corridor link between death receptor– dependent and independent pathways, was also markedly increased in glioma cells with test supernatant compared with those without (Fig 2). The in vivo expression of Fas, FasL, and caspase 8 was analyzed in glioma cells xenografted into nude mice. The glioma cells were treated with test supernatant before transplantation. Immunohistochemical data showed that the levels of Fas (3.62 ⫾ 0.87 vs 2.18 ⫾ 0.60, P ⬍ .01), FasL (2.11 ⫾ 0.63 vs 1.08 ⫾ 0.46, P ⬍ 0.05), and caspase-8 (3.18 ⫾ 0.69 vs 1.89 ⫾ 0.52, P ⬍ 0.01) were all increased in glioma cells treated with test supernatant compared with those treated with control medium (Fig 3, A-F). A similar pattern was obtained on Western-blot analy-
sis (Fig 4). Data similar to those obtained with test supernatant and control supernatant were obtained in the samples treated with activated macrophages directly and those without treatment, respectively, on Western-blot or immunochemical analysis (data not shown). Expression of death receptor–independent proteins.
Bcl-2 family members are the main player in the death receptor–independent pathway. The level of Bax protein was increased in glioma cells treated with test supernatant, as demonstrated on immunohistochemical staining (3.37 ⫾ 0.99 vs 2.01 ⫾ 0.68, P ⬍ .01) or Western-blot analysis (Fig 3, G and H, and Fig 4). On the contrary, the expression of Bcl-2 protein was significantly decreased in glioma cells treated with test supernatant (1.83 ⫾ 0.51 vs 2.89 ⫾ 0.52, P ⬍ .05) compared with those without treatment or treated with control supernatant (Fig 3, I and J, and Fig 4). We did not find a significant difference in the level of Bcl-xL between glioma cells treated with test supernatant and those treated with control supernatant (1.67 ⫾ 0.34 vs 1.83 ⫾ 0.47, P ⬎ .05) (Fig 3, K and L, and Fig 4).
196
J Lab Clin Med March 2003
Chen et al
Fig 4. Western-blot analysis of death receptor– dependent and independent proteins. Proteins were isolated from tumors formed by glioma cells treated with test supernatant or with control supernatant or without treatment. Anti-Fas, caspase-8, Bax, Bcl-2, and Bcl-xL antibodies were used to detect their protein levels on Western-blot analysis. Lane 1, tissue samples obtained from tumors treated with test supernatant. Lane 2, samples from tumors treated with control supernatant. Lane 3, samples from tumors without treatment.
Fig 5. Apoptosis of glioma cells. Glioma cells were cultured in the presence of test supernatant (10%) for 2 days, after which the cells were collected and stained with annexin V and propidium iodine. The percentage of positive cells was analyzed on flow cytometry. *P ⬍ .05, n ⫽ 3, paired Student t test, vs tumor cells treated with test supernatant.
Similar expression patterns in test supernatnat and control supernatant were obtained in the samples treated with activated macrophages directly and the samples without treatment, respectively, on either Western-blot or immunochemical analysis (date not shown). Apoptosis of glioma cells. The percentage of apoptotic glioma cells treated with test supernatant was significantly higher (21.5%) than that in the cells without treatment or treated with control supernatant, as demonstrated by annexin V and propidium iodine staining with the aid of flow cytometry (Fig 5). In situ detection
also indicated that the number of apoptotic glioma cells was markedly increased in the cells treated with test supernatant, compared with the cells without treatment or treated with control supernatant (Fig 6). We set up a positive control for in situ apoptotic detection in which apoptosis was induced by DNase I that was known to cause fragmentation of the chromosomal DNA. A significant number of cells became apoptotic in the positive-control tissue treated with DNase I (Fig 6, C). LPS itself did not show any apopototic effect on glioma, and therefore the direct effect of LPS on glioma was ruled out. In vivo tumor growth. The growth of glioma in vivo was significantly slower in the tumor cells receiving test supernatant or macrophage treatment directly than in those without treatment or treated with control supernatant (Fig 7). We ruled out the growth of macrophages in vivo because we detected no signs of growth in the mice injected with macrophages only. The glioma tumor in the mice without treatment and those treated with control supernatant was significantly larger than those with treatment after just 6 days’ transplantation of the tumor cells. By the end of experiment (36 days), the size of the glioma tumors was about five times different between those receiving treatment and those without. The rate of tumor growth was not significant different between those treated with test supernatant and those with macrophages directly, although the tumor was slightly larger in the former than in the latter. TNF-␣ and IFN-␥ levels. Both TNF-␣ and IFN-␥ levels were significantly higher in test supernatant (11.25 ⫾ 1.02 ng/mL and 0.94 ⫾ 0.03 pg/mL, respectively) than in the control supernatant (8.61 ⫾ 0.43 ng/mL and 0.54 ⫾ 0.02 pg/mL, respectively; both P ⬍ .05, paired Student t test, n ⫽ 5). Glioma cells alone did not respond significantly to LPS stimulation for TNF-␣ and IFN-␥ production (data not shown). DISCUSSION
Macrophages require a series of extracellular signals to become activated. The activated macrophages will undergo a set of phenotypic changes, including cytokine production and tumoricidal activity.22 In this study, we showed that macrophages could be activated by LPS, a glycolipid that constitutes the major portion of the outermost membrane of gram-negative bacteria,23 and that activated macrophages were able to generate a significant level of TNF-␣ and IFN-␥. This finding is in line with previous evidence indicating that activated macrophages synthesize and secrete a wide variety of lytic or cytotoxic molecules that are capable of destroying tumor cells.22,24,25 However, how macrophage-generated molecules exert their tumoricidal
J Lab Clin Med Volume 141, Number 3
Chen et al
197
Fig 6. In situ detection of apoptosis in glioma tissues. Apoptotic cells stained brown (original magnification ⫻ 400). We detected very few apoptotic cells in glioma tissues treated with control supernatant (A), whereas of apoptosis was much more frequent in glioma tissues treated with test supernatant (B). Tissue treated with DNase I was used as a positive control (C). DNase I treatment is known to result in fragmentation of the chromosomal DNA.
