Gossypol effects on endothelial cells and tumor blood flow

Life Sciences, Vol. 49, pp. PL-67 - PL-72 Printed in the U.S.A.

Pergamon Press

PHARMACOLOGY LETTERS Accelerated Communication

GOSSYPOL EFFECTS ON ENDOTHELIAL CELLS AND TUMOR BLOOD FLOW C.C. Benz, S.B. lyer, H.S. Asgari, S.A. M~lin, F.R. Aronson, and A. Barohowsky 1 Cancer Research Institute, University of California, San Francisco, CA 94143 and 1Clinical Pharmacology, Thomas Jefferson University, Philadelphia, PA 19107 (Submitted June 14, 1991; accepted June 24, 1991; received in final form July 16, 1991)

Abstract. Isomers (-,+) of the antitumor agent gossypol (G) were studied for their ability to reduce tumor ATP and blood flow in rats bearing subcutaneously implanted pancreatic tumors. A 50% reduction in tumor ATP/Pi within Ih of a single injection of -G was associated with a 60% decline in tumor blood flow. To determine if these changes in tumor physiology could be due to a direct drug effect on tumor endothelium, G isomers were compared for their ability to alter protein (1251-BSA)permeability and metabolic (32p) labelling of cultured endothelial cells. Treatments lor Ih produced no endothelial cell leakage, but 24h exposures to either -G(511M) or +G(50t~M) produced complete permeability of the monolayers to 12SI-BSA. In contrast, 0.5-1.0h exposures to -G(4pM) or +G(10pM) produced 2 to 3-fold increases in phosphorylated 27kDa heat-shock protein, hsp-27. Hsp-27 phosphoprotein isoforms were differentially labelled following -G and +G exposures with the phosphorylation profile of -G appearing most similar to that of oxyradical producing agents known to induce hsp-27 and injure endothelial cells. We postulate that the tumor ischemic effects of -G are mediated by endothelial response to oxyradical Droduction in a mechanism similar to that of tissue ischemia-reD~rfusion injury, INTRODUCTION Gossypol (G) is a plant-derived binaphthol with antiproliferative activity against epithelial carcinomas of breast, ovarian, colon, and pancreatic origin (1). G has antiturnor properties similar to those of lipophilic antimitochondrial agents (like rhodamine and platinum-rhodamine), and we have demonstrated that phosphorus magnetic resonance spectroscopy (31 P - M R S ) prowdes a rapid means of noninvasively monitoring tumors for the uncoupling ATPase effects of such antimitochondrial agents (2-5). G is also a potent inhibitor of various non-mitochondrial enzymes such as calmodulin (CAM) stimulated cyclic nucleotide phosphodiesterase and protein kinase C (PKC), raising the question of whether the antitumor properties of G are due to its antimitochondrial or enzyme inhibitory mechanisms (1). Of the two naturally occuring G enantiomers (-, +) studied in vitro, -G produces 10-fold greater antiproliferative and antimitochondrial effects while both isomers are equally potent inhibitors of CaM and PKC (1). At present, little is known about the in vivo effects of these two isomers on tumor bioenergetics and physiology. Modest growth inhibiting doses of some antimitochondrial agents produce prominent in vivo reductions in tumor ATP levels measured by 31P-MRS as early as 60 min after intraperitoneal (ip) drug injection (4,5). Studies with the metabolic inhibitor tlavone acetic acid (FAA) indicate that early reductions in tumor ATP are associated with a concomitant decline in tumor blood flow, as monitored by deuterium spectroscopy (6). This finding prompted our analysis of G effects on tumor ATP and blood flow as well as a comparison of G enantiomer effects on cultured endothelial cells. M ETHODS Druas, Cells. and Assays G (-,+) enantiomers were isolated and purified as described earlier and stock solutions were prepared in DMSO/ethanol or Tween80 vehicles (1). Human saphenous vein endothelial cells and porcine aortic endothelial cells (EC) were prepared to measure drug effects on the permeability of confluent EC monolayers (7) and the phosphorylation pattern of EC proteins (8). EC leakage was measured by CORRESPONDING AUTHOR: ChristopherBenz, M.D. M-1282 CancerResearchlnstitute, University of California, San Frandsco, CA 94143-0128. 0024-3205/91 $3.00 +.00 Copyright (c) 1991 Pergamon Press plc

