©Copyright 1992 by The Humana Press Inc. All rights of any nature whatsoever reserved. 1044-7393;92i1702-0169 502.60
.
.
t ... s :- I sitiïsi .w-
JOSEPH R. MOSKAL,'* MARK SINNETT, 2 PAUL L. KORNBLITH, 2 PATRICK LASALA, 3 DANIEL A. LEVINE, 4 THOMAS S. PARKER, 4 AND HARRY LANDER 4 'The Chicago Institute for Neurosurgery and Neuroresearch, 428 W. Deming Pl. Chicago, IL 60614; 2 Department of Neurosurgery, Montefiore Medical Center, Bronx, NY; 3 Department of Surgery, St. Lukes— Roosevelt Hospital, New York, NY; 4 Rogosin Institute, New York, NY
Received February 4, 1992; Accepted April 3, 1992
Experiments were performed using an established human glioblastoma cell line to determine the effect of lipoproteins on regulating their growth. It was found that synthetic and natural human high density lipoproteins (HDL) were effective in inhibiting tumor cell growth in a nontoxic, dose-dependent manner, and that the LD50 was 10-fold lower than that for normal rat astrocytes grown under identical conditions. In the presence of the antioxidant, glutathione, essentially all of the growth-inhibiting properties of HDL could be reversed suggesting that oxidized lipids from the HDL interacting with the plasma membranes of the glioblastoma cells were responsible for the growth-inhibiting effect observed. The markedly lower concentration of HDL required to inhibit glioblastoma cells in culture compared to normal astrocytes suggested that the mechanism of HDL-induced inhibition may be important for tumor growth in vivo. One possible mechanism under investigation is the possibility of HDL modulation of a membrane-associated, tumor-specific phosphatase. Index Entries: Lipoproteins: high density; glioblastoma growth regulation; receptor-mediated processes; regulation of cell-cycle by phosphatases; autocrine growth control; membrane-membrane interactions. *Author to whom all correspondence and reprint requests should be addressed. Molecular and Chemical Neuropathology
169
Vol. 17, 1992
170
Moskal et al.
Many of the structural, functional, and regulatory aspects of lipoproteins have been elucidated in recent years (Chapman, 1980; Bilheimer, 1988). Briefly, lipoproteins consist of chylomicron particles, very low density lipoproteins (VLDL), low density lipoproteins (LDL), ß-VLDL, and high density lipoproteins (HDL). VLDL, /3-VLDL, and LDL are associated with the transport of cholesterol away from the liver to the periphery, whereas HDL carry cholesterol from peripheral cells to the liver (see Fig. 1). There are a variety of proteins that are associated with lipoproteins. Perhaps the best characterized apolipoprotein is Apo E (Mahley and Innerarity, 1983; Havel et al., 1980). It is a 30-kDa, lipophilic protein that binds to two receptors found on both hepatic and nonhepatic cells (Hui et al., 1981; Mahley et al., 1981). Its function is to coordinate lipoprotein internalization by hepatocytes via speciic receptor-mediated processes and to help in the regulation of cholesterol homeostasis. Apo E is also involved in removing cholesterol from cells by facilitating its packaging into HDL (Gordon et al., 1983). Here we report our results based on the working hypothesis that lipoproteins play a key role in regulating the growth of brain tumors. This hypothesis emerged based on the following considerations. First, in a now classic study, Barclay and coworkers (1970) showed that men, women, and children with cancer had significantly lower HDL levels than controls. Moreover, they also showed a direct relationship between low levels of HDL in serum of normal subjects and a high family incidence of cancer. Second, in vivo and in vitro studies have shown that LDL itself is toxic to endothelial cells, whereas HDL is protective (Stein and Stein, 1976; Reckless et al., 1978; Vlodaysky et al., 1978; Tauber et al., 1980). Third, recent studies have demonstrated the presence of trypsinsensitive, high-affinity binding sites for HDL on brain capillary epithelial cells. Competition studies between HDL and apolipoprotein A-I suggested that apolipoprotein A-I was involved in HDL-receptor complex formation (Martin-Nizard et al., 1989). Interestingly, a number of highaffinity binding sites for HDL on a variety of cell types have been reported, but their physiological significance remains to be determined (Jurgens et al., 1989). Fourth, Boyles and coworkers (1989) have shown that during regeneration of rat peripheral nerve that the appropriate components and temporal sequencing of their expression are in place for a complete, functioning cholesterol transfer system necessary for nerve regeneration and remyelination. Finally, Eishourbagy et al. (1984) have shown that the brain was among the tissues with the highest concentration of apo E. More recently Boyles et al. (1985) demonstrated that apo E was present in all astrocytes but absent from neurons, oligodendroglia, and microglia. The data presented in the present report demonstrate that purified fractions of human HDL can inhibit the growth of an established Molecular and Chemical Neuropathology
Vol. 17, 1992
Lipoproteins Inhibit Glioma Growth
UJ.
W ;
O
Q
o_ VI 0
U.
