Gd-BOPTA transport into rat hepatocytes: pharmacokinetic analysis of dynamic magnetic resonance images using a hollow-fiber bioreactor

(1)

Article

Reference

Gd-BOPTA transport into rat hepatocytes: pharmacokinetic analysis of dynamic magnetic resonance images using a hollow-fiber

bioreactor

PLANCHAMP MESSEILLER, Corinne, et al .

Abstract

To investigate the transport of the hepatobiliary magnetic resonance (MR) imaging contrast agent Gd-BOPTA into rat hepatocytes.

PLANCHAMP MESSEILLER, Corinne, et al . Gd-BOPTA transport into rat hepatocytes:

pharmacokinetic analysis of dynamic magnetic resonance images using a hollow-fiber bioreactor. Investigative Radiology , 2004, vol. 39, no. 8, p. 506-15

PMID : 15257212

DOI : 10.1097/01.rli.0000129156.16054.30

Available at:

http://archive-ouverte.unige.ch/unige:42524

Disclaimer: layout of this document may differ from the published version.

(2)

O

RIGINAL

A

RTICLE

Gd-BOPTA Transport Into Rat Hepatocytes:

Pharmacokinetic Analysis of Dynamic Magnetic Resonance Images Using a Hollow-Fiber Bioreactor

Corinne Planchamp, PhD,* Marianne Gex-Fabry, MSc,† Christophe Dornier, PhD,*

Rafael Quadri, MSc,‡ Marianne Reist, PhD,§ Marko K. Ivancevic, PhD,*

Jean-Paul Valle´e, MD, PhD, Sibylle Pochon, PhD,¶ Franc¸ois Terrier, MD,* Luc Balant, PhD,†*

Bruno Stieger, PhD, 储 Peter J. Meier, MD, PhD, 储 and Catherine M. Pastor, MD, PhD*

Rationale and Objectives: To investigate the transport of the hepatobiliary magnetic resonance (MR) imaging contrast agent Gd-BOPTA into rat hepatocytes.

Materials and Methods:In a MR-compatible hollow-fiber bioreactor containing hepatocytes, MR signal intensity was measured over time during the perfusion of Gd-BOPTA. For comparison, the perfusion of an extracellular contrast agent (Gd-DTPA) was also studied. A compartmental pharmacokinetic model was developed to describe dynamic signal intensity-time curves.

Results:The dynamic signal intensity-time curves of the hepatocyte hollow-fiber bioreactor during Gd-BOPTA perfusion were adequately fitted by 2 compartmental models. Modeling permitted to discriminate between the behaviors of the extracellular contrast agent (Gd-DTPA) and the hepatobiliary contrast agent (Gd-BOPTA). It allowed the successfully quantification of the parameters involved in such differences.

Gd-BOPTA uptake was saturable at high substrate concentrations.

Conclusions:The transport of Gd-BOPTA into rat hepatocytes was successfully described by compartmental analysis of the signal intensity recorded over time and supported the hypothesis of a transporter-mediated uptake.

Key Words:MRI, contrast agent, liver, Gd-BOPTA, pharmacokinetic model

(Invest Radiol2004;39: 506 –515)

G

d-BOPTA is a new hepatobiliary contrast agent (CA) for magnetic resonance imaging (MRI) whose structure differs from the well-known extracellular CA Gd-DTPA by an additional lipophilic moiety (Fig. 1). With this lipophilic substituent, part of Gd-BOPTA is taken up specifically by hepatocytes and is partially excreted into the bile. Conse- quently, Gd-BOPTA has a dual imaging capability. On the one hand, it is used as an extracellular agent for MRI similarly to Gd-DTPA, and on the other hand, it can be used in the liver to facilitate MRI detection and characterization of focal and diffuse hepatic diseases.^{1– 4}

Gd-BOPTA is taken up into hepatocytes and excreted without biotransformation.^5,6In rats, 50% of the dose injected is excreted into the bile, the remaining being excreted into urine.^6,7 In contrast, only 2–7% of the dose is excreted into the bile in humans.^6,8

The transport of Gd-BOPTA into hepatocytes is not fully understood. The hepatic transport of Gd-BOPTA has been mainly studied in vivo using pharmacological antago- nists.^7–9 A single study found that Gd-BOPTA crosses rat sinusoidal hepatocyte membrane by passive diffusion.¹⁰ In the isolated perfused rat liver, however, evidence exists that Gd-BOPTA enters into hepatocytes through the same transporter as bilirubin and bromosulfophthalein, the organic anion transporting polypeptide 1 (Oatp1) localized on the sinusoidal membrane of hepatocytes.¹¹ After its intracellular transport, Gd-BOPTA is eliminated into the bile through the ATP-dependent multidrug resistance-associated protein 2 (Mrp2) at the canalicular side of the hepatocyte.^7,12

Additional studies are necessary to better describe Gd- BOPTA transport into hepatocytes. Isolated cells are of great value for the study of transport mechanisms. Hollow-fiber bioreactors (HFBs) have the advantage to restore a physiological environment that ensures the viability and function- ality of cells.¹³ The aim of the study was to compare the kinetics of the MR signal intensity (SI) over time during the

From the *Geneva University Hospitals, Radiology Department, Geneva,

†Geneva University Hospitals, Psychiatry Department, Cheˆne-Bourg;

‡Geneva University Hospitals, Gastroenterology Department, Geneva;

§University of Lausanne, Section of Pharmacy, BEP, Lausanne; ¶Bracco Research SA, Plan-Les-Ouates; and㛳Klinische Pharmakologie und Toxi- cologie, Universita¨tsSpital Zu¨rich, Switzerland.

Supported by the “Fonds National Suisse de la Recherche Scientifique” N 3200-100 868 to CM Pastor.