PL-68

Antitumor Endothelial Effects of Gossypol

Vol. 49, No. 12, 1991

assaying radiolabelled bovine serum albumin (t251-BSA)diffusion through EC monolayers grown on filter disks after treatment with either vehicle (DMSO/ethanol, 0.005/0.045 v/v%) or G enantiomers, as previously detailed (7). Briefly, confluent human EC monolayers grown on fibronectin-coated 0.4 micron-pore filter disks in dual chamber Transweil plates were treated for Ih or 24h as indicated, f2SI-BSA was then added to the inner chamber of the wells, samples of medium were periodically aspirated from the outer chamber, and radioactivity was measured in a gamma counter. A NonidetP-40 solution or plain medium were used as positive and negative controls, respectively, and all data were normalized to the detergent treated cells. Results were expressed as the mean % of positive control counts from duplicate wells of at least two different experiments. To measure G effects on intracellular phosphorylation, porcine or human EC cultures in first to third passage were grown to confluence in 0.5% gelatin-coated 35mm dishes with DMEM media containing 10% fetal bovine serum. EC cultures were rinsed with serumfree medium (MEM with 10p.Morthophosphate) and then incubated with I ml of medium containing 0.05 mCi[32p]-orthophosphate at 37°C for Ih. After this incubation period cultures were treated for 0.5-1.0h, washed with homogenization buffer (10mM Tris-HCI, pH 7.4, 10mM EDTA, 5mM EGTA, 0.1M NaF, 0.2 M sucrose, 101~Msodium orthovanadate, and 5mM pyrophosphate), and then scraped into buffer containing 0.1% TritonX-100. Following sonication (60 mHz for 15 seo x 2), an aliquot of cell lysate was taken for protein determination and the remaining sample diluted 1:5 (v/v) in either Laemmli buffer for one-dimensional (I-D) gel electrophoresis (7.5-17.5% polyacrylamide), or isoelectric focusing buffer for two-dimensional (2-D) gel electrophoresis as previously described (9). Radiolabelled phosphoproteins were detected by autoradiography (Kodak XRP film) and I-D autoradiographs were quantitated by scanning densitometry at 550nm. Implanted Tumors and in vivo SDectroscoDv. Male Lewis rats (180 grams) received bilateral subcutaneous flank injections of cultured syngeneic AR42j pancreatic cancer cells (106), or minced fragments of serially transplanted tumors. Spectroscopy was performed before and after drug injection in anesthetized animals whose tumors had reached 2 cm diameter; 31p-MRS was performed using a 1.4 cm 3-turn surface coil (tuned to 34.6 MHz) and a 2.0T G.E.CSI imaging spectrometer as described previously (4,5). All phosphorus spectra were acquired using a 10t~secpulse, 4K data points, 4000 Hz spectral width, 2.5 sec interpulse delay and 512 scans. Quadrature cycling was used and the signal was filtered with exponential multiplication using 20 Hz line broadening before Fourier transformation. Using G.E. software, computer generated peaks were adjusted until coincident with experimental spectra and peak areas of ATP and inorganic phosphate (Pi) were determined for calculation of ATP/Pi ratios. ATP/Pi values after treatment were expressed as % of pretreatment values (100%), with each animal serving as its own control, and statistical significance was assessed by T-test. Deuterium spectroscopic tumor blood flow measurements were made in accordance with previous studies using a horizontal bore 4.7T Narolac spectrometer (6). Anesthetized rats with subcutaneously implanted pancreatic tumors had their body temperatures maintained by warm water circulation through a heating pad. A single-tuned (30.7 MHz) 2-turn surface coil of 10mm diameter was used. A baseline spectrum was first obtained, intratumor injection of 100~1 isotonic saline-D20 was administered, and then additional spectra acquired. Following ip injection of G (60-90 min) in 0.3 ml volume, a repeat baseline spectrum was obtained, the animal was given another intratumor injection of isotonic saline-D20 (5001~1),and another spectrum acquired. All deuterium spectra were acquired using a 27 p.sec pulse, relaxation delay of 100msec, 2K data points and a spectral width of 2500 Hz. Tumor blood flow (TBF) was calculated using the equation TBF=I00 ~T, where ~. is the tissue-to-blood partition coefficient defined as the ratio of water weight per unit tissue mass to water weight per unit blood volume, and T is the exponential time constant governing tracer washout. T was obtained by applying a monoexponentialcurve-fitting routine to the tracer residue washout time course data, and ~.was assumed to be 0.9 (6). R ESULTS Table I shows the time dependent and tissue specific effects of -G on ATP/Pi levels after a single ip injection (15mg/kg) into tumor implanted rats. +G given at the same dose produced no decline in tumor ATP/Pi, while -G reduced tumor ATP/Pi by 25% within 30 min of injection and produced an overall 50% reduction within 60 min (p < 0.05). In the same animals that showed drug induced declines in tumor ATP/Pi, no significant spectral changes were detected in the hind limb muscles over the same time course. In four other tumor implanted rats, deuterium spectroscopy was used to determine tumor blood flow (TBF). Pretreatment TBF appeared to vary with tumor size, resulting in a calculated mean (+ SE) value = i7.29 (+ 8.24) ml/100g tissue/rain. When posttreatment TBF was measured and expressed relative to

