European Journal of Pharmacology 626 (2010) 193–199

Contents lists available at ScienceDirect

European Journal of Pharmacology

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

Molecular and Cellular Pharmacology

Palonosetron triggers 5-HT3 receptor internalization and causes prolonged inhibitionof receptor function

Camilo Rojas a,⁎, Ajit G. Thomas a, Jesse Alt a, Marigo Stathis a, Jie Zhang b, Edward B. Rubenstein c,Silvia Sebastiani d, Sergio Cantoreggi e, Barbara S. Slusher a

a Johns Hopkins University School of Medicine, Brain Science Institute Neurotranslational Center, Baltimore, Maryland, USAb Profectus Biosciences, Baltimore, Maryland, USAc Medical & Commercial Development Solutions Inc., USAd Business Unit Oncology & Cancer Supportive Care, Helsinn Healthcare SA, Lugano, Switzerlande Research and Development, Helsinn Healthcare SA, Lugano, Switzerland

⁎ Corresponding author. Brain Science Institute NeHopkins University School of Medicine, 6611 Tributary6515, USA. Tel.: +1 410 662 7435, +1 443 801 0592 (M

E-mail address: crojas2@jhmi.edu (C. Rojas).

0014-2999/$ – see front matter © 2009 Elsevier B.V. Aldoi:10.1016/j.ejphar.2009.10.002

a b s t r a c t

a r t i c l e i n f o

Article history:Received 28 August 2009Accepted 7 October 2009Available online 15 October 2009

Keywords:5-HT3 receptorPalonosetronGranisetronOndansetronReceptor internalization

Palonosetron is a 5-HT3 receptor antagonist that has demonstrated superiority in preventing both acute anddelayed emesis when compared to older first generation 5-HT3 receptor antagonists. The objective of thiswork was to determine if palonosetron exhibits unique molecular interactions with the 5-HT3 receptor thatcould provide a scientific rationale for observed clinical efficacy differences. Previously, we showed thatpalonosetron exhibits allosteric binding and positive cooperativity to the 5-HT3 receptor in contrast toondansetron and granisetron which exhibit simple bimolecular binding. The present work shows, throughseveral independent experiments, that palonosetron uniquely triggers 5-HT3 receptor internalization andinduces prolonged inhibition of receptor function. After 24 h incubation followed by dissociation conditions,[3H]palonosetron remained associated with whole cells but not to cell-free membranes (P<0.001). [3H]Palonosetron's binding to cells was resistant to both protease and acid treatments designed to denature cellsurface proteins suggesting that the receptor complex was inside the cells rather than at the surface. Cellspretreated with unlabeled palonosetron subsequently exhibited reduced cell surface 5-HT3 receptor binding.Palonosetron-triggered receptor internalization was visualized by confocal fluorescence microscopy usingcells transfected with 5-HT3 receptor fused to enhanced cyan fluorescent protein. In contrast, granisetronand ondansetron showed minimal to no effect on receptor internalization or prolonged inhibition of receptorfunction. These experiments may provide a pharmacological basis for differences noted in published clinicaltrials comparing palonosetron to other 5-HT3 receptor antagonists.

urotranslational Center, JohnsStreet, Baltimore, MD 21224-obile).

l rights reserved.

© 2009 Elsevier B.V. All rights reserved.

1. Introduction

Serotonin 5-HT3 receptor antagonists are widely used alone or incombination with other agents to prevent or treat chemotherapy-induced nausea and vomiting and postoperative nausea and vomiting.5-HT3 receptor antagonists are competitive antagonists of serotonin, thenaturally occurring ligand for the 5-HT3 receptor; their therapeuticactivity is thought to be the result of a similar mechanism of action(Hesketh, 2008). Interestingly, palonosetron, the latest 5-HT3 receptorantagonist, introduced in 2003, has demonstrated superiority for theprevention of emesis in multiple randomized, prospective clinical trialswhen compared to first generation 5-HT3 receptor antagonists.

Palonosetron was superior to ondansetron in preventing acute (0–24 h) and delayed (24–120 h) emesis in one randomized phase IIIclinical trial, and superior to dolasetron in a second phase III trial ofpatients receiving moderately emetogenic chemotherapy (Eisenberget al., 2003; Gralla et al., 2003). Palonosetronwas also significantlymoreeffective in preventing emesis vs. ondansetron throughout the delayedpost-chemotherapy period in a third phase III randomized trial inpatients receiving highly emetogenic agents (Aapro et al., 2006). In arecent fourth phase III trial, palonosetron provided superior delayedemesis prevention vs. granisetron in patients receiving highly emeto-genic chemotherapy (Yoshizawa et al., 2008). Additionally, palonose-tron has significantly reduced the severity of nausea to a greater extentthan other 5-HT3-receptor antagonists (Decker et al., 2006; Eisenberget al., 2003; Gralla et al., 2003).

Palonosetron's clinical results could be partly due to its potentbindingaffinity (Wonget al., 1995) and longhalf-life (Stoltz et al., 2004).However, these characteristics do not entirely explain palonosetron'sclinical differentiation throughout the 5 days following emetogenic

194 C. Rojas et al. / European Journal of Pharmacology 626 (2010) 193–199

chemotherapy. Higher doses of a drug with lower binding affinity for areceptor could make-up for increased binding affinity of another drug.However, palonosetron exhibits greater efficacy against acute emesisassociated with moderately emetogenic chemotherapy compared tohigher doses of dolasetronor ondansetron (Eisenberg et al., 2003; Grallaet al., 2003). Similarly, more frequent administration of a drug with ashort half-life could compensate for the longer-half-life of another.Interestingly, when ondansetron is administered beyond 24h afterchemotherapy, it does not provide substantial protective action indelayed emesis (Geling and Eichler, 2005).