Reprints: Corinne Planchamp, PhD, Geneva University Hospitals, Radiology Department, Rue Micheli-du-Crest 24, CH-1211 Geneva 14, Switzer- land. E-mail: corinne.planchamp@hcuge.ch

DOI: 10.1097/01.rli.0000129156.16054.30

Investigative Radiology • Volume 39, Number 8, August 2004

506

(3)

perfusion of Gd-DTPA and Gd-BOPTA in an MR-compatible HFB containing freshly isolated rat hepatocytes in suspension and to determine whether the hepatocyte transport of Gd-BOPTA might be better defined by a compartmental pharmacokinetic analysis.

MATERIALS AND METHODS Chemicals

Gd-BOPTA was provided by Bracco Research S.A.

(Geneva, Switzerland). Gd-DTPA is commercially available (Magnevist, Schering, Germany). Bovine serum albumin (BSA, fraction V, 97%) and ethyleneglycol-O,O⬘-bis(2-aminoethyl)-N,N,N⬘,N⬘-tetraacetic acid (EGTA) were purchased from Fluka. Collagenase fromClostridium histolicumtype IV (collagen digestion activity: 295 U/mg solid), bovine pan- creas insulin, 4-(2-aminoethyl) benzenesulfonyl fluoride hy- drochloride (AEBSF), aprotinin, leupeptin, bestatin, pepstatin A, and N-(transepoxysuccinyl)-L-leucine 4-guanidinobutyl- amide (E-64) were obtained from Sigma (Buchs, Switzer- land). All other chemicals were of analytical grade.

Animals

Adult male Sprague-Dawley rats (250 –350 g) were housed in standard cages in a temperature-controlled room

(21–25°C) with a 12-hour light/dark cycle. Standard labora- tory chow and water were given ad libitum. The protocol was approved by our local veterinary office.

Isolation of Hepatocytes

Hepatocytes were isolated by a 2-step collagenase perfusion of the liver as already described by Seglen.¹⁴Briefly, after rat anesthesia (intraperitoneally, sodium pentobarbital, 80 mg/kg), a midline incision was performed and the portal vein was cannulated. Then, the liver was perfused with a buffer solution containing 0.50 mM EGTA at 37°C with a 30 mL/min flow rate. After 10 minutes, collagenase (0.05%) and CaCl₂(2.00 mM) were added in the buffer solution (pH 7.55) and the solution was perfused during 10 minutes. Buffer solution included 151 mM NaCl, 5.37 mM KCl, 0.63 mM Na₂HPO₄, 0.44 mM KH₂PO₄, 4.2 mM NaHCO₃, and 5.55 mM glucose.

After collagenase perfusion, the liver was removed and transferred to a cold buffer solution containing 0.5% BSA and 0.41 mM MgSO₄, pH 7.4. The hepatocytes were gently released by shaking after disruption of the Glisson capsule.

Cell suspension was filtered (100-␮m nylon mesh) and the viable cells were allowed to sediment. The supernatant containing debris and dead cells was discarded. With this tech- nique, cell viability was greater than 90% (trypan blue exclusion). An average of 5⫻ 10⁸cells were isolated from a single liver, which is about 50% of the total hepatocyte population of the liver.¹⁵

Expression of Oatps and Mrp2 in Hepatocytes To ensure the expression of hepatic transporters in freshly isolated hepatocytes, samples containing 5⫻10⁷cells were snap-frozen in liquid nitrogen and kept at – 80°C until use (n ⫽ 3). Samples were homogenized (Dounce tissue grinder) with ice-cold Tris buffer (100 mM Tris-HCl, pH 7.6) containing protease inhibitors, AEBSF, aprotinin, leupeptin, bestatin, pepstatin A, and E-64 and centrifuged (10,000g, 10 minutes, 4°C). This first supernatant was used for the detection of Oatp1 and Oatp4 because no Oatp1 and Oatp4 were present in the pellet. Because no Mrp2 was present in this first supernatant, the pellet was homogenized with the same buffer containing 1% Triton X-100. After centrifugation (10,000g, 10 minutes, 4°C), this second supernatant was used for the detection of Mrp2. To increase the extraction of Oatp2 samples were extracted a second time with Tris buffer (50 mM Tris-HCl, 0.1 mM EGTA, 0.1 mM ethylenedinitrilotet- raacetic acid, 0.1% sodium dodecyl sulfate, 0.1% deoxy- cholic acid, 1% Triton X-100, pH 7.5). Protein concentration was determined according to Bradford (Bio-Rad, Glattbrugg, Switzerland).

Rabbit polyclonal antibodies against rat Oatp1 (Slc21a1),^16,17rat Oatp2 (Slc21a5),¹⁸rat Oatp4 (Slc21a10),¹⁹ and rat Mrp2 (Abcc2)²⁰ were used. Sodium dodecyl sulfate FIGURE 1.Chemical structure of the 2 contrast agents inves-

tigated. MGH^⫹stands for meglumine.

(4)

polyacrylamide gel electrophoresis and Western blotting were performed according to standard procedures. Protein extracts (100␮g) were separated on a 7.5% polyacrylamide gel. After the gel had been transferred to a polyvinylidene diflouride membrane (Millipore, Volketswil, Switzerland), the membrane was blocked with 5% nonfat dry milk in phosphate-buffered saline (PBS; 1 hour at room temperature) and incubated with the specific antibodies. Staining with Ponceau Red before blocking assessed that equal quantities of proteins were loaded on each line. The membrane was incubated overnight at 4°C with anti-Oatp1 (1:1,000), anti-Oatp2 (1:5,000), anti-Oatp4 (1:2,000) or anti-Mrp2 (1:2,000) antibody in PBS containing 5% nonfat dry milk. The membrane was washed 4 times with PBS-0.2% Tween-20 and incubated for 1 hour with an alkaline phosphatase-conjugated goat antirabbit IgG antibody at 1:5,000 (Stressgen, Victoria, Canada). Then, the membrane was washed 4 times with PBS-Tween buffer. Development was performed using an Immune-Star chemiluminescent protein-detection system (Bio-Rad). Molecular weight markers and controls were included in each experiment.

