Life Sciences 114 (2014) 93–101

Contents lists available at ScienceDirect

Life Sciences

j ourna l homepage: www.e lsev ie r .com/ locate / l i fesc ie

Transport of ginkgolides with different lipophilicities based on anhCMEC/D3 cell monolayer as a blood–brain barrier cell model

Shuwei Ma a,⁎, Xingyan Liu b, Qingrun Xu c, Xiantao Zhang d

a Pharmaceutical Engineering, Institute of Chemistry and Chemical Engineering, Qiqihar University, Qiqihar 161006, PR Chinab Department of Complex Prescription of TCM, China Pharmaceutical University, Nanjing 210038, PR Chinac Institute of TCM, Heilongjiang Academy of Traditional Chinese Medicine, Harbin, 150036, PR Chinad Guangdong Research Institute of Traditional Chinese Medicine, Guangzhou, 510520 PR China

⁎ Corresponding author. Tel./fax: +86 452 2738310.E-mail address: mashuwei_2004@163.com (S. Ma).

http://dx.doi.org/10.1016/j.lfs.2014.08.0060024-3205/© 2014 Elsevier Inc. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 21 March 2014Accepted 8 August 2014Available online 17 August 2014

Keywords:TransportBBBGinkgolidehCMEC/D3 cellApparent permeability coefficientsEfflux ratioLipophilicity

Aims: In this report, the transport of ginkgolides with different lipophilicities was investigated using an hCMEC/D3 cell monolayer as a blood–brain barrier (BBB) cell model in vitro in an attempt to explain ginkgolide transportpath mediated by lipophilicity.Main methods: The log P values of ginkgolides were determined by measuring the distribution of the moleculebetween oil and water. Additionally, the cytotoxicity of ginkgolides on hCMEC/D3 cells was assayed with theMTT method. Ginkgolide contents were determined with an ultra performance liquid chromatograph equippedwith an evaporative light scattering detector (ULPC–ELSD) method. Apparent permeability coefficients (Papp)and efflux ratios (PappBL → AP/PappAP → BL) were then calculated to describe the transport characteristics ofginkgolide.Key findings: The transport of ginkgolide A, ginkgolide B, ginkgolide C, and ginkgolide J across the hCMEC/D3 cellmonolayer was non-directional. Additionally, ginkgolide C transport on the cell monolayer was time- andconcentration-dependent in the paracellular pathway controlled by cytochalasin D (a tight junction modulator).

The transport of ginkgolide N, ginkgolide L, and ginkgolide K across the cell monolayer displayed clear direction-ality at low ginkgolide concentrations. This behavior indicated that the transport of ginkgolide N, ginkgolide L,and ginkgolide K was influenced by the transcellular pathway containing an efflux protein accompanied by theparacellular pathway for passive diffusion. Additionally, the transport of ginkgolide K was increased significantlyby co-culturing with a P-gp inhibitor.Significance: These findings provide important information for elucidating ginkgolide transport pathways andmay be beneficial for the design of ginkgolide molecules with high neuroprotective effects.

© 2014 Elsevier Inc. All rights reserved.

Introduction

Ginkgo biloba extract obtained from Ginkgo biloba leaves has beenreported to be an antioxidant and neuroprotective agent in a varietyof conditions, including ischemia (Sung et al., 2012), oxidative stress(Zhao et al., 2014) and neuronopathies (Saleem et al., 2008). Up tonow, flavones glycosides, terpene lactones and other chemical compo-nents have been separated from and identified in Ginkgo biloba extracts(Wang et al., 2013; Dew et al., 2013). Previous studies have shown thatginkgolide A, ginkgolide B, ginkgolide C, ginkgolide J, and ginkgolide Kall have neuroprotective effects and antioxidative properties, especiallyon neuron injuries induced by cerebral ischemia (Chen et al., 2012; Qinet al., 2014; Ma et al., 2012a,b; Numa et al., 2007; Vitolo et al., 2009). In-terestingly, ginkgolide activities, particularly in protecting against cere-bral ischemia injury,were found tobe affected by the chemical structure

of ginkgolide (Ma et al., 2012). Drugs to treat cerebral ischemia mustfirst penetrate the blood–brain barrier (BBB) to be effective. Obviously,the ability of a ginkgolide to cross the BBB is a determining factor whenevaluating its neuroprotective action in cerebral neurons and itsdruggability. Therefore, there is a need to clarify whether the ability ofginkgolide to cross the BBB is related to its chemical structure.

In previous studies, primary and immortalized cells were used asBBBmodels in vitro to study the transport of drugs across the BBB. How-ever, all model BBBs established to date have exhibited significantshortcomings, ranging from failure to exhibit the BBB phenotype to ge-netic instability (Chen et al., 2013; Zehendner et al., 2014; Hsiao et al.,2008; Förster et al., 2008). As a consequence, it is critical to constructan in vitro BBB model to evaluate drug transport across the BBB. ThehCMEC/D3 cell line is an immortalized human endothelial cell linethat retains most of the morphological and functional characteristicsof brain endothelial cells, even when not co-cultured with glial cells.Additionally, it has been proposed that hCMEC/D3 cells may constitutea reliable in vitro model of the human BBB, as these cells express ATP-

94 S. Ma et al. / Life Sciences 114 (2014) 93–101

blinding cassette (ABC) efflux transporters, such as P-glycoprotein(P-gp),multidrug resistanceproteins (MRP) and the breast cancer resis-tance protein (BCRP) (Weksler et al., 2013). Therefore, this cell line wasselected as our in vitro BBB model. The log P value, which indicateslipophilicity, is an important criterion for evaluating the efficacy of asubstance, especially brain agents that must be able to penetrate theBBB. Therefore, in our experiments, seven ginkgolideswere investigatedto determine the relationship between transport and lipophilicity.