Since palonosetron also differs in chemical structure from other5-HT3-receptor antagonists (Thompson and Lummis, 2007), wewondered whether improved clinical efficacy could be the result ofdifferences at the molecular level. Recent work in our laboratory hasshown that palonosetron exhibits allosteric interactions and positivecooperativity with the 5-HT3 receptor, characteristics that are notdisplayed by ondansetron and granisetron. We chose ondansetronand granisetron for our comparative studies because together withpalonosetron they constitute the majority of the 5-HT3 receptorantagonists used in worldwide clinical practice (Rojas et al., 2008).In the present work we show that palonosetron's differential 5-HT3receptor binding results in 5-HT3 receptor internalization andprolonged inhibition of receptor function. These actions wereminimal in the case of granisetron and non-existent in the case ofondansetron. These molecular interactions provide a rationale thatmay help explain palonosetron's differentiation in the clinic.

2. Materials and methods

2.1. Plasmid preparation and cell transfections

Plasmid preparation and cell transfections have been describedbefore (Rojas et al., 2008).

2.2. Dissociation of antagonists from cells and from cell-free membranes

5-HT3A-HEK293 cells in 35 mmdisheswere incubatedwith5-foldKd

concentrations of [3H]ondansetron (30 nM), [3H]granisetron (5 nM) or[3H]palonosetron (1 nM) for 24 h. At the end of this period, antagonist-containing media were replaced with antagonist-free media anddissociation of radiolabeled antagonist at 37 °C was followed at 0, 2, 4,6, 8, 15, 30, 60 and 120 min. After removing the medium, cells werescraped into 100 µl of fresh ice-cold buffer and the radioactivity presentin the scraped material at each time point was measured using ascintillation counter. Student's t test was used for statistical analyses ofthe results. Preparation of cell-free membranes and kinetic dissociationexperiments using these membranes have been described previously(Rojas et al., 2008; Wong et al., 1995).

2.3. Protease treatment

The protease treatment protocol was adapted from the literature(Simantov and Sachs, 1973). Briefly, 5-HT3A transfected HEK293 cellswere incubated with [3H]ondansetron (30 nM), [3H]granisetron(5 nM) or [3H]palonosetron (1 nM) for 24 h. At the end of this period,media were removed and cells were incubated with HEPES bufferedsaline containing trypsin (2.5 mg/ml) or chymotrypsin (5.0 mg/ml)for 5 min at 37 °C. Digestion by proteases was terminated by washingcells twice with ice-cold HEPES buffered saline containing soybeantrypsin inhibitor (50 µg/ml). Radioactivity present in each wash andin the cells was determined with a scintillation counter and percentradioactivity in the cell fraction was calculated. A control experimentwas carried out to measure dissociation of antagonists from cells, inthe absence of proteases, under similar experimental conditions.Student's t test was used for statistical analysis.

2.4. Acid treatment

The acid treatment protocol was based on published methodology(Haigler et al., 1980). 5-HT3A transfected HEK293 cells were incubatedwith [3H]ondansetron (30 nM), [3H]granisetron (5 nM) or [3H]palonosetron (1 nM) for 24 h. At the end of this period, media wereremoved and cells were incubatedwith saline (0.5 MNaCl) containingacetic acid (0.2 M, pH 2.5) for 6 min on ice. Acid denaturation of cellsurface proteins was terminated with the addition of one volume ofice-cold HEPES buffered saline (pH 7.4). Cells were then washed oncewith the same buffer. Radioactivity present in each wash and in thecells was measured with a scintillation counter and percentradioactivity in the cell fraction was calculated. Student's t test wasused for statistical analysis of the results.

2.5. Binding of [3H]palonosetron after incubation of cells with unlabeledantagonists

The dependence of [3H]palonosetron binding to cells after differentpreincubation times with unlabeled palonosetron was carried out asoutlined previously (Rojas et al., 2008). Briefly, 5-HT3A-transfectedHEK293 cells were pretreated with unlabeled ondansetron (30 nM),granisetron (5 nM), palonosetron (1 nM) or saline for various times.Subsequent to each treatment, antagonists were removed and freshgrowthmedia were added to the cells; 2 h later, cells were incubated infresh HEPES buffered saline at room temperature for an additional30 min. The buffer was removed and cells were incubated with [3H]palonosetron (1 nM) for 40 min at room temperature. At the end of theincubation period the label was removed and the cells were washedwith ice-cold buffer. Following the wash, cells were scraped into 100 µlof fresh ice-cold buffer and the radioactivity associated with the cellswasmeasured using a scintillation counter. Student's t test was used forstatistical analyses of the results.

2.6. Confocal fluorescence microscopy

HEK293 cells were plated at 500,000 cells/well in six well plates.After 18–24 h p5-HT3R-ECFP was transfected into cells using theFuGENE® HD Transfection Reagent. The plasmid was a kind gift fromDr. Vogel's laboratory (Ilegems et al., 2004). The following day, cellswere plated onto glass bottom dishes at 3000 cells/well. Experimentswere carried out within 3 days following transfection with anOlympus Fluoview laser scanning confocal microscopy system using488 and 568 laser lines. On the day of imaging, cells were treated withvehicle, palonosetron (10 nM), granisetron (50 nM) or ondansetron(300 nM). Cells were imaged 2 h after the addition of antagonists. Cellmembrane stain (Alexa Fluor 594 wheat germ agglutinin 0.1 mg/ml)was added to the media 10 min before imaging. Immediately beforeimaging, cell culture media were replaced with Hanks balanced saltsolution. Quantification was carried out using ImageJ software.Student's t test was used for statistical analysis of the results.

