In Vivo Biodistribution of Ginkgolide B, a Constituent of Ginkgo Biloba

Makiko Suehiro1,4 Norman R. Simpson2 Mark D. Underwood3 In vivo Biodistribution of Ginkgolide B, a Constituent of John Castrillon3 Koji Nakanishi1 Ginkgo biloba, Visualized by MicroPET Ronald van Heertum2 rgnlPaper Original

Abstract by various organs including the liver, the intestine and possibly the stomach. Consequently, in plasma, the proportion of the io- The in vivo dynamic behavior of ginkgolide B (GB), a terpene lac- nized form of GB increases dramatically with time. Thereafter tone constituent of the Ginkgo biloba extracts, in the living ani- the ratio between the 2 forms appears to shift slowly towards mal was visualized by positron emission tomographic (PET) ima- equilibrium. The results suggest that more attention needs to be ging using a GB analogue labeled with the positron emitter 18F. focused on in vivo dynamics between the 2 forms of GB. The in vivo imaging studies, combined with ex vivo dissection experiments, reveal that GB exists in 2 forms in the body: the origi- Key words nal GB with its lactone rings closed and a second form with one Ginkgolide B ´ positron emission tomography ´ in vivo biodistri- of the rings open. The original GB in plasma is taken up rapidly bution of ginkgolide B ´ microPET ´ Ginkgo biloba ´ Ginkgoaceae

Introduction the role of ginkgolides in both physiological and pathological conditions. We report here the in vivo behavior of 18F-labeled 622 Extracts from leaves of the Ginkgo tree (Ginkgo biloba L.) such as ginkgolide B in the rodent visualized by the PET technique for EGb 761 have been utilized as complimentary medications for small animals, microPET. To confirm that the 18F-labeled GB decades [1], [2], and in vitro bioactivities of ginkgolides, terpene (Fig.1a) behaves in a similar fashion to GB itself in vivo, we com- lactone constituents of the G. biloba extracts, are well document- pared its biodistribution with that of 3H-labeled GB (Fig.1b). ed: they are thought to be platelet activating factor (PAF) antagonists [3], glucocorticoid synthesis regulators [4], [5], [6], in- hibitors of production of inflammatory markers such as tumor Materials and Methods necrosis factor-a, nitric oxide, interleukin-1 [7], [8], and glycine- 18 18 3 or GABAA-gated anion channel blockers [9], [10]. However, their The radiosyntheses of the F-labeled GB, 7- FGB, and HGB are in vivo behavior in the living body has not been reported. Visua- described elsewhere [11], [12]. The animal experiments were lization of their in vivo dynamic distribution by means of an ima- performed according to the animal protocol approved by the ging technique such as positron emission tomography (PET) with Columbia University Medical Center Institutional Animal Care an analogue of ginkgolides labeled with a positron emitter such and Use Committee. SD rats (male) weighing 250±350 g were as 11Cor18F provides unique information which sheds light on used.

Affiliation 1 Department of Chemistry, Columbia University, New York, NY, USA 2 Department of Radiology, Columbia University, New York, NY, USA 3 Department of Psychiatry, Columbia University, New York, NY, USA 4 Citigroup Biomedical Imaging Center, Weill Medical College of Cornell University, New York, NY, USA

Correspondence Makiko Suehiro, Ph. D. ´ Department of Radiology ´ Citigroup Biomedical Imaging Center ´ Weill Medical College of Cornell University ´ 516 East 72nd Street ´ New York ´ NY 10021 ´ USA ´ Phone: +1-212-746-5853 ´ E-mail: [email protected]

Received August 24, 2004 ´ Accepted May 17, 2005 Bibliography Planta Med 2005; 71: 622±627 ´ Georg Thieme Verlag KG Stuttgart ´ New York DOI 10.1055/s-2005-871267 ´ Published online July 18, 2005 ISSN 0032-0943 Fig. 1 Chemical structures of the 18F- labeled analogue of ginkgolide B and 3H- ginkgolide B. rgnlPaper Original

