Journal of Cerebral Blood Flow and Metabolism
10:580--587 © 1990 Raven Press, Ltd., New York
Imaging Peripheral Benzodiazepine Receptors in Brain Tumors in Rats: In Vitro Binding Characteristics
Kiyonobu Ikezaki, Keith L. Black, * Arthur W. Toga, *Emily M. Santori, Donald P. Becker, and Mayumi L. Smith Division of Neurosurgery and Brain Research Institute and *Laboratory of Neuro Imaging and Department of Neurology, University of California, Los Angeles, School of Medicine, Los Angeles, California, U.S.A.
Summary: Peripheral benzodiazepine binding constants for transplanted RG-2 gliomas and HK and LK Walker 256 tumors (metastatic breast carcinoma) were deter mined in Wistar rats using autoradiography. In addition, Kd and Bmax parameters for peripheral benzodiazepine receptors on RG-2 tumors were directly visualized using digital image analysis of autoradiograms. High specific binding of [3H]PKI1195, a selective peripheral benzodi azepine ligand, had excellent topographical correlation to areas of histologically verified tumor. Scatchard analysis suggested a single class of peripheral binding sites with similar binding affinities in RG-2 and LK Walker 256 tu mors and normal cortex. Bmax was 20-fold greater in glial tumors and 11.6- and lO.6-fold greater in LK and HK Walker 256 tumors, respectively, compared to normal
cortex. The location of metastatic tumors, either intra cerebrally or subcutaneously, did not effect their Kd or Bmax values. Kd and Bmax values for RG-2 tumors were similar whether determined densitometrically or by direct visualization with image analysis. Binding parameters within normal brain were difficult to visualize by image analysis due to the low level of specific binding. The abil ity to label specifically intracerebral tumor cells and to characterize the binding parameters shown in this study suggest that peripheral benzodiazepine receptor ligands could be utilized by PET to analyze directly a variety of tumors in humans. Key Words: Brain tumor-Benzo diazepine receptor-Autoradiography-Digital image analysis-PK 11195.
The binding kinetics for peripheral benzodiaze pine receptors in normal brain structures and a va riety of intracranial tumors in humans have recently been demonstrated (Black et al. , 1990). There is little specific binding of peripheral-type ligands to normal brain structures with the exception of the choroid plexus, ependyma cells, or the pineal gland in rats (Braestrup and Squires, 1977; Marangos et al., 1982). Specific binding is high in intracerebral tumors and higher binding is correlated with in creasing malignancy in gliomas in humans (Black et al., 1990). Although the biologic role of these recep-
tors remains unclear, they are normally expressed in systemic tissues, such as the kidney (De Souza et al. , 1985; Schoemaker et al. , 1983). In this report, peripheral benzodiazepine receptor binding in three experimental brain tumors is characterized. In ad dition, the possibility that tumor growth within the brain might influence peripheral benzodiazepine binding constants is investigated. We also directly visualized Kd and Bmax values in RG-2 tumors using digital image analysis. This tech nique, unlike previous ligand binding methods, al lows for the visual display of specific binding, and Bmax and Kd constants obtained from an Eadie Hofstee analysis of the saturation data for an entire tissue section (Toga et al. , 1986). The relationships between binding parameters and anatomic (spatial) data or histological changes can thus be more readily analyzed and quantified. This report illus trates the utility and limitations of this technology in visually displaying receptor binding constants in brain tumors.
Received May 29, 1989; revised August 31, 1989; accepted September 11, 1989. Address correspondence and reprint requests to Dr. K. L. Black at Division of Neurosurgery, UCLA School of Medicine, Los Angeles, CA 90024, U.S.A. Abbreviations used: DMEM, Dulbecco's modified eagle me dium; GABA, -y-aminobutyric acid; PET, positron emission to mography; PK11195, 1-(2-chlorophenyl)-N-methyl-N-(1-methyl propyl)-3-isoquinoline carboxamide.
580
BENZODIAZEPINE BINDING IN BRAIN TUMOR MATERIALS AND METHODS eH]PK l l195 (75.2 Ci/mmol) was obtained from New England Nuclear. PK l l195 was kindly donated by Dr. G. Le Fur of Phamuka Laboratories. Ethylnitrosourea induced RG-2 rat glioma cell lines (Wechsler et aI., 1972) were obtained from Dr. D. Bigner (Duke University) and maintained in a monolayer culture in Dulbecco's modified eagle medium (DMEM) with 10% calf serum (Gibco). LK and HK Walker 256 metastatic breast carcinomas (Earle, 1935) were donated by Dr. R. Blasberg (NIH, Bethesda, MD, U.S.A.) and maintained subcutaneously in Wistar rats (Charles River). Wistar female rats (80-100 g) were anesthetized with ketamine (50 mg/kg) and xylazine (0.8 mglkg) i.m. Expo nentially growing RG-2 glioma cells (5 x 104) in 10 IJ-I of F lO serum free medium (Gibco) were implanted intrace rebrally into the left hemisphere, 4 mm lateral to the mid line and just behind the coronal suture. For Walker 256 tumor studies, 1 mm3 of HK or LK tumor was implanted intracerebrally into the left hemisphere. In some studies, either HK or LK tumors were simultaneously trans planted subcutaneously and intracerebrally in the same rat. Mter 14 days for RG-2 tumors and 11 days for Walker 256 tumors, rats were decapitated and their brains were removed and frozen in isopentane chilled on dry ice. Sixty continuous 20 IJ-m thick sections of brain were sliced using a cryostat and thaw-mounted on gelatin coated coverglasses, kept in a freezer overnight, and dried in desiccator before binding studies. Sections were subsequently prewashed twice for 2 min in an ice-cold binding buffer (50 roM Tris-HCl, pH 7.4) and dried by cool air. They were then incubated in binding buffer with various concentrations of eH]PKI1195 (0.25-10 