Materials and methods

Chemicals and reagents

The hCMEC/D3 cell line was supplied by the Guangzhou JINIAOBiological Technology Co., LTD, a business agent in China. Ketoconazole,chloramphenicol, 3-(4, 5-dimethylthiazol-2yl)-2, 5-diphenyl tetrazoli-um bromide (MTT), cyclosporin A, Ko143, MK-571, cytochalasin D,dimethyl sulfoxide (DMSO), penicillin G, amphotericin B, streptomycin,and trypsin-EDTA were purchased from Sigma-Aldrich (Shanghai,China). Fetal bovine serum (FBS) and Hank's Balanced Salt Solution(HBSS) were purchased from Sijiqing Biotech Co. Ltd. (Hangzhou,China). Basal medium-2 (EBM-2), VEGF, IGF-1, EGF, basic FGF, hydro-cortisone and ascorbate were purchased from Gibco Life Technologies(Grand Island, NY, USA). Gentamycinwasmanufactured by Lunan Phar-maceutical Co. Ltd. (Shandong, China). N-octyl alcohol was obtainedfrom Better Chemical Co. Ltd. (Shandong, China). Verapamil was pur-chased from the National Institute for the Control of Pharmaceuticalsand Biological Products (Beijing, China). Methanol was of HPLC gradeand was obtained from Fisher Scientific Products (Fair Lawn, NJ, USA).Ginkgolide A, ginkgolide B, ginkgolide C and ginkgolide J were pur-chased from Nanjing Chunqiu Biotechnology Co. Ltd. (Nanjing, China).Ginkgolide L, ginkgolide N and ginkgolide K were isolated and purifiedfrom the ginkgo leaf; molecule identities were confirmed by Dr. ZhangXiantao from the Guangdong Institute of Traditional Chinese Medicine(TCM). The chemical structures of all ginkgolides used in our experi-ment are shown in Fig. 1. The purities of these ginkgolides were allabove 98% as determined by an ultra performance liquid chromato-graph equipped with an evaporative light scattering detector(UPLC–ELSD). All other reagents and chemicals used were of analyticalgrade. Water was triple distilled. Transwell polycarbonate filters(surface area, 0.33 cm2; pore size, 0.4 μm), Transwell™ 24-well plates

Fig. 1. Chemical structures o

and the electrical resistance system (ERS) were purchased from theMillipore Corporation (Bedford, MA, USA).

Instrumentation and assays of ginkgolides by UPLC–ELSD

All analyses were performed on an Agilent ultra performance liquidchromatographic system (Series 4200, Agilent Technology, Palo Alto,CA, USA), which consisted of a G4220B 1290 quaternary solvent deliv-ery pump, a G1379A on-line degasser, a G1316 C column heater, aG4226A auto-sampler and an Alltech 3300 evaporative light scatteringdetector (ELSD) (W.R. Grace Company, USA). Chromatographic datawere recorded and processed with Agilent Chemstation software. Theanalytical column was an Agilent XDB-C18 column (5 μm, 50 mm ×4.6 mm). The mobile phase was a mixture of solvent A (methanol)and B (water containing 1% acetic acid) employing gradient elution(from 20% A: 80% B to 60% A: 40% B, v/v) at a flow rate of 0.2 ml/min.The flow rate and temperature of nitrogen gas were controlled at1.6 l/min and at 48 °C, respectively. The column temperaturewasmain-tained at 55 °C. Peak area measurements were used to generate stan-dard calibration curves to determine the concentrations of ginkgolidecompounds in experimental samples.

Measurement of log P values

Ginkgolides were dissolved in n-octyl alcohol (5 ml) saturated withdeionized water to a final concentration of 100 μg/ml and shakenat 160 rpm and 30 °C for 10 min. When equilibrium between thetwo phases was achieved, samples were centrifuged at 3500 rpm tocompletely delaminate the phases. The aqueous layer was then collect-ed and dried under vacuum, then resuspended inmethanol. The samplefor ULPC–ELSD analysis was filtered through a 0.22 μmMillipore mem-brane. The N-octanol/water partition coefficient was calculated accord-ing to the following formula:

logP ¼ logρoρw

where ρo and ρw indicate the contents in the n-octanol and in thewater,respectively.

f the tested ginkgolides.

95S. Ma et al. / Life Sciences 114 (2014) 93–101

hCMEC/D3 cell cultures

Cells were grown in EBM-2 medium in a humidified atmosphere at37 °C with 5% CO2. The medium was supplemented with VEGF, IGF-1,EGF, basic FGF, hydrocortisone, ascorbate, gentamycin, 2.5% fetal bovineserum (FBS), 100 U/ml penicillin G, 0.25 mg/ml amphotericin B and100 mg/ml streptomycin as recommended by the manufacturer. Thecell medium was changed every 48 h, and the cells reached confluenceafter 5–6 days of incubation. For subculturing, cells were dissociatedwith 0.25% trypsin-EDTA, diluted 1:5 and subcultured in petri dishescollagen-coated with a 21 cm2 growth area (Corning Costar®,Badhoevedorp, The Netherlands).

For transport experiments (Yang et al., 2014), hCMEC/D3 cells wereseeded at a density of 2 × 104 cells/well on transwell inserts (polycar-bonate membrane; effective growth area, 0.33 cm2; pore size, 0.4 μm;diameter, 6.5 mm) coated with collagen. The cells were cultured usingan EBM-medium, and the medium was changed daily. Monolayerswere formed after culturing for 9–10 days. The integrity of the cellmonolayers was evaluated by measuring the transepithelial electricalresistance (TEER) with an electrical resistance system. Monolayerswith TEER values greater than 600 Ω/cm2 were used for transportexperiments.

Cytotoxicity of ginkgolide on hCMEC/D cells

The cell viability was assayed by theMTTmethod (Lang et al., 2012).Briefly, hCMEC/D3 cells were counted and seeded onto 96-well cultureplates at a density of 1 × 104 cells/well and cultured with EBM-2 medi-um (150 μl) for 48 h. The cells were treated with ginkgolides at variousconcentrations (0, 20, 40, 80, 120, 160, 200 and 240 μM) for 48 h. Then,20 μl ofMTT (10%, 5 mg/ml)was added into eachwell followed by incu-bation for 4 h. Following incubation, the culture mediumwas removed,and 150 μl DMSOwas added to eachwell. Plates were then vortexed for5min. The optical density (OD) of eachwellwas immediatelymeasuredon an ELISA microplate reader (Bio-Rad 680) at 570 nm to assess cellu-lar viability.