Vol. 49, No. 12, 1991

Antitumor Endothelial Effects of Gossypol

PL-69

pretreatment TBF in the same animal 60-90 min after ip -G injection (15mg/kg), the average effect of -G was a reduction in TBF to 38+2% of pretreatment control (p < 0.05). TABLE

I

Spectral Measurement of ATP/Pi in Rat Tumor and Normal Tissue after 15 mg/kg ip Gossypol (G) Injection Condition

Time after injection

% ATP/Pi level 96.6 + 2.0

(n)*

vehicle treated tumor

60 min

+G treated tumor

60 min

-G treated tumor

30 min 60 rain

75.3 + 1.7 49.5 + 4.0

(4) (7)

-G treated muscle

60 min

101.5 + 6.2

(4)

119.5

(2)

*Mean percent (+SE) of pretreatment ATP/Pi ratios for n different tumor bearing animals.

100

= ~ °~

80

~

~- ~

I='-

60

=

r~ ~-~

40

e~

20

~

0

:

+

-

+

~tM:

5

5

5

50

5

h:

1

1

24

24

24

Gossypol(+,-)

(7)

+

FIG. I Effect of gossypol e na ntiomers on the permeability of EC monolayers to

125I-BSA

PL-70

Antitumor Endothelial Effects of Gossypol

Vol. 49, No. 12, 1991

The in vivo spectroscopy studies suggested that the G induced reductions in tumor ATP and TBF might be related. Thus, in vitro studies were initiated to determine if G could injure or metabolically alter cultured EC within 60 min of drug exposure. Figure I compares the time and dose dependent leakage induced by -G and +G on cultured human EC which were otherwise impermeable to radiolabelled albumin (12SI-BSA). Treatment with either enantiomer (5t~M) for only 60 min had no significant effect on the leakiness of these confluent cells. After 24h treatment, -G but not +G enhanced EC leakiness to nearly the same degree as seen with the detergent treated positive control. The dose dependence of this stereospecific drug effect was apparent since a 24h treatment with a 10-fold greater dose of +G produced a similar degree of EC injury as the 51~Mdose of -G. Although the in vivo effects of -G were apparent within 60 min of drug injection, in vitro EC leakiness appeared to require more prolonged treatment with growth suppressing drug concentrations. Therefore, we looked for changes in the intracellular phosphorylation of EC proteins that might occur within the first hour of drug treatment. Figure 2 shows autoradiographs of metabolically labelled EC

A

FIG. 2 2-D gel electrophoresis followed by autoradiography of G treated and metabolically (32p) labelled porcine aortic endothelial cells (EC). The portion of the gel shown here encompasses the molecular weight range of 14-31kDa and a pH range of 5 - 7 (left to right). Control EC cultures in A were treated with vehicle only while those in B and C were treated for l hour with 4 pM (-)G or 10 I.dVl(+)G, respectively. Arrows indicate the location of three 27 kDa phosphoprotein isoforms with distinct isoelectric points. These autoradiographs are representative of four replicate experiments.