nM) in the absence or presence of 10 IJ-M PK11195 for 60 min at O°C. The sections were again washed twice for 2 min in the ice-cold buffer, dried, and then coexposed with tissue calibrated 3H standards (Amersham) to Hiperfilm-3H (Amersham) for 3 to 7 days. Aliquots of binding buffer were taken to calculate the amount of free radioactive ligand after binding. After exposure, sections were stained with thionin or hematoxylin and eosin for histo logical verification and localization of the tumor. The re sultant autoradiograms and histological sections were an alyzed with a digital image analysis system using a mod ification of our previously described method (Toga et aI., 1986). Scatchard analysis was performed on data collected by discrete measurements of individual autoradiographic im ages. Nonspecific binding values were subtracted from the total binding values of adjacent sections. Specific binding data obtained from three sections were then av eraged. For image analysis, slightly out of focus digital images of total and nonspecific binding were obtained at two concentrations (2.5 and 10 nM). We have found that optical blurring prior to digitization reduces the graininess or noise of autoradiographic data and facilitates pixel by-pixel image manipUlations. Specific binding images were obtained by subtracting nonspecific images from to tal binding images, pixel by pixel. Specific binding images at two concentrations of [3H]PKII195 were then aligned to perform the Eadie-Hofstee analysis. Pixel-by-pixel re gression of bound/free vs. bound images yielded Kd (slope) and Bmax (intercept) images. To examine regional tissue quenching of 13 emissions from tritium-labeled PK l1195, a modified version of the
581
tissue overlay technique described by Kuhar and Unner stall (1985) was employed. This technique only provides a qualitative assessment of differential tissue quenching since the quantitative relationship between quenching in labeled tissue sections and quenching observed with tis sue overlays has not been established. Ten micron sec tions of unlabeled tumor were thaw-mounted and dried onto a block of tritium-impregnated methacrylate (do nated by American Radiolabeled Chemicals, St. Louis, MO, U.S.A.). LKB film was exposed to these tissue overlays for 3 days. In addition, dark-field photographs of adjacent sections heat dried onto coverslips and im mersed in water were obtained to help delineate the white matter and tumor portions of the section. Statistical anal ysis was performed using the Student's t test.
RESULTS In all three tumors, areas of increased binding of peripheral benzodiazepine ligands on autoradio grams had high topographical correlation with areas of histologically verified tumor. There was little binding to necrotic areas within tumor or normal brain tissue with the exception of the choroid plexus, pineal gland, and ependymal structures, which showed moderate binding (Fig. O. Sections incubated in the presence of excess PK11195 (non specific binding) appeared to show higher back ground binding in the cerebral cortex and the deep nuclei than in the tumor and white matter. One rea son for this heterogeneous pattern of nonspecific binding could be the problem of differential tritium quenching. The tissue overlay autoradiogram shown in Fig. 2 does reveal a greater absorbance of the background 13 emissions by white matter but the dark image cast through the tumor suggests less quenching by this tissue. Thus, the lower nonspe cific binding of the tumor can not be explained by an increase in tissue absorbance. The saturation curve and Scatchard plot for a rep resentative RG-2 glioma and a HK Walker 256 tu mor analyzed densitometrically are shown in Fig. 3. The binding of eH]PK11195 to tumors and normal brain was reversible and saturable and appeared to involve a single class of binding sites. Kd and Bmax values determined densitometrically for tumor and normal cortex are shown in Table 1. Although Kd values of RG-2 tumors and cortex were similar, Bmax was 20-fold higher in tumor compared to cor tex. The binding affinity was similar between tumor and normal cortex for both RG-2 and Walker 256 tumors. In contrast, the binding affinity was signif icantly higher (p < 0.05) and Bmax lower (p < 0.05) in HK Walker 256 tumors either in the brain or subcutaneously compared to RG-2 or LK Walker 256 tumors. Kd and Bmax images produced by digital image analysis of a single saturation experiment are shown J Cereb Blood Flow Metab, Vol.
10,
No. 4,
1990
582
K. IKEZAKI ET AL. A
FIG. 1. Digitized a u t o r a d i o g r a m of
[3H]PK11195 binding to a HK Walker 256 tumor (A) and corresponding histological section stained with thionine (8). A close correlation between areas of verified tu mor and binding was observed.
8
A
FIG. 2. (A) Autoradiographic image cast by a tritium-impregnated methacrylated back ground overlayed with a 10 I-l-m section of brain/tumor tissue. (8) Dark-field photograph of an adjacent section highlighting the tumor, white matter, and gray matter regions of the specimen.
8
J Cereb Blood Flow Metab, Vol. 10, No.4, 1990
BENZODIAZEPINE BINDING IN BRAIN TUMOR
A
600� :2
c
500
300
�
0'
E �
200
400
100
0
E
ILiJ UJ Il LL
�, 0 Z =:J
0 OJ
•
300
o�--�--��--�--�__� o 10 3H-PK11195(nM)
200
Added
•
100 • 0
0
100
200
300
FIG. 3. Scatchard plots and satu
400
500
BOUND(fmol/mg tissue)
ration curves of [3H]PK11195 binding to a HK Walker 256 tumor (A) and a RG-2 tumor (B). Sec tions were incubated for 60 min at O°C w i t h 0.2 5 to 10 nM [3H]PK11195 in the presence or absence of 10 11M PK11195. Auto radiograms were analyzed densi tometrically. Each point repre sents the mean of triplicate sec tions.
600
B
500
900 800
:2 ,
c
" 0'
� ·2
400
E
� 0
E
583
•
!
•
I §?
�
300
700 600 500 400 300 200 100 2
w w Il LL
200
� 0 z ::::J 0 OJ
100
3
4 5 6 7 3H_PK 11195 (nM)
8
9
10
0 0
200
400
600
BOUND
800
1000
1200
(fmol/mg tissue)
TABLE 1.