2.7. Measurement of prolonged inhibition of receptor function

Calcium-ion influx was used as a measure of receptor function.Calcium-ion influx measurements were based on published procedures(Dubin et al., 1999; Kao, 1994). HEK293 cells transfectedwith the human5-HT3A homomeric serotonin receptor were plated onto 96-well platestreated with poly-D-lysine and allowed to grow for at least 4days inRPMI-1640 media supplemented with 10% heat-inactivated fetal bovineserum and 2mM L-glutamine. Cells were treated with ondansetron(30 nM), granisetron (5 nM) and palonosetron (1 nM) for 24 h or withpalonosetron (1 nM) for 15, 30, 60 and 120 min. Groups of control cellswhere vehicle saline was added were carried out at 24 h or at 15 and120 min. Subsequent to treatment, antagonist-containing media wereremoved and antagonist-free media were added to the cells. One hour

Fig. 1. Dissociation of [3H]ondansetron, [3H]granisetron and [3H]palonosetron fromcells vs. cell-free membranes — each panel compares dissociation from cells (opencircles) vs. cell-free membranes (X) for (A) [3H]ondansetron, (B) [3H]granisetron and(C) [3H]palonosetron. For experiments with whole cells, 5-HT3A-HEK293 cells wereincubated with [3H]palonosetron (1 nM), [3H]granisetron (5 nM) or [3H]ondansetron(30 nM) for 24 h. Media were then replaced with antagonist-free media anddissociation of radiolabeled antagonist from cells was followed over time. Values arethe average of four independent experiments. For experiments with cell-freemembranes, dissociation of [3H]antagonist was initiated by the addition of excessunlabeled antagonist following the association phase. Dissociation profiles from cell-free membranes are similar to those reported earlier (Rojas et al., 2008) except that theprofiles shown here correspond to the average of a larger set of single determinations(Table 1). Error bars correspond to S.E.M. values. Control HEK293 cells that do notexpress the receptor did not bind any of the [3H]antagonists.

195C. Rojas et al. / European Journal of Pharmacology 626 (2010) 193–199

later, cells were rinsed with F-12/Dulbecco's modified Eagle's medium(DMEM) and incubated for 1 h at room temperature with theacetoxymethyl ester form of the fluorescent Ca2+ indicator Fluo-4(2 µM). Pluronic acid (0.04%) was added as nonionic surfactant tosequester theAMester intomicelles for cellular uptake. Stock solutions ofthe acetoxymethyl ester and pluronic acid were made up in DMSO andreconstituted intoHEPESbuffered saline (130 mMNaCl, 2 mMKCl, 1 mMMgCl2, 2 mM CaCl2 and 20 mMHEPES, pH 7.4). Cells were then allowedto recover in fresh buffer for 30 min in order to allow for de-esterificationof the dye into the polycarboxylate form. The polycarboxylate formof thedye binds Ca2+ when it enters the cell to form a fluorescent complex.Following the recovery period, cells were challenged with serotonin(10 µM) and changes in fluorescence intensity at 470 nmwere capturedusing a FLIPR® system (Molecular Devices). Student's t test was used forstatistical analysis of the results.

3. Results

3.1. [3H]Palonosetron dissociates completely from cell-free membraneswhereas dissociation from whole cells is significantly reduced

Earlier work had suggested the possibility that palonosetrontriggered receptor internalization into cells (Rojas et al., 2008). If thiswas the case in binding studies, the [3H]palonosetron that went insidethe cells aspart of a [3H]palonosetron-5HT3 receptor complexwouldnotbe available for dissociation. Accordingly,we studieddissociation of [3H]palonosetron from 5-HT3A-HEK293 cells after allocating time forreceptor internalization (Section 2.2). The amount of [3H]palonosetronthat remained bound to cells after 1h under dissociation conditionswas60±2% of the total radioligand; dissociation leveled off with noadditional [3H]palonosetron dissociating during the second hour. Thisresult was in marked contrast to results with cell-free membraneswhere only 2±0.8% of [3H]palonosetron remained bound after 1 h. Thelarge percentage of radioactive material remaining with cells wasunique to [3H]palonosetron. Only 15±3% [3H]granisetron remainedbound to cells after dissociation leveled off at 60 min and virtually no[3H]ondansetron (3±0.4%) remained bound to cells after 15 min(Fig. 1). The rate constant of dissociation of the 40% [3H]palonosetronthat dissociated from cells was the same within experimental error asthe 100% [3H]palonosetron that dissociated from cell-free membranes(Table 1). Similarly, the dissociation rate constants of [3H]granisetronand [3H]ondansetron that dissociated fromcells vs. cell-freemembraneswere indistinguishable within experimental error (Fig. 1, Table 1).

Separate control experiments usingHEK293 cells that donot expressthe 5-HT3 receptor did not bind any of the [3H]antagonists after 24 hincubation. The radioactivity counts recovered in the cell fractionswerebackground level. These results indicate that the [3H]palonosetronpresent in the cell fraction when 5-HT3A-HEK293 cells were used musthave been taken up by the cells through a 5-HT3A-receptor-mediatedprocess. Clog P of ondansetron, granisetron and palonosetron is 2.1±0.5, 1.5±0.5 and 2.6±0.5 respectively (Glennon and Dukat, 2008); thesimilarity in Clog P values indicates that palonosetron's uniquecharacteristics could not derive from differences in lipid solubility andit is consistent with the results of the control experiments mentionedabove.