MicroPET imaging with 7± 18FGB A rat was anesthetized with isoflurane (2% in compressed air) blown over the nose and placed in the center of the field of view of the microPET camera (Concord R4, Microsystems Inc.), where the spatial resolution is the highest. The animal was injected via 623 the tail vein with approximately 300 mCi of 7-18FGB with specific activity of approximately 1 Ci/mmol, and serial images were ac- quired from the time of injection through to the termination of the study. Images of the animal were analyzed by placing regions of interest (ROIs) and generating time activity curves. At 60 and 90 minutes, two hundred mL of blood were drawn from the rat through a catheter inserted into the femoral vein and centrifuged at 3000 rpm for 5 minutes at room temperature. The plasma was separated and treated with 200 mL of acetonitrile, and the protein sediment was removed by centrifugation. The upper layer was passed through a 0.22 mm pore filter and injected into an analyti- Fig. 2A:MicroPET image (projection) of the rat body showing the cal C18 HPLC column (4.6 mm i.d. 25 cm, Microsorb, Varian) distribution of 7-18FGB for the first 2.75 minutes post injection of 7- eluted with a mobile phase of acetonitrile and water (30/70) con- 18FGB. White spots representing the ªhottestº regions, where 7-18FGB taining 0.1 M ammonium formate at a flow rate of 2 mL/min. The accumulated, are the kidneys. The broad gray region is the liver. The effluent was monitored with an NaI gamma detector (Bioscan, animal was positioned face down in the camera. B: Time course of 7- Flow-Count) and collected at every one minute using a fraction 18FGB in rat kidneys. Squares and circles represent right and left kid- collector. The radioactivity in the fractions was measured by neys, respectively. Data are means 1 s.d. (n = 9±10 pixels). gamma counting. To confirm the information obtained by micro- PET, dissection experiments were performed in parallel to the imaging studies. Briefly, the animals injected with 7-18FGB intra- ment was repeated 3 times to re-confirm the in vivo behavior of venously were sacrificed at 10, 30, 60 and 90 minutes (n = 3 ani- 7-18FGB. mals per time point) or at the end of the image acquisition, and the organs were dissected, lightly rinsed with saline and blotted. Biodistribution studies with 3HGB Blood clots were removed from the heart, and stomach contents Rats received approximately 1 mCi of 3HGB via the tail vein. After were separated from the stomach body. 7-18FGB in the organs 10, 30, 60, 90 minutes and 5 hours post injection, the animals was determined by gamma counting after measurement of organ (n = 3 animals per time point) were sacrificed, and the tissues weight. The imaging study combined with a dissection experi- were taken out, lightly rinsed with saline, blotted and weighed.

Suehiro M et al. In vivo biodistribution ¼ Planta Med 2005; 71: 622±627 Radioactivity concentrations in the tissues were determined by philic component derived from 3HGB and circulating in plasma liquid scintillation counting after the tissues were solubilized at 60 and 90 minutes post injection, the first 2±4 fractions from with 0.5 to 2 mL of Soluene-350 (Packard) at 58 8C for a few the HPLC column, where the radioactivity peak appeared, were hours and dissolved in 15 mL of the scintillation cocktail Hionic- collected. The combined eluate was diluted with 15 mL of water 3 Fluor (Packard). HGB in plasma was analyzed by reversed-phase and passed through a C18 Sep-Pak (Waters), which was washed

HPLC using an analytical C18 HPLC column (4.6 mm i.d.25 cm, with 10 mL of water. The radioactivity retained on the Sep-Pak Vydac) with a mobile phase of acetonitrile and water (20/80) was eluted with 1.5 mL of methanol. After the solvent was evapo- containing 0.1 M ammonium formate at a flow rate of 2 mL/ rated under reduced pressure, the residue was treated with 1 N min. The effluent was fractioned at every one minute and count- HCl in methanol at 58 8C for 5 days and applied to the above- ed for 3H radioactivity. Under these conditions, the retention mentioned HPLC system again. time of the authentic GB was 8.6 minutes. To identify the hydro- rgnlPaper Original

624

Fig. 3A:MicroPET image (projection) of the rat body taken at 2.75±6.5 minutes post 7-18FGB injection, showing a surge of radioactivity in the duodenum near the jejunum. The animal positioning is the same as Fig. 2A. B: MicroPET image (projection) of the rat body taken at 19±56 minutes after 7-18FGB injection, showing accumulation of the radiotracer into the gastric antrum, the pylorus, the duodenum, and part of the intestine. The animal positioning is the same as Fig. 2A. C: Time course of 7-18FGB in 4 regions of interests (ROIs): the proximal duodenum (green), the gastric antrum (red), the liver (blue) and the bile duct (yellow). Data are means 1 s.d. (n = 5±12 pixels). D: Distribution of 7-18FGB at 90 minutes post injection determined by dissection. LG = lung, ST&DD = stomach and duodenum, LV = liver, KD = kidney, HT = heart, PAN = pan- creas, SP = spleen, BR = brain, BL = blood. Data are means 1 s.d. (n = 3).