0
_Z 4
W Ü
á Z
¢ q
Q Z
171
0 M
W J
C11
99% purity was needed, the product was redissolved in Tris 6M urea buffer and chromatographed on Sephacryl-S 200 as described by Brewer et al. (1986). Reconstituted apo-HDL: Apo HDL was reconstituted by detergent dialysis with the following modifications to increase scale. Egg lecithin (Matreya, Pleasant Gap, PA) (3.4 g) was dissolved in 205 mL of 108 mM sodium cholate. The clear solution was then brought to a final volume of 720 mL with Tris-EDTA buffer (31 mM cholate). This solution was dialyzed against T-E buffer (31 mM cholate). The preparation of apo HDL protein and reconstitution steps were as described by Matz and Jonas (1982). Removal of cholate was essentially by the method of Bonoma and Swaney (1988) adapted to the 2-4 L scale. The concentration of cholate in the final R-HDL preparation was < 100 µM when the phospholipid concentration is -20 mg/mL: a cholate to PC molar rate of -1:500. This level of cholate is not toxic in the neutral red cytotoxicity test. Synthetic Peptides: A peptide analogous to the concensus sequence of apo A-I was prepared by the solid-phase peptide synthesis core facility at the State University of New York at Stony Brook. The sequence, designated as peptide 18A by Anantharamaiah et al. (1985), was designed to form an amphipathic helix with polar and nonpolar faces of approximately equal surface area and is long enough to permit cooperative interactions between multiple and amphipathic helical domains. Purified peptide was reconstituted with phospholipid by the same method described earlier for apo-HDL.
Figure 1 depicts the effects of various lipoproteins on SNB-19 growth. Each lipoprotein fraction used in this study contained 0.85-1.0 mg of protein/mL. This study was performed in the presence of serum-containing
Molecular and Chemical Neuropathology
Vol. 17, 1992
Lipoproteins Inhibit Glioma Growth
175
medium. There was a significant inhibition of cell growth with human HDL without any significant (< 5%) uptake of Trypan blue. LDL inhibition of growth, however, was toxic. Interestingly, synthetic human apo Al (which contained the lecithin: cholesterol acyl transferase activator domain), cholesterol-free, reconstituted Al and All particles, and a fraction containing the apoproteins found in HDL without the lipid components also showed significant inhibition of growth of SNB-19 cells. These data demonstrated that HDL was capable of inhibiting glioblastoma cell growth in vitro, that LDL were toxic, and that these phenomena occurred in the reported concentration range reported for the modulation of endothelial cell growth in culture (Tauber et al., 1980). Furthermore, it appeared that HDL apoproteins could markedly inhibit cell growth suggesting that this process might be receptor-mediated. In the second set of experiments, seen in Fig. 2, SNB-19 cells were treated with a variety of HDL concentrations and compared to human glioblastoma cell lines of low tumorigenic capability. These experiments were performed in the absence of serum. First, it can be seen that SNB-19 cells, in the absence of serum, show 50% inhibition of growth at approximately 5 µg/ mL of HDL. The two low-tumor forming glioblastoma cell lines each showed a LD5o of approximately 15-20 µg/ mL of HDL. All three cell lines responded to HDL in a dose-dependent manner further suggesting a specific, possibly receptor-mediated mechanism for the growth inhibition of HDL. Moreover, the shift in LD50 in the low tumorforming glioblastoma cell lines suggested a relationship between tumorigenicity and the concentration of HDL necessary to inhibit growth. To investigate this phenomenon further, HDL concentration curves were performed on SNB-19 cells and primary cultures of normal rat astrocytes. These experiments were performed in the presence of serum since normal rat astrocyte growth is markedly reduced in the absence of serum. This was not the case with the glioblastoma cultures. As can be seen in Fig. 3, the ability of HDL to inhibit normal rat astrocyte growth was dramatically different than SNB-19 glioblastoma cells. The LD5o for SNB-19 cells, grown in the presence of serum was approximately 40-50 ttg/mL compared to an LD50 of approximately 500 µglmL for the normal astrocytes in culture. Figure 4 shows the LD50 for SNB-19 cells grown in serum-free conditions compared to SNB-19 cells growth in the presence of serum compared to normal rat astrocytes grown in the presence of serum. We assume that the significant, almost 10-fold, difference in the LD50 for SNB-19 cells grown in the absence of serum compared to the LD5o of SNIB -19 cells grown in the presence of serum is likely HDL's proclivity for serum proteins. However, the striking difference in SNB-19 compared to the normal astrocyte LD,o, as well as the absolute amount of HDL required to inhibit 50% of the growth of normal astrocytes, suggests that the ability of HDL to modulate the growth of glioblastoma cells in culture of biological significance.
Molecular and Chemical Neuropathology
Vol. 17, 1992
176
Moskal et al.
100
60
60
40
20
10
20
30
40
50
HDL [ tg!ml]
Fig. 2. The effect of human high density lipoprotein (HDL) on in vitro cell growth of various human glioblastoma cell lines. Three cell lines were examined at various concentrations of HDL. High tumor (s); highly tumorigenic when injected into nude mice. Low tumor (A) and low tumor (®) refer to two separate glioblastoma cell lines that produce significantly fewer tumors in nude mice than the high tumor cell line. The LD50 for the high tumor cell line was approximately 5µg/mL, whereas the LD50 for the two low tumor cell lines was approximately 15-20 µg/mL. These experiments were performed in serum-free medium. All experimental points depicted represent the average of at least six measurements made in at least two independent experiments. Error bars represent the range of values obtained for each data point. No value exceeds 10% of the average. The procedure for computing dose responses was: 100% Inhibition = 100 - (response -blank response/maximum response -blank response x 100). At this point several hypotheses emerge. The data indicate that HDL could be inhibiting the growth of the tumor cells by a receptor-mediated mechanism further suggesting that the glioblastoma cells might possess altered lipoprotein receptors compared to normal glia. A second hypothesis is that HDLs sequester extracellular factors, such as growth factors, that act in an autocrine fashion to control tumor growth. Morrison and coworkers (Morrison et al., 1990) have recently reported that basic fibroblast growth factor (bFGF) is synthesized by SNB-19 cells, and that when it is added to cultures grown in serum-free conditions cell growth is stimulated. Moreover, bFGF receptors were also present on these cells. A
Molecular and Chemical Neuropathology
Vol. 17, 1992
177
Lipoproteins Inhibit Glioma Growth
100
80
60 U
z
0 0 40
20
100 200
300
400
500
HDL [µghn!]