Fig 7. Growth of the xenograft glioma in nude mice. BALB/c nude mice were subcutaneously injected with 1 ⫻ 106 glioma cells that were subjected to different treatments. The growth of glioma in tumor-bearing mice was monitored along the course of experiment (36 days). Lump size was recorded every 6 days and expressed as square millimeters. Each group comprised 5 mice. Statistical comparisons between groups are described in the text. P ⬍ .05, n ⫽ 5, vs glioma only or glioma ⫹ control supernatant; P ⬍ .01, n ⫽ 5, vs glioma only or glioma ⫹ control supernatant.
effects is not yet completely understood, especially in the situation of interaction between macrophages and brain tumors. Our results indicate that the activated macrophages are able to induce apoptosis in a significant number of glioma cells and inhibit the growth of glioma through the production of the tumoricidal molecules, probably mainly TNF-␣ and IFN-␥. Both TNF-␣ and IFN-␥ are well known for their proapoptotic functions by way of a death receptor– dependent pathway.26,27 In this study, we demonstrated that the supernatant (rich in these two cytokines) from the activated macrophages could significantly stimulate the expression of Fas and FasL in the glioma cells. Fas and FasL are the most important
death-receptor system identified so far. It has been documented that glioma cells express functional Fas and FasL that induce apoptosis in tumor cells.8,28 Ligation of Fas by FasL recruits the adaptor molecule Fas-associating death domain (FADD) to the receptor by way of mutual interaction of their death domains. FADD engages procaspase-8, and subsequently procaspase-8 is autoproteolytically activated. Activated caspase-8 serves as an enzyme for downstream effector caspases including 3, 6, and 7. The activated effector caspase, in turn, cleaves a number of cytoplasmic and nuclear substrates, degrades chromosomal DNA, and induces cell death. The evidence to support the involvement of the death receptor– dependent path-
198
J Lab Clin Med March 2003
Chen et al
way also came from the alteration of caspase-8 in both in vitro and in vivo experiments. The activity of caspase-8 was significantly increased in glioma cells treated with the supernatants derived from the activated macrophages, and the level of this protein was overexpressed in xenograft glioma formed from tumor cells treated with the molecules generated by the activated macrophages. Without amplification by caspase-8, it is not possible for cells to undergo the death receptor– dependent pathway. Another finding of this study was that the percentage of glioma cells positive for Fas and FasL was much lower than the percentage of tumor cells with signs of apoptosis, suggesting that multiple channels are involved in induction of apoptosis of glioma cells when the cells were treated with the supernatant from the activated macrophages. After analysis of the Bcl-2 system, we found that the expression of Bax was significantly increased, whereas the level of Bcl-2, a conversed homologue of Bax, was much lower in xenograft glioma formed from tumor cells treated with the molecules generated by the activated macrophages compared with that treated with control medium or the supernatant from inactivated macrophages. Bax is well known as a proapoptotic member of Bcl-2 family, whereas Bcl-2 is an antiapoptotic agent. Both Bax and Bcl-2 play an important role in the regulation of apoptosis by way of the death receptor–independent pathway. On being stimulated, Bax is translocated to the mitochondria and forms a heterodimer with Bcl-2, which results in the release of cytochrome c. Cytochrome c will function to induce activation of caspases 9 and 3.9 It has not been reported how Bcl-2 family members behave in glioma when the tumor cells encounter macrophages. However, it has been demonstrated that alteration of Bax and Bcl-2 is associated with ceramide-induced apoptosis of glioma cells.9 Macrophage-related cytokines such as TNF-␣ and IFN-␥ are known to affect the expression of Bcl-2 family members.29,30 It is reported that a cross-talk between the death receptor– dependent and independent pathways occurs in certain types of cells by way of caspase-8 and proapoptotic Bid, a BH3-domain Bcl-2 family member.31 Although this is likely the case in the model we tested, this remains to be proved. The xenograft glioma experiment revealed that the growth of glioma was significantly slower and the size of the tumor much smaller in the transplantation treated with the supernatant of the activated macrophages. In situ apoptosis analysis indicated that the frequency of cell death was high in the macrophage-treated glioma tissue, along with the altered expression of apoptosisrelated proteins of both death receptor– dependent and independent pathways. Obviously the apoptosis in-
duced by way of both the death receptor– dependent and independent pathways is, at least in part if not completely, responsible for inhibition of tumor growth. On the basis of whether Bcl-2 system is involved in the apoptotic pathway, apoptotic cells can be classified as type I or type II cells.32 In type I cells, activated caspase-8 directly triggers the cascade of effector caspases without the involvement of mitochondria and Bcl-2 family members. However, in addition to Fas and caspase-8, apoptosis of type II cells involves mitochondria and Bcl-2 family members. According to this concept, the glioma cells we tested here are of type II. However, whether a subgroup of tumor cells functions as type I cells cannot be ruled out. In fact, a study by Knight et al indicated that glioma cells were heterogeneous and behaved in both type I and type II manners.28 Identification of the death receptor– dependent and independent pathways in glioma apoptosis induced by activated macrophages may provide some useful information for designing an efficient intervention against tumor growth. Understanding the contribution of individual molecules in both pathways following the interaction between macrophages and glioma is not only a fascinating scientific puzzle; it may aid in the design of pharmacologic strategies to interfere more specifically with the aim of enhancing apoptosis, thereby avoiding the troublesome unwanted side effects of immunotherapy for cancer. REFERENCES
1. Leibel SA, Scott CB, Loeffler JS. Contemporary approaches to the treatment of malignant gliomas with radiation therapy. Semin Oncol 1994;21:198-219. 2. Lesser GJ, Grossman S. The chemotherapy of high-grade astrocytomas. Semin Oncol 1994;21:220-35. 3. Mahaley MS Jr. Neuro-oncology index and review (adult primary brain tumors): radiotherapy, chemotherapy, immunotherapy, photodynamic therapy. J Neurooncol 1991;11:85-147. 4. Rinehart JJ, Vessella R, Lange P, Kaplan ME, Gormus BJ. Characterization and comparison of human monocyte- and macrophage-induced tumor cell cytotoxicity. J Lab Clin Med 1979;93: 361-9. 5. Keller R, Keist R, Wechsler A, Leist TP, van der Meide PH. Mechanisms of macrophage-mediated tumor cell killing: a comparative analysis of the roles of reactive nitrogen intermediates and tumor necrosis factor. Int J Cancer 1990;46:682-6. 6. Shinoda J, Whittle IR. Nitric oxide and glioma: a target for novel therapy? Br J Neurosurg 2001;15:213-20. 7. Duan L, Aoyagi M, Tamaki M, Nakagawa K, Nagashima G, Nagasaka Y, et al. Sensitization of human malignant glioma cell lines to tumor necrosis factor–induced apoptosis by cisplatin. J Neurooncol 2001;52:23-36. 8. Glaser T, Wagenknecht B, Weller M. Identification of p21 as a target of cycloheximide-mediated facilitation of CD95-mediated apoptosis in human malignant glioma cells. Oncogene 2001;20: 4757-67. 9. Sawada M, Nakashima S, Banno Y, Yamakawa H, Hayashi K, Takenaka K, et al. Ordering of ceramide formation, caspase
J Lab Clin Med Volume 141, Number 3
10.
11.
12. 13. 14.
15.
16.
17.
18. 19. 20.