Transport of ginkgolides across hCMEC/D3 cell monolayer

Transport experiments on hCMEC/D cell monolayer were conductedas described previously with minor modifications (Qiu et al., 2012).After hCMEC/D cell monolayers were observed on the transwells of 24wells using a hemocytometer under a microscope, hCMEC/D3 cellmonolayer integrities were examined by measuring transepithelialelectrical resistances (TEER) with an electrical resistance system. Priorto the experiment, hCMEC/D3 cell monolayers were washed with awarm HBSS solution and preincubated with a transport medium for 1h at 37 °C and under 8% CO2 before TEER measurements. A monolayerwas only used if its TEER value was greater than 600Ω/cm2. In addition,the transport abilities of hCMEC/D3 cell monolayers were examinedusing standard assays with ketoconazole as the paracellular fluxmarkerand chloramphenicol as the transcellularfluxmarker (Yang et al., 2014).Transport experiments were conducted by adding HBSS solutions con-taining different final ginkgolide concentrations (5 μM, 10 μM, 20 μM,40 μM and 80 μM of ginkgolide A, ginkgolide B, ginkgolide C, ginkgolideJ, ginkgolide N, ginkgolide L, and ginkgolide K), to the apical side(AP, 0.4 ml) and by injecting HBSS solution into the basolateral side(BL, 0.6 ml). Cell culture inserts were shaken at 55 rpm and 37 °C in agas bath. An aliquot (100 μl) of medium from the BL side was taken at120 min for Papp (AP to BL) measurements. A corresponding volumeof HBSS buffer was simultaneously added to the BL side. In addition,transport experiments from the BL to the AP sidewere executed similarto transport experiments from the AP to the BL side, except that the BLside was used as the donor. An aliquot (100 μl) of medium from the APside was collected at 120 min for Papp (BL to AP) measurements, and a

corresponding volume of HBSS buffer was immediately added to theAP side.

Sample preparation

Each standard ginkgolide stock solution (10 mg/ml), includingginkgolide A, ginkgolide B, ginkgolide C, ginkgolide J, ginkgolide N,ginkgolide L, and ginkgolide K, was prepared by dissolving ginkgolidesin DMSO. Initial solutions of ginkgolides were prepared by dilutingstock solutions with Hank's balanced salt solution (HBSS, PH 7.4) fortransport experiments and with EBM-2 medium for hCMEC/D3 cellviability measurements. Stock solutions for seven ginkgolides were dis-solved in methanol to a final concentration of 1 mM. To build a calibra-tion curve for each analyte, working standard solutions were preparedby diluting stock solutions with methanol to provide a series of analyt-ical standards ranging from 1 μM to 100.0 μM. All solutions werevortexed, sonicated and stored at 4 °C prior to use.

Determination of the time and concentration dependence of ginkgolide Cand ginkgolide K transport across the hCMEC/D3 cell monolayer

Ginkgolide Cwas chosen to be a representative compoundbecause ithad the lowest Papp value relative to the other ginkgolides and alsoexhibited an efflux ratio below 1.5 for concentrations between 5 and80 μM. In contrast, ginkgolide K exhibited an efflux ratio above 2 forconcentrations between 5 and 40 μM. Therefore, ginkgolide C andginkgolide K were used in more detailed transport assays.

To observe the time dependence of ginkgolide C and ginkgolide Ktransport, 20 μM of ginkgolide C or ginkgolide K was added to the AP,and an HBSS solution was added to the BL. Samples were shaken(37 °C, 55 rpm). Aliquots (100 μl) were taken from the BL at 5, 10, 20,40, 80 and 120 min. Equivalent volumes (100 μl) of HBSS were addedto the BL after each sampling. To determine the concentration depen-dence of ginkgolide C and ginkgolide K transport, various concentra-tions of ginkgolide C or ginkgolide K (5 μM, 10 μM, 20 μM, 40 μM and80 μM) were added to the AP side. After shaking at 55 rpm for120 min in a 37 °C shaking gas bath, aliquots (100 μl) were collectedfrom AP for transport measurements. Immediately, 100 μl of HBSSsolution was supplied to the BL.

Measurement of the effect of cytochalasin D on ginkgolide C transport

Ginkgolide C was observed to permeate the cell monolayer via pas-sive diffusion. Hence, it was necessary to determinewhether ginkgolideC passes through the cell monolayer paracellularly. Cytochalasin D is acell-permeable mycotoxin that has been reported to decrease intestinalbarrier function by disrupting actin filaments and inhibiting actinpolymerization (Qiu et al., 2012). If the transport mechanism of a druginvolves passive paracellular diffusion, its transport would be boostedwhen the tight junction structure is opened or destroyed. To investigatethe transport of ginkgolide C in the absence or presence of cytochalasinD, cell monolayers were treated with different concentrations of cyto-chalasin D (10 μM, 20 μMand 40 μM) for 20min prior to ginkgolide C ad-dition. Papp(A → B) values for ginkgolide C were measured at 120 minaccording to the method described in the Transports of ginkgolidesacross hCMEC/D3 cell monolayer section.

Determination of the effect of efflux protein inhibitors on ginkgolide Ktransport

The effect of efflux protein inhibitors on ginkgolide K (20 μM)transport was further explored. Applying the cell monolayermethod de-scribed in the Transports of ginkgolides across hCMEC/D3 cellmonolayersection, the effects of P-glycoprotein (P-gp) antagonist (cyclosporin A,10 μM), BCRP antagonist (Ko143, 20 μM) and MRP antagonist(MK-571, 50 μM) on the Papp and efflux ratio of ginkgolide K were

96 S. Ma et al. / Life Sciences 114 (2014) 93–101

investigated. (Sha and Fang, 2004; Juvale et al., 2013; Vaidyanathan andWalle, 2001).