Suehiro M et al. In vivo biodistribution¼ Planta Med 2005; 71: 622±627 Results and Discussion

In the in vivo PET images of the first 2.75 minutes post injection, 7-18FGB was observed to be excreted into the kidneys and washed-out of the regions quickly (Figs. 2a and b). Between 2.75 and 20 minutes, there was a surge of radioactivity in the duodenum near the jejunum (Fig. 3a). A time-activity curve of a region of interest (ROI) placed over the proximal duodenum (Fig. 3c, green line) suggests that 7-18FGB starts appearing in the region at 4 minutes post injection, peaks at 17 minutes and clears quickly. Thereafter, the radiotracer was found predominantly in the gastric antrum, the pylorus, and the duodenum (Fig. 3b). The distribution was confirmed by dissection (Fig. 3d). It was also revealed

that most of the radioactivity in the stomach region was asso- Paper Original ciated with its contents rather than the stomach body. The 7- 18FGB accumulation in the region started at about 10 minutes, gradually increased and then plateaued (Fig. 3c, red line). At 90 Fig. 4 Distribution of 7-18FGB vs. that of 3H-GB in rat determined by minutes, the region had 9.5 and 19 times more radioactivity dissection (correlation coefficient: 0.98). than the regions corresponding to the proximal duodenum and the liver (Fig. 3c, blue line), respectively. In contrast, a time-activity curve for the bile duct (Fig. 3c, yellow line) indicates that the secretion of GB into the bile occurs continuously throughout the and the ring-opened GB dramatically decreased with time to 15 observation time. A comparison between the biodistribution of to 85 at 60 minutes. Since at 5 hours post administration we still 7-18FGB with that of 3H-labeled GB revealed a good correlation found 10% of GB circulating in plasma in the original form, this (correlation coefficient: 0.98) (Fig. 4) confirming that 7-18FGB be- may mean that ionized GB in plasma is continuously converted haves similarly to GB itself in vivo. Thus we concluded that the to the original form, as GB in the original form is utilized in the dynamic distribution of 7-18FGB visualized by microPET repre- body, until a new equilibrium is reached. This may also explain sents that of GB. the long retention of GB in human plasma with an elimination half-life of 10.6 hours [15]when GB is given orally in humans, Analysis of 7-18FGB and 3HGB in plasma by reversed-phase HPLC the most frequent route of G. biloba extract administration. The indicated that while at 30 minutes after injection, 80±85% of the ratio of the two forms of GB, the ionized to the original, found in radiolabeled GB is intact having the original form, at 60 and 90 plasma at 5 hours, 9 to 1, also suggests that in vivo the ionization minutes, a more hydrophilic component, which elutes near the constant may fall into a lower range than the value measured in 625 void volume, becomes predominant (Fig. 5a). We collected the vitro. radioactive eluate near the void volume, treated it with methanol containing HCl and then analyzed the sample by HPLC again. The plasma analysis data and the microPET images, taken to- After the acid treatment, 75% of the radioactivity appeared at the gether, suggest that the distribution of 7-18FGB visualized by retention time of the original GB (Fig. 5b), suggesting that the microPET is most likely that of the original form of GB because compound eluting near the void volume (Fig. 5a) was, in fact, the events visualized by microPET (Fig. 3) coincide with the rapid GB but in a different form, presumably GB with one of the lac- disappearance of the original GB from the plasma (Fig. 5a). This tone rings open (Fig. 5c). This result strongly implicates that in implicates that a) GB in the original form present in plasma is ra- plasma, and possibly also in the body, GB exists in 2 forms, i.e., pidly taken up by various organs such as the kidneys, the intes- the form in which its 3 lactone rings are closed (the original GB) tine, and the liver, b) it accumulates into the gastric juice, and c) and a second form with one of the rings open. We suspected that it is also continuously excreted into the bile. the radiolabeled compound eluted in the void volume might be a different form of GB since we had observed that while 3H-labeled The high accumulation and long retention of 7-18FGB in the gas- GB stored in ethanol maintained its intact form, 3H-labeled GB tric juice (Fig. 3c red line), with a gastric juice-to-blood ratio of stored in normal saline-pH 5.8±6.5-transformed slowly into a 20 to 1 at 90 minutes, may be attributable to the physiological hydrophilic form, which could be converted back to intact GB duodenogastric reflux, where 7± 18FGB secreted in the proximal after addition of an acid in an organic solvent such as ethanol. duodenum is propelled into the gastric antrum together with bi- carbonate and IgA, which is thought to be associated with de- The lactone ring opening of GB in a basic medium was first re- fense mechanisms of the antral mucosal barrier function [16], ported in 1967 by Maruyama [13]. Recently, Zekri has reported [17]. Another possibility is that GB penetrates the blood-antrum the ionization constant (pK) of GB in an aqueous solution, where barrier. The significance of the finding that 7-18FGB is concen- one of the three lactone rings opens, to be 6.9 to 7.4 [14], which is trated in the gastric antrum needs to be further assessed. None- close to a physiological pH. This suggests that, theoretically, GB theless, it is interesting to note that the transfer of some drugs, exists in 2 forms in a ratio of approximately 1 to 1 in plasma if such as metroindazole or clarithromycin, from blood into the they are in equilibrium. In vivo, however, we observed that the stomach has been a focus of attention in association with the radiolabeled GB in the original form disappeared quickly from treatment of chronic gastritis caused by Helicobacter pylori [18]. the plasma and, as a consequence, the ratio between the original It is also noteworthy that ascorbic acid, an antioxidant and