3.2. Protease and acid treatments do not relinquish [3H]palonosetron'sassociation with cells

One explanation to [3H]palonosetron lingering with cells is thatbinding of ligand to receptor could evoke a conformational change thatlocks the ligand–receptor complex in an irreversible state at the cellsurface. This possibility was explored by limited exposure of cells toproteases (Simantov and Sachs, 1973) or acid (Haigler et al., 1980).Limited exposure to proteases or acid denatures proteins on the cellsurfacewithout affecting the inside of the cell. If radioligand is locked to

the receptor at the cell surface, receptor denaturation would terminatethis association and release ligands into the medium. 5HT3A-HEK293cells were incubated with excess [3H]ondansetron, [3H]granisetron and[3H]palonosetron for 24 h; media were then removed and cells wereexposed to proteases or acid (Sections 2.3 and 2.4). Cells exposed to [3H]granisetron or [3H]ondansetron exhibited minimal radioactivity afterreceptor denaturation by proteases or acid (Table 2). Radioactivity wasfully accounted for in the medium (data not shown). The resultsindicated [3H]ondansetron and [3H]granisetron bound to the 5-HT3receptor and remained at the cell surface. In contrast, cells exposed to[3H]palonosetron contained significant amount of radioactivity afterexposure to proteases (trypsin: 62%±8; chymotrypsin: 64%±7)or acid(53%±2); 100% radioactivity was defined as the total radioactivity

Table 1Rate and extent of dissociation from cells vs. cell-free membranes.

Antagonist koff cell-freemembranes(min−1)

koff cells(min−1)

% [3H] antagonist associatedwith cells after dissociationlevels off

[3H]Ondansetron 0.54±0.05 0.34±0.01 3±0.4[3H]Granisetron 0.13±0.01 0.15±0.03 15±3[3H]Palonosetron 0.080±0.003 0.079±0.014 60±2

Values of dissociation rate constants (koff) correspond to the data in Fig. 2. Graph PadPRISM® was used to obtain dissociation rate constants using the best fit of a one phasesingle exponential decay. Errors correspond to ±S.E.M. Data is the average of at leastfour independent determinations.

196 C. Rojas et al. / European Journal of Pharmacology 626 (2010) 193–199

found in cells before incubation with proteases or acid (Table 2). [3H]Palonosetron was only partially accessible to the effects of protease oracid exposure indicating it was sequestered inside the cell.

Fig. 2. (A) Effect of pre-incubation of antagonists on subsequent 5-HT3 receptor cellsurface binding. 5-HT3-HEK293 cells were preincubated with ondansetron (30 nM),granisetron (5 nM) and palonosetron (1 nM) for 24 h. Media were then removed andcells were incubated in antagonist-free media for 2.5 h with one additional mediachange. Cells were then incubated with [3H]palonosetron and binding was measured.Results are the average of three independent experiments (⁎⁎P<0.01). (B) Binding of[3H]palonosetron to 5-HT3-HEK293 cells after various times of preincubation withunlabeled palonosetron. Cells were preincubated with unlabeled palonosetron (1 nM)for various times as indicated. Work-up of experiment was the same as in (A). Resultsare the average of four independent experiments (⁎⁎⁎P<0.001). Error bars correspondto S.E.M.

3.3. Preincubationwith palonosetron reduces subsequent [3H]palonosetronbinding to cells

Oneway to demonstrate ligand-induced receptor internalization isto pretreat cells with unlabeled ligand to trigger internalization, washthe cells and show that binding of radiolabeled ligand to these cells isreduced in comparison to the binding to cells that have not beenpretreated with ligand (Li et al., 1999). After 24 h preincubations,palonosetron selectively reduces binding of [3H]palonosetron to 5-HT3-HEK293 cells to 52±7% (P<0.01 compared to control) whereasgranisetron and ondansetron do not (Fig. 2A). In order to determinethe minimum time required to observe reduced binding at the surfaceafter pretreatment with palonosetron, a similar experiment wascarried out using different preincubation times (Section 2.5). [3H]Palonosetron binding was reduced to 69±2% (P<0.001 compared tocontrol) after 15 min preincubation and it gradually reduced overtime to 41±1% (P<0.001 compared to control) after 2 h preincuba-tion (Fig. 2B). In short, palonosetron-triggered internalization beganwithin minutes of exposure of the cells to the antagonist and itreached its maximum at about 2 h.

3.4. Palonosetron-triggered receptor internalization was visualized usingfluorescence microscopy

In order to visualize receptor internalization, HEK293 cells weretransfected with green fluorescent 5-HT3A receptor (ECFP-5-HT3A)(Ilegems et al., 2004). Cells were then incubated with ondansetron,granisetron and palonosetron for 2 h, stained with red wheat germagglutinin dye and observed under the confocal fluorescencemicroscope (Section 2.6).

Table 2Protease and acid treatments.

% [3H] antagonist remaining with cells

Protease treatment Acid treatment

Trypsin Chymotrypsin Acetic acid (pH 2.5)

[3H]Ondansentron (30 nM) 3±1 4±2 2±0.1[3H]Granisetron (5 Nm) 11±5 10±1 3±0.6[3H]Palonosetron (1 nm) 62±8 64±7 53±2

5-HT3A-HEK293 cells were incubated with 5-fold Kd [3H]ondansetron, [3H]granisetronand [3H]palonosetron for 24 h. Media were then removed and cells were exposed tosaline containing trypsin or chymotrypsin for 5 min at 37 °C or acetic acid pH 2.5 for6 min on ice. After each treatment, cells were washed twice and radioactivity releasedinto washes and that remaining with cells was measured. Recovery of radioactivity wasclose to 100% each time. Values listed are percent of each antagonist remaining withcells. Data shown here are the average of four independent determinations ±S.E.M.