Fig. 3. The effect of human high density lipoprotein on normal rat astrocytes. Normal rat astrocytes (A) were prepared and cultured as described in the text. The highly tumorigenic human glioblastoma cell line (b), SNB-19, was also grown under identical conditions as a comparison. The LD50 for SNB-19 cells grown in serum was approximately 50-75 µg/mL, whereas for normal astrocytes it was > 500 µglmL. All experimental points depicted represent the average of at least six measurements made in at least two independent experiments. Error bars represent the range of values obtained for each data point. No value typically exceeds 10% of the average. [ 3 H]thymidine incorporation was used to monitor cell growth as described in the text.
third hypothesis, and the most straightforward to test, is that oxidized lipids from the HDL interact with tumor cell membranes to modulate their growth in culture. Based on this an experiment was performed in which the antioxidant, glutathione, was mixed with the HDL fraction and its effect on HDL-inhibition of cell growth was determined. Figure 5 shows the effect of varying the concentration of glutathine in the presence of a concentration of HDL equivalent to its LD 50 (an average of 5 µg/mL of HDL in serum-free experiments with SNB-19 cells). It is clear that at a concentration of approximately 50 pM, glutathione can block essentially all of the growth inhibitory properties of HDL. Glutathione alone, at this concentration, showed no growth inhibitory properties (data not shown). This is strong support for the idea that oxidized lipids from the HDL
Molecular and Chemical Neuropathology
Vol. 17, 1992
178
Moskal et al.
500
400
O
300
x E
a 200
Z!11í
gifoma
gifoma
(-serum)
(+serum)
astrocytes (+serum)
Fig. 4. A comparison of the effects of human high density lipoprotein on glioblastoma cell line SNB-19 in the absence or presence of serum in the growth medium and normal rat astrocytes also grown in the presence of serum. Each value graphed is the LD30 for HDL. These data were taken from the data depicted in Figs. 3 and 4.
cause the growth inhibition of the glioblastoma cells in culture. These data do not rule out the possibility that HDL could inhibit tumorigenesis in vivo by sequestering growth factors. Irrespective of the mechanism by which HDL inhibited glioblastoma growth in vitro, the difference between the concentration of HDL necessary to inhibit tumor growth compared to normal astrocytes is dramatic. Interestingly, data using a lymphocyte model have shown that HDL inhibit a specific membrane-associated phosphatase (unpublished observations). Although purely conjectural, perhaps our studies involve an altered phosphatase in glioblastoma cells. Recent work on the molecular mechanisms that control the cell cycle has demonstrated that after motisis the protein cyclin falls below a threshold level, which then leads to the inactivation of maturation promoting factor (MPF). The loss of MPF activity in effect increases phosphatase activity, which removes phosphate groups that were added to key proteins during mitosis thereby allowing cyclin levels to build up culminating in another round of mitosis. Clearly, alterations in the molecular machinery of the cell cycle could lead to loss of growth control and tumorigenesis (Murray and Kirschner, 1989; Nurse, 1990). Another recently described mechanism Molecular and Chemical Neuropathology
Vol. 17, 1992
179
Lipoproteins Inhibit Glioma Growth
100
80 0 2 0 U 60
40
HOL
0
25
50
75
100
200
GLUTATHIONE [uM]
Fig. 5. The effect of glutathione on the growth inhibitory properties of human high density lipoprotein. The human glioblastoma cell line SNB-19 was used in these experiments. The concentration of HDL used was that which gave approximately the LD5o (in these studies 5 µg/mL). The studies were carried out in serum-free growth medium as described in the text. Between 50 and 100 gg/mL of glutathione significantly reversed the effects of HDL on glioblastoma cell growth. Glutathione alone at these concentrations showed no effect on inhibition of cell growth (data not shown). Glutathione, at a concentration of 200 µglmL, was toxic. [ 3 H]thymidine incorporation was used to monitor cell growth as described in the text.
in which phosphatases may control cell growth was reported by Kreuger et al. (1990). These investigators found that some tyrosine phosphatase receptors are structurally similar to cell-surface molecules that mediate contact inhibition of growth. Experiments are in progress to determine if HDL can alter protein phosphorylation in tumor cells and how this is related to normal growth control.
This work was supported in part by a grant from the Health Foundation (New York, NY). Molecular and Chemical Neuropathology
Vol. 17, 1992
180
Moskal et al.