21.
activation, and Bax/Bcl-2 expression during etoposide-induced apoptosis in C6 glioma cells. Cell Death Differ 2000;7:761-72. Tani E, Kitagawa H, Ikemoto H, Matsumoto T. Proteasome inhibitors induce Fas-mediated apoptosis by c-Myc accumulation and subsequent induction of FasL message in human glioma cells. FEBS Lett 2001;504:53-8. Kondo S, Barna BP, Morimura T, Takeuchi J, Yuan J, Akbasak A, et al. Interleukin-1 beta-converting enzyme mediates cisplatin-induced apoptosis in malignant glioma cells. Cancer Res 1995;55:6166-71. Ashkenazi A, Dixit VM. Death receptors: signaling and modulation. Science 1998;281:1305-8. Gupta S. Molecular steps of cell suicide: an insight into immune senescence. J Clin Immunol 2000;20:229-39. Tzeng JJ, Barth RF, Orosz CG. Quantitation of glioma-reactive cytolytic T lymphocyte precursors by means of limiting dilution analysis. J Immunol Methods 1992;146:177-84. Tzeng JJ, Barth RF, Orosz CG, James SM. Phenotype and functional activity of tumor-infiltrating lymphocytes isolated from immunogenic and nonimmunogenic rat brain tumors. Cancer Res 1991;51:2373-8. Bouwens L, Baekeland M, Wisse E. Cytokinetic analysis of the expanding Kupffer-cell population in rat liver. Cell Tissue Kinetics 1986;19:217-26. Jones EA, Summerfield JA. Kupffer Cells. In: Arias IM, Jakoby WB, Popper H, et al, eds. The liver: biology and pathobiology, ed 2. New York: Raven Press, 1988:683-704. Olynyk JK, Clarke SL. Isolation and primary culture of rat Kupffer cells. J Gastroenterol Hepatol 1998;13:842-5. Alpini G, Phillips JO, Vroman B, LaRusso NF. Recent advances in the isolation of liver cells. Hepatology 1994;20:494-514. Dijkstra CD, Dopp EA, Joling P, Kraal G. The heterogeneity of mononuclear phagocytes in lymphoid organs: distinct macrophage subpopulations in the rat recognized by monoclonal antibodies ED1, ED2 and ED3. Immunology 1985;54:589-99. Casciola-Rosen L, Nicholson DW, Chong T, Rowan KR, Thornberry NA, Miller DK, et al. Apopain/CPP32 cleaves proteins that are essential for cellular repair: a fundamental principle of apoptotic death. J Exp Med 1996;183:1957-64.
Chen et al
199
22. Askew D, Yurochko AD, Burger CJ, Elgert KD. Normal and tumor-bearing host macrophage responses: variability in accessory function, surface markers, and cell-cycle kinetics. Immunol Lett 1990;24:21-9. 23. Raetz CR, Ulevitch RJ, Wright SD, Sibley CH, Ding A, Nathan CF. Gram-negative endotoxin: an extraordinary lipid with profound effects on eukaryotic signal transduction. FASEB J 1991; 5:2652-60. 24. Shimoda O, Takeda Y, Woo HJ, Shimada S, Higuchi M, Osawa T. A human macrophage hybridoma producing a cytotoxic factor distinct from TNF, LT, and IL-1. Cancer Immunol Immunother 1988;6:101-8. 25. Erickson KL, Hubbard NE. Dietary fish oil modulation of macrophage tumoricidal activity. Nutrition 1996;12(suppl 1):S34-8. 26. Naujokat C, Sezer O, Possinger K. Tumor necrosis factor-alpha and interferon-gamma induce expression of functional Fas ligand on HT29 and MCF7 adenocarcinoma cells. Biochem Biophys Res Commun 1999;264:813-9. 27. Choi C, Park JY, Lee J, Lim JH, Shin EC, Ahn YS, et al. Fas ligand and Fas are expressed constitutively in human astrocytes and the expression increases with IL-1, IL-6, TNF-alpha, or IFN-gamma. J Immunol 1999;162:1889-95. 28. Knight MJ, Riffkin CD, Muscat AM, Ashley DM, Hawkins CJ. Analysis of FasL and TRAIL induced apoptosis pathways in glioma cells. Oncogene 2001;20:5789-98. 29. Saldeen J. Cytokines induce both necrosis and apoptosis via a common Bcl-2-inhibitable pathway in rat insulin-producing cells. Endocrinology 2000;141:2003-10. 30. Boldrini L, Calcinai A, Samaritani E, Pistolesi F, Mussi A, Lucchi M, et al. Tumour necrosis factor-alpha and transforming growth factor-beta are significantly associated with better prognosis in non-small cell lung carcinoma: putative relation with BCL-2-mediated neovascularization. Br J Cancer 2000;83:480-6. 31. Zimmermann KC, Bonzon C, Green DR. The machinery of programmed cell death. Pharmacol Ther 2001;92:57-70. 32. Scaffidi C, Fulda S, Srinivasan A, Friesen C, Li F, Tomaselli KJ, et al. Two CD95 (APO-1/Fas) signaling pathways. EMBO J 1998;17:1675-87.
Apoptosis of glioma may represent a promising intervention for tumor treatment. Macrophages are able to induce apoptosis in a number of tumor cells, including glioma. It is known that apoptosis of cells is executed on either a death receptor– dependent or independent pathway. Whether and how apoptosis of glioma cells induced by activated macrophages is involved in these two pathways simultaneously are not known. Using in vitro and in vivo experimental models, we investigated Bcl-2 system and Fas/FasL channel, representing the death receptor– dependent and independent pathways, respectively, in glioma cells treated with the supernatant from the activated macrophages, which was rich in tumor necrosis factor-␣ and interferon-␥. We found that levels of Fas and FasL were up-regulated both in vitro and in vivo, accompanying an increase in the expression of caspase-8. The number of apoptotic cells was also increased significantly, although the percentage of death cells exceeded the number of tumor cells positive for Fas or FasL. It was also evident that the expression of Bax was increased, whereas the level of Bcl-2 was decreased, in glioma cells treated with the supernatant from the activated macrophages. The alteration of molecules related to both death pathways led to apoptosis of glioma and the inhibition of xenograft glioma growth in mice. Apoptosis of glioma induced by the activated macrophage is executed by way of both death receptor– dependent and independent pathways, and such an apoptosis-induced approach can effectively inhibit the growth of glioma in vivo. (J Lab Clin Med 2003;141:190-9) Abbreviations: DMEM ⫽ Dulbecco’s modified Eagle medium; ELISA ⫽ enzyme-linked immunosorbent assay; FCS ⫽ fetal calf serum; IETD-pNA ⫽ N-acetyl-Ile-Glu-Thr-Asp-p-nitroanilide; IFN ⫽ interferon; LPS ⫽ lipopolisaccharide; PBS ⫽ phosphate-buffered saline solution; pNA ⫽ p-nitroanilide; TNF ⫽ tumor necrosis factor
M
alignant glioma is the most common malignant primary brain tumor in human beings. Despite aggressive therapeutic strategies of surgical debulking, radiation, and chemotherapy, the median survival for patients afflicted with this malignancy remains just 9 months.1,2 Less than 1 year after
chemotherapy or irradiation, tumor recurs in most patients, and because of intrinsic drug resistance or acquired resistance, these patients respond poorly and only briefly to conventional therapy. Increasing evidence suggests that immunologic modulation and genetic approach are two logical choices to improve
From the Departments of Surgery and Anatomical and Cellular Pathology, the Sir Y. K. Pao Center for Cancer at Prince of Wales Hospital, and the Chinese University of Hong Kong. Supported by an RGC direct grant (2040833), Chinese University of Hong Kong. Submitted for publication April 22, 2002; revision submitted September 9, 2002; accepted November 13, 2002.