Binding constants offH1PKll195 Bmax (fmollmg of tissue)
Tumor RG-2 LKW256 Brain Subcutaneous HKW256 Brain Subcutaneous
Tumor
Normal
Tumor
Normal
2.29
±
0.30
1.80
±
0.14
1,027.3
±
78.7Q
52.0
±
9.3
2.17 2.24
±
0.23 0.21
2.41
±
1.23
924.3 845.7
±
183.7Q 61.7
79.8
±
5.3
1.27 1.64
±
0.27 0.30
1.91
549.2 564.5
±
49.5Q 205.5
52.0
±
19.0
±
±
±
0.68
±
±
Scatchard analysis of [3H1PKll195 binding to normal cortex, RG-2 gliomas, and LK and HK Walker 256 metastatic tumors grown either subcutaneously or intracerebrally. Data were obtained from discrete measurement of three consecutive sections per rat and expressed as mean ± SD (n = 3 rats). Q p < 0.005.
J Cereb Blood Flow Metab, Vol.
10,
No.4,
1990
FIG. 4. Thionin-stained section (A), specific binding image (8), and visual display of Bmax (C) and Kd (0) in RG-2 tumor using digital image analysis. Kd and Bmax values within tumor closely correlated with values obtained densitometrically. Within normal tissues where specific binding was low, accurate values for Kd and Bmax could not be determined. Images of infiltrated tissue at tumor border zones were likewise felt not to represent true Kd or Bmax values. J Cereb Blood Flow Metab. Vol. 10. No.4. 1990
FIG. 4. Continued.
J Cereb Blood Flow Metab, Vol. 10, No. 4, 1990
586
K. IKEZAKI ET AL.
in Fig. 4. The image of Bmax (Fig. 4C) is composed of the x intercept values of the linear regression of bound vs. bound/free. Within the tumor, the Bmax pixel values average 1,044.2 fmollmg of tissue, which was consistent with the Bmax values obtained densitometrically (1,055.6 fmollmg of tissue). The Kd image shown in Fig. 4D is a display of the Eadie Hofstee slope. The average pixel Kd value in tumor of 2.41 nM was also in close agreement with Kd values determined densitometrically (2.39 nM). A visual survey of Kd and Bmax images revealed ho mogeneity within RG-2 tumors, which was not readily appreciated without image analysis. Although consistent Kd and Bmax values for pe ripheral benzodiazepine binding within tumors could be determined by image analysis, images of these parameters could not be generated for normal tissue. Low specific binding by peripheral ligands in normal tissue resulted in low signal-to-noise ratios. This noise prevented accurate pixel-by-pixel trans formations of saturation data and led to an inability to generate reliable Eadie-Hofstee slopes or inter cepts. High specific binding appears to be essential to the generation of Kd and Bmax images with cur rent technology. Comparison of Walker 256 tumors grown in brain or subcutaneously revealed no significant differ ences in either Kd or Bmax (Table 1). DISCUSSION An understanding of peripheral benzodiazepine receptor binding in brain tumors may be important in improving our ability to diagnose and manage these lesions. Unlike the "central" binding site, which is located on neurons, closely linked to -y-aminobutyric acid (GABA) receptors, and modu lates the GABA-regulated anion channel (Wastek et al., 1978; Tallman et al. , 1980; Haefely et al. , 1982; Tallman and Gallager, 1985), the "peripheral" re ceptor is sparse in normal nervous tissue, but prom inent in many other tissues such as the kidney (Braestrup and Squire, 1977; Marangos et al., 1982, De Souza et al. , 1985). Binding by peripheral ligands is increased significantly in experimental brain tumors (Starosta-Rubinstein et al. , 1987; Black et al. , 1989) and brain tumors in humans (Black et al. , 1990). Our initial studies suggest that increased binding in glial tumors in humans may be correlated with increased tumor malignancy. Posi tron-labeled peripheral benzodiazepine ligands could therefore allow better delineation of tumor borders in the brain and the identification of subtle areas of tumor infiltration with positron emission tomography (PET). J Cereb Blood Flow Metab, Vol. 10, No.4, 1990
The biological role of these receptors remain un known. Curran and Morgan (1985) showed that spe cific induction of c-fos messenger RNA and protein by nerve growth factor is enhanced more than 100fold in the presence of peripherally active benzodi azepines, suggesting an action of these receptors at the level of gene expression. Peripheral-type ben zodiazepines will also inhibit the proliferation of thymoma cells in vitro (Wang et al., 1984a), en hance melanogenesis in Blb/C3 melanoma cells (Matthew et al. , 1981), induce the synthesis of he moglobin in Friend erythroleukemia cells (Wang et al. , 1984b), and stimulate phospholipid methylation in C6 astrocytoma cells (Strittmatter et al. , 1979). These effects are observed, however, at micromo lar concentrations while the affinity for these mol ecules at the receptor occurs at nanomolar levels. In contrast, PK11195 will stimulate cell growth of C6 glioma and DNA synthesis of mouse 3T3 cells (Ikezaki and Black, 1990), induce the enhancement of the respiratory burst of macrophages at nanomo lar concentrations (Zavala and Lenfant, 1987), and stimulate human mononcyte chemotaxis (Ruff et al., 1985). Why peripheral benzodiazepine receptors are ex pressed less in normal brain tissue is unclear. One possibility is that the genes encoding for the expres sion of these receptors are inhibited by factors re leased by the brain itself or that systemic factors stimulate the expression of these receptors. If either of these hypotheses were true, brain tumors would be predicted to have different binding constants when grown intracerebrally than when grown out side the brain. Our findings, however, are that bind ing constants for Walker 256 tumors are not influ enced by growth in an intracerebral location. Why background "nonspecific" binding is less in tumors compared to cortex remains unclear. Our data, however, rule out quenching of tritium emissions as a possible explanation of this finding. We could not image "peripheral" receptor pa rameters by digital image analysis in normal brain because specific binding was very low, which re sulted in low signal-to-noise ratios between total and nonspecific binding images. Binding sites for peripheral benzodiazepine ligands could be imaged in experimental tumors where specific binding was high. For RG-2 gliomas, both Kd and Bmax appear to be relatively homogeneous within the tumor but analysis of binding parameters at the tumor-normal brain interface should be viewed with caution due to lower signal-to-noise ratios. Despite these limi tations, digital imaging technology works well when specific binding is moderate or high. The use of positron-labeled peripheral benzodiaz-
BENZODIAZEPINE BINDING IN BRAIN TUMOR epine receptor ligands should allow brain tumors to be specifically imaged in humans and may give bet ter definition of tumor borders than current imaging technology. More importantly, an understanding of the role of peripheral benzodiazepine receptors in brain tumors may provide new and important clues to tumor biology. Acknowledgment: This work was supported by a NIH FIRST award (l-R29 NS 26523-01), a Biomedical Re search Support Grant, and a Robert Wood Johnson Foun dation Grant.