REFERENCES Anatharamaiah G. M., Jones J. L., Brouillette C. G., Schmidt C. F. A., Chung B. H., Hughes T. A., Bhown A. S., and Segrest J. P. (1985) Structure of complexes with dimyristoyl phosphatidylcholine. J. Biol. Chem. 260, 10248-10255. Banker G. and Goslin K. (1991) Culturing Nerve Cells, MIT, Cambridge, MA. Barclay M., Skipski V. P., Terebus-Kekish 0., Greene E. M., Kaufman R. J., and Stock C. C. (1970) Effects of cancer upon high-density and other lipoproteins. Cancer Res. 30, 2420-2430. Bilheimer D. W. (1988) The lipoprotein receptor concept. Drugs 36, 55-62. Bonomo E. A. and Swaney J. B. (1988) A rapid method for the synthesis of protein-lipid complexes using adsorption chromatography. J. Lipid. Res. 29, 380-384. Boyles J. K., Pitas R. E., Wilson E., Mahley R. W., and Taylor J. M. (1985) Apolipoprotein E associated with astrocytic glia of the central nervous system and with nonmyelinating glia of the peripheral nervous system. J. Clin. Invest. 76, 1501-1513. Brewer H. B., Ronan R., Meng M., and Bishop C. (1986) Isolation and characterization of apolipoprotein A-I, A-II, and A-IV. Methods Enzymol. 128, 22 3-246. Burstein M. and Scolnick H. R. (1973) Lipoprotein-polyanion-metal interactions. Adv. Lipid Res. 11, 67-108. Chapman M. J. (1980) Animal lipoproteins: chemistry, structure, and comparative aspects. J. Lipid Res. 21, 789-853. Chung B. H., Segrest J. P., Ray M. J., Brunzell J. D., Hokanson J. E., Krauss R. M., Beaudrie K., and Cone J. T. (1986) Single vertical spin density gradient ultracentrifugation. Methods Enzymol. 128, 181-209. Elshourbagy N. S., Liao W. S., Mahley R. W., and Taylor J. M. (1984) Apolipoprotein E mRNTA is abundant in the brain and adrenals, as well as in the liver, and is present in other peripheral tissues of rats and marmosets. Proc. Natl. Acad. Sci. USA 82, 203-207. Gordon V., Innerarity T. L., and Mahley R. W. (1983) Formation of cholesteroland apoprotein E-rich high density lipoproteins in vitro. J. Biol. Chem. 258, 6202-6212. Gross J. L., Behrens D. L., Mullins D. E., Kornblith P. L., and Dexter D. L. (1988) Plasminogen activator and inhibitor activity in human glioma cells and modulation by sodium butyrate. Cancer Res. 48, 291-296. Havel R. J., Goldstein J. L., and Brown M. S. (1980) Lipoproteins and lipid transport, in Metabolic Control and Disease (Bodny P. K. and Rosenberg L. E., eds.), pp. 393-494, Saunders, Philadelphia. Hui D. Y., Innerarity T. L., and Mahlet' R. W. (1981) Lipoprotein binding to canine hepatic membranes: metabolically distinct apo-E and apo-B, E receptors. J. Biol. Chem. 256, 5646-5655. Ignatius M. J., Gebicke-Harter P. J., Skene J. H. P., Schilling J. W., Weisgraber K. H., Mahley R. W, and Shooter E. M. (1986) Expression of apolipoprotein E during nerve degeneration and regeneration. Proc. Natl. Acad. Sci. USA 83, 1125-1129.
Molecular and Chemical Neuropathology
Vol. 17, 1992
Lipoproteins Inhibit Glioma Growth
181
Jurgens G., Xu Q., Huber L. A., Bock G., Howanietz H., Wick G., and Trail K. N. (1989) Promotion of lymphocyte growth by high density lipoproteins (HDL). J. Biol. Chem. 264, 8549-8556. Koizumi J., Kano M., Okabayashi K., Jadhav A., and Thompson G. R. (1989) Behavior of human apolipoprotein A-I: phospholipid and apoHDL:phospholipid complexes in vitro and after injection into rabbits. J. Lipid Res. 29, 1405-1415. Kreuger N. X., Streuli M., and Saito H. (1990) Structural diversity and evolution of human receptor-like tyrosine phosphatases. EMBO J. 9, 3241. Mahley R. W. and Innerarity T. L. (1983) Lipoprotein receptors and cholesterol homeostasis. Biochim. Biophys. Acta 737, 197-222. Mahley R. W., Hui D. Y., Innerarity T. L., and Weisgraber (1981) Two independent lipoprotein receptors on hepatic membranes of dog, swing, and man. Apo-B,E and apo-E receptors. J. Clin. Invest. 68, 1197-1206. Martin-Nizard F., Meresse S., Cecchelli R., Fruchart J. C., and Delbart C. (1989) Interactions of high-density lipoprotein 3 with brain capillary endothelial cells. Biochim. Biophys. Acta 1005, 201-208. Matz C. E. and Jonas A. (1982) Reaction of human lecithin cholesterol acyltransferase with synthetic micellar complexesl of apolipoprotein A-I, phosphatidylcholine, and cholesterol. J. Biol. Chem. 257, 4541-4546. Morrison R. S., Gross J. L., Herblin W., Reilly T. M., Lasala P. A., Alterman R. L., Moskal J. R., Kornblith P. L., and Dexter D. L. (1990) Basic fibroblast growth factor-like activity and receptors are expressed in a human glioma cell line. Cancer Res. 50, 2524-2529. Murray W. and Kirschner M. W. (1989). Dominoes and Clocks: The union of two views of the cell cycle. Science 246, 614-621. Nichols A. V., Blanche P. J., and Gong E. L. (1983) Handbook of Electrophoresis. CRC, Boca Raton, FL. Nurse P. (1990) Universal control mechanism regulating onset of M-Phase. Nature 344, 503-508. Reckless J. P. D., Weinstein D. B., and Steinberg D. (1978) Lipoprotein and cholesterol metabolism in rabbit bacterial endothelial cells in culture. Biochim. Biophys. Acta 529, 475-487. Schumaker V. N. and Puppione D. L. (1986) Sequential ultracentrifugation. Methods Enzymol. 128, 155-170. Stein O. and Stein Y. (1976). High density lipoproteins reduce the uptake of low density lipoproteins by human endothelial cells in culture. Biochim. Biophys. Acta 23, 563 -568. Tauber J. P., Cheng J., and Gospodarowicz D. (1980) Effect of high and low density lipoproteins on proliferation of cultured bovine vascular endothelial cells. J. Clin, Invest. 66, 696-708. Vlodaysky I., Fielding P. E., Feilding C. J., and Gospadarowicz D. (1978) Role of contact inhibition in the regulation of receptor-mediated uptake of low density lipoprotein in cultured vascular endothelial cells. Proc. Nati. Acad. Sci, USA 75, 356-360.