Uptake of Gd-BOPTA by Hepatocytes in Suspension

Suspensions containing 5⫻10⁷freshly isolated hepatocytes were gently mixed with ice-cold incubation solution (10 mL) containing 76.0 mM NaCl, 5.37 mM KCl, 0.63 mM Na₂HPO₄, 0.44 mM KH₂PO₄, 17.9 mM NaHCO₃, 2.00 mM CaCl₂, 0.41 mM MgSO₄, 40.0 mM HEPES Na, 5.55 mM glucose, 99,420 U/L penicillin G, 77,700 U/L streptomycin sulfate, 280 U/L insulin, and 0.5% BSA, pH 7.2. The suspensions were centrifuged (50g, 2 minutes, 4°C) and the supernatant was discarded. The cell pellets were suspended in the incubation solution (100 mL) with no CA, 200 ␮M Gd-BOPTA, or 200 ␮M Gd-DTPA and incubated for 60 minutes (37°C, shaking, 5% CO₂ in air atmosphere). Cell viability was checked every 30 minutes by trypan blue exclusion. After the incubation period, the suspensions were centrifuged and the cell pellets were washed twice with ice-cold incubation solution free of CA. The cell pellets were kept at⫺80°C until the measurement of gadolinium (Gd^3⫹) concentration by inductively coupled plasma atomic emission spectrometry (ICP-AES). Intracellular concentration of Gd^3⫹

was estimated considering that the cell pellets contained 5⫻ 10⁷hepatocytes with a mean volume of 8181␮m³.^21–23 Kinetics of CAs in the MR-Compatible Hepatocyte HFBs

The MR-compatible perfusion system was described in details elsewhere.²⁴ Briefly, it includes a closed perfusion circuit, a heating device, a silicone membrane oxygenator, and a HFB consisting of a network of semipermeable artifi- cial capillaries bundled together within a transparent plastic

shell (Fig. 2). Perfusion solution was continually pumped from a reservoir, heated by circulating a refrigerant through a glass spiral while hot water flowed outside the spiral, charged with O₂ and CO₂ through a silicone membrane oxygenator.

Then it flowed within the intracapillary space (ICS) of the HFB before diffusing through the semipermeable membrane into the extracapillary space (ECS) and returned to the reservoir. To avoid the entry of gas bubbles into the HFB, a plastic bubble trap was installed just before its inlet. To avoid interference with the magnetic field, all material used in the magnet was free of metal and made of plastic and glass. The heating device, the O₂-CO₂ tank and the pump were installed in an adjacent room. All elements inside the magnet were fixed to a PVC support built to fix the HFB at the iso-center of the magnet.

A suspension containing 1⫻10⁹freshly isolated hepatocytes was centrifuged (50g, 2 minutes, 4°C) and the cell pellet was suspended in ice-cold modified University of Wisconsin (UW) solution (30 mL) containing 80 mM lacto- bionic acid, 30 mM raffinose, 25 mM KH₂PO₄, 25 mM NaOH, 10 mM glycine, 5 mM MgSO₄, 5 mM adenosine, 1 mM allopurinol, 165,700 U/L penicillin G, 5% PEG 800, and 0.5% BSA (a 0.5% BSA solution was used instead of a physiological concentration to avoid clotting of the fibers).

The pH of the solution was adjusted to 7.4 with 2.5 N KOH (⬃55 mM). The UW solution was aseptically filtered before use (Millex GS, 0.22␮m, Millipore Co., Bedford, MA). After the ECS and ICS of the HFB had been filled with UW solution, the cell suspension was injected in the ECS (ECS volume: 25 mL, fiber porosity 0.5␮m; Minikros® Sampler M15E 260 01N, Spectrum, Rancho Dominguez). Considering that the hepatocyte mean volume is 8181 ␮m,^321–23 the volume of the cells in the ECS was 8.2 mL. The hepatocyte HFB was preserved at 4°C until MRI.

Buffer solutions (same composition as the incubation solution used for suspensions) with and without CA were successively perfused in the hepatocyte HFB (flow rate: 100 mL/min, temperature: 37°C). The system was perfused with buffer (30 minutes), buffer⫹Gd-DTPA (20 minutes), buffer (20 minutes), buffer⫹Gd-BOPTA (30 minutes), and buffer (30 minutes). Thus, the total time course of an experiment was about 130 minutes. CA concentrations in the perfusion solution were 50, 100, 200, and 400 ␮M (n ⫽ 2 for each concentration).

The viability of hepatocytes was determined during experiment by measuring lactate dehydrogenase and aspartate aminotransferase release in the perfusate (Roche Diagnostics, Rotkreuz, Switzerland) and by determining the hepatocyte O₂ consumption using a blood gas analyzer (ABL 500, Radiom- eter Copenhagen, Brunson, DK).

MRI

Dynamic T₁-weighted imaging of the HFBs was performed using fast gradient echo sequence FAST²⁵ with the

Planchamp et al Investigative Radiology • Volume 39, Number 8, August 2004

508

(5)

following imaging parameters: 90°–180° magnetization preparation, TI/TR/TE 28/10.52/4 milliseconds, FA 90°, FOV 10 cm, matrix size 256⫻256, slice thickness 5 mm, interimage delay 20 seconds.

The sequence was optimized using Gd-DTPA solutions in water.²⁶A linear relationship was found between concentrations and signal up to 0.5 mM. The r₁relaxivity was 4.1 mM^⫺¹s^⫺¹. Similar results were found with Gd-BOPTA in water as well as with Gd-DTPA and Gd-BOPTA in solutions containing 0.5% BSA. Hence, the concentration of BSA used in this study has no influence on the Gd-BOPTA relaxivity in comparison to a 4% BSA solution that increases Gd-BOPTA relaxivity.²⁷

The HFB was included in a wrist coil. During perfusion, 8 cross-sections were imaged on an Eclipse 1.5-T MR system (Marconi Medical Systems Inc., Cleveland, OH).