Assay of samples

Samples from both transwell sides of 24 plates were frozen, lyophi-lized and preserved below−20 °C for subsequent UPLC–ELSD analysis.Lyophilized samples from the receiving side of transwell insert filterswere resuspended in a volume of methanol corresponding to the origi-nal sample volume and centrifuged at 10,000 rpm for 5 min to precipi-tate proteins. A 2 μl aliquot of the supernatant was assayed by UPLC–ELSD. The apparent permeability coefficients (Papp) were calculatedfrom Eq. (1).

Papp ¼ ΔQ=ΔtAC0

ð1Þ

where ΔQ/Δt is the linear appearance rate of the compound on thereceiver (in μM/s), A is the membrane surface area (cm2), and C0 isthe initial concentration in the donor chamber (μM/cm3).

The efflux ratio was calculated from Eq. (2):

Efflux ratio ¼ Papp B →APappA → B

: ð2Þ

Statistical analysis

Statistical analysis was performed using SPSS Software (SPSS Inc.,version 13, Chicago, IL, USA). An independent sample t-test andone-way analysis of variance were used to analyze the data. Data arepresented as themeans± S.D., and P b 0.05was considered to be statis-tically significant.

Results

Ginkgolide log P values

Ginkgolide log P values are shown in Table 1. Our results indicatedthe order of ginkgolide log P values to be ginkgolide C, ginkgolide A,ginkgolide B, ginkgolide J, ginkgolide N, ginkgolide L and ginkgolide K.

Effect of ginkgolides on hCMEC/D3 cell viability

The effects of ginkgolides on hCMEC/D3 cell viabilities are shown inFig. 2. Ginkgolide A, ginkgolide B, ginkgolide C, ginkgolide J all observ-ably decreased the viability of hCMEC/D3 cells compared to controlcells at concentrations varying from 200 μM to 240 μM. Cytotoxicitywas clearly observed for ginkgolide K, ginkgolide L and ginkgolide N atconcentrations between120 μMand 240 μM,with inhibition rates risingsignificantly from 7.59% to 57.79% as concentrations were increasedfrom 120 μM to 240 μM. The cytotoxicities of ginkgolides on hCMEC/D3 cells were found to increase as follows: ginkgolide C, ginkgolide A,ginkgolide B, ginkgolide J, ginkgolide N, ginkgolide L and ginkgolide K.

Table 1The log P values of the different ginkgolides obtained in the oil–water partition coefficient method.

Compounds log P value

Ginkgolide A 1.291 ± 0.035Ginkgolide B 1.658 ± 0.096Ginkgolide C 1.145 ± 0.031Ginkgolide J 1.737 ± 0.128Ginkgolide K 2.152 ± 0.199Ginkgolide L 1.814 ± 0.108Ginkgolide N 1.802 ± 0.186

The cytotoxicities of ginkgolide C, ginkgolide A, ginkgolide B andginkgolide J were not significantly different, nor were the cytotoxicitiesof ginkgolide N, ginkgolide L and ginkgolide K. Notably, the cytotoxic-ities of ginkgolide N, ginkgolide L and ginkgolide K were significantlygreater than the cytotoxicities of ginkgolide C, ginkgolide A, ginkgolideB and ginkgolide J, indicating that ginkgolides with high lipophilicitieswere highly cytotoxic. Based on themeasured inhibitory concentrationsof these compounds, a safe ginkgolide concentration (less than 120 μM)was used for transmembrane transport experiments.

Papp value and efflux ratio of ginkgolides in the hCMEC/D3 cell monolayer

The Papp values and the abilities of ginkgolide A, ginkgolide B,ginkgolide C, ginkgolide J, ginkgolide N, ginkgolide L and ginkgolide Kto cross the cell monolayer were investigated. The Papp values and effluxratios of various ginkgolides at 120 min were measured over a non-cytotoxic concentration range (5 μM–80 μM). Bilateral Papp values andefflux ratios for different ginkgolides at different concentrations on thehCMEC/D3 cell monolayer are listed in Table 2. As shown in Table 2,both the PappA → B and PappB→ A of ginkgolide A, ginkgolide B, ginkgolideC and ginkgolide J did not change significantly at five different concen-trations (5 μM, 10 μM, 20 μM, 40 μMand 80 μM). Additionally, the effluxratios of ginkgolide A, ginkgolide B, ginkgolide C and ginkgolide J wereall below 1.5 for concentrations ranging between 5 μM and 80 μM.

The PappA → B values of ginkgolide N, ginkgolide L and ginkgolideK were increased in a concentration-dependent manner while thePappB → A values of these ginkgolides were decreased. Simultaneously,the efflux ratios of these ginkgolides decreased as their concentrationswere increased from 5 μM to 80 μM. Notably, ginkgolide K at a concen-tration of 5 μM exhibited a high efflux ratio of 21.63.

Effect of time and concentration on ginkgolide C and ginkgolide K transport

The amount of ginkgolide C transported increased linearlywith time(Fig. 3A), with a correlation coefficient of 0.9997. The rates ofmembranepermeation (AP to BL) essentially increased linearly with increasingconcentration over the range of concentrations tested (5–80 μM), witha correlation coefficient of 0.9979 (Fig. 4A). Papp values remained virtu-ally unchanged over the concentration range tested (Table 2). Theseresults indicate that ginkgolide C transport may proceed via a non-directional passive transport.

The amount transported and the rates of membrane permeation(AP to BL) of ginkgolide K increased linearly with time and concentra-tion over the range of concentrations tested (5–80 μM),with correlationcoefficients of 0.9758 and 0.9419, respectively. (Figs. 3B and 4B). The ef-flux ratio of ginkgolide K declined from 21.63 to 1.34 as ginkgolide Kconcentration was increased from 5 μM to 80 μM, as shown in Table 2.