Molecular and Chemical Neuropathology
Vol. 17, 1992
.
.
t ... s :- I sitiïsi .w-
JOSEPH R. MOSKAL,'* MARK SINNETT, 2 PAUL L. KORNBLITH, 2 PATRICK LASALA, 3 DANIEL A. LEVINE, 4 THOMAS S. PARKER, 4 AND HARRY LANDER 4 'The Chicago Institute for Neurosurgery and Neuroresearch, 428 W. Deming Pl. Chicago, IL 60614; 2 Department of Neurosurgery, Montefiore Medical Center, Bronx, NY; 3 Department of Surgery, St. Lukes— Roosevelt Hospital, New York, NY; 4 Rogosin Institute, New York, NY
Received February 4, 1992; Accepted April 3, 1992
Experiments were performed using an established human glioblastoma cell line to determine the effect of lipoproteins on regulating their growth. It was found that synthetic and natural human high density lipoproteins (HDL) were effective in inhibiting tumor cell growth in a nontoxic, dose-dependent manner, and that the LD50 was 10-fold lower than that for normal rat astrocytes grown under identical conditions. In the presence of the antioxidant, glutathione, essentially all of the growth-inhibiting properties of HDL could be reversed suggesting that oxidized lipids from the HDL interacting with the plasma membranes of the glioblastoma cells were responsible for the growth-inhibiting effect observed. The markedly lower concentration of HDL required to inhibit glioblastoma cells in culture compared to normal astrocytes suggested that the mechanism of HDL-induced inhibition may be important for tumor growth in vivo. One possible mechanism under investigation is the possibility of HDL modulation of a membrane-associated, tumor-specific phosphatase. Index Entries: Lipoproteins: high density; glioblastoma growth regulation; receptor-mediated processes; regulation of cell-cycle by phosphatases; autocrine growth control; membrane-membrane interactions. *Author to whom all correspondence and reprint requests should be addressed. Molecular and Chemical Neuropathology
169
Vol. 17, 1992
170
Moskal et al.
Many of the structural, functional, and regulatory aspects of lipoproteins have been elucidated in recent years (Chapman, 1980; Bilheimer, 1988). Briefly, lipoproteins consist of chylomicron particles, very low density lipoproteins (VLDL), low density lipoproteins (LDL), ß-VLDL, and high density lipoproteins (HDL). VLDL, /3-VLDL, and LDL are associated with the transport of cholesterol away from the liver to the periphery, whereas HDL carry cholesterol from peripheral cells to the liver (see Fig. 1). There are a variety of proteins that are associated with lipoproteins. Perhaps the best characterized apolipoprotein is Apo E (Mahley and Innerarity, 1983; Havel et al., 1980). It is a 30-kDa, lipophilic protein that binds to two receptors found on both hepatic and nonhepatic cells (Hui et al., 1981; Mahley et al., 1981). Its function is to coordinate lipoprotein internalization by hepatocytes via speciic receptor-mediated processes and to help in the regulation of cholesterol homeostasis. Apo E is also involved in removing cholesterol from cells by facilitating its packaging into HDL (Gordon et al., 1983). Here we report our results based on the working hypothesis that lipoproteins play a key role in regulating the growth of brain tumors. This hypothesis emerged based on the following considerations. First, in a now classic study, Barclay and coworkers (1970) showed that men, women, and children with cancer had significantly lower HDL levels than controls. Moreover, they also showed a direct relationship between low levels of HDL in serum of normal subjects and a high family incidence of cancer. Second, in vivo and in vitro studies have shown that LDL itself is toxic to endothelial cells, whereas HDL is protective (Stein and Stein, 1976; Reckless et al., 1978; Vlodaysky et al., 1978; Tauber et al., 1980). Third, recent studies have demonstrated the presence of trypsinsensitive, high-affinity binding sites for HDL on brain capillary epithelial cells. Competition studies between HDL and apolipoprotein A-I suggested that apolipoprotein A-I was involved in HDL-receptor complex formation (Martin-Nizard et al., 1989). Interestingly, a number of highaffinity binding sites for HDL on a variety of cell types have been reported, but their physiological significance remains to be determined (Jurgens et al., 1989). Fourth, Boyles and coworkers (1989) have shown that during regeneration of rat peripheral nerve that the appropriate components and temporal sequencing of their expression are in place for a complete, functioning cholesterol transfer system necessary for nerve regeneration and remyelination. Finally, Eishourbagy et al. (1984) have shown that the brain was among the tissues with the highest concentration of apo E. More recently Boyles et al. (1985) demonstrated that apo E was present in all astrocytes but absent from neurons, oligodendroglia, and microglia. The data presented in the present report demonstrate that purified fractions of human HDL can inhibit the growth of an established Molecular and Chemical Neuropathology
Vol. 17, 1992
Lipoproteins Inhibit Glioma Growth
UJ.
W ;
O
Q
o_ VI 0
U.
0
_Z 4
W Ü
á Z
¢ q
Q Z
171
0 M
W J
C11
99% purity was needed, the product was redissolved in Tris 6M urea buffer and chromatographed on Sephacryl-S 200 as described by Brewer et al. (1986). Reconstituted apo-HDL: Apo HDL was reconstituted by detergent dialysis with the following modifications to increase scale. Egg lecithin (Matreya, Pleasant Gap, PA) (3.4 g) was dissolved in 205 mL of 108 mM sodium cholate. The clear solution was then brought to a final volume of 720 mL with Tris-EDTA buffer (31 mM cholate). This solution was dialyzed against T-E buffer (31 mM cholate). The preparation of apo HDL protein and reconstitution steps were as described by Matz and Jonas (1982). Removal of cholate was essentially by the method of Bonoma and Swaney (1988) adapted to the 2-4 L scale. The concentration of cholate in the final R-HDL preparation was < 100 µM when the phospholipid concentration is -20 mg/mL: a cholate to PC molar rate of -1:500. This level of cholate is not toxic in the neutral red cytotoxicity test. Synthetic Peptides: A peptide analogous to the concensus sequence of apo A-I was prepared by the solid-phase peptide synthesis core facility at the State University of New York at Stony Brook. The sequence, designated as peptide 18A by Anantharamaiah et al. (1985), was designed to form an amphipathic helix with polar and nonpolar faces of approximately equal surface area and is long enough to permit cooperative interactions between multiple and amphipathic helical domains. Purified peptide was reconstituted with phospholipid by the same method described earlier for apo-HDL.