C

Vol. 49, No. 12, 1991

Antitumor Endothelial Effects of Gossypol

PL-71

phosphoproteins separated by 2-D gel electrophoresis, showing 32p-phosphoproteins ranging in size from 14-31kDa. Untreated EC contained three related 27kDa protein isoforms with different isoelectric points of which only the most basic isoform was preferentially phosphorylated under confluent culture conditions. After 60 min of treatment by either -G or +G, this paltem of 27kDa protein phosphorylation was dramatically altered. +G treatment resulted in the phosphorylation of all three 27kDa isoforms, while -G treatment resulted in loss of phosphorylation by the most basic phosphoprotein isoform and enhanced phosphorylation by the most acidic 27kDa isoform. TABLE II Phosphorylation of 27kDa Endothelial Cell Proteins Condition

Treatment time

Labelled 27kDa phosphoproteins* OD units % Control

Control (vehicle)

60 min

8,930

100%

-G (4p.M)

30 min 60 min

16,891 27,163

189% 304%

+G (10pM)

30 min 60 min

29,808 23,073

334% 258%

Menadione (201iM)

60 min

50,820

569%

Porcine EC were drug treated in serum-free media and metabolically labelled with 32p-orthophosphate as described in Methods. Autoradiographs from 1-D SDS-polyacrylamide gels were scanned and the peak areas of optical density (OD) units recorded for the 27kDa phosphoprotein bands from duplicate experiments. Quantitative treatment effects on total 27kDa protein phosphorylation are shown in Table I1. While both enantiomers increased total 27kDa phosphorylation by up to 3-fold, there appeared to be different time courses for the two enantiomers. -G showed nearly 2-fold phosphorylation at 30 min and 3-fold enhancement of 27kDa phosphorylation by 60 min. Menadione (vitamin K3), a related napthoquinone known to produce intracellular free radicals in EC at sublethal drug concentrations (8), similarly increased 27kDa protein phosphorylation in cultured porcine or human EC. Interestingly, menadione produced a profile of 27kDa isoform phosphorylation similar to that of -G rather than +G. The identity of the 27kDa phosphoproteins was determined to be low molecular weight (27-28kDa) heat-shock protein, hsp-27, based on immunoblot analysis using a high titre anti-hsp27/28 antisera supplied by W.J. Welch (10). DISCUSSION We have previously shown that various antimitochondrial agents including 43 inhibit tumor cell proliferation in culture alter producing structural and functional mitochondrial damage that is most evident 24h after drug administration (I,2). Some of these agents, like the bioflavinoid FAA, have been shown to maximally reduce tumor ATP levels within I-6h of in vivo injection, suggesting that they can produce rapid alterations in tumor physiology and blood flow (4-6). The present study utilized phosphorus and deuterium spectroscopic techniques to demonstrate that -G reduces tumor ATP and blood flow by 50% and 60%, respectively, within 60 min of in vivo administration. If these early tumor ischemic effects are mediated by direct endothelial injury within the newly vascularized tumors, then -G might be expected to metabolically alter or injure cultured EC within a similar time frame. Both enantiomers enhanced EC leakiness but -G did so at a 10-fold lower concentration than +G, an effect similar to its greater antimitochondrial potency. Furthermore, the enhanced endothelial leakiness was prominent only after a 24h exposure to drug, comparable to the kinetics of -G induced antimitochondrial effects on cultured tumor cells (I). Thus, it appears as if this delayed form of EC damage does not account for the early reduction in tumor blood flow observed in vivo. In contrast, -G produced a 2 to 3-fold increase in phosphorylation of 27kDa hsp-27 protein within 30-60 min of drug exposure. The multiple isoforms of hsp-27 have been shown to be rapidly affected in cultured EC exposed to tumor necrosis factor-(~, with a more acidic hsp-27 isoform showing enhanced