REFERENCES Black KL, Ikezaki K, Arthur WT (1989) Imaging of brain tumors using peripheral benzodiazepine receptor ligands. J Neuro surg 71:113-118 Black KL, Ikezaki K, Santori EM, Becker DP, Vinters HV (1990) Specific high-affinity binding of peripheral benzodiaz epine receptor ligands to brain tumors in rat and man. Can cer 65:93-97 Braestrup C, Squires RF (1977) Benzodiazepine receptors in rat brain. Nature (Lond) 266:732-734 Curran T, Morgan 11 (1985) Superinduction of c-fos by nerve growth factor in the presence of peripherally active benzo diazepines. Science 229:1265-1268 De Souza EB, Anholt RRH, Murphy KMM, Snyder SH, Kuhar MJ (1985) Peripheral-type benzodiazepine receptors in en docrine organs: autoradiographic localization in rat pitu itary, adrenal, and testis. Endocrinology 116:567-573 Earle WR (1935) A study of the Walker rat mammary carcinoma 256, in vivo and in vitro. Am J Cancer 24:566-612 Haefely W, Pole P, Pieri L, Schaffner R, Laurent JP (1982) From molecular biology to clinical practice. In: The Benzodiaz epines (Costa E, ed), New York, Raven Press, pp 21-66 Ikezaki K, Black KL (1990) Stimulation of cell growth and DNA synthesis by peripheral benzodiazepine. Cancer Lett 49: 115-120 Kuhar MJ, Unnerstall JR (1985) Quantitative receptor mapping by autoradiography: some current technical problems. Trends Neurosci 8:49-53 Marangos PJ, Patel J, Boulenger JP, Clark-Rosenberg R (1982)
587
Characterization of peripheral-type benzodiazepine binding sites in brain using [3H1RO 5-4864. Mol PharmacoI22:26-32 Matthew E, Laskin JD, Zimmerman EA, Weinstein IB, Hsu KC, Engelhardt DL (1981) Benzodiazepines have high-affinity binding sites and induce melanogenesis in B16/C3 melanoma cells. Proc Natl Acad Sci USA 78:3935-3939 Ruff MR, Pert CB, Weber RJ, Wahl SM, Paul SM (1985) Ben zodiazepines: receptor-mediated chemotaxis of human monocytes. Science 229:1281-1283 Schoemaker H, Boles RB, Horst WD, Yamamura HI (1983) Spe cific high affinity binding sites for [3H1RO 5-4864 in rat brain and kidney. J Pharmacol Exp Ther 225:61-69 Starosta-Rubinstein S, Ciliax BJ, Penney JB, McKeever P, Young AB (1987) Imaging of a glioma using peripheral ben zodiazepine receptor ligands. Proc Natl Acad Sci USA 84:891-895 Strittmatter WJ, Hirata F, Axelrod J, Mallorga P, Tallman JF, Henneberry RC (1979) Benzodiazepine and �-adrenergic re ceptor ligands independently stimulate phospholipid meth ylation. Nature (Lond) 282:857-859 Tallman JF, Gallager DW (1985) The GABA-ergic system: a lo cus of benzodiazepine action. Annu Rev Neurosci 8:21-44 Tallman JF, Thomas JW, Gallager DW (1978) GABA-ergic mod ulation of benzodiazepine binding site sensitivity. Nature (Lond) 274:383-385 Tallman JF, Paul SM, Skolnick P, Gallagher DW (1980) Recep tors for the age of anxiety: pharmacology of the benzodiaz epines. Science 207:274-281 Toga AW, Santori EM, Samaie M (1986) Regional distribution of flunitrazepam binding constants: visualizing Kd and Bmax by digital image analysis. J Neurosci 6:2747-2756 Wang JKT, Morgan 11, Spector S (1984a) Benzodiazepines that bind at peripheral sites inhibit cell proliferation. Proc Natl Acad Sci USA 81:753-756 Wang JKT, Morgan 11, Spector S (1984b) Differentiation of Friend erythroleukemia cells induced by benzodiazepines. Proc Natl Acad Sci USA 81:3770-3772 Wastek GJ, Speth RC, Reisine TD, Yamamura HI (1978) The effect of -y-aminobutyric acid on 3H-flunitrazepam binding in rat brain. Eur J Pharmacol 50:445-447 Wechsler W, Ramadan MA, Geister A (1972) Isogenic transplan tation of ethylnitrosourea-induced tumors and peripheral nervous system in two different inbred rat strains. Natur wissenschaften 59:474-478 Zavala F, Lenfant M (1987) Peripheral benzodiazepines enhance the respiratory burst of macrophage-like P338Dl cells stim ulated by arachidonic acid. Int J ImmunopharmacoI9:269274
J
Cereb Blood Flow Metab,
Vol. 10, No.4, 1990
10:580--587 © 1990 Raven Press, Ltd., New York
Imaging Peripheral Benzodiazepine Receptors in Brain Tumors in Rats: In Vitro Binding Characteristics
Kiyonobu Ikezaki, Keith L. Black, * Arthur W. Toga, *Emily M. Santori, Donald P. Becker, and Mayumi L. Smith Division of Neurosurgery and Brain Research Institute and *Laboratory of Neuro Imaging and Department of Neurology, University of California, Los Angeles, School of Medicine, Los Angeles, California, U.S.A.
