Nonalcoholic fatty liver diseases include a large spectrum of abnormalities characterized by an increased intrahepatic triglyceride content (steatosis) with or without inflammation and fibrosis . The transition from fatty livers to more severe disease (or steatohepatitis) is triggered by inflammation. Cholestasis (or decreased bile flow) is an important marker of this increased severity. Cholestasis is associated with the dysregulation of proteins involved in the transport of bile acids, phospholipids, sterols and organic anions through hepatocytes [2-5]. Interestingly, when the bile excretion of xenobiotics and endobiotics or their metabolites is impaired, the upregulation of transporters located in the sinusoidal membrane may favor their efflux back into sinusoidal blood, preventing a potentially harmful accumulation within hepatocytes . The expression of hepatic transporters during cholestasis is modified, but the consequences on the hepatic accumulation of hepatobiliary compounds are largely unknown. Interestingly, various liver imaging techniques have been developed to assess fatty livers in patients [7-9]. Few experimental and clinical studies have used magnetic resonance imaging (MRI) with hepatobiliary contrast agents to characterize these livers [10-12].
We previously showed that hepatic concentrations of 153Gd–BOPTA can be measured on-line using a gamma scintillation probe placed over rat livers . Gd–BOPTA, the active moiety of gadobenate dimeglumine (MultiHance®, Bracco Milan), is an MR contrast agent that distributes within the extracellular hepatic space and enters into hepatocytes through the sinusoidal transporters organic anion transporting peptides (Oatp1a1, Oatp1a4, and Oatp1b2) . Gd–BOPTA is not metabolized during its transport to the canalicular membrane and is excreted unchanged into bile through Mrp2 [14, 15]. At liver MRI, Gd–BOPTA remains within the extracellular space in rat livers with severe cirrhosis that lack sinusoidal transporters .
Because cholestatic fatty livers have a decreased expression of Oatp1a4 [17, 18] and Mrp2 [17, 19], we hypothesized that Gd–BOPTA would differently accumulate within cholestatic hepatocytes. Moreover, cholestasis may also modify the transport of Gd–BOPTA through the transporters. The aim of the present study was then to compare the pharmacokinetics of Gd–BOPTA concentrations in normal and cholestatic rat fatty livers and to understand how the transport through cell membrane proteins generates the concentrations within cholestatic hepatocytes.
2 MATERIALS AND METHODS
Before liver isolation, Zucker fat (fa/fa) and Zucker lean (fa/+) rats were anesthetized with pentobarbital (50 mg kg−1 i.p.). The protocol was approved by the animal welfare committee of the University of Geneva and the veterinary office and followed the guidelines for the care and use of laboratory animals.
2.2 Perfusion of Isolated Rat Livers
Livers were isolated and perfused in situ with a Krebs–Henseleit-bicarbonate (KHB) solution ± contrast agents using a nonrecirculating system (Fig. 1) . In each experiment, the flow rate was 30 ml min−1 and the bile duct was cannulated for bile sampling collection.
2.3 Quantification of Hepatic Concentrations
Gd–BOPTA was prepared by adding 153GdCl3 to a 0.5 M Gd–BOPTA solution (1 MBq ml−1), which contained a slight excess of the ligand BOPTA [20, 21]. To assess the hepatic concentrations induced by the perfusion of the extracellular contrast agent Gd–DTPA (gadopentetate dimeglumine, Magnevist®, Bayer, Zürich), Gd–DTPA was also prepared by adding 153GdCl3 to a 0.5 m solution of Magnevist®. Gd–DTPA is an extracellular contrast agent, which has the same extracellular distribution as Gd–BOPTA but does not enter into hepatocytes [13, 14]. Both contrast agents were then diluted in KHB solution to obtain a 200 µ m concentration to be perfused in rat livers.
To quantify the on-line hepatic concentrations of 153Gd-labeled contrast agents, a gamma-scintillation probe that measures radioactivity every 20 s was placed 1 cm above the liver (Fig. 1A) . To transform radioactivity count rates into contrast agent amounts, the radioactivity in the entire liver at the end of each experiment was measured (Activimeter Isomed 2000, MCD Nuklear Medizintechnik, Germany) and related to the last count rates measured by the probe.
