Facilitated Uptake of Fentanyl, but Not Alfentanil, by Human Pulmonary Endothelial Cells

Human lung microvascular endothelial cells were obtained from Clonetics (San Diego, CA) and maintained in a supplemented endothelial growth medium (EGM-2-MV) supplied by Clonetics that contained human epidermal growth factor (hEGF), hydrocortisone, vascular endothelial growth factor (VEGF), human basic fibroblast growth factor (hFGF-B with heparin), long recombinant 3-insulin–like growth factor-1 (R3-IGF-1), ascorbic acid, heparin, gentamicin, amphotericin B, and 5% fetal bovine serum. Cells between passages 4 and 8 were seeded onto 12-well culture plates (3.8 cm²/well) 2 days before experimentation. For experimentation, cells were more than 90% confluent, and the cell density was approximately 7 × 10⁴cells/well, as evaluated by use of a Coulter counter (Model ZM; Coulter Electronics, Hialeah, FL) for representative wells from each plate.

Each experiment was performed in triplicate with use of cells from at least two different donors. At the beginning of the experiment, the cells were rinsed twice with 1 ml of Hank’s balanced salt solution that contained 0.5% bovine serum albumin (HBSS; GIBCO, Grand Island, NY) and then incubated for 10 min at 37°C, with HBSS containing varying concentrations of verapamil, ¹⁴C-labeled (Sigma Chemical, St. Louis, MO) and unlabeled (Sigma Chemical) antipyrine, and ³H-labeled (Research Diagnostics, Flanders, NJ) and unlabeled (Sigma Chemical) fentanyl or ³H-labeled (Research Diagnostics) and unlabeled (Janssen Pharmaceutica, Piscataway, NJ) alfentanil. To study the effects of verapamil on the uptake of antipyrine and the opioids, we measured uptake during control conditions (i.e. , the HBSS contained no verapamil) and in the presence of 1.0, 10, and 100 μm verapamil. ¹⁴C-Antipyrine (0.0015 mm) was used alone and with unlabeled antipyrine at concentrations ranging from 0.00106 to 10.6 mm. ³H-fentanyl (0.0071 μm) was used alone and with unlabeled fentanyl at concentrations ranging from 0.0946 to 946 μm. ³H-Alfentanil (0.005 μm) was used alone and with unlabeled alfentanil at concentrations ranging from 0.024 to 94.5 μm. Preliminary experiments showed that equilibrium between the cells and the supernatant was achieved by 10 min.

After 10 min of incubation, the supernatant was removed from the wells and the cells were rinsed twice with HBSS that contained excess unlabeled antipyrine and fentanyl or alfentanil to prevent back-diffusion of cell-associated, labeled drug. The cells were then dissolved using 1 N NaOH. The cell-associated and free ¹⁴C-antipyrine and ³H-fentanyl or ³H-alfentanil were measured using a dual-label, liquid-scintillation technique, with external standard quench correction, as previously described. 16Counts that were less than twice the background count were considered to be below the lower limit of detection; the coefficient of variation of replicate sample analysis by the assay was less than 3%. Cell-associated antipyrine and fentanyl or alfentanil data were normalized for antipyrine and fentanyl or alfentanil per cell by dividing the antipyrine and fentanyl or alfentanil concentrations by the mean number of cells/well for that day’s experiments.

To enable description of drug uptake as passive, facilitated, or both, we evaluated HMVE-L cellular drug uptake with use of a model that includes both a diffusional pathway and a saturable facilitated uptake pathway, as previously described 10and as illustrated in figure 1. Cellular uptake by passive diffusion is characterized by rapid partitioning between the supernatant fluid (C^S) and a cellular diffusion compartment (C^D) with use of a partition coefficient H:

At equilibrium, specific, facilitated drug uptake into the cellular transporter compartment (C^T) is characterized by a bimolecular reaction rate constant (k^t) for binding to the number of available transporter sites (R, which is the difference between the total transport capacity, R^max, and the drug in the cell arriving by the transport mechanism, C^T), assuming dissociation of drug from the transporter, characterized by k^o, results in immediate equilibration with the drug partitioned into the cell by diffusion alone, C^D:

The equilibration between cell-associated drug and extracellular drug can be given by where the ratio k^o/k^tis the free drug concentration that leads to 50% occupancy of the transporters. 10 

The ratio (K^EQ) of total (labeled plus unlabeled) cell-associated drug (C^D+ C^T), normalized to the number of cells/well, to supernatant drug concentration (C^S) is plotted as a function of supernatant drug concentration (C^S). Data from each experiment were fit to equation 3using a constant weight and TableCurve2D (SPSS, Chicago, IL). From the nonlinear least-squares fit, values for H, R^max, and the ratio k^o/k^twere obtained. If the nonlinear least-squares fit (equation 3) did not reach the criterion for rejection of the null hypothesis, P < 0.05, the data were pooled for the concentration range studied and represented by the average value for K^EQ.