Summary: Peripheral benzodiazepine binding constants for transplanted RG-2 gliomas and HK and LK Walker 256 tumors (metastatic breast carcinoma) were deter mined in Wistar rats using autoradiography. In addition, Kd and Bmax parameters for peripheral benzodiazepine receptors on RG-2 tumors were directly visualized using digital image analysis of autoradiograms. High specific binding of [3H]PKI1195, a selective peripheral benzodi azepine ligand, had excellent topographical correlation to areas of histologically verified tumor. Scatchard analysis suggested a single class of peripheral binding sites with similar binding affinities in RG-2 and LK Walker 256 tu mors and normal cortex. Bmax was 20-fold greater in glial tumors and 11.6- and lO.6-fold greater in LK and HK Walker 256 tumors, respectively, compared to normal
cortex. The location of metastatic tumors, either intra cerebrally or subcutaneously, did not effect their Kd or Bmax values. Kd and Bmax values for RG-2 tumors were similar whether determined densitometrically or by direct visualization with image analysis. Binding parameters within normal brain were difficult to visualize by image analysis due to the low level of specific binding. The abil ity to label specifically intracerebral tumor cells and to characterize the binding parameters shown in this study suggest that peripheral benzodiazepine receptor ligands could be utilized by PET to analyze directly a variety of tumors in humans. Key Words: Brain tumor-Benzo diazepine receptor-Autoradiography-Digital image analysis-PK 11195.
The binding kinetics for peripheral benzodiaze pine receptors in normal brain structures and a va riety of intracranial tumors in humans have recently been demonstrated (Black et al. , 1990). There is little specific binding of peripheral-type ligands to normal brain structures with the exception of the choroid plexus, ependyma cells, or the pineal gland in rats (Braestrup and Squires, 1977; Marangos et al., 1982). Specific binding is high in intracerebral tumors and higher binding is correlated with in creasing malignancy in gliomas in humans (Black et al., 1990). Although the biologic role of these recep-
tors remains unclear, they are normally expressed in systemic tissues, such as the kidney (De Souza et al. , 1985; Schoemaker et al. , 1983). In this report, peripheral benzodiazepine receptor binding in three experimental brain tumors is characterized. In ad dition, the possibility that tumor growth within the brain might influence peripheral benzodiazepine binding constants is investigated. We also directly visualized Kd and Bmax values in RG-2 tumors using digital image analysis. This tech nique, unlike previous ligand binding methods, al lows for the visual display of specific binding, and Bmax and Kd constants obtained from an Eadie Hofstee analysis of the saturation data for an entire tissue section (Toga et al. , 1986). The relationships between binding parameters and anatomic (spatial) data or histological changes can thus be more readily analyzed and quantified. This report illus trates the utility and limitations of this technology in visually displaying receptor binding constants in brain tumors.
Received May 29, 1989; revised August 31, 1989; accepted September 11, 1989. Address correspondence and reprint requests to Dr. K. L. Black at Division of Neurosurgery, UCLA School of Medicine, Los Angeles, CA 90024, U.S.A. Abbreviations used: DMEM, Dulbecco's modified eagle me dium; GABA, -y-aminobutyric acid; PET, positron emission to mography; PK11195, 1-(2-chlorophenyl)-N-methyl-N-(1-methyl propyl)-3-isoquinoline carboxamide.
580
BENZODIAZEPINE BINDING IN BRAIN TUMOR MATERIALS AND METHODS eH]PK l l195 (75.2 Ci/mmol) was obtained from New England Nuclear. PK l l195 was kindly donated by Dr. G. Le Fur of Phamuka Laboratories. Ethylnitrosourea induced RG-2 rat glioma cell lines (Wechsler et aI., 1972) were obtained from Dr. D. Bigner (Duke University) and maintained in a monolayer culture in Dulbecco's modified eagle medium (DMEM) with 10% calf serum (Gibco). LK and HK Walker 256 metastatic breast carcinomas (Earle, 1935) were donated by Dr. R. Blasberg (NIH, Bethesda, MD, U.S.A.) and maintained subcutaneously in Wistar rats (Charles River). Wistar female rats (80-100 g) were anesthetized with ketamine (50 mg/kg) and xylazine (0.8 mglkg) i.m. Expo nentially growing RG-2 glioma cells (5 x 104) in 10 IJ-I of F lO serum free medium (Gibco) were implanted intrace rebrally into the left hemisphere, 4 mm lateral to the mid line and just behind the coronal suture. For Walker 256 tumor studies, 1 mm3 of HK or LK tumor was implanted intracerebrally into the left hemisphere. In some studies, either HK or LK tumors were simultaneously trans planted subcutaneously and intracerebrally in the same rat. Mter 14 days for RG-2 tumors and 11 days for Walker 256 tumors, rats were decapitated and their brains were removed and frozen in isopentane chilled on dry ice. Sixty continuous 20 IJ-m thick sections of brain were sliced using a cryostat and thaw-mounted on gelatin coated coverglasses, kept in a freezer overnight, and dried in desiccator before binding studies. Sections were subsequently prewashed twice for 2 min in an ice-cold binding buffer (50 roM Tris-HCl, pH 7.4) and dried by cool air. They were then incubated in binding buffer with various concentrations of eH]PKI1195 (0.25-10 nM) in the absence or presence of 10 IJ-M PK11195 for 60 min at O°C. The sections were again washed twice for 2 min in the ice-cold buffer, dried, and then coexposed with tissue calibrated 3H standards (Amersham) to Hiperfilm-3H (Amersham) for 3 to 7 