These sections included HFB inlet tubing; reference vials containing 0.5, 2, and 4 mM of Gd-DTPA in water; and the HFB. Results were expressed as the mean SI in a whole cross-section in the center of the HFB over time normalized to the SI obtained in the 4 mM reference vial.

MRI Data: Compartmental Pharmacokinetic Analysis

Pharmacokinetic Model of MRI Signal

Development of the model proceeded in 3 steps. First, we considered a classic compartmental pharmacokinetic model, which described the relationship between time and amount of CA in the system. Because the MRI signal and not the amount of CA was analyzed, the relationship between SI and amount of CA was then modeled. Finally, the SI versus time model was implemented in a pharmacokinetic software.

Pharmacokinetic Model: Amount of CA Versus Time

A 2-compartment pharmacokinetic model was used to represent the hepatocyte HFB (Fig. 3), where central compartment described the amount of CA in extracellular space A₁(t) and peripheral compartment described the amount in hepatocytes A₂(t). First-order rate constants k₁₀, k₁₂, and k₂₁ reflected elimination rate from central compartment and exchange rates towards and from peripheral compartment, re- FIGURE 2. (A), Schematic illustra- tion of the hollow-fiber bioreactor with a cross-sectional view showing the extracapillary space (ECS) and the intracapillary space (ICS). Illus- tration of the hollow-fiber bioreactor before (B) and after (C) injection of 10⁹hepatocytes in the ECS.

(6)

spectively. Entry of CA into the system was modeled as a zero-order infusion rate K_inover a time period ␶. Biphasic accumulation in the central compartment during infusion period and subsequent decrease were described by the rate constants␣and␤and the corresponding half-lives t_1/2(␣) and t_1/2(␤). Equations describing this model can be found in the literature.²⁸

Relationship Between the Measured SI and CA Amount

The full equation relating MRI signal intensity SI(t) to CA concentration C(t) has been investigated by different groups.^29,30 It depends on tissue characteristics and MR imaging sequence. In the present study, a linear relationship was demonstrated between concentrations and relative signal enhancement over baseline (data not shown). Signal intensi- ties in a unit volume i reflecting extracellular space and in a unit volume j reflecting hepatocytes were written as:

si(t)⫽s0⫹a1䡠C1(t) (1) sj(t)⫽s0⫹a2䡠C2(t) (2) where s₀is baseline signal. C₁(t) and C₂(t) are concentrations in extracellular space and hepatocytes, respectively, and proportionality constants a₁ and a₂generally depend on tissue properties. Average SI in the cross section of the HFB was:

SI(t)⫽s0⫹a1䡠v1

V 䡠C1(t)⫹a2䡠v2

V 䡠C2(t) (3) where v₁ and v₂ are volumes corresponding to extracellular space and hepatocytes, respectively, and V⫽v₁⫹v₂.

Equation 3 can be rewritten in terms of amounts in central and peripheral compartments:

SI(t)⫽s0⫹g1䡠A1(t)⫹g2䡠A2(t) (4) with new proportionality constants g₁⫽a₁/V and g₂⫽a₂/V generally depending on tissue properties but not on true volumes.

Model Implementation and Evaluation

The model was implemented in the NONMEM software (version V, University of California, San Francisco, CA). Originally developed for population pharmacokinetics, ie, analysis of interindividual variability on the basis of sparse data in large patient populations, NONMEM offers a flexible programming environment. As discussed in the first subsec- tion, a predefined two-compartment model was considered (subroutines ADVAN3 and TRANS1), with observed signal intensity SI_obs(t) modeled as:

SIobs(t)⫽SIpred(t)䡠(1⫹⑀) (5) where SI_pred(t) is model-predicted SI according to equation 4 and ⑀ is a random residual error, assumed to be normally distributed with mean 0 and variance␴.²Primary parameters of the model were rate constants k₁₀, k₁₂, k₂₁, and a single proportionality constant g₁⫽ g₂(all analyses were compatible with a single proportionality constant). Secondary parameters were rate constants ␣ and ␤, with corresponding half-lives t_1/2(␣) and t_1/2(␤).

Infusion duration ␶ and baseline signal s₀were fixed parameters determined prior to modeling for each experiment.␶was calculated as the difference between start and end of infusion which were determined as the time points where mean SI differed byⱖ2% from the preceding and following windows (window width⫽3 time points). S₀was determined as the average SI over time points preceding the start of infusion.

Signal versus time curves corresponding to each indi- vidual experiment were analyzed according to 3 different submodels: (1) a 1-compartment model, with no access of CA to peripheral compartment (ie, k₁₂⫽k₂₁⫽0); (2) a 2-compartment model, with an unidirectional exchange with peripheral compartment (ie, k₂₁⫽0); and (3) a full 2-compartment model with bidirectional exchange with peripheral compartment. To identify the best model, statistical comparison of models was based on a likelihood ratio test, ie, a␹²test of the difference in the objective function (the lower the value, the better the fit), with P ⬍ 0.01 considered to represent a FIGURE 3.Pharmacokinetic model-

ing. Entry of contrast agents into the hepatocyte hollow-fiber bioreactor was modeled as a zero-order infusion rateK_in. First-order rate constants k₁₀, k12, and k21reflected elimination from central compartment (extracellular space) and exchange towards and from peripheral compartment (hepatocyte pool), respectively.

510

(7)

significant improvement of fit (ie, ␹² ⬎ 6.6). In addition, standard errors of parameter estimates and correlations between them were considered. Graphical assessment of good- ness of fit was performed through analysis of residuals (ie, differences between measured and predicted SI). When 2 models performed similarly, the simplest model was selected.

