Vascular permeability induced by protein product of malignant brain tumors: inhibition by dexamethasone

Transcription

1 J Neurosurg 67: , 1987 Vascular permeability induced by protein product of malignant brain tumors: inhibition by dexamethasone JEFFRE N. BRUCE, M.D., GREGOR R. CRISCUOLO, M.D., MARSHA J. MERRILL, PH.D., Ross R. MOQUIN, M.D., J. BoB BLACKLOCK, M.D., AND EDWARD H. OLDFIEED~ M.D. Surgical Neurology Branch, National Institute of Neurological and Communicative Disorders and Stroke, Bethesda, Maryland v, Serum-free conditioned medium derived from confluent monolayer cultures of malignant human astroglial tumors contains a substance that rapidly increases capillary vascular permeability after intradermal injection into guinea pigs. Accumulation of vascular permeability factor (VPF) activity occurs with increasing duration of tumor incubation in vitro. Expression of this activity is inhibited by incubation of cell cultures with cycloheximide or dexamethasone. This VPF is an acid-stable heat-labile macromolecule that is inactivated by trypsin and pepsin and binds immobilized heparin. Activity is retained by ultrafiltration with 30,000-dalton cut-off microconcentrators. Pretreatment of test animals with systemic dexamethasone prior to intradermal injection of VPF diminishes microvascular permeability. Furthermore, VPF activity is not inhibited by antihistamines. Secretion of VPF may cause the vasogenic brain edema that is frequently associated with malignant primary and metastatic intracerebral tumors. Inhibition by dexamethasone of both VPF expression in tissue culture, and VPF activity at the microvascular level in test animals, is in keeping with the known efficacy of this agent in treating the vasogenic edema associated with brain tumors. KE WORDS ~ brain neoplasm 9 vascular permeability factor 9 astrocytoma 9 glioblastoma C EREBRAL edema is a significant cause of the neurological deficits and elevated intracranial pressure associated with malignant brain tumors. This form of cerebral edema is produced by disruption of the blood-brain barrier with increased vascular permeability and excessive interstitial fluid accumulation. 6 Prior investigations into the origin of tumorinduced increased cerebrovascular permeability implicate biochemical mediators, 2'4'~4 structural alterations in tumor-induced blood vessels, 5 destruction of the cerebral capillary endothelium, 7 increased capillary hydrostatic pressure, ~ and nonspecific tumor secretory productsj ~ Glucocorticoids such as dexamethasone effectively reduce neurological deficits and intracranial hypertension associated with peritumoral brain edema. The pharmacological basis for their efficacy is, however, largely empirical. 3 The purpose of this study was to investigate the possibility that brain tumors produce and release a specific substance that evokes cerebral edema by increasing vascular permeability. To accomplish this, conditioned medium of confluent monolayer cultures derived from a variety of human brain tumors was harvested and the Miles assay ~ was used to measure capillary permeability in normal skin following intradermal injection of test substances. Materials and Methods Tissue Culture Techniques Cells derived from tissue explants of human brain tumors were cultured at 37~ in 95% room air/5% CO2 and 100% humidity in Dulbecco's modified Eagle's medium (DMEM) supplemented with 1% L-glutamine, i % penicillin ( 10,000 U/ml), 1% streptomycin (10,000 ug/ml), and 10% fetal calf serum. Confluent cultures grown in Falcon tissue-culture flasks (175-sq cm growth area) were washed three times with Hanks' balanced salt solution without the addition of calcium, magnesium, or phenol red, and incubated with 25 ml of serum-free DMEM for 24 hours. Flasks typically contained to 4 l0 7 cells. The washing process was then repeated, and conditioned serum-free DMEM was collected 5 to 7 days later and either 1) assayed for vascular permeability factor (VPF) activity, 2) concentrated fivefold by vacuum dialysis using collodion mem- 880 J. Neurosurg. / Volume 67~December, 1987