Effect of cytochalasin D on ginkgolide C transport

Cytochalasin D is able to enhance paracellular permeability andincrease the transport of paracellular drugs by opening tight cell junc-tions. To explore the effect of cytochalasin D on ginkgolide C transport,different concentrations (10 μM, 20 μM and 40 μM) of cytochalasin Dwere added to the apical side (AP) of the cell monolayers to reduceparacellular tightness. The results in Table 3 indicate that ginkgolide Cpermeability significantly increased in a concentration-dependentmanner with cytochalasin D addition.

Influences of P-gp, BCRP and MRP inhibitors on ginkgolide K transport

To determine the roles of P-gp, BCRP andMRP in ginkgolide K trans-port, a P-gp inhibitor (cyclosporin A), a BCRP inhibitor (Ko143) or anMRP inhibitor (MK-571) was added to medium containing 20 μMginkgolide K. Bidirectional transport of ginkgolide K in the presence orabsence of P-gp, BCRP and MRP inhibitors was investigated. Bilateral

Fig. 2. Effect of ginkgolides on hCMEC/D3 cell viability. The data are presented as themeans± S.D. (n= 6). ⁎P b 0.05 and ⁎⁎P b 0.01, significantly different compared to the control group.

97S. Ma et al. / Life Sciences 114 (2014) 93–101

Papp values for ginkgolideK in the presence or absence of P-gp, BCRP andMRP inhibitors are summarized in Table 4.

The PappA → B of ginkgolide K increased approximately 5.13-fold(from 0.65 to 3.34) when cyclosporin A was added, as shown inTable 4. When the MRP inhibitor MK-571 and the BCRP inhibitorko143 were added at concentrations of 50 μM and 10 μM, respectively,into a medium containing ginkgolide K (20 μM), the PappA → B ofginkgolide K was unchanged. The PappB → A of ginkgolide K decreasedwhen a P-gp inhibitor was added to a medium containing 20 μMginkgolide K, while the PappB → A of ginkgolide K remained unchangedwhenMRP inhibitorMK-571 or BCRP inhibitorwas present in themedi-um. The efflux ratio of ginkgolide K alone was 10.04, and a coculture ofginkgolide K with cyclosporin A reduced the efflux ratio (0.95).Additionally, the efflux ratios of ginkgolide K treated with MK-571 orko143 were barely altered compared to a treatment with ginkgolide Kalone.

Discussion

The lipophilicity of a drug can affect its transmembrane transport,bioavailability, pharmacological activity and toxicity (Chan andStewart, 1996; Konsoula and Barile, 2005). In particular, the lipophilicityof a drug plays a vital role in determining its diffusion path through theBBB (Perioli et al., 2004). Generally, the log P value iswidely used to rep-resent the lipophilicity of a molecule and to indicate the tendency ofthat molecule to partition into the lipophilic membrane or its capacityto form hydrogen bonds with water molecules. Therefore, an accuratelog P valuemeasurement is verymeaningful for the evaluation and pre-diction of the BBB penetrability of a compound. In our experiment,ginkgolide A, ginkgolide B, ginkgolide C and ginkgolide J, which havelow lipophilicity, were found to transport non-directionally across thehCMEC/D3 cell monolayer. Moreover, the PappA → B values of ginkgolide

A, ginkgolide B, ginkgolide C and ginkgolide J decreasedwith decreasinglog P values from 1.06 to 2.34, as shown in Table 2.

The cytotoxicities of seven ginkgolides on hCMEC/D3 cells weremeasured to determine appropriate ginkgolide concentrations to beused for transport experiments in an in vitro BBB model. The resultsshowed that the cytotoxicities of ginkgolides on hCMEC/D3 cells corre-lated with their log P values, indicating that ginkgolides with highlipophilicities possessed high cytotoxicities. Therefore, a concentrationof 80 μM was determined to be a safe concentration for ginkgolidetransport experiments hCMEC/D3 cell monolayers.

The BBB plays a vital role in maintaining homeostasis of the brainparenchymal microenvironment and is regarded as a shield betweendrugs in the blood and the central nervous system (CNS) (Biegel et al.,1995). For this reason, understanding the transport of drug candidatesacross the BBB is critical to predicting and evaluating the absorptionand bioavailability of drugs designed to combat CNS disease and injury(Liu et al., 2014). Ginkgolides have been shown to protect against cere-bral neuron cell injury in vitro and in vivo (Chen et al., 2012; Qin et al.,2014; Ma et al., 2012a,b; Numa et al., 2007; Vitolo et al., 2009), and thetransport of ginkgolide A, ginkgolide B and ginkgolide C across MDCKcell monolayers has been investigated. It has been shown that thechemical structure of ginkgolide could affect the ability of this moleculeto penetrate the BBB in vitro. In addition, ginkgolides were found to behighly able to enter the BBB only in cerebral ischemia/reperfusion rats(Fang et al., 2010). Therefore, to evaluate the permeability characteris-tics of ginkgolides through BBB cells and into the CNS, hCMEC/D3 cellswere used to investigate the ability of ginkgolides to cross the cellmonolayer. Our results suggested that transport of ginkgolide A,ginkgolide B, ginkgolide C and ginkgolide J were due to passive diffu-sion, although the molecules' permeabilities across the hCMEC/D3 cellmonolayer were still dependent on their log P values (i.e., the lipidsolubility of these compounds) (Gnoth et al., 2010). The efflux ratios

Table 2Papp values and efflux ratio of ginkgolides in hCMEC/D3 cell monolayer (mean ± S.D.,n = 3).