2.4 Experimental Protocol
We perfused five livers isolated from fa/+ (control Zucker) rats and five livers isolated from fa/fa (fat Zucker) rats with KHB solution (15 min, recovery period), 153Gd–DTPA (10 min), KHB solution (rinse period, 30 min), 153Gd–BOPTA (30 min) and KHB solution (rinse period, 30 min; Fig. 1). Both contrast agents were perfused with a 200 µ m concentrations. During each perfusion, we measured 153Gd–BOPTA bile concentration (nmol g−1), 153Gd–BOPTA bile excretion rate (nmol min−1 g−1), 153Gd–BOPTA concentration within the liver (nmol g−1) and bile flow (µl min−1 g−1).
2.5 Liver Tests and Histology
Alanine amino transferases, cholesterol and triglycerides in serum were determined using a serum multiple analyzer. Liver tissue was fixed, embedded in paraffin (4 µ m thick sections), and analyzed under light microscopy after hematoxylin and eosin staining.
2.6 Simulations in the Pharmacokinetic Model
In perfused livers isolated from control fa/+ rats, the compartment model that best described the amounts of contrast agents over time in perfusate, bile and liver is illustrated in Fig. 1B [13, 21]. Each compartment describes the amount of contrast agent over time. Compartment 1 reflects the mixing between KHB and contrast agent solutions when the solutions of perfusion are changed and accounts for the progressive changes of concentrations before the final concentrations (200 or 0 µ m) are reached. Compartment 2 (C2) describes the amount of contrast agent in the extracellular space and compartments 4 (C4) and 5 (C5) the amounts in hepatocytes from two distinct compartments. Indeed, the model distinguishes two compartments of hepatocytes C4 and C5 with different concentrations over time [13, 21]. The transfer between C4 and C5 is unidirectional. From both compartments, a transfer into compartment 6 (C6 or bile compartment) occurs. Compartment 3 represents the amount of contrast agent in the outflow perfusate leaving the liver. Entry of contrast agent into the system is modeled by a zero-order input rate Kin over a time period. In the first-order rate constant kij, i represents the compartment of origin and j the compartment of convergence. The model was implemented in the MATLAB Software (2010b). It was assumed that the rate constant k24 is associated with Oatp-mediated hepatic uptake of 153Gd–BOPTA. k46 and k56 describe Mrp2-dependent canalicular excretion from both hepatocellular compartments and k42 is associated with the efflux of 153Gd–BOPTA from hepatocytes back to sinusoids. Pharmacokinetic modeling of contrast agent transport was compared between fa/fa and fa/+ rats.
Additionally, we conducted mathematical simulations and evaluated how modifications of a single parameter representing 153Gd–BOPTA entry into hepatocytes (k24), efflux back from hepatocytes to the extracellular space (k42), or bile excretion from both hepatocellular compartments (k46 and k56) modify the hepatic concentration of 153Gd–BOPTA. These simulations mimic the abnormalities that may be observed in cholestatic hepatocytes.
Parameters are means ± SD. A Mann–Whitney test compared means between fa/+ and fa/fa rats. Two-way ANOVA compared the evolution over time of parameters between the experimental groups.
3.1 Characteristics of fa/+ and fa/fa Rats
The fa/fa rats (8 week old) are overweight and have significantly higher serum triglyceride and cholesterol concentrations than fa/+ rats (Table 1). Liver weights were greater in obese rats. During liver perfusion, portal pressure was similar in both groups. Moreover, bile flow was significantly lower in fa/fa (0.6 ± 0.1 µl min−1 g−1) than fa/+ (1.3 ± 0.1 µl min−1 g−1) rats (p < 0.008, data not shown). Histology was normal in fa/+ rats but steatosis was present in fatty livers (Fig. 1C).
Table 1. Characteristics of fa/fa and fa/+ rats
Data are means ± SD.