2 Malignant brain tumor vascular permeability factor branes* with a 25,000-molecular weight (MW) cut-off, or 3) concentrated 20- to 40-fold by wet-tubing dialysist (MW cut-off 25,000) against 0.05 molar ammonium bicarbonate titrated to ph 7.4 with glacial acetic acid. The medium was then subjected to shell freezing in super-cooled acetone and lyophilization for 48 hours. The lyophilized product was then reconstituted using Dulbecco's phosphate-buffered saline (PBS) without calcium or magnesium and was filtered through a u membrane; it was then subjected to assay for VPF activity or biochemical characterization steps. Flasks of confluent cells were subcultured every 5 to 20 days as necessary using standard procedures. Miles Assay The Miles assay 8'9 is a biological assay of induction of capillary permeability in normal skin following the intradermal injection of test substances. Modifications of the original assay were used to more precisely quantify VPF activity. After induction of anesthesia with diethyl-ether, previously shaved male Hartley strain guinea pigs,~ each weighing 450 to 500 gm, were given an intracardiac injection of 1% Evans blue dye and 10 uci (2.22 x 107 disintegrations/min) of iodine-125- labeled bovine serum albumin (125I-BSA)w in a total volume of 1.0 ml. Test samples (0.1 ml) of conditioned medium from benign and malignant human brain tumors, fibroblasts, and control substances were then injected intradermally using No. 30 needles. All samples were tested in triplicate. Histamine standards (1.0 ug) were included when indicated. Test samples that induce cutaneous vascular permeability cause a visible blue stain at the injection site due to extravasation of the circulating Evans blue dye. After 20 minutes the animals were exsanguinated, and the injection sites were excised and counted on a Beckman Gamma 4000 counter]l to measure ~25I-BSA extravasation. The tissue tags were then allowed to dry for 5 days, after which they were weighed. Activity of VPF was expressed as counts per minute per milligram drytissue weight (cpm/mg). Activity of uninjected (blank) skin tags was subtracted as background to provide values for the increase in vascular permeability induced by the sample. Brain Tumor-Derived Vascular Permeability Factor Studies Cycloheximide Study. Tissue culture medium derived from malignant glial tumors was assayed after 12 * Collodion membranes obtained from Schleicher and Schuell, Inc., Keene, New Hampshire. t Dialysis system obtained from Spectrum Medical Industries, Inc., Los Angeles, California. $ Hartley strain guinea pigs obtained from Charles River Laboratories, Wilmington, Massachusetts. w Iodine-125-labeled bovine serum albumin obtained from Dupont-New England Nuclear, Boston, Massachusetts. II Gamma 4000 counter manufactured by Beckman Instruments, Fullerton, California. hours of incubation with cycloheximide (20 ug/ml). Control studies were performed on plain DMEM, conditioned medium from 12-hour cultures not exposed to cycloheximide, and plain DMEM to which cycloheximide was added just before intradermal injection of the test substances. Dexamethasone Studies. Tissue culture medium derived from malignant glial tumors was assayed after incubation for 40 and 72 hours with graded concentrations of dexamethasone ( 10-5, 10-7, 10-9, and 10-Jl M). Controls included 40- and 72-hour preparations of medium not exposed to dexamethasone and conditioned medium to which similar concentrations of steroid were added just before intradermal injection of the test substances. To determine the effect of pretreatment of the animals with systemic glucocorticoid on VPF activity, dexamethasone (2 mg/kg) was administered intraperitoneally 6 hours and 1 hour before intradermal injection of the test substances. Control animals received an intraperitoneal injection of an equivalent volume of Dulbecco's PBS. Pyrilamine/Cimetidine Study. To determine if the