Reprint requests: George G. Chen, MD, PhD, Department of Surgery, Prince of Wales Hospital, Chinese University of Hong Kong, Shatin, NT, Hong Kong; e-mail: [email protected].
190
Copyright © 2003 by Mosby, Inc. All rights reserved. 0022-2143/2003 $30.00 ⫹ 0 doi:10.1067/mlc.2003.22
J Lab Clin Med Volume 141, Number 3
the current medications or to provide alternative therapy. Destruction of tumor cells by macrophages represents an important defensive system in human beings. Compared with systemic neoplasms, glioma has a high macrophage content,3 suggesting that a special channel exists to recruit and keep macrophages at the site of brain tumors. This makes macrophage-based immunotherapy of glioma particularly attractive. Macrophages have been recognized as potent tumor-cell killers for many years. As early as 1978, experiments by Reinhart et al4 showed that macrophages were more cytotoxic to tumor cells than to monocytes or lymphocytes and that they did not attack nonneoplastic cells such as fibroblasts and lymphocytes. Macrophage-tumoricidal function is a multiple-step process involving the activation of macrophages by a variety of stimulating agents. The cytotoxicity of macrophages to tumor cells can be accomplished through a direct contact or indirect killing or both. It is believed that the effectiveness of the tumoricidal pathway depends on the type of target and the channel of macrophage activation.5 Apoptosis is a normal, active, genetically controlled process of cell elimination. The induction of apoptosis has been shown to be an important determinant of the response of tumors to therapy. It can be induced by several different stimuli, including ligation of death receptors, cellular stresses, DNA damage, growth-factor withdrawal, and pathogen assault. Increasing evidence has accumulated to indicate that apoptosis occurs in glioma cells when these cells are treated with a variety of agents, including nitric oxide, ceramide, cycloheximide, cisplatin, proteasome inhibitors and interleukin-1.6-11 It is now known that whatever the stimuli are, they must go through death receptor– dependent or independent pathways.12,13 However, in no report have the questions of whether and how both pathways function simultaneously in apoptosis of glioma cells induced by macrophages. In this study we demonstrate that the apoptosis of glioma induced by activated macrophages involves both death receptor– dependent and independent pathways. METHODS Chemicals. Cell-culture medium, DMEM, and FCS were
purchased from Life Technologies (Grand Island, NY). LPS, collagenase, pronase, DNase, and mouse anti–rabbit IgG with fluorescein conjugate were from Sigma Chemical Co (St Louis, Mo). Anti–TNF-␣ (clone TN3-19.12) and anti–IFN-␥ (clone H22) were obtained from BD Pharmingen (San Diego, Calif). Percoll solution was from Amersham Pharmacia Biotech (Buckinghamshire, England). Mouse anti–rat ED1 antibody was obtained from Serotec Ltd (Oxford, UK), and rabbit anti–rat Fas, FasL, Bax, Bcl-2, Bcl-xL, and caspase-8 were
Chen et al
191
from Santa Cruz Biotechnology (Santa Cruz, Calif). Fluorescein-conjugated annexin V and propidium iodine were from Molecular Probes (Eugene, Ore). ABC reagent was from Vector Laboratories (Burlingame, Calif). The sources for various assay kits used are indicated elsewhere in this section. Animals. BALB/c nude mice (6-7 weeks) were supplied by the Laboratory Animal Services Center of the Chinese University of Hong Kong. The animals were housed under specific pathogen–free conditions in air-controlled rooms, which were specifically designed for maintenance of nude mice. Mice were fed with chow and sterile water ad libitum. The care and use of the animals was in compliance with institutional guidelines. Glioma cell line and culture. The F98 glioma cell line was derived from an undifferentiated neoplasm transplacentally induced by N-ethyl-N-nitosourea in an inbred CD Fischer rat and has been propagated in vitro and in vivo since 1971. Its structure and growth characteristics closely resemble those of human glioblastoma multiforme, forming a progressively growing tumor with islands of tumor cells at varying distances from the centrally growing mass and being weakly immunogenic.14,15 F98 cells were grown in DMEM with 10% FCS at 37°C in 5% CO2/95% air and saturated humidity. Isolation and culture of macrophages. It is extremely difficult to isolate macrophages from the central nervous system. However, it is relatively convenient to obtain macrophages from other sources. One rich pool of macrophages is liver; 30% of the hepatic nonparenchymal cell population is Kupffer cells. Kupffer cells constitute 80% to 90% of the body total macrophage mass while retaining the property of macrophages.16,17 Therefore macrophages from liver were used in this experiment. Kupffer cells were isolated from rats (250-320 g); the procedure for the isolation has been described in detail elsewhere.18,19 We identified and assessed the purity of macrophages using ED1 antibody staining, which recognized a single-chain glycoprotein expressed by resident and recruited rat hepatic macrophages.20 The purity of Kupffer cells was greater than 95%. Macrophage-conditional supernatant. Macrophages were activated by means of incubation with 100 ng/mL LPS. The culture was maintained in DMEM with 10% FCS at 37°C for 24 hours. At the end of the culture period, supernatants were collected, aliquoted, and stored at 0°C for future experiments. This supernatant, macrophage-conditional supernatant, was called test supernatant. The supernatant from the macrophage culture without LPS stimulation was used as a control and called control supernatant. Establishment of tumor-bearing mice and observation of tumor growth. Groups of 25 BALB/c nude mice were used
to develop tumor-bearing mice. Mice were divided into 5 groups, each comprising 5 mice (Fig 1). Mice were subcutaneously injected with 1 ⫻ 106 malignant glioma cells, which had not been treated (glioma only) or treated with control supernatant, test supernatant, anti–TNF-␣, or anti–IFN-␥. Tumor size was recorded on the sixth day of injection and was thereafter checked every 6 days. Experimental and control animals were killed after 36 days of tumor inoculation. The
192
Chen et al
J Lab Clin Med March 2003
Fig 1. Expression of Fas and FasL on glioma cells. Tumor cells were cultured in the presence of test supernatant (10%) for 2 days, after which the cells were stained with anti-Fas or anti-FasL antibody and the percentage of positive cells was determined on flow cytometry. In neutralization experiments, test supernatant was incubated with anti–TNF-␣ (2.5 g/mL) or anti–IFN-␥ (0.25 g/mL) for 2 hours before it was applied to tumor cells. We set up a control using an isotype antibody (IgG) in each experiment. *P ⬍ .05, n ⫽ 3, paired Student t test, vs tumor cells treated with control supernatant.