8.5 ± 0.0
8.8 ± 0.5
Body weight (g)
247 ± 10
328 ± 15
Liver weight (g)
9.6 ± 0.6
15.5 ± 2.2
Portal pressure (mmHg)
7.7 ± 1.8
7.4 ± 0.5
Serum cholesterol (mmol l−1)
1.1 ± 0.1
2.2 ± 0.2
Serum triglycerides (mmol l−1)
0.4 ± 0.1
2.3 ± 0.6
Serum ALAT (UI l−1)
79 ± 10
135 ± 25
3.2 Concentrations of 153Gd–BOPTA in Hepatocytes from Normal and Fatty Livers (T0–T75)
During the perfusion with the extracellular contrast agent 153Gd–DTPA (Fig. 2A), the hepatic concentrations rapidly increased to a low steady state (<100 nmol g−1). 153Gd–DTPA was then rapidly washed out from the liver during the subsequent KHB perfusion and, as expected, no 153Gd–DTPA was detected in bile (Fig. 3C). At time 45 min (T45), 153Gd–BOPTA was perfused during 30 min before washing with KHB solution. 153Gd–BOPTA hepatic concentrations were much higher than that of 153Gd–DTPA and maximal at T75 (Fig. 2A).
To assess the concentrations of 153Gd–BOPTA within hepatocytes, we must withdraw the concentrations induced by the extracellular distribution of the contrast agent (see Fig. 3A). We can then measure an initial hepatocellular uptake index (IHUI, nmol min−1 g−1) from time 46 min (T46) to T49, a period with minimal 153Gd–BOPTA excretion adequate to measure the entry of 153Gd–BOPTA from sinusoids to hepatocytes (Fig. 3A). The IHUI was significantly decreased in fa/fa rats (27 ± 5 nmol min−1 g−1) in comparison to the control group (48 ± 11 nmol min−1 g−1; Fig. 3B, p = 0.008). After this initial period, concentrations inside hepatocytes reflect both hepatocyte uptake and exit.
In contrast to the hepatic concentrations of 153Gd–BOPTA shown in Fig. 2A (those obtained by the gamma probe), the accumulation and elimination concentrations in Fig. 3D and E only reflect 153Gd–BOPTA concentrations within hepatocytes. At T75, the maximal concentrations were similar in both experimental groups and a two-way ANOVA over time between both groups found no interaction (Fig. 3D, p = 1.00). At T75, 153Gd–BOPTA bile accumulation was lower in fa/fa rats while 153Gd–BOPTA uptake was impossible to measure at this time-point, accumulation reflecting the combination of entry and exit.
3.3 153Gd–BOPTA Elimination from Hepatocytes (T75–T105)
When 153Gd–BOPTA perfusion was replaced by a KHB solution, 153Gd–BOPTA concentrations within hepatocytes gradually decreased (Fig. 3E). In the absence of 153Gd–BOPTA uptake, the elimination from hepatocytes can be measured by an initial hepatocellular elimination index (IHEI, nmol min−1 g−1; Fig. 3E). The IHEI was measured from T76 to T79 and was significantly decreased in fa/fa rats (−19 ± 5 nmol min−1 g−1) vs fa/+ rats (−40 ± 12 nmol min−1 g−1, p < 0.008). Moreover, a two-way ANOVA of concentrations over time between both groups found a positive interaction (p = 0.05), the hepatic elimination of 153Gd–BOPTA in fa/fa rats being significantly lower over time. Interestingly, from T95 to T105, we were able to detect the efflux from hepatocytes back to sinusoids (Fig. 3E). During this period, livers had already been perfused with a KHB solution for 20 min (total volume perfusion: 600 ml into rat livers < 16 g) and we can be sure that the radioactivity measured in outflow perfusate during these last 10 min corresponds to the efflux back of 153Gd–BOPTA from hepatocytes to sinusoids: 64 ± 39 nmol in fa/+ rats and 381 ± 105 nmol in fa/fa rats (p < 0.008). During the same period, the bile excretion was 337 ± 69 nmol in fa/+ rats and 490 ± 115 nmol in fa/fa rats (p = 0.06). The higher amounts of 153Gd–BOPTA in perfusate and bile of fatty livers despite a decreased transport functions and/or protein expression are explained by increased residual concentrations within hepatocytes (see Discussion) [17-19]. At T105, the residual hepatic 153Gd–BOPTA concentrations were 214 ± 60 nmol g−1 (fa/fa rats) and 75 ± 21 nmol g−1 (fa/+ rats p < 0.008). In fa/+ rats, 13% of 153Gd–BOPTA elimination was directed to the sinusoids while 44% of 153Gd–BOPTA left hepatocytes through the sinusoidal membrane in fa/fa rats.