days. Aliquots of binding buffer were taken to calculate the amount of free radioactive ligand after binding. After exposure, sections were stained with thionin or hematoxylin and eosin for histo logical verification and localization of the tumor. The re sultant autoradiograms and histological sections were an alyzed with a digital image analysis system using a mod ification of our previously described method (Toga et aI., 1986). Scatchard analysis was performed on data collected by discrete measurements of individual autoradiographic im ages. Nonspecific binding values were subtracted from the total binding values of adjacent sections. Specific binding data obtained from three sections were then av eraged. For image analysis, slightly out of focus digital images of total and nonspecific binding were obtained at two concentrations (2.5 and 10 nM). We have found that optical blurring prior to digitization reduces the graininess or noise of autoradiographic data and facilitates pixel by-pixel image manipUlations. Specific binding images were obtained by subtracting nonspecific images from to tal binding images, pixel by pixel. Specific binding images at two concentrations of [3H]PKII195 were then aligned to perform the Eadie-Hofstee analysis. Pixel-by-pixel re gression of bound/free vs. bound images yielded Kd (slope) and Bmax (intercept) images. To examine regional tissue quenching of 13 emissions from tritium-labeled PK l1195, a modified version of the
581
tissue overlay technique described by Kuhar and Unner stall (1985) was employed. This technique only provides a qualitative assessment of differential tissue quenching since the quantitative relationship between quenching in labeled tissue sections and quenching observed with tis sue overlays has not been established. Ten micron sec tions of unlabeled tumor were thaw-mounted and dried onto a block of tritium-impregnated methacrylate (do nated by American Radiolabeled Chemicals, St. Louis, MO, U.S.A.). LKB film was exposed to these tissue overlays for 3 days. In addition, dark-field photographs of adjacent sections heat dried onto coverslips and im mersed in water were obtained to help delineate the white matter and tumor portions of the section. Statistical anal ysis was performed using the Student's t test.
RESULTS In all three tumors, areas of increased binding of peripheral benzodiazepine ligands on autoradio grams had high topographical correlation with areas of histologically verified tumor. There was little binding to necrotic areas within tumor or normal brain tissue with the exception of the choroid plexus, pineal gland, and ependymal structures, which showed moderate binding (Fig. O. Sections incubated in the presence of excess PK11195 (non specific binding) appeared to show higher back ground binding in the cerebral cortex and the deep nuclei than in the tumor and white matter. One rea son for this heterogeneous pattern of nonspecific binding could be the problem of differential tritium quenching. The tissue overlay autoradiogram shown in Fig. 2 does reveal a greater absorbance of the background 13 emissions by white matter but the dark image cast through the tumor suggests less quenching by this tissue. Thus, the lower nonspe cific binding of the tumor can not be explained by an increase in tissue absorbance. The saturation curve and Scatchard plot for a rep resentative RG-2 glioma and a HK Walker 256 tu mor analyzed densitometrically are shown in Fig. 3. The binding of eH]PK11195 to tumors and normal brain was reversible and saturable and appeared to involve a single class of binding sites. Kd and Bmax values determined densitometrically for tumor and normal cortex are shown in Table 1. Although Kd values of RG-2 tumors and cortex were similar, Bmax was 20-fold higher in tumor compared to cor tex. The binding affinity was similar between tumor and normal cortex for both RG-2 and Walker 256 tumors. In contrast, the binding affinity was signif icantly higher (p < 0.05) and Bmax lower (p < 0.05) in HK Walker 256 tumors either in the brain or subcutaneously compared to RG-2 or LK Walker 256 tumors. Kd and Bmax images produced by digital image analysis of a single saturation experiment are shown J Cereb Blood Flow Metab, Vol.
10,
No. 4,
1990
582
K. IKEZAKI ET AL. A
FIG. 1. Digitized a u t o r a d i o g r a m of
[3H]PK11195 binding to a HK Walker 256 tumor (A) and corresponding histological section stained with thionine (8). A close correlation between areas of verified tu mor and binding was observed.
8
A
FIG. 2. (A) Autoradiographic image cast by a tritium-impregnated methacrylated back ground overlayed with a 10 I-l-m section of brain/tumor tissue. (8) Dark-field photograph of an adjacent section highlighting the tumor, white matter, and gray matter regions of the specimen.
8
J Cereb Blood Flow Metab, Vol. 10, No.4, 1990
BENZODIAZEPINE BINDING IN BRAIN TUMOR
A
600� :2
c
500
300
�
0'
E �
200
400
100
0
E
ILiJ UJ Il LL
�, 0 Z =:J
0 OJ
•
300
o�--�--��--�--�__� o 10 3H-PK11195(nM)
200
Added
•
100 • 0
0
100
200
300
FIG. 3. Scatchard plots and satu
400
500
BOUND(fmol/mg tissue)
ration curves of [3H]PK11195 binding to a HK Walker 256 tumor (A) and a RG-2 tumor (B). Sec tions were incubated for 60 min at O°C w i t h 0.2 5 to 10 nM [3H]PK11195 in the presence or absence of 10 11M PK11195. Auto radiograms were analyzed densi tometrically. Each point repre sents the mean of triplicate sec tions.
600
B
500
900 800
:2 ,
c
" 0'
� ·2
400
E
� 0
E
583
•
!
•
I §?
�
300
700 600 500 400 300 200 100 2
w w Il LL
200
� 0 z ::::J 0 OJ
100
3
4 5 6 7 3H_PK 11195 (nM)
8
9
10
0 0
200
400
600
BOUND
800
1000
1200
(fmol/mg tissue)
TABLE 1.