tumors were resected and the tissues immediately processed for immunohistochemical staining and protein isolation. Fas and FasL detection. The expression of Fas and FasL proteins was detected by means of indirect immunofluorescence labeling and flow-cytometric analysis. Cells resuspended in 100 L of PBS with 0.1% sodium azide were incubated with 1000 ng of anti-Fas or anti-FasL antibody at 4°C for 30 minutes. Cells were washed and recovered by means of centrifugation at 1000g for 5 minutes at 4°C before being resuspended in 20 L of mouse anti–rabbit IgG with fluorescein conjugate (20 g/mL); cells were then incubated in dark for 30 minutes at 4°C. Finally, cells were analyzed with a Becton Dickinson FACScan machine (BD Biosciences, San Jose, Calif) for cell-associated fluorescence. Caspase-8 activity assay. We measured the activity of caspase-8 using a caspase-8 assay kit from Clontech (Palo Alto, Calif). The assay is based on spectrophotometric detection of the chromophore pNA after cleavage from the labeled
substrate IETD-pNA.21 The pNA light emission can be quantified using a spectrophotometer at 405 nm. Comparison of the absorbance of pNA from an apoptotic sample with an uninduced control allows determination of the fold increase in caspase-8 activity. The result was expressed as nmol/50 g protein/hr. Immunohistochemical examination for Fas, FasL, Bax, Bcl-l, Bcl-xL, and Caspase-8. Tissues were subjected to
immunohistochemical processing and embedded in paraffin. Formalin fixation before embedding was of less than 30 hours’ duration throughout, which is important for preserving the antigenic determinants analyzed in this study. Tissues were sectioned at a thickness of 4 m. Immunostaining was performed on paraffin sections in accordance with the instructions of the ABC kit from Vector Laboratories. In brief, tissue sections were deparaffinized and rehydrated through three changes of xylene and graded alcohol. Tissue sections were then boiled in citrate-based antigen-unmasking solution for 1
J Lab Clin Med Volume 141, Number 3
Chen et al
193
Table I. Scoring standard for immunohistochemical staining Grade
Scoring by tissue
0 1 2
None Slight Moderate
3
Strong
4
Extremely strong
Scoring by cell
None Part of the cytoplasm Whole cytoplasm without reticular deposits Reticular deposits in less than half of cytoplasm Reticular deposits in more than half of cytoplasm
minute and cooled in Milli-Q water (Millipore, Billerica, Mass), the endogenous peroxidase activity in tissue sections was quenched with 3% hydrogen peroxide solution for 5 minutes. Normal blocking serum (1.5%) supplemented with avidin solution (avidin/biotin blocking kit; Vector Laboratories) was used to block tissue sections for 30 minutes. Afterward, preparations were incubated with a primary antibody overnight at 4°C. The primary antibody was prepared in 1.5% normal blocking serum supplemented with biotin solution from the Vector Laboratories avidin/biotin blocking kit and used at a working dilution of 1:200. After tissue sections were washed with PBS, a biotinylate-labeled secondary antibody, IgG, was applied for 30 minutes. Tissue sections were then washed with PBS, ABC reagent (avidin/biotin kit; Vector Laboratories) conjugated with horseradish peroxidase was applied for 30 minutes. Bid antigen staining was visualized with the use of NovaRED or DAB substrate (Vector Laboratories). We terminated the reaction by rinsing the sections in tap water. All incubations were performed in a humidified environment at room temperature, except where indicated. Finally, sections were counterstained with Vector Gill hematoxylene. After dehydration through graded alcohol and being cleared with xylene, they were mounted with DPX permanent mountant (Vector Laboratories). Negative controls were prepared by replacing the primary antibody with PBS. Immunoreactivity for antigens in the tissues was examined with the use of the Eclipse TE300 microscope (Nikon, Tokyo, Japan). The image was analyzed with the MetaMorph Imaging System (Universal Imaging Corp, Downington, Pa) and scored in accordance with the standard described in Table I. Western blotting for Fas, FasL, Bax, Bcl-l, Bcl-xL, and caspase-8. Tissue samples were homogenized with ice-cold
PBS and then subjected to lysis in a solution containing 8 mol/L urea, 0.1 mol/L Na2H2PO4, and 0.01 mol/L Tris-HCl. Supernatants were obtained after centrifugation at 10,000g. After boiling, proteins were separated on 10% sodium dodecyl sulfate–polyacrylamide gels. Proteins were then electrophoretically transferred from the gel onto nitrocellulose membranes, and the membranes were blocked for 1 hour in PBSTween buffer containing 5% dry milk powder (fat-free) at room temperature. The membranes were then incubated with a primary antibody for 1 hour. After washing, the membranes were incubated with a secondary antibody, IgG-HRP. Finally they were treated with the reagents in the chemilumines-
Fig 2. Activity of caspase-8. Glioma cells were cultured in the presence of test supernatant (10%) for 2 days, after which the cell lysate was isolated for the measurement of caspase-8 activity. *P ⬍ .05, n ⫽ 6, paired Student t test, vs tumor cells treated with control supernatant.
cence-detection kit (ECL system; Amersham Pharmacia Biotech, Piscataway, NJ) in accordance with the manufacturer’s instructions. The densities of the protein bands were determined and analyzed with the Quantity One program (version 4.2; Bio-Rad Laboratories, Hercules, Calif). Anti-actin antibody (Santa Cruz Biotechnology, Santa Cruz, Calif) was used to detect actin, which was used as a control for equal loading. Apoptosis detection. Apoptosis was detected with the use of both flow cytometry and in situ detection. For flow cytometry, cells were washed in cold PBS and incubated with fluorescein-conjugated annexin V and propidium iodine. Flow cytometry of annexin V and propidium iodine-stained cells was performed on a Becton Dickinson FACScan machine; results were analyzed with CellQuest software (BD Biosciences). For in situ detection of apoptotic cells, the DeadEnd apoptosis detection kit (TUNEL assay) from Promega (Madison, Wis) was used to conduct in situ labeling of the 3⬘-end of the DNA fragments generated by apoptosisassociated endonucleases. In brief, the sections, after being dewaxed in xylene and rehydrated in ethanol, were incubated with 20 g/mL proteinase K at room temperature for 15 minutes. The slides were then incubated with a terminal transferase enzyme and biotinylated nucleotide mix in 37°C for 60 minutes to allow the end-labeling reaction to occur. We blocked the endogenous peroxidase activity by treating the slides with 0.3% hydrogen peroxide in PBS, pH 7.2. Horseradish-peroxidase–labeled streptavidin solution was then applied to the slides, which were incubated for 30 minutes at room temperature. After incubation, the color was developed with the peroxidase substrate, hydrogen peroxide, and the stable chromogen diaminobenzidine. The slides were then mounted and examined under a light microscope. Cells were defined as apoptotic if the whole nuclear area of the cell labeled positively. Apoptotic bodies were defined as small
194
J Lab Clin Med March 2003
Chen et al
Fig 3. Expression of death receptor– dependent and independent proteins in tumor tissues. The expression of Fas, FasL, caspase-8, Bax, Bcl-2, and Bcl-xL proteins in tumor tissues was examined with the use of immunohistochemical staining (original magnification ⫻400). A and B, Fas. C and D, FasL. E and F, caspase-8. G and H: Bax. I and J, Bcl-2. K and L, Bcl-xL. A, C, E, G, I, and K represent tissue samples obtained from tumors treated with control supernatant; B, D, F, H, J, and L represent tissue samples obtained from tumors treated with test supernatant. Bar ⫽ 18 mol/L.