Cell membranes from ondansetron and granisetron-treated cellswere similar in color to vehicle-treated cells; they exhibited acomposite of red and intense yellow colors indicating that the green5-HT3 receptor was present in the cell membrane. In contrast, the cellmembranes from the palonosetron-treated cells exhibited primarilyred fluorescence of the cell membrane dye indicating minimal 5-HT3receptor on the cell surface (Fig. 3, left column).

A corresponding increase in green ECFP-5-HT3 receptorfluorescenceinside the cells could not be detected. Fluorescence inside the cellprimarily reflected transfection efficiency and possibly fluorescent labeldebris after receptor turnover; this fluorescence dwarfed the minorfluorescence increases expected from receptor moving in from thesurface. Consequently, changes in green fluorescence inside the cellwere not representative of receptor internalization (Fig. 3, left column).

The fluorescence images obtained from the microscope exhibitthree colors: green for receptor, red for cell membrane and yellow forreceptor overlapping on the cell membrane. The software does notresolve color mixtures but treats them as individual colors so that itwas not possible to tease out receptor overlapping membrane frommembrane. In order to quantify receptor internalization under eachtreatment, receptor that was not overlapping withmembrane (green)was subtracted from the image. This subtraction left membrane (red)plus receptor overlapping with membrane (yellow) (Fig. 3, middlecolumn).When red and green were subtracted, the resulting yellow isproportional to the amount of receptor on the membrane under eachtreatment (Fig. 3, right column).

A cell population (8–12 cells) under each treatment was analyzedas illustrated in Fig. 3. The ratio of yellow intensity vs. yellow plus red

Fig. 3. Confocal fluorescence microscopy of cells after treatment with 5-HT3 receptorantagonists — HEK293 cells transfected with ECFP-5-HT3A receptor were incubatedwith vehicle (A), ondansetron (B), granisetron (C) and palonosetron (D) for 2 h. Wheatgerm agglutinin (red cell membrane marker) was added 10 min before imaging inorder to monitor appearance of cell membrane after each treatment. Left column: cellscontaining receptor inside cells (green), receptor overlapping membrane (yellow) andmembrane (red). Middle column: receptor inside cells (green) was subtracted from theimage leaving receptor overlapping membrane (yellow) and membrane (red). Rightcolumn: membrane (red) and receptor inside cells (green) was subtracted. Yellowpresent in right column is proportional to the amount of receptor on the membraneunder each treatment. (For interpretation of the references to color in this figurelegend, the reader is referred to the web version of this article.)

Table 3Quantification of fluorescent intensity.

Average ratio ±S.E.M. P value

Control 0.890 0.019 Not applicableOndansetron 0.895 0.015 NSGranisetron 0.864 0.010 NSPalonosetron 0.806 0.016 <0.01

A population of HEK293 cells transfectedwith fluorescent 5-HT3A receptor (ECFP-5-HT3A)was incubated with vehicle, ondansetron, granisetron and palonosetron for 2 h, stainedwith red wheat germ agglutinin dye and observed under the confocal fluorescencemicroscope (Section 2.6). Images were analyzed as illustrated in Fig. 3 using ImageJsoftware. Thefluorescent intensity fromeachanalyzed cell suchas those in Fig. 3, column3was divided by its corresponding fluorescent intensity in Fig. 3, column 2. These ratioswere obtained for 8–12 cells per treatment group and provide an indication of the relativeamount of receptor on the cell surface for each cell. The table lists an average ratio obtainedunder each treatment. The P value compares the average ratio obtained from antagonist-treated cells vs. control using student's t test. NS: not statistically significant.

197C. Rojas et al. / European Journal of Pharmacology 626 (2010) 193–199

intensities was quantified for each cell using ImageJ software(Table 3). This ratio is a measure of how much receptor is on themembrane with respect to the whole membrane in each cell. Theaverage of the ratios corresponding to receptor on the membrane inpalonosetron-treated cells was significantly lower than that ofvehicle, granisetron and ondansetron treated cells (Table 3). Inten-sities measured in cells after treatment with ondansetron andgranisetron were not statistically different to control cells.

Fig. 4. (A) Long-term inhibition of calcium-ion influx after 24 h preincubations withondansetron, granisetron and palonosetron. (B) Long-term inhibition of calcium-ioninflux after various times of preincubation with palonosetron. At the end of eachincubation period, media containing excess antagonist were removed and cells wereincubated for 2.5 h with three additional media changes. Cells were then challengedwith serotonin and calcium influx was measured (Section 2.7). Calcium influx wasnormalized to cells that were not incubated with antagonist (control). Results are theaverage of four independent experiments (⁎P<0.05; ⁎⁎P<0.01; ⁎⁎⁎P<0.001). Errorbars correspond to S.E.M.