Compounds Concentrations (μM) Papp (10−6 cms−1) Efflux ratio

A → B B → A

Ginkgolide A 5 1.49 ± 0.21 1.68 ± 0.46 1.1210 1.43 ± 0.82 1.79 ± 0.51 1.2120 1.49 ± 0.41 1.79 ± 0.66 1.1940 1.50 ± 0.42 1.74 ± 0.42 1.1680 1.52 ± 0.23 1.72 ± 0.24 1.14

Ginkgolide B 5 1.82 ± 0.42 2.56 ± 0.44 1.4110 1.87 ± 0.86 2.74 ± 0.58 1.4620 1.80 ± 0.11 2.59 ± 0.52 1.3940 1.79 ± 0.47 2.53 ± 0.74 1.4180 1.83 ± 0.22 2.56 ± 0.41 1.39

Ginkgolide C 5 1.06 ± 0.17 1.31 ± 0.39 1.2310 1.10 ± 0.12 1.32 ± 0.32 1.220 1.06 ± 0.12 1.35 ± 0.21 1.2840 1.05 ± 0.31 1.29 ± 0.55 1.2280 1.08 ± 0.26 1.38 ± 0.87 1.28

Ginkgolide J 5 2.34 ± 0.53 3.48 ± 0.93 1.4710 2.31 ± 0.88 3.37 ± 0.59 1.4520 2.29 ± 0.33 3.38 ± 0.82 1.4740 2.32 ± 0.38 3.45 ± 1.15 1.4880 2.32 ± 0.85 3.33 ± 1.32 1.43

Ginkgolide N 5 1.76 ± 0.45 7.41 ± 1.31 4.2110 2.22 ± 0.83 6.45 ± 1.05 2.9120 2.79 ± 0.72 5.78 ± 0.99 2.0740 3.11 ± 0.97 4.91 ± 1.41 1.5780 3.57 ± 1.03 4.49 ± 1.52 1.25

Ginkgolide L 5 1.22 ± 0.91 8.61 ± 1.15 7.0510 1.98 ± 0.53 7.24 ± 1.43 3.6520 2.18 ± 0.84 5.65 ± 0.72 2.5540 2.89 ± 0.42 4.89 ± 0.77 1.6880 3.06 ± 0.99 3.80 ± 1.06 1.24

Ginkgolide K 5 0.55 ± 0.12 11.90 ± 3.23 21.6310 0.62 ± 0.15 9.67 ± 2.92 15.5920 0.92 ± 0.32 7.49 ± 1.93 8.1440 1.47 ± 0.74 5.39 ± 1.45 3.6680 2.13 ± 0.92 2.87 ± 0.79 1.34

Fig. 3. Time course of 20 μM ginkgolide C and ginkgolide K transport in hCMEC/D3 cell monol

98 S. Ma et al. / Life Sciences 114 (2014) 93–101

of ginkgolide N, ginkgolide L and ginkgolide K increasedwith increasinglog P, particularly at low concentrations. The efflux ratios of ginkgolide Kwere all above 20, suggesting that the ability of ginkgolides with highlog P values to cross the BBB is directionally mediated by an effluxprotein (Zhao and Liang, 2009).

The paracellular route involves migration between adjacent endo-thelial cells and requires the transient disassembly of endothelial celljunctions, while the transcellular route occurs directly through an indi-vidual endothelial cell, likely requiring the formation of a channel orpore and transport (Wittchen ES. 2009). As shown in Fig. 5, the tight-ness of the cell monolayer, represented by its TEER value, affects theparacellular route. Ketoconazole, which passes through the cell mono-layer paracellularly, was used to demonstrate that the paracellularroute through the cell monolayer was existent. Chloramphenicol wasused to prove that the transcellular route through the cell monolayerwas fine in our experimental cells, as chloramphenicol can penetratethe cell monolayer transcellularly. Our results showed that the sevenginkgolides investigated were transported via both the paracellularand transcellular paths. The transcellular path accounted for a smallportion of the observed ginkgolide transport due to saturation of thetransporter. The paracellular pathwas responsible for a large proportionof the observed ginkgolide transport, especially at high ginkgolideconcentrations.

Ginkgolide Cwas used in unilateral transport assayswith a transwellsystem to determine the time and concentration dependence of theamount transported and the transport ratio. The results showed thatthe amount transported increased with time and that the transportratio increased linearly with concentration for concentrations rangingfrom 5 μM to 80 μM, suggesting that the transport mechanism ofginkgolide C is non-directional passive transport. The paracellular pathplays a vital role in controlling drug diffusion, particularly in passivetransport. As a consequence, cytochalasin D, an enhancer of cell perme-ability, was investigated at three concentrations (10 μM, 20 μM, 40 μM)to explore whether Papp values could be increased by opening tightjunctions in the cell monolayer. The results in our experiment showedthat cytochalasin D could increase the Papp value compared to a group

ayers. A. Time course of ginkgolide C transport; B. Time course of ginkgolide K transport.

Fig. 4. Transport rate of ginkgolide C and ginkgolide K for concentrations ranging from5 μMto 80 μM inhCMEC/D3 cellmonolayers. Panels A and B show the transport rates of ginkgolide Cand ginkgolide K, respectively.

99S. Ma et al. / Life Sciences 114 (2014) 93–101

treated with ginkgolide C alone. Remarkably, the Papp value ofginkgolide C soaredwhen cytochalasinD concentrationswere increasedfrom 10 μM to 40 μM, implying that paracellular tightness was thor-oughly destroyed by cytochalasin D. This loss of tightness likely resultedin free diffusion in the transwell.

Because ginkgolide K exhibited the highest log P value ofthe ginkgolides tested, the transport amount and transport ratio ofginkgolide K were obtained by measuring the transport amount atdifferent time points and the transport rate at different concentrations.The results implied that transport amount increases with time and thattransport ratio increaseswith concentration. Linear time and concentra-tion dependencies are obvious, with correlation coefficients of 0.9419and 0.9758, respectively. The results showed that the effect of the trans-cellular path on transport was minimal compared with the paracellularpath, especially at high ginkgolide concentrations. The dominance ofthe paracellular path resulted in a linear transport relationship forginkgolide K due to the saturation of the efflux protein.

The fact that the efflux ratio of ginkgolide K is greater than 2 at lowconcentrations suggests that ginkgolide K transport may be directionaland mediated by transporters. Therefore, the effect of P-gp, BCRP andMRP on ginkgolide transport was further investigated. The ABC trans-porters expressed in human BBBs that are involved in the efflux ofchemicals from BBB cells include P-gp, MRP and BCRP (Sha and Fang,2004; Juvale et al., 2013; Vaidyanathan and Walle, 2001). These

Table 3The effect of cytochalasin D on transport of ginkgolide C in hCMEC/D3 cell monolayer(mean ± S.D., n = 3).