Binding constants offH1PKll195 Bmax (fmollmg of tissue)
Tumor RG-2 LKW256 Brain Subcutaneous HKW256 Brain Subcutaneous
Tumor
Normal
Tumor
Normal
2.29
±
0.30
1.80
±
0.14
1,027.3
±
78.7Q
52.0
±
9.3
2.17 2.24
±
0.23 0.21
2.41
±
1.23
924.3 845.7
±
183.7Q 61.7
79.8
±
5.3
1.27 1.64
±
0.27 0.30
1.91
549.2 564.5
±
49.5Q 205.5
52.0
±
19.0
±
±
±
0.68
±
±
Scatchard analysis of [3H1PKll195 binding to normal cortex, RG-2 gliomas, and LK and HK Walker 256 metastatic tumors grown either subcutaneously or intracerebrally. Data were obtained from discrete measurement of three consecutive sections per rat and expressed as mean ± SD (n = 3 rats). Q p < 0.005.
J Cereb Blood Flow Metab, Vol.
10,
No.4,
1990
FIG. 4. Thionin-stained section (A), specific binding image (8), and visual display of Bmax (C) and Kd (0) in RG-2 tumor using digital image analysis. Kd and Bmax values within tumor closely correlated with values obtained densitometrically. Within normal tissues where specific binding was low, accurate values for Kd and Bmax could not be determined. Images of infiltrated tissue at tumor border zones were likewise felt not to represent true Kd or Bmax values. J Cereb Blood Flow Metab. Vol. 10. No.4. 1990
FIG. 4. Continued.
J Cereb Blood Flow Metab, Vol. 10, No. 4, 1990
586
K. IKEZAKI ET AL.
in Fig. 4. The image of Bmax (Fig. 4C) is composed of the x intercept values of the linear regression of bound vs. bound/free. Within the tumor, the Bmax pixel values average 1,044.2 fmollmg of tissue, which was consistent with the Bmax values obtained densitometrically (1,055.6 fmollmg of tissue). The Kd image shown in Fig. 4D is a display of the Eadie Hofstee slope. The average pixel Kd value in tumor of 2.41 nM was also in close agreement with Kd values determined densitometrically (2.39 nM). A visual survey of Kd and Bmax images revealed ho mogeneity within RG-2 tumors, which was not readily appreciated without image analysis. Although consistent Kd and Bmax values for pe ripheral benzodiazepine binding within tumors could be determined by image analysis, images of these parameters could not be generated for normal tissue. Low specific binding by peripheral ligands in normal tissue resulted in low signal-to-noise ratios. This noise prevented accurate pixel-by-pixel trans formations of saturation data and led to an inability to generate reliable Eadie-Hofstee slopes or inter cepts. High specific binding appears to be essential to the generation of Kd and Bmax images with cur rent technology. Comparison of Walker 256 tumors grown in brain or subcutaneously revealed no significant differ ences in either Kd or Bmax (Table 1). DISCUSSION An understanding of peripheral benzodiazepine receptor binding in brain tumors may be important in improving our ability to diagnose and manage these lesions. Unlike the "central" binding site, which is located on neurons, closely linked to -y-aminobutyric acid (GABA) receptors, and modu lates the GABA-regulated anion channel (Wastek et al., 1978; Tallman et al. , 1980; Haefely et al. , 1982; Tallman and Gallager, 1985), the "peripheral" re ceptor is sparse in normal nervous tissue, but prom inent in many other tissues such as the kidney (Braestrup and Squire, 1977; Marangos et al., 1982, De Souza et al. , 1985). Binding by peripheral ligands is increased significantly in experimental brain tumors (Starosta-Rubinstein et al. , 1987; Black et al. , 1989) and brain tumors in humans (Black et al. , 1990). Our initial studies suggest that increased binding in glial tumors in humans may be correlated with increased tumor malignancy. Posi tron-labeled peripheral benzodiazepine ligands could therefore allow better delineation of tumor borders in the brain and the identification of subtle areas of tumor infiltration with positron emission tomography (PET). J Cereb Blood Flow Metab, Vol. 10, No.4, 1990
The biological role of these receptors remain un known. Curran and Morgan (1985) showed that spe cific induction of c-fos messenger RNA and protein by nerve growth factor is enhanced more than 100fold in the presence of peripherally active benzodi azepines, suggesting an action of these receptors at the level of gene expression. Peripheral-type ben zodiazepines will also inhibit the proliferation of thymoma cells in vitro (Wang et al., 1984a), en hance melanogenesis in Blb/C3 melanoma cells (Matthew et al. , 1981), induce the synthesis of he moglobin in Friend erythroleukemia cells (Wang et al. , 1984b), and stimulate phospholipid methylation in C6 astrocytoma cells (Strittmatter et al. , 1979). These effects are observed, however, at micromo lar concentrations while the affinity for these mol ecules at the receptor occurs at nanomolar levels. In contrast, PK11195 will stimulate cell growth of C6 glioma and DNA synthesis of mouse 3T3 cells (Ikezaki and Black, 1990), induce the enhancement of the respiratory burst of macrophages at nanomo lar concentrations (Zavala and Lenfant, 1987), and stimulate human mononcyte chemotaxis (Ruff et al., 1985). Why peripheral benzodiazepine receptors are ex pressed less in normal brain tissue is unclear. One possibility is that the genes encoding for the expres sion of these receptors are inhibited by factors re leased by the brain itself or that systemic factors stimulate the expression of these receptors. If either of these hypotheses were true, brain tumors would be predicted to have different binding constants when grown intracerebrally than when grown out side the brain. Our findings, however, are that bind ing constants for Walker 256 tumors are not influ enced by growth in an intracerebral location. Why background "nonspecific" binding is less in tumors compared to cortex remains unclear. Our data, however, rule out quenching of tritium emissions as a possible explanation of this finding. We could not image "peripheral" receptor pa rameters by digital image analysis in normal brain because specific binding was very low, which re sulted in low signal-to-noise ratios between total and nonspecific binding images. Binding sites for peripheral benzodiazepine ligands could be imaged in experimental tumors where specific binding was high. For RG-2 gliomas, both Kd and Bmax appear to be relatively homogeneous within the tumor but analysis of binding parameters at the tumor-normal brain interface should be viewed with caution due to lower signal-to-noise ratios. Despite these limi tations, digital imaging technology works well when specific binding is moderate or high. The use of positron-labeled peripheral benzodiaz-
BENZODIAZEPINE BINDING IN BRAIN TUMOR epine receptor ligands should allow brain tumors to be specifically imaged in humans and may give bet ter definition of tumor borders than current imaging technology. More importantly, an understanding of the role of peripheral benzodiazepine receptors in brain tumors may provide new and important clues to tumor biology. Acknowledgment: This work was supported by a NIH FIRST award (l-R29 NS 26523-01), a Biomedical Re search Support Grant, and a Robert Wood Johnson Foun dation Grant.