3.4 Pharmacokinetic Modeling of 153Gd–BOPTA Transport in Perfused Rat Livers
To better understand the transport of 153Gd–BOPTA in cholestatic hepatocytes, we performed pharmacokinetic modeling of 153Gd–DTPA and 153Gd–BOPTA transport over time using the compartmental model illustrated in Fig. 1B and described in previous publications [20, 21]. The model successfully fits the experimental data for both fa/fa and fa/+ rats (Fig. 2A) with three compartments in the liver: an extracellular compartment (C2, depicted by 153Gd–DTPA) and two hepatocellular compartments (C4 and C5). 153Gd–BOPTA from C2 can enter into C4 but not into C5, 153Gd–BOPTA in C5 being provided by C4 without possible return from C5 to C4.
The concentrations of 153Gd–BOPTA in C2 were lower in fa/fa than fa/+ rats (Fig. 2A–C). Because the same concentration of Gd–BOPTA was perfused in both groups, only a larger extracellular distribution volume can explain the decreased hepatic concentrations measured by the gamma probe. In fa/+ rats, the accumulation in C4 rapidly increased until a plateau is reached by T60 (Fig. 2B). When 153Gd–BOPTA solution was switched to a KHB solution, 153Gd–BOPTA was rapidly eliminated from C4. In C5, 153Gd–BOPTA had a linear accumulation and its maximal accumulation remained lower than that observed in C4. Moreover, the concentrations in C5 remained high during the rinse period with KHB.
The evolution of 153Gd–BOPTA concentrations was different in fa/fa rats (Fig. 2C). In fatty livers, the accumulation in C4 was more gradual and, from T75, the elimination was slower than in fa/+ rats. At T105, the concentrations in C5 were high and corresponded to the increased residual hepatic concentrations.
From the model, rate constants were calculated (Table 2). The hepatic uptake of 153Gd–BOPTA through Oatps (k24) was lower in fa/fa rats without reaching statistical significance. The bile excretions from C4 (k46) and C5 (k56) were significantly lower in fa/fa rats. Exchange from C4 to C5 (k45) was unidirectional and low in both groups. Efflux from hepatocytes back to sinusoids (k42) was also significantly decreased in fatty livers.
Table 2. Hepatic pharmacokinetics by compartmental analysis
Rate constants (min−1)
The rate constant k24 is associated with 153Gd-BOPTA uptake. k46 and k56 describe bile excretion in C6 from the hepatic compartments C4 and C5. k42 is associated with the efflux of 153Gd-BOPTA from hepatocytes back to sinusoids.
1.19 ± 0.13
1.33 ± 0.36
9.60 ± 4.52
9.27 ± 1.44
0.96 ± 0.39
0.79 ± 0.17
0.11 ± 0.02
0.05 ± 0.01
0.05 ± 0.02
0.03 ± 0.01
0.065 ± 0.011
0.025 ± 0.007
0.045 ± 0.008
0.013 ± 0.006
3.5 Pharmacokinetic Modeling and Mathematical Simulations
To understand why the maximal hepatic concentration (at T75) was similar in fa/fa and fa/+ rats, we performed mathematical simulations decreasing a single parameter k24 (entry into hepatocytes), k46 (bile excretion from C4), k56 (bile excretion from C5) or k42 (efflux from C4 to sinusoids), while the three others remained unchanged (Fig. 4). With these simulations, we understand how changing a single parameter interferes with hepatic concentrations. Each parameter was modified from the value obtained in fa/+ rats to 0 (10% change at each step). The values calculated in fa/fa livers were found between these extreme values. We showed that, by decreasing k24, the maximal hepatic concentration at T75 decreased until the extracellular 153Gd–BOPTA concentration was reached for k24 = 0. When k46, k56, or k42 were decreased, the maximal hepatic accumulation at T75 increased, and the increase was high with k42 set to 0, emphasizing the importance of efflux from hepatocytes to control the intracellular concentrations of 153Gd–BOPTA.
To better understand the hepatic residual concentration, we also performed simulations at T105. The residual hepatic accumulation of 153Gd–BOPTA was moderately modified by the decrease in k46 and k42 but decreasing k56 greatly increased the residual concentration, C5 being the only hepatocellular compartment to retain 153Gd–BOPTA at the end of the protocol.