positively labeled globular bodies in the cytoplasm of the cells, which were detected either singly or in groups. To estimate the apoptotic index that represented the percentage of apoptotic cells in a given area, we counted apoptotic cells and bodies in 10 high-power fields, then divided this figure by the number of cells in the same high-power fields. TNF-␣ and IFN-␥ determination. The levels of TNF-␣ and IFN- released from cultured cells into the medium were determined with the use of ELISA kits from Chemicon (Temecula, Calif) in accordance with the manufacturer’s instructions. In brief, samples were added to the assay plate, which was precoated with an antibody to capture TNF-␣ or IFN-␥. The biotinylated TNF-␣ or IFN-␥ conjugate and sample/ standard compete for TNF-␣ or IFN-␥, and thus the amount of TNF-␣ or IFN-␥ in sample/standard could be determined. The assay was visualized with the use of a streptavidin– alkaline phosphatase conjugate and an ensuring chromagenic substrate reaction.
Statistical analysis. All values are expressed as mean ⫾ SD. We analyzed statistical comparisons with the Student t or Mann-Whitney U test, using InStat software (GraphPad Software, San Diego, Calif). A P value of less than .05 was taken as statistically significant.
RESULTS Expression of death receptor– dependent proteins. Fas was the best-characterized death receptor. We first examined the expression of Fas and its ligand FasL in glioma cells. The glioma cells were stimulated with test supernatant from the activated macrophages. Fas and FasL were recognized by their antibodies, and the positive signals were detected on flow cytometry. We found that concentrations of both Fas and FasL were much higher in glioma cells with test supernatant than
J Lab Clin Med Volume 141, Number 3
Chen et al
195
Fig 3. Continued
in those without (Fig 1). The effect of test supernatant on Fas expression was significantly blocked by the application of a TNF-␣– or IFN-␥–neutralizing antibody (Fig 1). However, the high level of FasL expression could be inhibited only by the TNF-␣–neutralizing antibody, not by the IFN-␥ neutralizing antibody. The activity of caspase-8, which is an upstream caspase and a corridor link between death receptor– dependent and independent pathways, was also markedly increased in glioma cells with test supernatant compared with those without (Fig 2). The in vivo expression of Fas, FasL, and caspase 8 was analyzed in glioma cells xenografted into nude mice. The glioma cells were treated with test supernatant before transplantation. Immunohistochemical data showed that the levels of Fas (3.62 ⫾ 0.87 vs 2.18 ⫾ 0.60, P ⬍ .01), FasL (2.11 ⫾ 0.63 vs 1.08 ⫾ 0.46, P ⬍ 0.05), and caspase-8 (3.18 ⫾ 0.69 vs 1.89 ⫾ 0.52, P ⬍ 0.01) were all increased in glioma cells treated with test supernatant compared with those treated with control medium (Fig 3, A-F). A similar pattern was obtained on Western-blot analy-
sis (Fig 4). Data similar to those obtained with test supernatant and control supernatant were obtained in the samples treated with activated macrophages directly and those without treatment, respectively, on Western-blot or immunochemical analysis (data not shown). Expression of death receptor–independent proteins.
Bcl-2 family members are the main player in the death receptor–independent pathway. The level of Bax protein was increased in glioma cells treated with test supernatant, as demonstrated on immunohistochemical staining (3.37 ⫾ 0.99 vs 2.01 ⫾ 0.68, P ⬍ .01) or Western-blot analysis (Fig 3, G and H, and Fig 4). On the contrary, the expression of Bcl-2 protein was significantly decreased in glioma cells treated with test supernatant (1.83 ⫾ 0.51 vs 2.89 ⫾ 0.52, P ⬍ .05) compared with those without treatment or treated with control supernatant (Fig 3, I and J, and Fig 4). We did not find a significant difference in the level of Bcl-xL between glioma cells treated with test supernatant and those treated with control supernatant (1.67 ⫾ 0.34 vs 1.83 ⫾ 0.47, P ⬎ .05) (Fig 3, K and L, and Fig 4).
196
J Lab Clin Med March 2003
Chen et al
Fig 4. Western-blot analysis of death receptor– dependent and independent proteins. Proteins were isolated from tumors formed by glioma cells treated with test supernatant or with control supernatant or without treatment. Anti-Fas, caspase-8, Bax, Bcl-2, and Bcl-xL antibodies were used to detect their protein levels on Western-blot analysis. Lane 1, tissue samples obtained from tumors treated with test supernatant. Lane 2, samples from tumors treated with control supernatant. Lane 3, samples from tumors without treatment.
Fig 5. Apoptosis of glioma cells. Glioma cells were cultured in the presence of test supernatant (10%) for 2 days, after which the cells were collected and stained with annexin V and propidium iodine. The percentage of positive cells was analyzed on flow cytometry. *P ⬍ .05, n ⫽ 3, paired Student t test, vs tumor cells treated with test supernatant.
Similar expression patterns in test supernatnat and control supernatant were obtained in the samples treated with activated macrophages directly and the samples without treatment, respectively, on either Western-blot or immunochemical analysis (date not shown). Apoptosis of glioma cells. The percentage of apoptotic glioma cells treated with test supernatant was significantly higher (21.5%) than that in the cells without treatment or treated with control supernatant, as demonstrated by annexin V and propidium iodine staining with the aid of flow cytometry (Fig 5). In situ detection
also indicated that the number of apoptotic glioma cells was markedly increased in the cells treated with test supernatant, compared with the cells without treatment or treated with control supernatant (Fig 6). We set up a positive control for in situ apoptotic detection in which apoptosis was induced by DNase I that was known to cause fragmentation of the chromosomal DNA. A significant number of cells became apoptotic in the positive-control tissue treated with DNase I (Fig 6, C). LPS itself did not show any apopototic effect on glioma, and therefore the direct effect of LPS on glioma was ruled out. In vivo tumor growth. The growth of glioma in vivo was significantly slower in the tumor cells receiving test supernatant or macrophage treatment directly than in those without treatment or treated with control supernatant (Fig 7). We ruled out the growth of macrophages in vivo because we detected no signs of growth in the mice injected with macrophages only. The glioma tumor in the mice without treatment and those treated with control supernatant was significantly larger than those with treatment after just 6 days’ transplantation of the tumor cells. By the end of experiment (36 days), the size of the glioma tumors was about five times different between those receiving treatment and those without. The rate of tumor growth was not significant different between those treated with test supernatant and those with macrophages directly, although the tumor was slightly larger in the former than in the latter. TNF-␣ and IFN-␥ levels. Both TNF-␣ and IFN-␥ levels were significantly higher in test supernatant (11.25 ⫾ 1.02 ng/mL and 0.94 ⫾ 0.03 pg/mL, respectively) than in the control supernatant (8.61 ⫾ 0.43 ng/mL and 0.54 ⫾ 0.02 pg/mL, respectively; both P ⬍ .05, paired Student t test, n ⫽ 5). Glioma cells alone did not respond significantly to LPS stimulation for TNF-␣ and IFN-␥ production (data not shown). DISCUSSION
Macrophages require a series of extracellular signals to become activated. The activated macrophages will undergo a set of phenotypic changes, including cytokine production and tumoricidal activity.22 In this study, we showed that macrophages could be activated by LPS, a glycolipid that constitutes the major portion of the outermost membrane of gram-negative bacteria,23 and that activated macrophages were able to generate a significant level of TNF-␣ and IFN-␥. This finding is in line with previous evidence indicating that activated macrophages synthesize and secrete a wide variety of lytic or cytotoxic molecules that are capable of destroying tumor cells.22,24,25 However, how macrophage-generated molecules exert their tumoricidal
J Lab Clin Med Volume 141, Number 3
Chen et al
197
Fig 6. In situ detection of apoptosis in glioma tissues. Apoptotic cells stained brown (original magnification ⫻ 400). We detected very few apoptotic cells in glioma tissues treated with control supernatant (A), whereas of apoptosis was much more frequent in glioma tissues treated with test supernatant (B). Tissue treated with DNase I was used as a positive control (C). DNase I treatment is known to result in fragmentation of the chromosomal DNA.