VPF acted via a pathway in common with histamine, guinea pigs were pretreated with pyrilamine (5 ~mole/ kg, subcutaneously) and cimetidine (500 umole/kg, intraperitoneally) prior to intradermal injection of the test substances. Experiments to Determine Physical Characteristics of VPF Separate solutions containing lyophilized conditioned medium in Dulbecco's PBS were treated with heat (950C for 20 minutes), acidification (ph 7.4 to 3.0 for 10 minutes followed by titration to ph 7.4), and trypsinization (incubation at 37"C with 1.0 mg trypsin/ ml VPF solution) followed by neutralization of trypsin with soybean trypsin inhibitor. Peptic degradation was achieved by acidifying the 1.0 mg/ml pepsin-vpf mixture to ph 3.4 and incubating at 37"C for 30 minutes followed by retitration to ph 7.4 to inactivate the pepsin's proteolytic capability. Soybean trypsin inhibitor (1000 ug/ml) was also coinjected with conditioned medium to determine whether VPF activity was related to Hageman factor activation (clotting Factor XIIa). Results Conditioned medium from low-passage confluent monolayer cell cultures of human malignant astrocytoma (glioblastoma multiforme) lines evoked increased microvascular permeability, whereas medium from benign astrocytoma, meningioma, and fibroblasts had little or no activity (Table 1). Fluid aspirated from a cystic glioblastoma had very high VPF activity, whereas no activity was evident in samples of cerebrospinal fluid from a normal volunteer, a patient with a sacral chordoma, or a patient with a malignant cerebral glioma (Table 1). Induction of capillary permeability by glioblastoma J. Neurosurg. / Volume 67/December,

3 J. N. Bruce, et al. TABLE 1 Expression of vascular permeability factor (VPF) activity* Source of Test Sample VPF Activity Experiment A control (DMEM) fibroblast meningioma meningioma astrocytoma 15.2 _ 5.8 glioblastoma glioblastoma _ Experiment B control (DPBS) CSF (normal volunteer) CSF (sacral chordoma) CSF (malignant glioma) cyst fluid (glioblastoma) _ cyst fluid (glioblastoma) * The VPF activity for all Miles assays is expressed as mean counts per minute/rag tissue standard error of the mean. Experiment A: VPF activity in conditioned medium from human tissue lines; Experiment B: VPF activity in cerebrospinal fluid (CSF) and cystic glioblastoma fluid. DMEM = Dulbecco's modified Eagle's medium; DPBS = Dulbecco's phosphate-buffered saline Hour Incubation ] 777/ 100 ~/ N 2o, l x *p <0.01 *'p < ~ 72 Hour Incubation **'p < ,,- 160 ~ I00,- 8O " m 300 t/) + x r IO0 * p <0.05 ** p <0.01 T 0 Media Control I Incubated With J I Dexamethasone Added I Dexamethasone After Incubation [ Glioblastoma #1 } SOURCE OF MEDIA INJECTED FIG. 2. Histograms showing glioblastoma-derived vascular permeability factor (VPF) activity after 40 and 72 hours' incubation of line 1 tumor cultures with decrementing quantities of dexamethasone. The VPF activity is expressed as mean counts per minute (CPM)/mg + 1 standard error of the mean (SEM) Hours Hours Hours Days Days DURATION OF GLIOBLASTOMA INCUBATION FIG. 1. Histogram showing induction of vascular extravasation of ~25I-bovine serum albumin as a function of cell culture incubation time for glioblastoma line 1 samples. Vascular permeability factor activity is expressed as mean counts per minute (CPM)/mg + 1 standard error of the mean (SEM). line 1 increased as the duration of culture incubation was lengthened (Fig. 1). Incubation of this glioblastoma multiforme line with dexamethasone exhibited reduced expression of VPF activity as did glucocorticoid pretreatment of test animals (Figs. 2 and 3). Pretreatment of test animals with indomethacin or antihistamines had no effect upon VPF activity (data not shown). Additional chemical and physical characterization of VPF (Table 2) revealed an insensitivity of VPF to soybean trypsin inhibitor (1000 ug/ml) which indicates that VPF is not Hageman factor (clotting Factor XIIa, a permeability factor that is unmasked when serum is diluted). The VPF activity