3.5. Preincubation of palonosetron with cells causes prolonged inhibitionof receptor function

Cells were preincubated with ondansetron, granisetron and palo-nosetron for 24 h and the effect on functionwasmeasured after 2.5 h ofinfinite dilutions and dissociation. Cells preincubatedwith palonosetronexhibited 37±6% of calcium influx compared to control cells(P<0.001). Cells preincubated with granisetron exhibited 79±8% ofcalcium influx compared to control cells (P<0.05) whereas cellspreincubated with ondansetron readily gained full functionality once

the antagonists were allowed to dissociate (Fig. 4A). Cells were thenpreincubated with palonosetron for shorter times than the 24 h timepoint to determine the minimum time of exposure required to observethe prolonged inhibition effect. Preincubation with palonosetron for15 min significantly reduced receptor function to 67±8.2% (P<0.01compared to control). Preincubation for 30 min further reducedcalcium-ion influx to 43±9.6% (P<0.01 compared to control); thisinhibition of function was the same within experimental error as that

198 C. Rojas et al. / European Journal of Pharmacology 626 (2010) 193–199

obtained after the 24 h preincubation mentioned above. The 1 and 2 hpreincubations confirmed that receptor function had been reduced to aplateau in the 40–50% range (Fig. 4B).

4. Discussion

Palonosetron is a 5-HT3 receptor antagonist used for the preventionof chemotherapy-induced nausea and vomiting and postoperativenausea and vomiting that has shown an improved clinical profile overolder 5-HT3 receptor antagonists (Aapro et al., 2006; Eisenberg et al.,2003; Gralla et al., 2003). The mechanism behind palonosetron'sapparent clinical differentiation has been difficult to ascertain giventhat all these drugs act on the same receptor. Recent work in ourlaboratory has shown that palonosetron exhibits allosteric interactionsandpositive cooperativitywith the5-HT3 receptorwhile twoother5-HT3receptor antagonists, ondansetron and granisetron, were shown toexhibit simple bimolecular competitive binding (Rojas et al., 2008). Thepresentwork showspalonosetron triggers 5-HT3 receptor internalizationand causes prolonged inhibition of 5-HT3 receptor function that persistslong after the antagonist has been allowed to dissociate from the cellsurface.

Given the differences in binding, we wondered if these threeantagonists could exert a different effect on the 5-HT3 receptor. In thisreport we describe four independent biochemical experiments whichindicate that palonosetron triggers receptor internalization. First,incubation of cells with [3H]palonosetron followed by exposure todissociation conditions showed that 60% [3H]palonosetron remainedwith cells. In contrast, virtually all [3H]palonosetron dissociated fromcell-free membranes (Fig. 1, Table 1). Second, over 60% of [3H]palonosetron remained associated with cells after protease treatmentdesigned to denature cell surface proteins (Table 2). Third, over 50% of[3H]palonosetron remained associated with cells after acid treatmentdesigned to denature cell surface proteins (Table 2). Finally,preincubation of cells with unlabeled palonosetron caused subse-quent reduced cell surface binding of [3H]palonosetron (Fig. 2).

These results stood in clear contrast with those obtained forgranisetron and ondansetron. Incubation of cells with [3H]granisetronfollowed by exposure to dissociation conditions showed 15% [3H]granisetron remaining with cells. Incubation with [3H]ondansetron ledto no association with cells (Fig. 1). Treatment of cells with proteases leftminimal amounts of [3H]granisetron (10–11%) and virtually no [3H]ondansetron (3–4%) associated with cells (Table 2). Treatment of cellswith acid left no [3H]granisetron or [3H]ondansetron associated withthe cells (Table 2). The bulk of both [3H]granisetron and [3H]ondansetronwere found in the media after treatment with either proteases or acid(data not shown). Neither preincubation of cells with ondansetronor granisetron caused a subsequent reduction of binding to [3H]palonosetron (Fig. 2A).

Resistance to protease (Simantov and Sachs, 1973) or acid treatment(Haigler et al., 1980) and reduction in binding following preincubationwith ligand (Li et al., 1999) provide strong evidence for receptorinternalization. Taken together, the biochemical evidence indicatespalonosetron triggered substantial receptor internalization (50–60%),while granisetron and ondansetron have minimal to no effect.

In addition to the biochemical evidence presented above, we usedHEK293 cells transfected with a fluorescent tag, enhanced cyanfluorescent protein (ECFP), incorporated into the receptor (ECFP-5-HT3A) to visualize receptor internalization. This system has been usedpreviously to observe agonist-triggered 5-HT3 receptor internalization(Ilegems et al., 2004). When using a red cell membrane stain,membranes from cells incubated with ondansetron and granisetronhad a composite yellow color; this color was the mixture of fluorescentred membrane stain and fluorescent green 5-HT3 receptor. In contrast,the membrane of the cell incubated with palonosetron was mostly red,indicating that there was a reduction of the green 5-HT3 receptor at thecell surface (Fig. 3, left column). Image quantification obtained from a

larger cell population for each treatment showed palonosetron-treatedcells exhibited significantly lower receptor on the membrane thanvehicle, ondansetron or granisetron-treated cells (Table 3). Both, theresults of the biochemical experiments and the confocal images stronglysupported palonosetron-triggered receptor internalization.

Previous reports have demonstrated agonist-triggered 5-HT3 recep-tor internalization using serotonin (Freeman et al., 2006) or mCPBG(Ilegems et al., 2004). Even though antagonist-triggered internalizationhas been reported to occurwith other receptors (Lin et al., 2000; Roettgeret al., 1997), to our knowledge, this is the first report on antagonist-triggered 5-HT3 receptor internalization. Interestingly, palonosetron and5-HT3 receptor agonists exhibit allosteric interactions (Bonhaus et al.,1995; Neijt et al., 1989; Rojas et al., 2008) and they also trigger receptorinternalization. In contrast, granisetron and ondansetron exhibit simplebimolecular binding, bind only to one site (Nelson and Thomas, 1989;Rojas et al., 2008; Vitalis et al., 2001) and do not trigger significantreceptor internalization.