Fig 7. Growth of the xenograft glioma in nude mice. BALB/c nude mice were subcutaneously injected with 1 ⫻ 106 glioma cells that were subjected to different treatments. The growth of glioma in tumor-bearing mice was monitored along the course of experiment (36 days). Lump size was recorded every 6 days and expressed as square millimeters. Each group comprised 5 mice. Statistical comparisons between groups are described in the text. P ⬍ .05, n ⫽ 5, vs glioma only or glioma ⫹ control supernatant; P ⬍ .01, n ⫽ 5, vs glioma only or glioma ⫹ control supernatant.
effects is not yet completely understood, especially in the situation of interaction between macrophages and brain tumors. Our results indicate that the activated macrophages are able to induce apoptosis in a significant number of glioma cells and inhibit the growth of glioma through the production of the tumoricidal molecules, probably mainly TNF-␣ and IFN-␥. Both TNF-␣ and IFN-␥ are well known for their proapoptotic functions by way of a death receptor– dependent pathway.26,27 In this study, we demonstrated that the supernatant (rich in these two cytokines) from the activated macrophages could significantly stimulate the expression of Fas and FasL in the glioma cells. Fas and FasL are the most important
death-receptor system identified so far. It has been documented that glioma cells express functional Fas and FasL that induce apoptosis in tumor cells.8,28 Ligation of Fas by FasL recruits the adaptor molecule Fas-associating death domain (FADD) to the receptor by way of mutual interaction of their death domains. FADD engages procaspase-8, and subsequently procaspase-8 is autoproteolytically activated. Activated caspase-8 serves as an enzyme for downstream effector caspases including 3, 6, and 7. The activated effector caspase, in turn, cleaves a number of cytoplasmic and nuclear substrates, degrades chromosomal DNA, and induces cell death. The evidence to support the involvement of the death receptor– dependent path-
198
J Lab Clin Med March 2003
Chen et al
way also came from the alteration of caspase-8 in both in vitro and in vivo experiments. The activity of caspase-8 was significantly increased in glioma cells treated with the supernatants derived from the activated macrophages, and the level of this protein was overexpressed in xenograft glioma formed from tumor cells treated with the molecules generated by the activated macrophages. Without amplification by caspase-8, it is not possible for cells to undergo the death receptor– dependent pathway. Another finding of this study was that the percentage of glioma cells positive for Fas and FasL was much lower than the percentage of tumor cells with signs of apoptosis, suggesting that multiple channels are involved in induction of apoptosis of glioma cells when the cells were treated with the supernatant from the activated macrophages. After analysis of the Bcl-2 system, we found that the expression of Bax was significantly increased, whereas the level of Bcl-2, a conversed homologue of Bax, was much lower in xenograft glioma formed from tumor cells treated with the molecules generated by the activated macrophages compared with that treated with control medium or the supernatant from inactivated macrophages. Bax is well known as a proapoptotic member of Bcl-2 family, whereas Bcl-2 is an antiapoptotic agent. Both Bax and Bcl-2 play an important role in the regulation of apoptosis by way of the death receptor–independent pathway. On being stimulated, Bax is translocated to the mitochondria and forms a heterodimer with Bcl-2, which results in the release of cytochrome c. Cytochrome c will function to induce activation of caspases 9 and 3.9 It has not been reported how Bcl-2 family members behave in glioma when the tumor cells encounter macrophages. However, it has been demonstrated that alteration of Bax and Bcl-2 is associated with ceramide-induced apoptosis of glioma cells.9 Macrophage-related cytokines such as TNF-␣ and IFN-␥ are known to affect the expression of Bcl-2 family members.29,30 It is reported that a cross-talk between the death receptor– dependent and independent pathways occurs in certain types of cells by way of caspase-8 and proapoptotic Bid, a BH3-domain Bcl-2 family member.31 Although this is likely the case in the model we tested, this remains to be proved. The xenograft glioma experiment revealed that the growth of glioma was significantly slower and the size of the tumor much smaller in the transplantation treated with the supernatant of the activated macrophages. In situ apoptosis analysis indicated that the frequency of cell death was high in the macrophage-treated glioma tissue, along with the altered expression of apoptosisrelated proteins of both death receptor– dependent and independent pathways. Obviously the apoptosis in-
duced by way of both the death receptor– dependent and independent pathways is, at least in part if not completely, responsible for inhibition of tumor growth. On the basis of whether Bcl-2 system is involved in the apoptotic pathway, apoptotic cells can be classified as type I or type II cells.32 In type I cells, activated caspase-8 directly triggers the cascade of effector caspases without the involvement of mitochondria and Bcl-2 family members. However, in addition to Fas and caspase-8, apoptosis of type II cells involves mitochondria and Bcl-2 family members. According to this concept, the glioma cells we tested here are of type II. However, whether a subgroup of tumor cells functions as type I cells cannot be ruled out. In fact, a study by Knight et al indicated that glioma cells were heterogeneous and behaved in both type I and type II manners.28 Identification of the death receptor– dependent and independent pathways in glioma apoptosis induced by activated macrophages may provide some useful information for designing an efficient intervention against tumor growth. Understanding the contribution of individual molecules in both pathways following the interaction between macrophages and glioma is not only a fascinating scientific puzzle; it may aid in the design of pharmacologic strategies to interfere more specifically with the aim of enhancing apoptosis, thereby avoiding the troublesome unwanted side effects of immunotherapy for cancer. REFERENCES
1. Leibel SA, Scott CB, Loeffler JS. Contemporary approaches to the treatment of malignant gliomas with radiation therapy. Semin Oncol 1994;21:198-219. 2. Lesser GJ, Grossman S. The chemotherapy of high-grade astrocytomas. Semin Oncol 1994;21:220-35. 3. Mahaley MS Jr. Neuro-oncology index and review (adult primary brain tumors): radiotherapy, chemotherapy, immunotherapy, photodynamic therapy. J Neurooncol 1991;11:85-147. 4. Rinehart JJ, Vessella R, Lange P, Kaplan ME, Gormus BJ. Characterization and comparison of human monocyte- and macrophage-induced tumor cell cytotoxicity. J Lab Clin Med 1979;93: 361-9. 5. Keller R, Keist R, Wechsler A, Leist TP, van der Meide PH. Mechanisms of macrophage-mediated tumor cell killing: a comparative analysis of the roles of reactive nitrogen intermediates and tumor necrosis factor. Int J Cancer 1990;46:682-6. 6. Shinoda J, Whittle IR. Nitric oxide and glioma: a target for novel therapy? Br J Neurosurg 2001;15:213-20. 7. Duan L, Aoyagi M, Tamaki M, Nakagawa K, Nagashima G, Nagasaka Y, et al. Sensitization of human malignant glioma cell lines to tumor necrosis factor–induced apoptosis by cisplatin. J Neurooncol 2001;52:23-36. 8. Glaser T, Wagenknecht B, Weller M. Identification of p21 as a target of cycloheximide-mediated facilitation of CD95-mediated apoptosis in human malignant glioma cells. Oncogene 2001;20: 4757-67. 9. Sawada M, Nakashima S, Banno Y, Yamakawa H, Hayashi K, Takenaka K, et al. Ordering of ceramide formation, caspase
J Lab Clin Med Volume 141, Number 3
10.