If a drug is equilibrated with cells by simple diffusion, the relative amount of uptake (i.e. , the partition coefficient) should be constant with increasing concentrations. As shown in figure 2, the uptake of antipyrine was constant for the concentration range, suggesting uptake by simple diffusion alone. However, the relative uptake of fentanyl is much higher at the lower concentrations. At the higher concentrations, the relative fentanyl uptake decreases and approaches that of antipyrine, suggesting saturation of facilitated uptake. Similar to antipyrine, the relative uptake of alfentanil was constant for the entire concentration range (fig. 3).

As indicated in equation 3, at the highest concentrations, the value of the equilibration constant K^EQapproaches the value of the diffusional partition coefficient H. The diffusional partition coefficient in this relation can be interpreted similarly to a volume of distribution for the cells. The diffusional partition coefficient H for fentanyl based on the best fit of equation 3to the control data was 57.9 picoliters/cell, whereas the H derived from the average K^EQfor antipyrine was 66.8 picoliters/cell. The H derived from the average K^EQfor alfentanil was 64.6 picoliters/cell. The similarity of the H for fentanyl and the average K^EQfor antipyrine and alfentanil would suggest that the drugs exhibit similar nonspecific partitioning, but that there is an additional mechanism that facilitates fentanyl uptake and becomes saturated at high concentrations.

The best-fit curve for fentanyl also gave the number of transporters/cell (R^max= 1.55 × 10⁸/cell) and the concentration for 50% occupancy of transporters (k^o/k^t= 1.84 μm).

If a transporter facilitates fentanyl uptake by HMVE-L cells, it should be possible to block that uptake with verapamil, a nonspecific competitive inhibitor of drug transport. Although verapamil partially inhibited facilitated fentanyl uptake at the lowest concentration tested (1 μm), higher concentrations (10 and 100 μm) substantially shifted the uptake curves downward (fig. 4). The downward shift indicates a loss of available transporter sites because the data for the highest concentration of verapamil could not be fit with the dual pathway model. Treatment of cells with verapamil had no effect on the uptake of either antipyrine or alfentanil (data not shown).

Because verapamil significantly blocked facilitated fentanyl uptake, we investigated whether P-glycoprotein was responsible for the enhanced uptake in HMVE-L cells with use of a blocking antibody. The UIC2 antibody (Immunotech, Marseille, France) was used as a blocking antibody against P-glycoprotein to determine whether enhanced fentanyl uptake could be blocked. UIC2 is highly specific and a potent inhibitor of P-glycoprotein. In an early study, 15UIC2 reacted with all tested cell lines that expressed the human MDR1  gene product, but not with P-glycoprotein–negative cells or with cells expressing the closely related human MDR2  gene product. In addition, UIC2 strongly inhibited P-glycoprotein–mediated efflux of several cytotoxic drugs, including vinblastine, vincristine, colchicine, taxol, doxorubicin, etoposide, actinomycin D, puromycin, and gramicidin D. UIC2 had maximal inhibitory effect at 10 μg/ml for cells expressing high levels of P-glycoprotein. 15When HMVE-L cells were incubated with 0, 0.4, 4.0, and 40.0 μg UIC2 antibody/ml, there was no effect on fentanyl uptake (data not shown). These results suggest that a mechanism other than the P-glycoprotein transport is responsible for facilitating the uptake of fentanyl.

Various mechanisms have developed in animals to protect them from xenobiotics to which they are exposed. 17Some of the most widely known of these protective mechanisms are the xenobiotic-metabolizing enzymes, including the cytochrome P-450 3A enzymes, that have wide substrate specificity and are responsible for molecular oxidation in preparation for conjugation with hydrophilic moieties and subsequent excretion. Another protective mechanism is the mammalian xenobiotic transporting P-glycoprotein, located on the plasma membrane of the cell, that can transport xenobiotics against a concentration gradient, thereby reducing intestinal toxin absorption, limiting drug distribution into the brain, and facilitating drug elimination into the bile and urine. 17,18The cytochrome P-450 3A enzymes and the P-glycoproteins have considerable overlap in substrate specificity. 19Because the extensive pulmonary uptake of autacoids and xenobiotics is potentially protective, 2,20the possibility exists that P-glycoprotein is involved in facilitating their uptake. Similarly, because fentanyl and alfentanil are metabolized by cytochrome P-450 3A4, 21,22the possibility exists that they are both substrates for P-glycoprotein transport. Verapamil, a P-450 3A4 and 3A5 substrate, 23is not only a substrate for P-glycoprotein, but is also the prototypical P-glycoprotein competitive inhibitor. 24 