Figure 1 depicts the effects of various lipoproteins on SNB-19 growth. Each lipoprotein fraction used in this study contained 0.85-1.0 mg of protein/mL. This study was performed in the presence of serum-containing
Molecular and Chemical Neuropathology
Vol. 17, 1992
Lipoproteins Inhibit Glioma Growth
175
medium. There was a significant inhibition of cell growth with human HDL without any significant (< 5%) uptake of Trypan blue. LDL inhibition of growth, however, was toxic. Interestingly, synthetic human apo Al (which contained the lecithin: cholesterol acyl transferase activator domain), cholesterol-free, reconstituted Al and All particles, and a fraction containing the apoproteins found in HDL without the lipid components also showed significant inhibition of growth of SNB-19 cells. These data demonstrated that HDL was capable of inhibiting glioblastoma cell growth in vitro, that LDL were toxic, and that these phenomena occurred in the reported concentration range reported for the modulation of endothelial cell growth in culture (Tauber et al., 1980). Furthermore, it appeared that HDL apoproteins could markedly inhibit cell growth suggesting that this process might be receptor-mediated. In the second set of experiments, seen in Fig. 2, SNB-19 cells were treated with a variety of HDL concentrations and compared to human glioblastoma cell lines of low tumorigenic capability. These experiments were performed in the absence of serum. First, it can be seen that SNB-19 cells, in the absence of serum, show 50% inhibition of growth at approximately 5 µg/ mL of HDL. The two low-tumor forming glioblastoma cell lines each showed a LD5o of approximately 15-20 µg/ mL of HDL. All three cell lines responded to HDL in a dose-dependent manner further suggesting a specific, possibly receptor-mediated mechanism for the growth inhibition of HDL. Moreover, the shift in LD50 in the low tumorforming glioblastoma cell lines suggested a relationship between tumorigenicity and the concentration of HDL necessary to inhibit growth. To investigate this phenomenon further, HDL concentration curves were performed on SNB-19 cells and primary cultures of normal rat astrocytes. These experiments were performed in the presence of serum since normal rat astrocyte growth is markedly reduced in the absence of serum. This was not the case with the glioblastoma cultures. As can be seen in Fig. 3, the ability of HDL to inhibit normal rat astrocyte growth was dramatically different than SNB-19 glioblastoma cells. The LD5o for SNB-19 cells, grown in the presence of serum was approximately 40-50 ttg/mL compared to an LD50 of approximately 500 µglmL for the normal astrocytes in culture. Figure 4 shows the LD50 for SNB-19 cells grown in serum-free conditions compared to SNB-19 cells growth in the presence of serum compared to normal rat astrocytes grown in the presence of serum. We assume that the significant, almost 10-fold, difference in the LD50 for SNB-19 cells grown in the absence of serum compared to the LD5o of SNIB -19 cells grown in the presence of serum is likely HDL's proclivity for serum proteins. However, the striking difference in SNB-19 compared to the normal astrocyte LD,o, as well as the absolute amount of HDL required to inhibit 50% of the growth of normal astrocytes, suggests that the ability of HDL to modulate the growth of glioblastoma cells in culture of biological significance.
Molecular and Chemical Neuropathology
Vol. 17, 1992
176
Moskal et al.
100
60
60
40
20
10
20
30
40
50
HDL [ tg!ml]
Fig. 2. The effect of human high density lipoprotein (HDL) on in vitro cell growth of various human glioblastoma cell lines. Three cell lines were examined at various concentrations of HDL. High tumor (s); highly tumorigenic when injected into nude mice. Low tumor (A) and low tumor (®) refer to two separate glioblastoma cell lines that produce significantly fewer tumors in nude mice than the high tumor cell line. The LD50 for the high tumor cell line was approximately 5µg/mL, whereas the LD50 for the two low tumor cell lines was approximately 15-20 µg/mL. These experiments were performed in serum-free medium. All experimental points depicted represent the average of at least six measurements made in at least two independent experiments. Error bars represent the range of values obtained for each data point. No value exceeds 10% of the average. The procedure for computing dose responses was: 100% Inhibition = 100 - (response -blank response/maximum response -blank response x 100). At this point several hypotheses emerge. The data indicate that HDL could be inhibiting the growth of the tumor cells by a receptor-mediated mechanism further suggesting that the glioblastoma cells might possess altered lipoprotein receptors compared to normal glia. A second hypothesis is that HDLs sequester extracellular factors, such as growth factors, that act in an autocrine fashion to control tumor growth. Morrison and coworkers (Morrison et al., 1990) have recently reported that basic fibroblast growth factor (bFGF) is synthesized by SNB-19 cells, and that when it is added to cultures grown in serum-free conditions cell growth is stimulated. Moreover, bFGF receptors were also present on these cells. A
Molecular and Chemical Neuropathology
Vol. 17, 1992
177
Lipoproteins Inhibit Glioma Growth
100
80
60 U
z
0 0 40
20
100 200
300
400
500
HDL [µghn!]