was eliminated by treatment with the proteases trypsin and pepsin. Acidification had no significant effect upon VPF activity, whereas heating attenuated activity by 46% to 67%. Exposure of brain tumor-derived VPF to polyclonal immunoglobulin G raised against the VPF derived from guinea pig line 10 tumor eliminated induction of vascular extravasation. Furthermore, both factors have been shown to bind heparin-sepharose (data not shown). Discussion Several substances which increase the permeability of normal blood vessels have been derived from normal and tumorous tissues. Previously described VPF's include histamine (MW 111), serotonin (MW 222), leukotriene (MW 336 to 522), prostaglandin (MW 350 to 355), prostacyclin (MW 352), thromboxane (MW 352 to 370), bradykinin (MW 1060), kallidin (MW 1188), leukokinin (MW 2500), lymphocyte permeability factors (MW 12,000 and 39,000), and kallikrein (MW 108,000). We have demonstrated that human malignant gliomas produce and release a substance in vitro that rapidly induces microvascular permeability in the Miles assay with onset of activity by 1 to 2 minutes, 882 J. Neurosurg. / Volume 67~December, 1987

4 Malignant brain tumor vascular permeability factor A =E U.I (/) O) E =[ a. r ,oo [] Saline Controls [] Dexamethasone o ~ Media Media Line Line 1 Line 2 Line 2 SOURCE OF INJECTED SAMPLE FIG. 3. Histogram showing the influence of dexamethasone on the in vivo activity of the vascular permeability factor (VPF) derived from glioblastoma multiforme lines 1 and 2 and from control media. These findings were determined by the Miles assay. The VPF activity is expressed as mean counts per minute (CPM)/mg + 1 standard error of the mean. TABLE 2 Physical and chemical characterization of VPF* Sample Identification VPF Activity Experiment 1 control (DPBS) 0 conditioned medium conditioned medium (heat) 26.3 _+ 3.5 conditioned medium (acid) conditioned medium (SBTI) Experiment 2 control (DPBS) 0 conditioned medium conditioned medium (trypsin) conditioned medium (pepsin) 0 * Vascular permeability factor (VPF) activity for all Miles assays is expressed as mean counts per minute/mg tissue + standard error of the mean. Experiment 1: exposure of conditioned medium from glioblastoma line 1 to heat, acid, and soybean trypsin inhibitor (SBTI); Experiment 2: proteolytic digestion of VPF. DPBS = Dulbecco's phosphate-buffered saline. peak activity by 10 minutes, and little residual activity by 20 to 30 minutes. Lymphocyte permeability factors display different kinetic features such as a delayed onset of 30 to 60 minutes; peak permeability occurs 2 to 4 hours after exposure. Vascular permeability factor accumulates over time. Its expression is diminished by coincubation of tissue cultures with cycloheximide and dexamethasone. Its activity at the microvascular level is inhibited by pretreatment of test animals with systemic dexamethasone. Activity of VPF is unlikely to be mediated by prostaglandin synthesis or release, as pretreatment of test animals with indomethacin (a cyclooxygenase inhibitor) failed to prevent an increase in capillary permeability by intradermal injection of VPFcontaining samples. Vascular permeability factor is distinguished from prostaglandins and leukotrienes by physical characteristics, such as the exquisitely acidand heat-labile nature of the latter agents. Moreover, H~ and H2 histaminergic blockage, while totally eliminating the cutaneous response to histamine, failed to attenuate VPF-induced extravasation. J 3 Additional distinguishing features include: 1 ) an MW of at least 30,000 daltons; 2) acid stability; 3) moderate heat-lability; 4) proteolytic inactivation; and 5) crossantigenicity with a VPF derived from a non-human malignant cell line that has previously been described and partially characterized. 