Suehiro M et al. In vivo biodistribution ¼ Planta Med 2005; 71: 622±627 rgnlPaper Original

626

Fig. 5A:3HGB in plasma at 60 minutes after injection. B: The fractions eluted near the void volume (a) with radioactivity (Fractions 2±4) were

collected and the radioactivity in the fractions was extracted using a C18 Sep-Pak column, and treated with 1 N HCl. After the acid treatment, 75% of the radioactivity was eluted at the retention time (9 minutes) of the authentic GB. C: The phenomenon depicted in Figs. 5A and B is attributable to the ring-opening of GB in plasma and possibly also in the body.

known for its anti-gastric cancer activity, is reported to be in a In the brain, we found an average 18F radioactivity of approxi- greater concentration in the antrum than in the body of the mately 0.02% dose/gram tissue at 10 minutes by dissection. The stomach [19]. Whether GB secreted in gastric juice is involved brain distribution showed some regional heterogeneity with the in the defense mechanism by which antral mucosa is protected, highest concentration of approximately 0.04% dose/gram in ol- or whether it acts as a PAF receptor antagonist needs also to be factory bulbs and the hypothalamus. However, the radioactivity further investigated. Interestingly, it has been reported that in- in the brain dropped quickly to 0.004% dose/gram tissue at 30 traperitoneally- or intravenously-administered GB attenuates minutes. These 18F activity concentrations in the brain at 10 and gastric damage mediated by PAF [20], [21]. 30 minutes were below the detection limit of the PET imaging device. We had assumed that the neuroprotective or memory- enhancing activities of ginkgolides stemmed from their direct