REFERENCES Black KL, Ikezaki K, Arthur WT (1989) Imaging of brain tumors using peripheral benzodiazepine receptor ligands. J Neuro surg 71:113-118 Black KL, Ikezaki K, Santori EM, Becker DP, Vinters HV (1990) Specific high-affinity binding of peripheral benzodiaz epine receptor ligands to brain tumors in rat and man. Can cer 65:93-97 Braestrup C, Squires RF (1977) Benzodiazepine receptors in rat brain. Nature (Lond) 266:732-734 Curran T, Morgan 11 (1985) Superinduction of c-fos by nerve growth factor in the presence of peripherally active benzo diazepines. Science 229:1265-1268 De Souza EB, Anholt RRH, Murphy KMM, Snyder SH, Kuhar MJ (1985) Peripheral-type benzodiazepine receptors in en docrine organs: autoradiographic localization in rat pitu itary, adrenal, and testis. Endocrinology 116:567-573 Earle WR (1935) A study of the Walker rat mammary carcinoma 256, in vivo and in vitro. Am J Cancer 24:566-612 Haefely W, Pole P, Pieri L, Schaffner R, Laurent JP (1982) From molecular biology to clinical practice. In: The Benzodiaz epines (Costa E, ed), New York, Raven Press, pp 21-66 Ikezaki K, Black KL (1990) Stimulation of cell growth and DNA synthesis by peripheral benzodiazepine. Cancer Lett 49: 115-120 Kuhar MJ, Unnerstall JR (1985) Quantitative receptor mapping by autoradiography: some current technical problems. Trends Neurosci 8:49-53 Marangos PJ, Patel J, Boulenger JP, Clark-Rosenberg R (1982)
587
Characterization of peripheral-type benzodiazepine binding sites in brain using [3H1RO 5-4864. Mol PharmacoI22:26-32 Matthew E, Laskin JD, Zimmerman EA, Weinstein IB, Hsu KC, Engelhardt DL (1981) Benzodiazepines have high-affinity binding sites and induce melanogenesis in B16/C3 melanoma cells. Proc Natl Acad Sci USA 78:3935-3939 Ruff MR, Pert CB, Weber RJ, Wahl SM, Paul SM (1985) Ben zodiazepines: receptor-mediated chemotaxis of human monocytes. Science 229:1281-1283 Schoemaker H, Boles RB, Horst WD, Yamamura HI (1983) Spe cific high affinity binding sites for [3H1RO 5-4864 in rat brain and kidney. J Pharmacol Exp Ther 225:61-69 Starosta-Rubinstein S, Ciliax BJ, Penney JB, McKeever P, Young AB (1987) Imaging of a glioma using peripheral ben zodiazepine receptor ligands. Proc Natl Acad Sci USA 84:891-895 Strittmatter WJ, Hirata F, Axelrod J, Mallorga P, Tallman JF, Henneberry RC (1979) Benzodiazepine and �-adrenergic re ceptor ligands independently stimulate phospholipid meth ylation. Nature (Lond) 282:857-859 Tallman JF, Gallager DW (1985) The GABA-ergic system: a lo cus of benzodiazepine action. Annu Rev Neurosci 8:21-44 Tallman JF, Thomas JW, Gallager DW (1978) GABA-ergic mod ulation of benzodiazepine binding site sensitivity. Nature (Lond) 274:383-385 Tallman JF, Paul SM, Skolnick P, Gallagher DW (1980) Recep tors for the age of anxiety: pharmacology of the benzodiaz epines. Science 207:274-281 Toga AW, Santori EM, Samaie M (1986) Regional distribution of flunitrazepam binding constants: visualizing Kd and Bmax by digital image analysis. J Neurosci 6:2747-2756 Wang JKT, Morgan 11, Spector S (1984a) Benzodiazepines that bind at peripheral sites inhibit cell proliferation. Proc Natl Acad Sci USA 81:753-756 Wang JKT, Morgan 11, Spector S (1984b) Differentiation of Friend erythroleukemia cells induced by benzodiazepines. Proc Natl Acad Sci USA 81:3770-3772 Wastek GJ, Speth RC, Reisine TD, Yamamura HI (1978) The effect of -y-aminobutyric acid on 3H-flunitrazepam binding in rat brain. Eur J Pharmacol 50:445-447 Wechsler W, Ramadan MA, Geister A (1972) Isogenic transplan tation of ethylnitrosourea-induced tumors and peripheral nervous system in two different inbred rat strains. Natur wissenschaften 59:474-478 Zavala F, Lenfant M (1987) Peripheral benzodiazepines enhance the respiratory burst of macrophage-like P338Dl cells stim ulated by arachidonic acid. Int J ImmunopharmacoI9:269274
J
Cereb Blood Flow Metab,
Vol. 10, No.4, 1990