11.
12. 13. 14.
15.
16.
17.
18. 19. 20.
21.
activation, and Bax/Bcl-2 expression during etoposide-induced apoptosis in C6 glioma cells. Cell Death Differ 2000;7:761-72. Tani E, Kitagawa H, Ikemoto H, Matsumoto T. Proteasome inhibitors induce Fas-mediated apoptosis by c-Myc accumulation and subsequent induction of FasL message in human glioma cells. FEBS Lett 2001;504:53-8. Kondo S, Barna BP, Morimura T, Takeuchi J, Yuan J, Akbasak A, et al. Interleukin-1 beta-converting enzyme mediates cisplatin-induced apoptosis in malignant glioma cells. Cancer Res 1995;55:6166-71. Ashkenazi A, Dixit VM. Death receptors: signaling and modulation. Science 1998;281:1305-8. Gupta S. Molecular steps of cell suicide: an insight into immune senescence. J Clin Immunol 2000;20:229-39. Tzeng JJ, Barth RF, Orosz CG. Quantitation of glioma-reactive cytolytic T lymphocyte precursors by means of limiting dilution analysis. J Immunol Methods 1992;146:177-84. Tzeng JJ, Barth RF, Orosz CG, James SM. Phenotype and functional activity of tumor-infiltrating lymphocytes isolated from immunogenic and nonimmunogenic rat brain tumors. Cancer Res 1991;51:2373-8. Bouwens L, Baekeland M, Wisse E. Cytokinetic analysis of the expanding Kupffer-cell population in rat liver. Cell Tissue Kinetics 1986;19:217-26. Jones EA, Summerfield JA. Kupffer Cells. In: Arias IM, Jakoby WB, Popper H, et al, eds. The liver: biology and pathobiology, ed 2. New York: Raven Press, 1988:683-704. Olynyk JK, Clarke SL. Isolation and primary culture of rat Kupffer cells. J Gastroenterol Hepatol 1998;13:842-5. Alpini G, Phillips JO, Vroman B, LaRusso NF. Recent advances in the isolation of liver cells. Hepatology 1994;20:494-514. Dijkstra CD, Dopp EA, Joling P, Kraal G. The heterogeneity of mononuclear phagocytes in lymphoid organs: distinct macrophage subpopulations in the rat recognized by monoclonal antibodies ED1, ED2 and ED3. Immunology 1985;54:589-99. Casciola-Rosen L, Nicholson DW, Chong T, Rowan KR, Thornberry NA, Miller DK, et al. Apopain/CPP32 cleaves proteins that are essential for cellular repair: a fundamental principle of apoptotic death. J Exp Med 1996;183:1957-64.
Chen et al
199
22. Askew D, Yurochko AD, Burger CJ, Elgert KD. Normal and tumor-bearing host macrophage responses: variability in accessory function, surface markers, and cell-cycle kinetics. Immunol Lett 1990;24:21-9. 23. Raetz CR, Ulevitch RJ, Wright SD, Sibley CH, Ding A, Nathan CF. Gram-negative endotoxin: an extraordinary lipid with profound effects on eukaryotic signal transduction. FASEB J 1991; 5:2652-60. 24. Shimoda O, Takeda Y, Woo HJ, Shimada S, Higuchi M, Osawa T. A human macrophage hybridoma producing a cytotoxic factor distinct from TNF, LT, and IL-1. Cancer Immunol Immunother 1988;6:101-8. 25. Erickson KL, Hubbard NE. Dietary fish oil modulation of macrophage tumoricidal activity. Nutrition 1996;12(suppl 1):S34-8. 26. Naujokat C, Sezer O, Possinger K. Tumor necrosis factor-alpha and interferon-gamma induce expression of functional Fas ligand on HT29 and MCF7 adenocarcinoma cells. Biochem Biophys Res Commun 1999;264:813-9. 27. Choi C, Park JY, Lee J, Lim JH, Shin EC, Ahn YS, et al. Fas ligand and Fas are expressed constitutively in human astrocytes and the expression increases with IL-1, IL-6, TNF-alpha, or IFN-gamma. J Immunol 1999;162:1889-95. 28. Knight MJ, Riffkin CD, Muscat AM, Ashley DM, Hawkins CJ. Analysis of FasL and TRAIL induced apoptosis pathways in glioma cells. Oncogene 2001;20:5789-98. 29. Saldeen J. Cytokines induce both necrosis and apoptosis via a common Bcl-2-inhibitable pathway in rat insulin-producing cells. Endocrinology 2000;141:2003-10. 30. Boldrini L, Calcinai A, Samaritani E, Pistolesi F, Mussi A, Lucchi M, et al. Tumour necrosis factor-alpha and transforming growth factor-beta are significantly associated with better prognosis in non-small cell lung carcinoma: putative relation with BCL-2-mediated neovascularization. Br J Cancer 2000;83:480-6. 31. Zimmermann KC, Bonzon C, Green DR. The machinery of programmed cell death. Pharmacol Ther 2001;92:57-70. 32. Scaffidi C, Fulda S, Srinivasan A, Friesen C, Li F, Tomaselli KJ, et al. Two CD95 (APO-1/Fas) signaling pathways. EMBO J 1998;17:1675-87.