Fig. 3. The effect of human high density lipoprotein on normal rat astrocytes. Normal rat astrocytes (A) were prepared and cultured as described in the text. The highly tumorigenic human glioblastoma cell line (b), SNB-19, was also grown under identical conditions as a comparison. The LD50 for SNB-19 cells grown in serum was approximately 50-75 µg/mL, whereas for normal astrocytes it was > 500 µglmL. All experimental points depicted represent the average of at least six measurements made in at least two independent experiments. Error bars represent the range of values obtained for each data point. No value typically exceeds 10% of the average. [ 3 H]thymidine incorporation was used to monitor cell growth as described in the text.
third hypothesis, and the most straightforward to test, is that oxidized lipids from the HDL interact with tumor cell membranes to modulate their growth in culture. Based on this an experiment was performed in which the antioxidant, glutathione, was mixed with the HDL fraction and its effect on HDL-inhibition of cell growth was determined. Figure 5 shows the effect of varying the concentration of glutathine in the presence of a concentration of HDL equivalent to its LD 50 (an average of 5 µg/mL of HDL in serum-free experiments with SNB-19 cells). It is clear that at a concentration of approximately 50 pM, glutathione can block essentially all of the growth inhibitory properties of HDL. Glutathione alone, at this concentration, showed no growth inhibitory properties (data not shown). This is strong support for the idea that oxidized lipids from the HDL
Molecular and Chemical Neuropathology
Vol. 17, 1992
178
Moskal et al.
500
400
O
300
x E
a 200
Z!11í
gifoma
gifoma
(-serum)
(+serum)
astrocytes (+serum)
Fig. 4. A comparison of the effects of human high density lipoprotein on glioblastoma cell line SNB-19 in the absence or presence of serum in the growth medium and normal rat astrocytes also grown in the presence of serum. Each value graphed is the LD30 for HDL. These data were taken from the data depicted in Figs. 3 and 4.
cause the growth inhibition of the glioblastoma cells in culture. These data do not rule out the possibility that HDL could inhibit tumorigenesis in vivo by sequestering growth factors. Irrespective of the mechanism by which HDL inhibited glioblastoma growth in vitro, the difference between the concentration of HDL necessary to inhibit tumor growth compared to normal astrocytes is dramatic. Interestingly, data using a lymphocyte model have shown that HDL inhibit a specific membrane-associated phosphatase (unpublished observations). Although purely conjectural, perhaps our studies involve an altered phosphatase in glioblastoma cells. Recent work on the molecular mechanisms that control the cell cycle has demonstrated that after motisis the protein cyclin falls below a threshold level, which then leads to the inactivation of maturation promoting factor (MPF). The loss of MPF activity in effect increases phosphatase activity, which removes phosphate groups that were added to key proteins during mitosis thereby allowing cyclin levels to build up culminating in another round of mitosis. Clearly, alterations in the molecular machinery of the cell cycle could lead to loss of growth control and tumorigenesis (Murray and Kirschner, 1989; Nurse, 1990). Another recently described mechanism Molecular and Chemical Neuropathology
Vol. 17, 1992
179
Lipoproteins Inhibit Glioma Growth
100
80 0 2 0 U 60
40
HOL
0
25
50
75
100
200
GLUTATHIONE [uM]
Fig. 5. The effect of glutathione on the growth inhibitory properties of human high density lipoprotein. The human glioblastoma cell line SNB-19 was used in these experiments. The concentration of HDL used was that which gave approximately the LD5o (in these studies 5 µg/mL). The studies were carried out in serum-free growth medium as described in the text. Between 50 and 100 gg/mL of glutathione significantly reversed the effects of HDL on glioblastoma cell growth. Glutathione alone at these concentrations showed no effect on inhibition of cell growth (data not shown). Glutathione, at a concentration of 200 µglmL, was toxic. [ 3 H]thymidine incorporation was used to monitor cell growth as described in the text.
in which phosphatases may control cell growth was reported by Kreuger et al. (1990). These investigators found that some tyrosine phosphatase receptors are structurally similar to cell-surface molecules that mediate contact inhibition of growth. Experiments are in progress to determine if HDL can alter protein phosphorylation in tumor cells and how this is related to normal growth control.
This work was supported in part by a grant from the Health Foundation (New York, NY). Molecular and Chemical Neuropathology
Vol. 17, 1992
180
Moskal et al.