The binding affinity of palonosetron is approximately 2500-foldhigher than that of serotonin (Rojas et al., 2008; Yan et al., 1999). Thismeans that after internalization, palonosetron would be expected toremain bound to the receptor much longer than serotonin, preventingrecycling and causing a reduction in receptor density at the cell surface.Pertinent to this point is the finding that after serotonin-inducedinternalization free receptor reappears at the surface within 1 h afterinternalization (Freeman et al., 2006)while in the case of palonosetron-induced receptor internalization, as measured by reduced binding of[3H]palonosetron, remained internalized for at least 2.5 h afterincubation with palonosetron (Fig. 2). The internalized receptor afterexposure to palonosetron could still be inside the cells or it may havebeen targeted for degradation. In short, palonosetron's binding to theallosteric site could produce a conformational change that facilitatesreceptor internalization and, in contrast to serotonin, it interferes withreceptor exocytosis.

In order to discriminate the long-term effect of receptorinternalization from simple binding of the three antagonists,measurements of calcium-ion influx were made after antagonistswere allowed to dissociate from the surface (Section 2.7). Under theseconditions, the only cells that exhibited substantial inhibition ofreceptor function were the cells that had been preincubated withpalonosetron. Preincubation of cells with granisetron or ondansetronhad minimal or no inhibition of receptor function (Fig. 4A). The onsetof receptor internalization as measured by reduced binding (Fig. 2B)coincided with the onset of long-term inhibition of receptor function(Fig. 4B) suggesting that internalization reduces receptor density atthe surface resulting in prolonged inhibition of receptor function.Functional inhibition could result from prolonged binding of palono-setron to the 5-HT3 receptor inside the cell or from degradation of theinternalized receptor.

These molecular findings provide a potential explanation of theclinicalfindingswherepretreatmentwithpalonosetronhas beenshownto prevent emesis in a substantial proportion of patients throughout the5 days following emetogenic chemotherapy (Eisenberg et al., 2003;Gralla et al., 2003; Yoshizawa et al., 2008). Since ondansetron andgranisetron remain at the cell surface, it is reasonable to speculate theseantagonists are gradually washed away by high levels of serotonin. As aresult, granisetron and ondansetron are effective only in the preventionof acute emesis while they remain bound to the 5-HT3 receptor. Incontrast, palonosetron-induced receptor internalization and subse-quent reduction of receptor density at the cell surface are not onlyeffective in counteracting the acute but also the delayed emetogenicresponse to chemotherapy.

In summary, palonosetron, in contrast to ondansetron and granise-tron, exhibits allosteric interactions, triggers receptor internalizationand exhibits prolonged inhibition of receptor function. These resultsmay provide a pharmacological basis for differences noted in clinicaltrials comparing palonosetron to other 5-HT3 receptor antagonists.

199C. Rojas et al. / European Journal of Pharmacology 626 (2010) 193–199

Acknowledgement

We thank Dr. Solomon H. Snyder at The Solomon H. SnyderDepartment of Neuroscience, Johns Hopkins University, School ofMedicine for helpful advice during the course of the work and duringthe preparation of the manuscript.

References

Aapro, M.S., Grunberg, S.M., Manikhas, G.M., Olivares, G., Suarez, T., Tjulandin, S.A.,Bertoli, L.F., Yunus, F., Morrica, B., Lordick, F., Macciocchi, A., 2006. A phase III,double-blind, randomized trial of palonosetron compared with ondansetron inpreventing chemotherapy-induced nausea and vomiting following highly emeto-genic chemotherapy. Ann. Oncol. 17, 1441–1449.

Bonhaus, D.W., Stefanich, E., Loury, D.N., Hsu, S.A., Eglen, R.M., Wong, E.H., 1995.Allosteric interactions among agonists and antagonists at 5-hydroxytryptamine3receptors. J. Neurochem. 65, 104–110.

Decker, G.M., DeMeyer, E.S., Kisko, D.L., 2006. Measuring the maintenance of daily lifeactivities using the functional living index-emesis (FLIE) in patients receivingmoderately emetogenic chemotherapy. J. Support. Oncol. 4 (35–41), 52.

Dubin, A.E., Huvar, R., D'Andrea, M.R., Pyati, J., Zhu, J.Y., Joy, K.C., Wilson, S.J., Galindo, J.E.,Glass, C.A., Luo, L., Jackson, M.R., Lovenberg, T.W., Erlander, M.G., 1999. Thepharmacological and functional characteristics of the serotonin 5-HT(3A) receptorare specifically modified by a 5-HT(3B) receptor subunit. J. Biol. Chem. 274,30799–30810.

Eisenberg, P., Figueroa-Vadillo, J., Zamora, R., Charu, V., Hajdenberg, J., Cartmell, A.,Macciocchi, A., Grunberg, S., 2003. Improved prevention of moderately emetogenicchemotherapy-induced nausea and vomiting with palonosetron, a pharmacologicallynovel 5-HT3 receptor antagonist: results of a phase III, single-dose trial versusdolasetron. Cancer 98, 2473–2482.

Freeman, S.L., Glatzle, J., Robin, C.S., Valdellon, M., Sternini, C., Sharp, J.W., Raybould, H.E.,2006. Ligand-induced 5-HT3 receptor internalization in enteric neurons in rat ileum.Gastroenterology 131, 97–107.