Finally, to understand the repartition of the contrast agent concentrations in C2, C4, and C5 over time in cholestatic livers, we modified each parameter while the three other parameters remain similar to that calculated in fa/+ rats (Fig. 5). When k24 was set at 0.768 (value close to that of fa/fa rats), while k46, k56, and k42 remained similar to fa/+ rats, the hepatic concentration decreased. Modifications of k46, k56 or k42 increased the hepatic concentration. Interestingly, setting k56 at 0.013 increased 153Gd–BOPTA concentrations in C5 over those observed in C4.
4.1 Relevance of Gd–BOPTA Concentrations within Hepatocytes
The transport of the organic anion Gd–BOPTA across hepatocytes relies on the expression and function of proteins located on the sinusoidal and canalicular membranes. It is relatively easy to measure a single transport parameter in in vivo or in vitro experimental studies but it is less frequent to measure simultaneously all the parameters (hepatocyte uptake, bile excretion, efflux back to perfusate) as well as the consequences of all parameters on the intracellular concentrations. The hepatic concentrations of Gd–BOPTA depend on the balance between uptake via sinusoidal transporters, biliary excretion and efflux back to sinusoids (either through bidirectional sinusoidal transporters or through sinusoidal Mrp transporters; Fig. 3).
In the experimental model of isolated perfused livers, we evaluate the extracellular space (C2) by measuring Gd–DTPA concentrations. The exact concentrations of Gd–BOPTA within hepatocytes are available (C4 and C5) and the IHUI measures the initial uptake function before the start of bile excretion. Bile excretion of Gd–BOPTA is available during the entire protocol while the efflux back from hepatocytes to sinusoids is evidenced at the end of the study (Fig. 3). During the protocol, the hepatic perfusate flow is constant in all experiments and set to 30 ml min−1, thus the decreased IHUI of fatty livers is not related to changes of hepatic perfusion. In our experimental model, liver isolation precludes interferences with extrahepatic organs. The composition of solutions is well controlled and does not include binding proteins. Because the transport of Gd–BOPTA through transporters is related to the concentration gradients of the contrast agent between sinusoids, hepatocytes and bile, we administered Gd–BOPTA at steady concentrations rather than as a bolus injection similarly to the clinical situation.
Understanding the function of transporters is important to evidence how Gd–BOPTA accumulates in hepatocytes. Cholestatic fatty livers from fa/fa rats have a decreased expression of Oatp1a4 [17, 18] and Mrp2 [17, 19]. Interestingly, Mrp3 and Mrp4 are increased in another experimental model of fatty livers . In addition to the expression of transporters, other function regulations were described. Phosphorylation of Mrp2 can retrieve the transporter from the canalicular membrane and trap Gd–BOPTA within rat hepatocytes [14, 22, 23]. Entry of Gd–BOPTA into hepatocytes is also impaired by drug–drug interaction when Gd–BOPTA is perfused together with bromosulfophthalein .
Thus, it is important to assess the function of transporters in injured hepatocytes. We show that fatty hepatocytes have a decreased uptake function as measured by the decreased uptake index (IHUI). The elimination function of Gd–BOPTA from hepatocytes is also impaired, as shown by the decrease in the IHEI, k46, and k56. Gd–BOPTA may also exit from hepatocytes by efflux back to sinusoids and this function is also impaired in fatty livers (decreased k42). However, the amounts of Gd–BOPTA that exit from hepatocytes during the period T95–T105 are higher in fatty than control livers because the residual concentrations are higher during this period. The amounts (A) of Gd–BOPTA transported through the sinusoidal membrane back to hepatic vessels are defined by the equation: A (nmol min−1 g−1) = k42 (min−1) × C (nmol g−1), where C is the concentration in hepatocytes. Thus, Gd–BOPTA amount transported back to sinusoids in fatty hepatocytes can be higher that in normal hepatocytes even though the transport function through Oatps is decreased. We previously showed, that the efflux back to sinusoids and bile excretion in normal hepatocytes are proportional to the concentrations inside livers . This point emphasizes the importance of concentrations within hepatocytes in the regulation of transport proteins. Of note, we do not distinguish the mechanisms that induce a decreased transport of Gd–BOPTA through membrane proteins in fatty hepatocytes: either decreased expression of transporters and/or dysregulation of protein function.