Suehiro M et al. In vivo biodistribution¼ Planta Med 2005; 71: 622±627 actions on their target tissues in the brain such as hippocampus. 5 Amri H, Drieu K, Papadopoulos V. Ex vivo regulation of adrenal cortical In fact, GB showed very low blood-brain barrier (BBB) permeabil- cell steroid and protein synthesis, in response to adrenocorticotropic hormone stimulation, by Ginkgo biloba Extract Egb 761 and isolated ity. The rat head images showed that GB is also secreted into sali- gingkolide B. Endocrinology 1997; 138: 5415±26 va. 6 Papadopoulos V, Widmaier EP, Amri H, Zili A, Li H, Culty M et al. In vivo studies on the role of the peripheral benzodiazepine receptor (PBR) in The results of this investigation suggest that the unionized form steroidogenesis. Endocrine Research 1998; 24: 479±87 7 Du Z-Y, Li X-Y. Effects of ginkgolides on interleukin-1, tumor necrosis of GB, where the three lactone rings are closed, is the bioactive factor-a and nitric oxide production by rat microglia stimulated with li- form of GB. Nevertheless, since the original GB seems short-lived popolysaccharides in vitro. Arzneim-Forsch/Drug Res 1998; 48: 1126± compared to the ionized GB, we cannot rule out the possibility 30 that the more prominent form of GB that exerts long-term ther- 8 Cheng F, Siow YL, Karmin O. Inhibition by gingkolides and bilobalide of the production of nitric oxide in macrophages (THP-1) but not in apeutic effects is the one with a lactone ring open. Further inves- endothelial cells (HUVEC). Biochem Biopharmacol 2001; 61: 503 ±10 tigation is warranted to determine the role of the ionized form of 9 Kondratskaya EL, Lishko PV, Chatterjee SS, Krishtal OA. BN52021, a GB in vivo. platelet activating factor antagonist, is a selective blocker of glycine- gated chloride channel. Neurochemistry International 2002; 40: rgnlPaper Original 647±53 In conclusion, our investigation on the in vivo behavior of GB has 10 Ivic L, Sands TTJ, Fishkin N, Nakanishi K, Kriegstein AR & Strùmgaard K. revealed that GB, a terpene trilactone derived from the Ginkgo Terpene trilactones from Ginkgo biloba are antagonists of cortical gly- leaves, is present in 2 forms in vivo. The nonionized form of GB cine and GABAA receptors. J Biol Chem 2003; 278: 49279±85 is rapidly taken up by the kidneys, the liver, the intestine and 11 Suehiro M, Simpson N, van Heertum R. Radiolabeling of ginkgolide B with 18F. J Labelled Compds Radiopharm 2004; 47: 485±91 possibly the stomach. The ionized form, with one of the lactone 12 Suehiro M, Strùmgaard K, Simpson N, Nakanishi K, van Heertum R. rings of GB open, remains longer in the body than the original Radiolabeling of ginkgolide B with 18F and 3H. J Labelled Compods form of GB and may play an important physiological role. More Radiopharm 2003; 46 (Suppl 1): S214 attention should be focused on the dynamic equilibrium be- 13 Maruyama M, Terahara A, Itagaki Y, Nakanishi K. Ginkgolide I. Tetrahe- dron Lett, 1967: 299 ±302 tween the original and the ring-opened forms of GB in vivo. 14 Zekri O, Boudeville P, Genay P, Perly B, Braquet P, Jouenne P, Burgot J-L. Ionization constants of ginkgolide B in aqueous solution. Anal Chem 1996; 68: 2598 ±604 Acknowledgements 15 Fourtillan JB, Brisson AM, Girault J, Ingrand J, Decourt JP, Drieu K et al. Proprietes pharmacocietiques du bilobalide et des ginkgolides A et B chez le sujet sain aprs administrations intraveineuses et orales d'ex- Financial supports were provided in part by grants from NARSAD trait de Ginkgo biloba (EGb 761). Therapie 1995; 50: 137±44 and NIH (MH68817, MH59198). We would like to thank Krsitian 16 Dalenback J, Abrahamson H, Bjornson E, Fandriks L, Mattson A, Olbe L Strùmgaard, Nathan Fishkin, Stanislav Jaracz and Young Jang for et al. Human duodenogastric reflux, retroperistalsis, and MMC. Am J Physiol 1996; 270: G113±22 their assistance in the dissection experiments. We also thank Ti- 17 Castedal M, Bjornson E, Gretarsdottir J, Fjalling H, Abrahamson H. gran Sinanian of PETNET Pharmaceuticals for 18F production. Scintigraphic assessment of interdigestive duodenogastric reflux in humans: distinguishing between duodenal and biliary reflux materi- 627 al. Scand J Gastroenterol 2000; 35: 590±8 18 Veldhuyzen van Zanten SJO, Goldie J, Hollingsworth J, Silletti C, Ri- References chardson H, Hunt RH. Secretion of intravenously administered anti- biotics in gastric juice: implication for management of Helicobacter 1 Curtis-Prior P, Vere D, Fray P. Therapeutic value of Gingko biloba in re- pylori. Clin Pathol 1992; 45: 225 ±7 ducing symptoms of decline in mental function. J Pharm Pharmacol 19 Rathbone BJ, Johnson AW, Wyatt JI, Heatley RV, Losowsky MS. Ascor- 1996; 51: 535±41 bic acid: a factor concentrated in human gastric juice. Clininal Science 2 Le Bars PL. Magnitude of effect and special approach to Ginkgo biloba 1989; 76: 237±41 extract EGb 761 in cognitive disorders. Pharmacopsychiatry 2003; 36 20 Filep J, Herman F, & Braquet P. Platelet-activating factor may mediate (Suppl 1): 544±9 dexamethasone-induced gastric damage in the rat. Lipids 1991; 26: 3 Braquet P. Ginkgolides: chemistry, biology, pharmacology and clinical 1356±8 perspectives. J. R. Prous Science Publishers, Barcelona: 1988 21 Wallace JL, Steel G, Whittle BJR, Lagente V, Vargaftic B. Evidence for 4 Amri H, Ogwuegbu SO, Boujrad N, Drieu K, Papadopoulos V. In vitro platelet-activating factor as a mediator of endotoxin-induced gastro- regulation of peripheral-type benzodiazepine receptor and glucocor- intestinal damage in the rat. Gastroenterol 1987; 93: 765±73 ticoid synthesis by Ginko biloba extract EGb 761 and isolated ginkgolides. Endocrinology 1996; 137: 5707±18

Suehiro M et al. In vivo biodistribution ¼ Planta Med 2005; 71: 622±627