Characterization of Gd-BOPTA Transport

To characterize Gd-BOPTA uptake into hepatocytes, SI predicted in the peripheral compartment at the end of infusion (␯⫽ SI value in the hepatocytes at 30 minutes – baseline value) was plotted against the concentration of Gd-BOPTA perfused (C). Using a simple Michaelis-Menten model (Eq.

6), theK_mandV_max were estimated (Prism 3.00, Graph Pad Software Inc., San Diego, CA).

v ⫽ Vmax[C]

(Km⫹[C]) (6)

Statistical Analysis

Data were expressed as mean⫾SD. Statistical analysis was performed by Mann-Whitney U test and by Kruskal-

Wallis Test when appropriate. APvalue less than 0.05 was considered statistically significant.

RESULTS

Suitability of Freshly Isolated Rat Hepatocytes for the Study of Gd-BOPTA Transport

Oatp1, Oatp2, and Mrp2 were detected in freshly isolated rat hepatocytes (Fig. 4). In contrast to liver extract, Oatp4 was not detected in isolated hepatocytes. When hepatocytes were incubated with 200␮M Gd-BOPTA, CA uptake was 32.6⫾4.6␮M (16.3⫾2.3% of the concentration in the incubation solution; Fig. 5). When hepatocytes were incubated with 200␮M Gd-DTPA, the intracellular concentration of Gd^3⫹was only 4.9⫾0.3␮M (2.4⫾0.2%). No CA was detected in hepatocytes incubated in the absence of CA.

Kinetics of CAs in the MR Compatible Hepatocyte HFB

Hepatocyte viability was routinely greater than 85%

during the entire protocols, based on the amount of aspartate aminotransferase and lactate dehydrogenase released in the perfusate (⬍15%). The O₂consumption was 663⫾52␮M/h during the first hour and then increased to 1037⫾27␮M/h (n ⫽ 3). This increase may reflect an augmented cellular activity during the rewarming (from 4°C to 37°C) that occurs at the initiation of perfusion.

During infusion of 200␮M Gd-DTPA (20 minutes), SI rapidly increased toward steady-state (Fig. 6). It returned rapidly to baseline value during the washing. During infusion of 200 ␮M Gd-BOPTA (30 minutes), the initial rapid SI increase was followed by a slower phase. During the wash- FIGURE 4. Protein expression in freshly isolated rat hepato-

cytes (n⫽3, lines 1–3). C, control liver extract; Oatp, organic anion transporting polypeptide; mrp, multidrug resistance- associated protein

FIGURE 5. Concentration of gadolinium (Gd^3⫹) in hepatocytes. Isolated hepatocytes were incubated during 60 minutes with 200␮M Gd-BOPTA, 200␮M Gd-DTPA, or in the absence of contrast agent (control). Gd³^⫹was measured in cell pellets by inductively coupled plasma atomic emission spectrometry.

Data are expressed as mean⫾SD, n⫽3 suspensions in each group (Kruskal-Wallis test,P⫽0.024).

(8)

ing, SI first decreased rapidly but did not return to baseline value as observed with Gd-DTPA. When the HFB was infused with various CA concentrations (50, 100, 200, and 400␮M), SI increased with CA concentrations but the shape of the kinetics remained similar.

For Gd-DTPA, the best model was the 2-compartment model with unidirectional exchange with peripheral compartment. This model was significantly better than the 1-compartment model with no access of CA to peripheral compartment (likelihood ratio test, P ⬍ 0.01). The full 2-compartment model with bidirectional exchange with peripheral compartment did not improve the fit. The precision of the parameter estimates was adequate, with standard errors less than 10%

for all parameters except for 2 experiments where standard error of k₁₂exceeded this limit. Moreover, the residual error was less than 3% for all experiments. Mean elimination rate from central compartment (k₁₀) was 0.454 minutes^⫺1 with 21.6% variability between the different experiments. Ex- change rate towards peripheral compartment was very slow

(k₁₂⫽0.00334 minutes^⫺1, 68.8% between-experiment variability). Monoexponential increase and decrease were described by a single half-life t_1/2(␣)⫽ 1.57 minutes (18.9%

between-experiment variability). Model predicted SI in the central compartment rapidly increased toward steady-state and returned to baseline value during the washing, while predicted SI in the peripheral compartment was tiny (Fig. 6).

For Gd-BOPTA, the 2-compartment model with bidirectional exchange with peripheral compartment was best.

Standard error of parameter estimates was less than 10% for all parameters, and the residual error was less than 3% for all experiments. Elimination rate from central compartment (k₁₀) was 0.443 minutes^⫺1 with 32.6% variability between the different experiments. Exchange rates towards (k₁₂) and from (k₂₁) peripheral compartment were 0.00980 minutes^⫺1 (53.9% between-experiment variability) and 0.0250 minutes^⫺1(37.0% between-experiment variability), respectively.

There was a rapid first phase (t_1/2(␣)⫽1.63 minutes, 21.8%

between-experiment variability) and a slow second phase FIGURE 6. Signal intensity(SI)-time curve of Gd-BOPTA and Gd-DTPA (200 ␮M) in the hepatocyte hollow-fiber bioreactor:

representative curves and curves predicted from the pharmacokinetic analysis with a two-compartment model.

512

(9)

(t_1/2(␤)⫽31.6 minutes, 34.7% between-experiment variability). Model predicted SI in the central compartment rapidly increased toward steady state and returned to baseline value during the washing (Fig. 6), as with Gd-DTPA. Predicted SI in the peripheral compartment, however, slowly increased during the infusion and remained steady during washing. The k₁₂ for Gd-BOPTA was statistically higher than for Gd- DTPA (Mann-Whitney U test: P ⫽ 0.007). There was no statistical difference for k₁₀. The Michaelis-Menten model showed that Gd-BOPTA uptake was saturable with an appar- entK_mvalue of 270⫾111␮M and aV_maxvalue of 6.62⫾ 1.38 SI units at 30 minutes (Fig. 7).