PL-72

Antitumor Endothelial Effects of Gossypol

Vol. 49, No. 12, 1991

phosphorylation and the most basic isoform showing decreased phosphorylation (11). This is similar to the -G induced phosphorylation pattern of hsp-27 isoforms we observed on 2-D gel electrophoresis. Under the serum-free conditions of this assay in which the +G effects could have been disproportionately amplified (12), +G appeared to increase phosphorylation of all hsp-27 isoforms while -G selectively reduced phosphorylation of the most basic isoform and enhanced phosphorylation of the most acidic hsp-27 isoform. This difference in isoform phosphorylation patterns is noteworthy since the isomers of G are equally potent inhibitors of PKC (I), and PKC function has been implicated in hsp-27 phosphorylation (13). The similarity between the -G pattern of hsp-27 phosphorylation and that reported for tumor necrosis factor-(~ is of some interest because this cytokine has been implicated in the ischemic tumor necrosis that results from FAA (14). The fact that menadione induces a profile of hsp-27 phosphorylation similar to -G is suggestive because menadione produces oxyradicals and H202 in cultured tumor cells and EC alike (8,15). Thus, local generation of oxyradicals by readily oxidizable naphthols like -G may be the initial stimulus for both hsp-27 phosphorylation and reduced regional blood flow in neovascularized tumors, much like the mechanism of oxyradical induced tissue damage that occurs with ischemia-reperfusion injury (16,17).

ACKNOWLEDGMENTS This work was supported by grants CA-46966 from the National Cancer Insitute, CH-235 from the American Cancer Society, and a Research Starter Grant from the Pharmaceutical Manufacturers Association Foundation, Inc. REFERENCES 1.

C.C.BENZ, M.A. KENIRY, JM. FORD, A.J. TOWNSEND, F.W. COX, S. PALAYOOR, S.A. MATLIN, W.N. HALT, and K.H. COWAN, Mol. Pharmacol. 37 840-847 (1990). 2. C. BENZ, C. HOLLANDER, M. KENIRY, T. JAMES, and M. MITCHEL, J. Clin. Invest. 79 517-523 (1987). 3. C. BENZ, M. KENIRY, and H. GOLDBERG, Contraception37 221-228 (1988). 4. S.B. IYER, H.I. GOLDBERG, and C.C. BENZ, Invest. Radiol. 25 1076-1084(1990). 5. S.B. IYER, M.W. DGREGORIO, H. ASGARI, M.A. KENIRY, J. LIU, B.A. TEICHER, and C.C. BENZ, Life Science Advances - Mol. Pharmacol. (in press, 1991). 6. J.L. EVELHOCH, M.C. BISSERY, G.G. CHABOT, N.E. SIMPSON, C.L. MCCOY, LK. HEILBRUN, and T.A. CORBETT, Cancer Res. 48 4749-4755 (1988). 7. J.W. MIER, E.P. BRANDON, P. LIBBY, M.W. JANICKA, and F.R. ARONSON, J. Immunol. 1432407-2414 (1989). 8. A. BARCHOWSKY, K. TABRIZI, R.S. KENT, and A.R. WHORTON, J. Clin. Invest. 83 1153-2259(1989). 9. P.H.O'FARRELL, J. Biol. Chem. 250 4007-4021 (1975). 10. A-P.ARRIGO, J.P. SUHAN, and W.J. WELCH, Mol. Cell. Biol. 8 5059-5071 (1988). I1. B. ROBAYE, A. HEPBURN, R. LECOCQ, and W. FIERS, Biochem. Biophys. Res. Comm. 163 301-308 (1989). 12. N. TANPHAICHITR, L.M. FITZGERALD, and S.A. MATLIN, J. Andro. 9. 270-277 (1988). 13. J.M. DARBON, M. ISSANDOU, J.F. TOURNIER, and F. BAYARD, Biochem. Biophys. Res. Commun. 168 527536 (1990). 14. J.C. ANDERSEN, and C.I. SHELLY, Proc. Amer. Assoc. Cancer Res. 32 397 (1991). 15. L.M. NUTTER, E.O. NGO, G.R. FISHER, and P.L. GUTIERREZ, Proc. Amer. Assoc. Cancer Res. 32 5 (1991). 16. C.E.CROSS, Ann. Intern. Med. 107 526-545 (1987). 17. A.M. LEFER, P.S. TSAO, D.J. LEFER, and M. XlN-LIANG., FASEB J. 5 2029-2034 (1991).