Group Papp AP → BL (10−6 cms−1)

Ginkgolide C (20 μM) 0.98 ± 0.31Ginkgolide C (20 μM) + cytochakain (10 μM) 2.24 ± 0.87⁎

Ginkgolide C (20 μM) + cytochakain (20 μM) 5.32 ± 1.32⁎⁎⁎

Ginkgolide C (20 μM) + cytochakain (40 μM) 7.88 ± 6.35⁎⁎⁎

Compared with ginkgolde C (20 μM).⁎ P b 0.05.

⁎⁎⁎ P b 0.001.

transporters are generally located specifically in the apical (blood/plasma side) or basolateral (cerebral luminal side) membranes ofbrain microvascular endothelial cells and drive compounds from insidethe cell back into the blood/plasma, preventing their absorption into thebrain. Drug transport is considered to be directional when the effluxratio is greater than 1.5 (Zhao and Liang, 2009), implying that drug ab-sorption is affected by efflux proteins such as P-gp, BCRP and MRP. Theefflux ratio of ginkgolide K at 20 μMwas 8.14, suggesting that the trans-port mechanism of ginkgolide K was directional. When a P-gp inhibitor(cyclosporin C, 10 μM) was added to a medium containing 20 μMginkgolide K, the efflux ratio decreased to less than 1.5. The results indi-cate that ginkgolide K transport in hCMEC/D3 cell monolayers mightproceed via active transport mediated by the P-gp transporter. WhenBCRP or MRP inhibitors were added to themedium, the ginkgolide K ef-flux ratio was not significantly changed. This result implied that P-gp isinvolved in ginkgolide K transport, but not BCRP orMRP. In addition, thefact that the ginkgolide K efflux ratio decreased as ginkgolide K concen-tration was increased from 5 μM to 80 μM suggests that the efflux pro-teins become saturated at high ginkgolide K concentrations.

Conclusions

This is the first report that evaluates and compares the transmem-brane transport of ginkgolides using a BBB cell model. The lipophilicitiesof these ginkgolides were the key determinant of the transportpath. Ginkgolides exhibited time- and concentration-dependence viaparacellular and transcellular diffusion with passive transport. Thetransport of ginkgolide K was found to be mediated by P-gp via a trans-cellular path at concentrations below 80 μM. An in vitro model of theBBB was established to quickly evaluate active compounds designedto target cerebral ischemia injuries.

Conflict of interest statement

None of the authors has conflicts of interest that interfere with the integrity of thecontent of the article.

Table 4The effect BCRP, P-gp and MRP inhibitor on transport of ginkgolide K in hCMEC/D3 cell monolayer (mean ± S.D., n = 3).

Group Papp (10−6 cms−1)

AP → BL BL → AP Efflux ratio

Ginkgolide K (20 μM) 0.65 ± 0.19 6.52 + 1.73 10.03Ginkgolide K (20 μM) + Ko143 (20 μM) 0.64 ± 0.25 7.56 ± 2.94 11.81Ginkgolide K (20 μM) + MK-571 (50 μM) 0.72 ± 0.33 7.23 ± 3.26 10.04Ginkgolide K (20 μM) + cyclosporin A (10 μM) 3.34 ± 0.82⁎⁎⁎ 3.19 ± 0.84⁎⁎ 0.95

Compared with ginkgolde K (20 μM) alone.⁎⁎⁎ P b 0.001.

blood

brainhCMCE/D3 cell monolayer

MRP

ginkdolide C, ginkgolide A, ginkgolide B or ginkgolide J

BCRP

P-gp

ginkdolide N, ginkgolide L or ginkgolide K

paracellular trtanscellular

Fig. 5. Architecture of the hCMEC/D3 cell monolayer as a BBBmodel and the transmigratory paths for ginkgolides across the hCMEC/D3 cell monolayer. Ginkgolide transport proceeds viatwo paths (paracellular and transcellular) through the hCMEC/D3 cell monolayer. Ginkgolide N, ginkgolide L and ginkgolide Kwere excreted transcellularlywhen P-gpwas present in thehCMEC/D3 cell monolayer, but not when MRP or BCRP were present.

100 S. Ma et al. / Life Sciences 114 (2014) 93–101

Acknowledgments

This work was supported by the Young Nature Science Fund ofHeilongjiang Province (QC2012C076), the Nature Science Fund ofGuangdong Province Program (S2011010001294) and the GuangzhouScience and Technology Plan Projects of GuangzhouCity (2011J2200012).

References

Biegel D, Spencer DD, Pachter JS. Isolation and culture of human brain microvesselendothelial cells for the study of blood–brain barrier properties in vitro. Brain Res1995;692(1–2):183–9.

Chan O, Stewart BH. Physicochemical and drug-delivery consideration for oral drugbioavailability. Drug Discov Today 1996;1:461–73.

Chen Y, Wang C, Hu M, Pan J, Chen J, Duan P, et al. Effects of ginkgolide A on okadaicacid-induced tau hyperphosphorylation and the PI3K–Akt signaling pathway in N2acells. Planta Med 2012;78(12):1337–41.

Chen ZZ, Lu Y, Du SY, Shang KX, Cai CB. Influence of borneol and muscone on geniposidetransport through MDCK and MDCK–MDR1 cells as blood–brain barrier in vitromodel. Int J Pharm 2013;456(1):73–9.

Dew TP, Wang G, Williamson G. Urinary excretion of ginkgolide terpene lactones follow-ing acute consumption of Ginkgo biloba extract. Biofactors 2013. http://dx.doi.org/10.1002/biof.1151.

Fang W, Deng Y, Li Y, Shang E, Fang F, Lv P, et al. Blood brain barrier permeability andtherapeutic time window of Ginkgolide B in ischemia–reperfusion injury. Eur JPharm Sci 2010;39(1–3):8–14.