DISCUSSION

Suitability of Freshly Isolated Rat Hepatocytes for the Study of Gd-BOPTA Transport

The expression of Oatp1, Oapt2, and Mrp2 was evidenced in freshly isolated hepatocytes, in contrast to Oatp4 (Fig. 4). Thus, although membrane polarity is altered during the isolation and the different poles of the cells (the canalicular and sinusoidal membranes) are no longer distinguished, Oatp1, Oatp2, and Mrp2 were not lost during the isolation.

The expression of Oatp1, Oapt2, and Mrp2 has already been evidenced in primary cultured rat hepatocytes.³¹

16.3% of the total extracellular concentration of Gd- BOPTA was measured in freshly isolated hepatocytes in suspension (Fig. 5), which indicated that Gd-BOPTA could enter into cells. Hence, freshly isolated hepatocytes can be injected in our MR compatible HFB system to study Gd- BOPTA transport kinetics.

The slight entry of Gd-DTPA in hepatocytes (2.4%) may be explained by CA sticking to the cell surface or uptake by pinocytosis or endocytosis. Such small contamination has already been shown with Gd-DTPA in isolated rat hepatocytes in suspension.³²

Kinetics of CAs in the MR-Compatible Hepatocyte HFB

To better define the transport mechanisms of Gd- BOPTA into hepatocytes, cells were injected in a HFB, and the SI over time was recorded and analyzed with a compartmental pharmacokinetic model. The SI-time curves of Gd- DTPA and Gd-BOPTA were adequately fitted by 2-compartmental models. Low standard error of parameter estimates indicated that the models were fully identifiable from the experimental data. In addition to fitting observations from the whole cross section of the HFB, pharmacokinetic modeling enabled to estimate the SI as a result of the presence of CA in 2 compartments postulated to reflect the extracellular space and the hepatocyte pool.

The elimination rate constant from central compartment was identical for Gd-DTPA and Gd-BOPTA, as expected from CA elimination by simple continuous flux of solution.

The transfer rate to the peripheral compartment was significantly higher for Gd-BOPTA than for Gd-DTPA, in keeping with Gd-BOPTA uptake by hepatocytes. Predicted SI-time curve in the central compartment was similar with the 2 CAs (Fig. 6), and was postulated to reflect the extracellular space filling. SI attributed to peripheral compartment was high for Gd-BOPTA but very small for Gd-DTPA. Thus, the model permitted to discriminate between the behaviors of an extracellular CA (Gd-DTPA) and a hepatobiliary CA (Gd- BOPTA).

Gd-BOPTA entered into the peripheral compartment slowly over the infusion period and was only slightly released back in the central compartment during washing. The biliary excretion of Gd-BOPTA is known to occur via the active transporter Mrp2.^7,12,33 In rats, 30% of the injected dose (1 mmol/kg intravenously) is eliminated into the bile after 168 hours.⁶ Although Mrp2 was expressed in freshly isolated hepatocytes (Fig. 4), the SI in the peripheral compartment remained high during the washing period (Fig. 6). Several hypotheses can be proposed. First, the excretion might be slow because excretion of Gd-BOPTA by Mrp2 into the bile is a rate-limiting step.³³ Secondly, Gd-BOPTA might be excreted by Mrp2 into an “excretory domain” between hepatocyte couplets which are reported to form 1 to 4 hours after hepatocytes isolation.^34,35Because these domains are closed and lack duct to the central compartment, Gd-BOPTA might be trapped with no access to the perfusion solution. Thirdly, the excretion could be impaired by (1) loss of Mrp2 function after cell isolation, (2) a defect of the energetic state of the hepatocytes, or (3) an unspecific binding of Gd-BOPTA to FIGURE 7. Predicted signal intensity (SI) in the peripheral

compartment after 30 minutes perfusion of the hepatocyte hollow-fiber bioreactor with a solution containing Gd-BOPTA at concentrations ranging from 50 to 400␮M (n⫽2 for each concentration).

(10)

hepatocytes. Of note in our HFB system, the cellular release of Gd-BOPTA back to the central compartment (ie, the solution) by Mrp2 is not distinguishable from the reversible transport through Oatps.³⁶

Gd-BOPTA uptake exhibited saturability at high substrate concentrations (Fig. 7) andK_m was estimated at 270

␮M. These observations support a transporter-mediated mechanism compatible with the Oatp hypothesis^37,38 rather than an uptake mechanism by free diffusion.¹⁰ Uptake of Gd-EOB-DTPA (another hepatobiliary CA similar to Gd- BOPTA) in Oatp1 cRNA-injectedX. laevisoocytes has aK_m value of⬃3300␮M.³⁷By comparison, aK_mvalue of 1.5␮M was reported for the Oatp1 uptake of bromosulfophthalein (BSP) inX. laevisoocytes.³⁹The lowerK_mof BSP compared with that of Gd-BOPTA and Gd-EOB-DTPA may explain why BSP inhibits the uptake of these CAs.^11,37In our experimental model, the transporter-mediated mechanism could also be confirmed by such pharmacological inhibition.

Compartmental pharmacokinetic models describing bi- ologic processes have been used for metabolic studies. In isolated perfused rat livers, modification of drug uptake,⁴⁰ metabolism⁴¹or excretion,⁴²determination of the rate limiting steps in drug disposition and identification of sites where drugs interact⁴³ have been investigated with compartmental analysis. In addition, pharmacokinetic analysis of MR images obtained with extracellular CAs has emerged as a promising method for diagnostic and therapeutic monitoring of cancer,^30,44for the measurement of blood brain-barrier permeabil- ity,^45,46 and for the assessment of organ perfusion such as the liver.^{47– 49} All these models used compartments representing the blood and the interstitial space to assess the distribution of extracellular CA between blood and tissue.⁵⁰In this study, we developed a model that includes the intracellular compartment.