REFERENCES Anatharamaiah G. M., Jones J. L., Brouillette C. G., Schmidt C. F. A., Chung B. H., Hughes T. A., Bhown A. S., and Segrest J. P. (1985) Structure of complexes with dimyristoyl phosphatidylcholine. J. Biol. Chem. 260, 10248-10255. Banker G. and Goslin K. (1991) Culturing Nerve Cells, MIT, Cambridge, MA. Barclay M., Skipski V. P., Terebus-Kekish 0., Greene E. M., Kaufman R. J., and Stock C. C. (1970) Effects of cancer upon high-density and other lipoproteins. Cancer Res. 30, 2420-2430. Bilheimer D. W. (1988) The lipoprotein receptor concept. Drugs 36, 55-62. Bonomo E. A. and Swaney J. B. (1988) A rapid method for the synthesis of protein-lipid complexes using adsorption chromatography. J. Lipid. Res. 29, 380-384. Boyles J. K., Pitas R. E., Wilson E., Mahley R. W., and Taylor J. M. (1985) Apolipoprotein E associated with astrocytic glia of the central nervous system and with nonmyelinating glia of the peripheral nervous system. J. Clin. Invest. 76, 1501-1513. Brewer H. B., Ronan R., Meng M., and Bishop C. (1986) Isolation and characterization of apolipoprotein A-I, A-II, and A-IV. Methods Enzymol. 128, 22 3-246. Burstein M. and Scolnick H. R. (1973) Lipoprotein-polyanion-metal interactions. Adv. Lipid Res. 11, 67-108. Chapman M. J. (1980) Animal lipoproteins: chemistry, structure, and comparative aspects. J. Lipid Res. 21, 789-853. Chung B. H., Segrest J. P., Ray M. J., Brunzell J. D., Hokanson J. E., Krauss R. M., Beaudrie K., and Cone J. T. (1986) Single vertical spin density gradient ultracentrifugation. Methods Enzymol. 128, 181-209. Elshourbagy N. S., Liao W. S., Mahley R. W., and Taylor J. M. (1984) Apolipoprotein E mRNTA is abundant in the brain and adrenals, as well as in the liver, and is present in other peripheral tissues of rats and marmosets. Proc. Natl. Acad. Sci. USA 82, 203-207. Gordon V., Innerarity T. L., and Mahley R. W. (1983) Formation of cholesteroland apoprotein E-rich high density lipoproteins in vitro. J. Biol. Chem. 258, 6202-6212. Gross J. L., Behrens D. L., Mullins D. E., Kornblith P. L., and Dexter D. L. (1988) Plasminogen activator and inhibitor activity in human glioma cells and modulation by sodium butyrate. Cancer Res. 48, 291-296. Havel R. J., Goldstein J. L., and Brown M. S. (1980) Lipoproteins and lipid transport, in Metabolic Control and Disease (Bodny P. K. and Rosenberg L. E., eds.), pp. 393-494, Saunders, Philadelphia. Hui D. Y., Innerarity T. L., and Mahlet' R. W. (1981) Lipoprotein binding to canine hepatic membranes: metabolically distinct apo-E and apo-B, E receptors. J. Biol. Chem. 256, 5646-5655. Ignatius M. J., Gebicke-Harter P. J., Skene J. H. P., Schilling J. W., Weisgraber K. H., Mahley R. W, and Shooter E. M. (1986) Expression of apolipoprotein E during nerve degeneration and regeneration. Proc. Natl. Acad. Sci. USA 83, 1125-1129.
Molecular and Chemical Neuropathology
Vol. 17, 1992
Lipoproteins Inhibit Glioma Growth
181
Jurgens G., Xu Q., Huber L. A., Bock G., Howanietz H., Wick G., and Trail K. N. (1989) Promotion of lymphocyte growth by high density lipoproteins (HDL). J. Biol. Chem. 264, 8549-8556. Koizumi J., Kano M., Okabayashi K., Jadhav A., and Thompson G. R. (1989) Behavior of human apolipoprotein A-I: phospholipid and apoHDL:phospholipid complexes in vitro and after injection into rabbits. J. Lipid Res. 29, 1405-1415. Kreuger N. X., Streuli M., and Saito H. (1990) Structural diversity and evolution of human receptor-like tyrosine phosphatases. EMBO J. 9, 3241. Mahley R. W. and Innerarity T. L. (1983) Lipoprotein receptors and cholesterol homeostasis. Biochim. Biophys. Acta 737, 197-222. Mahley R. W., Hui D. Y., Innerarity T. L., and Weisgraber (1981) Two independent lipoprotein receptors on hepatic membranes of dog, swing, and man. Apo-B,E and apo-E receptors. J. Clin. Invest. 68, 1197-1206. Martin-Nizard F., Meresse S., Cecchelli R., Fruchart J. C., and Delbart C. (1989) Interactions of high-density lipoprotein 3 with brain capillary endothelial cells. Biochim. Biophys. Acta 1005, 201-208. Matz C. E. and Jonas A. (1982) Reaction of human lecithin cholesterol acyltransferase with synthetic micellar complexesl of apolipoprotein A-I, phosphatidylcholine, and cholesterol. J. Biol. Chem. 257, 4541-4546. Morrison R. S., Gross J. L., Herblin W., Reilly T. M., Lasala P. A., Alterman R. L., Moskal J. R., Kornblith P. L., and Dexter D. L. (1990) Basic fibroblast growth factor-like activity and receptors are expressed in a human glioma cell line. Cancer Res. 50, 2524-2529. Murray W. and Kirschner M. W. (1989). Dominoes and Clocks: The union of two views of the cell cycle. Science 246, 614-621. Nichols A. V., Blanche P. J., and Gong E. L. (1983) Handbook of Electrophoresis. CRC, Boca Raton, FL. Nurse P. (1990) Universal control mechanism regulating onset of M-Phase. Nature 344, 503-508. Reckless J. P. D., Weinstein D. B., and Steinberg D. (1978) Lipoprotein and cholesterol metabolism in rabbit bacterial endothelial cells in culture. Biochim. Biophys. Acta 529, 475-487. Schumaker V. N. and Puppione D. L. (1986) Sequential ultracentrifugation. Methods Enzymol. 128, 155-170. Stein O. and Stein Y. (1976). High density lipoproteins reduce the uptake of low density lipoproteins by human endothelial cells in culture. Biochim. Biophys. Acta 23, 563 -568. Tauber J. P., Cheng J., and Gospodarowicz D. (1980) Effect of high and low density lipoproteins on proliferation of cultured bovine vascular endothelial cells. J. Clin, Invest. 66, 696-708. Vlodaysky I., Fielding P. E., Feilding C. J., and Gospadarowicz D. (1978) Role of contact inhibition in the regulation of receptor-mediated uptake of low density lipoprotein in cultured vascular endothelial cells. Proc. Nati. Acad. Sci, USA 75, 356-360.
Molecular and Chemical Neuropathology
Vol. 17, 1992