Förster C, Burek M, Romero IA, Weksler B, Couraud PO, Drenckhahn D. Differential effectsof hydrocortisone and TNFalpha on tight junction proteins in an in vitro model of thehuman blood–brain barrier. J Physiol 2008;586(7):1937–49.

Gnoth MJ, Sandmann S, Engel K, Radtke M. In vitro to in vivo comparison of the substratecharacteristics of sorafenib tosylate toward P-glycoprotein. Drug Metab Dispos 2010;38(8):1341–6.

Hsiao P, Bui T, Ho RJ, Unadkat JD. In vitro-to-in vivo prediction of P-glycoprotein-baseddrug interactions at the human and rodent blood–brain barrier. Drug Metab Dispos2008;36(3):481–4.

Juvale K, Gallus J, Wiese M. Investigation of quinazolines as inhibitors of breast cancerresistance protein (ABCG2). Bioorg Med Chem 2013;21(24):7858–73.

Konsoula R, Barile FA. Correlation of in vitro cytotoxicity with paracellular permeability inCaco-2 cells. Toxicol In Vitro 2005;19:675–84.

Lang Y, Chen D, Li D, Zhu M, Xu T, Zhang T, et al. Luteolin inhibited hydrogenperoxide-induced vascular smooth muscle cells proliferation and migration by

suppressing the Src and Akt signalling pathways. J Pharm Pharmacol 2012;64(4):597–603.

Liu D, Lin B, Shao W, Zhu Z, Ji T, Yang C. In vitro and in vivo studies on the transport ofPEGylated silica nanoparticles across the blood–brain barrier. ACS Appl MaterInterfaces 2014;6(3):2131–6.

Ma S, Liu H, Jiao H, Wang L, Chen L, Liang J, et al. Neuroprotective effect of ginkgolide K onglutamate-induced cytotoxicity in PC 12 cells via inhibition of ROS generation andCa(2+) influx. Neurotoxicology 2012a;33(1):59–69.

Ma S, Yin H, Chen L, Liu H, ZhaoM, Zhang X. Neuroprotective effect of ginkgolide K againstacute ischemic stroke onmiddle cerebral ischemia occlusion in rats. J NatMed 2012b;66(1):25–31.

Numa AM, Abbott FS, Chang TK. Effect of Ginkgo biloba extract on oxidativemetabolism ofvalproic acid in hepatic microsomes from donors with the CYP2C9*1/*1 genotype.Can J Physiol Pharmacol 2007;85(9):848–55.

Perioli L, Ambrogi V, Bernardini C, Grandolini G, Ricci M, Giovagnoli S, et al. Potentialprodrugs of non-steroidal anti-inflammatory agents for targeted drug delivery tothe CNS. Eur J Med Chem 2004;39(8):715–27.

Qin XF, Lu XJ, Ge JB, Xu HZ, Qin HD, Xu F. Ginkgolide B prevents cathepsin-mediated celldeath following cerebral ischemia/reperfusion injury. Neuroreport 2014;25(4):267–73.

Qiu J, Kitamura Y, Miyata Y, Tamaru S, Tanaka K, Tanaka T, et al. Transepithelial trans-port of theasinensins through Caco-2 cell monolayers and their absorption inSprague–Dawley rats after oral administration. J Agric Food Chem 2012;60(32):8036–43.

Saleem S, Zhuang H, Biswal S, Christen Y, Doré S. Ginkgo biloba extract neuroprotectiveaction is dependent on heme oxygenase 1 in ischemic reperfusion brain injury.Stroke 2008;39(12):3389–96.

Sha XY, Fang XL. Transport characteristics of 9-nitrocamptothecin in the humanintestinal cell line Caco-2 and everted gut sacs. Int J Pharm 2004;272(1–2):161–71.

Sung JH, Shah FA, Cho EH, Gim SA, Jeon SJ, Kim KM, et al. Ginkgo biloba extract (EGb 761)prevents the ischemic brain injury-induced decrease in parvalbumin expression. LabAnim Res 2012;28(2):77–82.

Vaidyanathan JB, Walle T. Transport and metabolism of the tea flavonoid (−)-epicatechin by the human intestinal cell line Caco-2. Pharm Res 2001;18(10):1420–5.

Vitolo O, Gong B, Cao Z, Ishii H, Jaracz S, Nakanishi K, et al. Protection againstbeta-amyloid induced abnormal synaptic function and cell death by ginkgolide J.Neurobiol Aging 2009;30(2):257–65.

Wang J, Cao F, Su E, Wu C, Zhao L, Ying R. Improving flavonoid extraction from Ginkgobiloba leaves by prefermentation processing. J Agric Food Chem 2013;61(24):5783–91.

Weksler B, Romero IA, Couraud PO. The hCMEC/D3 cell line as a model of the humanblood brain barrier. Fluids Barriers CNS 2013;10(1):16.

101S. Ma et al. / Life Sciences 114 (2014) 93–101

Wittchen ES. Endothelial signaling in paracellular and transcellular leukocyte transmigra-tion. Front Biosci (Landmark Ed) 2009 Jan 1;14:2522–45.

Yang Y, Bai L, Li X, Xiong J, Xu P, Guo C, et al. Transport of active flavonoids, based oncytotoxicity and lipophilicity: an evaluation using the blood–brain barrier cell andCaco-2 cell models. Toxicol In Vitro 2014;28(3):388–96.

Zehendner CM, White R, Hedrich J, Luhmann HJ. A neurovascular blood–brain barrierin vitro model. Methods Mol Biol 2014;1135:403–13.

Zhao J, Liang AH. Caco-2 Cell model and its application in the absorption and transporta-tion research of Chinese Materia Medica. Chin J Exp Tradit Med Formulae 2009;15:79–82.

Zhao M, Wang XX, Wan WH. Effects of the Ginkgo biloba extract on the superoxidedismutase activity and apoptosis of endothelial progenitor cells from diabetic periph-eral blood. Genet Mol Res 2014;13(1):220–7.