The present pharmacokinetic analysis supports a transporter- mediated uptake of Gd-BOPTA into hepatocytes.

In summary, the present study showed that freshly isolated rat hepatocytes injected into a hepatocyte HFB ex- press functional Oatps and Mrp2. The transport of Gd- BOPTA into hepatocytes was successfully described by compartmental analysis of the SI recorded over time and supported the hypothesis of a transporter-mediated uptake.

REFERENCES

1. Hamm B, Kirchin M, Pirovano G, et al. Cinical utility and safety of Multihance in magnetic resonance imaging of liver cancer: results of multicenter studies in Europe an the USA.J Comput Assist Tomogr.

1999;23(Suppl 1):S53–S60.

2. Manfredi R, Maresca G, Baron RL, et al. Gadobenate dimeglumine (BOPTA) enhanced MR imaging: patterns of enhancement in normal liver and cirrhosis.J Magn Reson Imaging. 1998;8:862– 867.

3. Schneider G, Maas R, Schultze Kool L, et al. Low-dose gadobenate dimeglumine versus standard dose gadopentetate dimeglumine for contrast-enhanced magnetic resonance imaging of the liver.Invest Radiol.

2003;38:85–94.

4. Kuwatsuru R, Kadoya M, Ohtomo K, et al. Comparison of gadobenate dimeglumine with gadopentetate dimeglumine for magnetic resonance imaging of liver tumors.Invest Radiol. 2001;36:632– 641.

5. Schuhmann-Giampieri G, Frenzel T, Schmitt-Willich H.

Pharmacokinetics in rats, dogs and monkeys of a gadolinium chelate used as a liver-specific contrast agent for magnetic resonance imaging.

Drug Res. 1993;43:927–931.

6. Lorusso V, Arbughi T, Tirone P, et al. Pharmacokinetics and tissue distribution in animals of gadobenate ion, the magnetic resonance imaging contrast enhancing component of gadobenate dimeglumine 0.5 M solution for injection (MultiHance).J Comput Assist Tomogr. 1999;

23(Suppl 1):181–194.

7. De Hae¨n C, Lorusso V, Tirone P. Hepatic transport of gadobenate dimeglumine in TR^⫺rats.Acad Radiol. 1996;3:452– 454.

8. Spinazzi A, Lorusso V, Pirovano G, et al. Multihance clinical pharma- cology.Acad Radiol. 1998;5(Suppl):86 – 89.

9. De Hae¨n C, Lorusso V, Luzzani F, et al. Hepatic transport of the magnetic resonance imaging contrast agent gadobenate dimeglumine in the rat.Acad Radiol. 1995;2:S232–S238.

10. Pascolo L, Cupelli F, Anelli PL, et al. Molecular mechanisms for the hepatic uptake of magnetic resonance imaging contrast agents.Biochem Biophys Res Commun. 1999;257:746 –752.

11. Pastor CM, Planchamp C, Pochon S, et al. Kinetics of gadobenate dimeglumine in isolated perfused rat liver: MR imaging evaluation.

Radiology. 2003;229:119 –125.

12. Pascolo L, Petrovic S, Cupelli F, et al. ABC protein transport of MRI contrast agents in canalicular rat liver plasma vesicles and yeast vacu- oles.Biochem Biophys Res Commun. 2001;282:60 – 66.

13. Planchamp C, Vu LT, Mayer JM, et al. Hepatocyte Hollow-fibre bioreactors: design, set-up, validation and applications.J Pharm Pharmacol.

2003;55:1181–1198.

14. Seglen O. Preparation of isolated rat liver cells. In: Prescott DM, ed.

Methods in Cell Biology. New York: Academic Press; 1976:29 – 83.

15. Quistorff B, Dich J, Grunnet N. Preparation of isolated rat liver hepatocytes. In: Pollard JW, Walker JM, eds.Animal Cell Culture. Totowa, NJ: Humana Press; 1989:151–160.

16. Eckhardt U, Schroeder A, Stieger B, et al. Polyspecific substrate uptake by the hepatic organic anion transporter Oatp1 in stably transfected CHO cells.Am J Physiol. 1999;276:G1037–G1042.

17. Dumont M, Jacquemin E, D’Hont C, et al. Expression of the liver Na^⫹-independent organic anion transporting polypeptide (oatp-1) in rats with bile duct ligation.J Hepatol. 1997;27:1051–1056.

18. Reichel C, Gao B, Van Montfoort JE, et al. Localization and function of the organic anion-transporting polypeptide Oatp2 in rat liver.Gastroen- terology. 1999;117:688 – 695.

19. Cattori V, Van Montfoort JE, Stieger B, et al. Localization of organic anion transporting polypeptide 4 (Oatp4) in rat liver and comparison of its substrate specificity with Oatp1, Oatp2 and Oatp3.Pflugers Arch.

2001;443:188 –195.

20. Madon J, Eckhardt U, Gerloff T, et al. Functional expression of the rat liver canalicular isoform of the multidrug resistance-associated protein.

FEBS Lett. 1997;406:75–78.

21. Greengard O, Federman M, Knox WE. Cytomorphology of developing rat liver and its application to enzyme differentiation. J Cell Biol.

1972;52:261–272.

22. Sasse D, Spornitz UM, Maly IP. Liver architecture.Enzyme. 1992;46:

8 –32.

23. Weibel ER, Staubli W, Gnagi HR, et al. Correlated morphometric and biochemical studies on the liver cell.J Cell Biol. 1969;42:68 –91.

24. Planchamp C, Ivancevic MK, Pastor CM, et al. Hollow fiber bioreactor:

new development for the study of contrast agent transport in hepatocytes by magnetic resonance imaging.Biotechnol Bioeng. 2004;85:656 – 665.