Geling, O., Eichler, H.G., 2005. Should 5-hydroxytryptamine-3 receptor antagonists beadministered beyond 24 hours after chemotherapy to prevent delayed emesis?Systematic re-evaluation of clinical evidence and drug cost implications. J. Clin.Oncol. 23, 1289–1294.

Glennon, R.A., Dukat, M., 2008. Serotonin receptors and drugs affecting serotonergicneurotransmission. In: Lemke, T.L., Williams, D.A., Roche, V.F., Zito, S.W. (Eds.),Foye's Principles of Medicinal Chemistry, sixth ed. Wolters Kluwer/LippincottWilliams & Wilkins, New York, Y, p. 434.

Gralla, R., Lichinitser, M., Van Der Vegt, S., Sleeboom, H., Mezger, J., Peschel, C., Tonini, G.,Labianca, R., Macciocchi, A., Aapro, M., 2003. Palonosetron improves prevention ofchemotherapy-induced nausea and vomiting following moderately emetogenicchemotherapy: results of a double-blind randomized phase III trial comparing singledoses of palonosetron with ondansetron. Ann. Oncol. 14, 1570–1577.

Haigler, H.T., Maxfield, F.R., Willingham, M.C., Pastan, I., 1980. Dansylcadaverine inhibitsinternalization of 125I-epidermal growth factor in BALB 3T3 cells. J. Biol. Chem. 255,1239–1241.

Hesketh, P.J., 2008. Chemotherapy-induced nausea and vomiting. N. Engl. J. Med. 358,2482–2494.

Ilegems, E., Pick, H.M., Deluz, C., Kellenberger, S., Vogel, H., 2004. Noninvasive imagingof 5-HT3 receptor trafficking in live cells: from biosynthesis to endocytosis. J. Biol.Chem. 279, 53346–53352.

Kao, J.P., 1994. Practical aspects of measuring [Ca2+] with fluorescent indicators.Methods Cell Biol. 40, 155–181.

Li, J., Xia, Y., Kuter, D.J., 1999. Interaction of thrombopoietin with the platelet c-mplreceptor in plasma: binding, internalization, stability and pharmacokinetics. Br. J.Haematol. 106, 345–356.

Lin, J.W., Ju, W., Foster, K., Lee, S.H., Ahmadian, G., Wyszynski, M., Wang, Y.T., Sheng, M.,2000. Distinct molecular mechanisms and divergent endocytotic pathways ofAMPA receptor internalization. Nat. Neurosci. 3, 1282–1290.

Neijt, H.C., Plomp, J.J., Vijverberg,H.P., 1989. Kinetics of themembrane currentmediatedbyserotonin 5-HT3 receptors in cultured mouse neuroblastoma cells. J. Physiol. 411,257–269.

Nelson, D.R., Thomas, D.R., 1989. [3H]-BRL 43694 (Granisetron), a specific ligand for 5-HT3binding sites in rat brain cortical membranes. Biochem. Pharmacol. 38, 1693–1695.

Roettger, B.F., Ghanekar, D., Rao, R., Toledo, C., Yingling, J., Pinon, D., Miller, L.J., 1997.Antagonist-stimulated internalization of the G protein-coupled cholecystokininreceptor. Mol. Pharmacol. 51, 357–362.

Rojas, C., Stathis,M., Thomas,A.G.,Massuda, E.B., Alt, J., Zhang, J., Rubenstein, E., Sebastiani, S.,Cantoreggi, S., Snyder, S.H., Slusher, B., 2008. Palonosetron exhibits unique molecularinteractions with the 5-HT3 receptor. Anesth. Analg. 107, 469–478.

Simantov, R., Sachs, L., 1973. Regulation of acetylcholine receptors in relation toacetylcholinesterase inneuroblastomacells. Proc.Natl.Acad. Sci. U. S.A. 70, 2902–2905.

Stoltz, R., Cyong, J.C., Shah, A., Parisi, S., 2004. Pharmacokinetic and safety evaluation ofpalonosetron, a 5-hydroxytryptamine-3 receptor antagonist, in U.S. and Japanesehealthy subjects. J. Clin. Pharmacol. 44, 520–531.

Thompson, A.J., Lummis, S.C., 2007. The 5-HT3 receptor as a therapeutic target. ExpertOpin. Ther. Targets 11, 527–540.

Vitalis, B., Sebestyen, L., Sike,M., Solyom, S., Harsing Jr., L.G., 2001. Binding characteristics ofGYKI-46 903, a non-competitive ligand at 5-HT3 receptors. Pharmacol. Res. 43,291–299.

Wong, E.H., Clark, R., Leung, E., Loury, D., Bonhaus, D.W., Jakeman, L., Parnes, H.,Whiting, R.L., Eglen, R.M., 1995. The interaction of RS 25259–197, a potent andselective antagonist, with 5-HT3 receptors, in vitro. Br. J. Pharmacol. 114, 851–859.

Yan, D., Schulte, M.K., Bloom, K.E., White, M.M., 1999. Structural features of the ligand-binding domain of the serotonin 5HT3 receptor. J. Biol. Chem. 274, 5537–5541.

Yoshizawa, H., Saito, M., Aogi, K., Sekine, I., Yanagita, Y., Sakai, H., Inoue, K., Kitagawa, C.,Ogura, T., Mitsuhashi, S., 2008. Palonosetron (Palo) is more effective thangranisetron (G) in preventing chemotherapy-induced nausea and vomiting(CINV) associated with cisplatin (CDDP)- or anthracycline plus cyclophosphamide(AC/EC)-based regimens: results of a phase III trial — Palo Japanese CooperativeStudy Group Annals of Oncology 19, viii275.