4.2 Hepatobiliary Contrast Agents and Liver Imaging
The clinical importance of our findings is provided by recent studies that investigated the hepatic accumulation of Gd–BOPTA and the other closely related compound Gd–EOB–DTPA (Primovist®, Bayer, Germany). In multinodular focal fatty infiltrations, Marin et al.  found a decreased signal intensities during the vascular and equilibrium phases following the injection of Gd–BOPTA, a finding similar to the decreased concentrations of the contrast agent in C2 from fa/fa livers. However, in contrast to our results, Gd–BOPTA did not enter human fatty hepatocytes and the lesions were hypointense in comparison to surrounding normal parenchyma. In rats with nonalcoholic steatohepatitis, the maximal intensity following a bolus of Gd–EOB–DTPA was similar to normal rats while the elimination was delayed . In this experimental model, the expression of Oatps and Mrp2 was similar in normal and fatty livers. However, the function of the transporters could not be determined.
4.3 Pharmacokinetic Modeling of Gd–BOPTA Transport
In previous studies, we described the pharmacokinetics of Gd–DTPA and Gd–BOPTA transport by compartmental analysis in normal rats [21, 26]. Such modeling is a valuable tool to understand the kinetic of drug concentrations in various compartments over time. The best model includes six compartments and eight rate constants. We show that this model perfectly fits the experimental data in both normal and fatty livers. The extracellular distribution volume is higher in fatty liver, and the evolution of Gd–BOPTA accumulation is different in both hepatocellular compartments C4 and C5. In C4 of normal livers, the initial accumulation rate is high until a plateau is reached. Gd–BOPTA is rapidly excreted from this compartment during the rinse period. Accumulation in C5 is more linear and late to disappear. In fatty livers, the accumulation in C4 is more gradual without reaching a plateau and the emptying is slower. Chandra et al.  also found two hepatocellular compartments in isolated rat livers perfused with 5-(and 6)-carboxy-2′,7′ dichlorofluorescein, which uses the same Oatps/Mrp2 pathway as Gd–BOPTA. The signification of two cellular compartments is puzzling but we can speculate that Gd–BOPTA accumulation in C4 is close to the canalicular membrane because its hepatic elimination is rapid.
4.4 Mathematical Simulations
Pharmacokinetic modeling with mathematical simulations has already been performed to better understand the hepatobiliary transport of drugs through the Oatps/Mrp2 pathway. Thus, in healthy volunteers, a pharmacokinetic model was developed based on blood, urine, and bile concentration-time profiles after the injection of 99mTc-mebrofenin . In this model, mathematical simulations were also performed to evaluate how decreasing the transport through OATPs, MRP2, or both OATPs and MRP2, would influence the hepatic accumulation . Interestingly, liver imaging with 99mTc-mebrofenin also represents the accumulation of tracers [29, 30].
Our simulations show that hepatic concentrations over time are the result of four biological processes: hepatocyte uptake through Oatps (k24), bile excretion through Mrp2 (two parameters: k46 and k56) and efflux back to sinusoids (k42). The most important parameters are those located on the sinusoidal membrane, a finding in line with the evidence that a compound that does not enter into hepatocytes will not be excreted in bile. Finally, to understand the repartition of the contrast agent over time in C2, C4 and C5 in cholestatic livers, we modified each parameter while the three other parameters remain similar to that calculated in normal livers rats. When k24 was set at 0.768 (value close to that of fatty livers), while k46, k56 and k42 remained similar to normal livers, the hepatic accumulation decreased. Decreasing k46, k56 or k42 increased the hepatic accumulation.
In summary, we show that cholestatic hepatocytes have similar maximal concentrations of Gd–BOPTA than normal cells despite a decreased function of Oatps and Mrp2. The study emphasizes the importance of concentrations of organic anions within hepatocytes for the transport function through membrane proteins. Although complex, such understanding is important to analyze liver imaging with hepatobiliary contrast agents in cholestatic fatty livers. Moreover, we describe an index of hepatocellular uptake function (IHUI) that may score the severity of fatty livers in clinical liver MRI associated with the injection of hepatobiliary contrast agents.
This work was supported by the Fonds National Suisse de la Recherche Scientifique to C.M. Pastor (grant 310030-126030).