25. Gyngell ML. The application of steady-state free precession in rapid 2DFT NMR imaging: FAST and CE-FAST sequences. Magn Reson Imaging. 1988;6:415– 419.

26. Ivancevic MK, Zimine Y, Lazeyras F, et al. FAST sequences optimiza- tion for contrast media pharmacokinetic quantification in tissue.J Magn Reson Imaging. 2001;14:771–778.

27. De Hae¨n C, Cabrini M, Akhnana L, et al. Gadobenate dimeglumine 0.5M solution for injection (Multihance): pharmaceutical formulation

514

(11)

and physicochemical properties of a new magnetic resonance imaging contrast medium.J Comput Assist Tomogr. 1999;23:S161–S168.

28. Wagner JG. Integrated equations for several pharmacokinetic models.

In:Biopharmaceutics and Relevant Pharmacokinetics. Hamilton: Drug Intelligence Publications; 1971:292–296.

29. Taylor JS, Reddick WE. Evolution from empirical dynamic contrast- enhanced magnetic resonance imaging to pharmacokinetic MRI.Adv Drug Deliv Rev. 2000;41:91–110.

30. Evelhoch JL. Key factors in the acquisition of contrast kinetic data for oncology.J Magn Reson Imaging. 1999;10:254 –259.

31. Rippin SJ, Hagenbuch B, Meier PJ, et al. Cholestatic expression pattern of sinusoidal and canalicular organic anion transport system in primary cultured hepatocytes.Hepatology. 2001;33:776 –782.

32. Thorstensen K, Romslo I. The interaction of gadolinium complexes with isolated rat hepatocytes.BioMetals. 1995;8:65– 69.

33. De Hae¨n C, La Ferla R, Maggioni F. Gadobenate dimeglumine 0.5 M solution for injection (Multihence) as contrast agent for magnetic resonance imaging of the liver: mechanistic studies in animals.J Comput Assist Tomogr. 1999;23(Suppl. 1):S169 –S179.

34. Gautam A, Ng O-C, Boyer JL. Isolated rat hepatocyte couplets in short-term culture: structural characteristics and plasma membrane re- organization.Hepatology. 1987;7:216 –223.

35. Roelofsen H, Soroka CJ, Keppler D, et al. Cyclic AMP stimulates sorting of the canalicular organic anion transporter (Mrp2/cMoat) to the apical domain in hepatocyte couplets.J Cell Sci. 1998;111:1137–1145.

36. Shi X, Bai S, Ford AC, et al. Stable inducible expression of a functional rat liver organic anion transport protein in HeLa cells.J Biol Chem.

1995;270:25591–25595.

37. Van Montfoort JE, Stieger B, Meijer DKF, et al. Hepatic uptake of the magnetic resonance imaging contrast agent gadoxetate by the organic anion transporting polypeptide Oatp1.J Pharmacol Exp Ther. 1999;290:153–157.

38. Moncelli MR, Becucci L, De Hae¨n C, et al. Interactions between magnetic resonance imaging contrast agents and phospholipid monolay- ers.Bioelectrochem Bioenerg. 1998;46:205–213.

39. Jacquemin E, Hagenbuch B, Stieger B, et al. Expression cloning of rat liver Na^⫹-independent organic anion transporter.Proc Natl Acad Sci USA. 1994;91:133–137.

40. Proost JH, Nijssen HMJ, Strating CB, et al. Pharmacokinetic modeling of the sinusoidal efflux of anionic ligands from the isolated perfused rat liver: the influence of albumin.J Pharmacokin Biopharm. 1993;21:375–

394.

41. Beaufort TM, Proost JH, Maring J-G, et al. Effect of hypothermia on the hepatic uptake and biliary excretion of vecuronium in the isolated perfused rat liver.Anesthesiology. 2001;94:270 –279.

42. Xiong H, Turner KC, Stacy Ward E, et al. Altered hepatobiliary disposition of acetaminophen glucuronide in isolated perfused livers from multidrug resistance-associated protein 2-deficient TR^⫺ rats.

J Pharmacol Exp Ther. 2000;295:512–518.

43. Booth CL, Pollack GM, Brouwer KLR. Hepatobiliary disposition of valproic acid and valproate glucuronide: Use of a pharmacokinetic model to examine the rate-limiting steps and potential sites of drug interactions.Hepatology. 1996;23:771–780.

44. Hawighorst H, Knapstein PG, Weikel W, et al. Angiogenesis of uterine cervical carcinoma: characterization by pharmacokinetic resonance parameters and histological microvessel density with correlation to lym- phatic involvement.Cancer Res. 1997;57:4777– 4786.

45. Port MD, Knopp MV, Hoffmann U, et al. Multicompartment analysis of gadolinium chelate kinetics: Blood-tissue exchange in mammary tumors as monitored by dynamic MR imaging.J Magn Reson Imaging. 1999;

10:233–241.

46. Tofts PS, Kermode AG. Measurement of the blood-brain barrier perme- ability and leakage space using dynamic MR imaging. 1. Fundamental concepts.Magn Reson Med. 1991;17:357–367.

47. Scharf J, Zapletal C, Hess T, et al. Assessment of hepatic perfusion in pigs by pharmacokinetic analysis of dynamic MR images.J Magn Reson Imaging. 1999;9:568 –572.

48. Materne R, Smith AM, Peeters F, et al. Assessment of hepatic perfusion parameters with dynamic MRI.Magn Reson Med. 2002;47:135–142.

49. Van Beers BE, Materne R, Annet L, et al. Capillarization of the sinusoids in liver fibrosis: noninvasive assessment with contrast-enhanced MRI in the rabbit.Magn Reson Med. 2003;49:692– 699.

50. Tofts PS. Modeling tracer kinetics in dynamic Gd-DTPA MR imaging.

J Magn Reson Imaging. 1997;7:91–101.