12 The affinity for heparin which brain tumor VPF shares with the VPF derived from guinea pig bile-duct carcinoma is particularly interesting, as heparin and similar polysaccharide moieties are known to exist on luminal endothelial membranes, thereby suggesting a potential site of action for VPF. Cerebral edema is associated with many primary and metastatic malignant brain tumors as well as with certain benign tumors. The mechanism of this "vasogenic" brain edema derives from a physiological and ultra- structural alteration in the blood-brain barrier. This results in increased microvascular permeability and subsequent accumulation of a plasma ultrafiltrate in the cerebral interstitial space. Our findings indicate that malignant glial tumors produce and release a substance that causes increased permeability in normal cutaneous capillaries. Furthermore, the production or secretion of VPF in vitro, as well as its action in vivo, is inhibited substantially by dexamethasone. The elaboration of the VPF may be responsible for the peritumoral cerebral edema frequently associated with malignant brain tumors. Suppression of this factor's production and its microvascular effects by dexamethasone suggests an explanation for the efficacy of glucocorticoids in symptomatic brain edema associated with cerebral neoplasms. This is consistent with the occurrence of steroid receptors in cerebral tumors 15 and the lack of steroid efficacy with ischemic, toxic, hemorrhagic, and traumatic cerebral edema. 3'~1 Acknowledgments The authors thank Mrs. Nancy Edwards and Mr. Calvin Hawkins for their technical assistance and Mrs. Ellie Frishman and Mrs. Doreen Quimby for preparing this manuscript. We also thank Dr. Don Senger for providing the antisera against his line 10 guinea pig bile-duct carcinoma-derived VPF. References 1. Casanova MF: Vasogenic edema with intraparenchymatous expanding mass lesions: a theory on its pathophysiology and mode of action of hyperventilation and corticosteroids. Med Hypotheses 13: , Chan PH, Fishman RA: The role of arachidonic acid in vasogenic brain edema. Fed Proc 43: , Fishman RA: Steroids in the treatment of brain edema. New Engl J Med 306: , Fishman RA, Chan PH: Hypothesis: membrane phospholipid degradation and polyunsaturated fatty acids play a J. Neurosurg. / Volume 6 7 / December,

5 J. N. Bruce, et al. key role in the pathogenesis of brain edema. Trans Am Neurol Assoc 106:58-6 i, Groothuis DR, Vick NA: Brain tumors and the bloodbrain barrier. Trends in Neurosci 5: , Klatzo I: Neuropathological aspects of brain edema. J Neuropathol Exp Neurol 26:1-14, Long DM: Capillary ultrastructure and the blood-brain barrier in human malignant brain tumors. J Neurosurg 32: , Miles AA, Miles EM: Vascular reactions to histamine, histamine-liberator and leukotaxine in skin of guinea pigs. J Physiol (Lond) 118: , Owen DAA, Poy E, Woodward DF, et al: Evaluation of the role of histamine H: and H2-receptors in cutaneous inflammation in the guinea-pig produced by histamine and mast cell degranulation. Br J Pharmacol 69: , Phillipon J, Foncin JF, Grob R, et al: Cerebral edema associated with meningiomas: possible role of a secretoryexcretory phenomenon. Neurosurgery 14: , Reulen H J: Vasogenic brain oedema. New aspects in its formation, resolution and therapy. Br J Anaesth 48: , Senger DR, Galli S J, Dvorak AM, et al: Tumor cells secrete a vascular permeability factor that promotes accumulation of ascites fluid. Science 219: , 1983 (Abstract) 13. Stoff JS, Rosa RM, Silva P, et al: Indomethacin impairs water diuresis in the DI rat: role of prostaglandins independent of ADH. Am J Physiol 241:F231-F237, Unterberg A, Baethmann AJ: The kallikrein-kinin system as mediator in vasogenic brain edema. Part 1: Cerebral exposure to bradykinin and plasma. J Neurosurg 61: 87-96, u Z, Wrange O, Bo6thius J, et al: A study of glucocorticoid receptors in intracranial tumors. J Neurosnrg 55: , 1981 Manuscript received May 11, Address reprint requests to: Edward H. Oldfield, M.D., Surgical Neurology Branch, National Institute of Neurological and Communicative Disorders and Stroke, 9000 Rockville Pike, Building 10 A-3E68, Bethesda, Maryland J. Neurosurg. / Volume 67~December, 1987