To test the hypothesis that fentanyl and alfentanil are transported by P-glycoprotein into pulmonary endothelial cells, we began by measuring the uptake of each drug in isolated HMVE-L cells. The equilibrium binding of antipyrine and alfentanil was consistent with a purely diffusion-driven uptake mechanism: as more drug was added to the extracellular medium, the fractional uptake (i.e., partition coefficient) remained constant (figs. 2 and 3). In the case of fentanyl, however, there was significantly higher fractional uptake at the lower supernatant concentrations than at the higher concentrations (figs. 2 and 3). As the fentanyl concentration was increased in the medium, there was a decrease in the fractional uptake that approached the uptake levels of antipyrine and alfentanil. This decrease can be interpreted as a saturation of the specific uptake-facilitating mechanism and can be explained by our model. In equation 3, as the C^Sof fentanyl becomes much greater than k^o/k^t, the second term on the right becomes smaller and contributes less to the overall equilibrium. The equilibrium partitioning is then largely controlled by the diffusional partition coefficient H. Conversely, when C^Sis small or comparable to k^o/k^t, facilitated uptake contributes significantly to the overall cellular uptake. Thus, despite the expectation that fentanyl and alfentanil might be substrates for facilitated cellular uptake because they are substrates for CYP 3A4, they are not; the overlap of CYP 3A and P-glycoprotein substrate specificities has been shown recently to be more fortuitous than indicative of a fundamental relation. 25As discussed in the next paragraph, the relative uptake of fentanyl, alfentanil, and antipyrine by isolated HMVE-L cells is consistent with the differences in pulmonary uptake observed in vivo.

Recirculatory compartmental pharmacokinetic modeling 26of the disposition of fentanyl, alfentanil, and antipyrine in human volunteers quantitated the difference in the pulmonary uptake of these drugs in terms of pulmonary distribution volume (unpublished observations). Although the pulmonary distribution volume of alfentanil (0.5 l) was similar to the extravascular lung water volume defined by antipyrine (0.3 l), that of fentanyl (8.4 l) was 28 times as big. Differences in pulmonary uptake may be a result of the fact that fentanyl is a basic amine, whereas alfentanil is not. Although fentanyl and alfentanil are structurally similar lipophilic amines (their octanol–water partition coefficients are 816 and 128, respectively 27), with similar plasma protein binding (84 and 92%, respectively 27), at physiologic pH the degree of ionization of fentanyl (pK^a= 8.4) is 91.5%, whereas that of alfentanil (pK^a= 6.5) is only 11%. 27The requirement that a molecule be a lipophilic basic amine for efficient pulmonary uptake to occur is well-known 2and is shown by contrasting the extensive uptake of the basic lipophilic amine methadone with the minimal uptake of the nonbasic lipophilic amine diazepam by the isolated perfused rat lung. 28 

With the recent discoveries of drug transporters that function to establish concentration gradients of drugs across cellular interfaces, 17including the blood–brain barrier, 29we hypothesized that such a transport mechanism may exist in the pulmonary endothelium. Because many lipophilic compounds are substrates for P-glycoprotein, 30including several opioids, 31we hypothesized that P-glycoprotein may transport fentanyl. Although its functional role has not been elucidated, there have been two previous reports of P-glycoprotein expression in the human lung, 32,33including a report of expression in bronchial capillary endothelial cells. 33Verapamil, a calcium-channel blocker, is a prototypical competitive blocker of P-glycoprotein and other drug transporters. 13,14When we incubated isolated cells with use of verapamil, we found a verapamil concentration–dependent decrease in fentanyl uptake (fig. 4). At the highest verapamil concentration (100 μm) the fentanyl HMVE-L cellular uptake curve was flat and resembled the uptake of alfentanil and antipyrine.

These results support facilitated fentanyl uptake by a cellular transporter such as P-glycoprotein. However, when we incubated cells with a blocking antibody specific for P-glycoprotein, UIC2, 15it did not affect the facilitated uptake of fentanyl (data not shown). Therefore, we hypothesize that an unidentified, facilitated uptake mechanism exists in human pulmonary endothelium that is capable of mediating significant uptake of fentanyl and other drugs. Henthorn et al.  13reached a similar conclusion in a recent study of fentanyl uptake by isolated bovine brain microvascular endothelial cells. They found that, although there was outwardly directed P-glycoprotein–mediated extrusion of fentanyl, it was small compared with the facilitated uptake by an unidentified transporter, and the transport of fentanyl by both could be blocked by verapamil. This is not unexpected because P-glycoprotein is primarily responsible for active removal of drugs from cells rather than for transport into them. 34Numerous transporters responsible for substrate import and export have been identified, and many more remain to be identified, 35one of which may be responsible for transport of fentanyl into pulmonary vascular endothelial cells. However, other mechanisms exist whereby cellular drug uptake is facilitated and may be responsible for the facilitated uptake of fentanyl.

Data from the current study were interpreted in terms of linear, passive diffusion and saturable, facilitated uptake of drug into pulmonary vascular endothelial cells. Although the model fit our data well, the data could just as easily be interpreted as representing a combination of weak, linear, cellular binding and saturable, high-affinity binding of amphiphilic, high pK^abases to subcellular components such as lysosomes 36or mitochondria. 37In the absence of studies of initial uptake rate, it is difficult to distinguish between active transport and subcellular trapping. 36However, data from our bovine pulmonary artery endothelial cell (BPAEC) column study 10showed rapid fentanyl efflux from the cells, findings that are consistent with in vivo  human data obtained with use of frequent arterial blood sampling. In that study, monolayers of the bovine pulmonary artery endothelial cells were grown on microcarrier beads and placed in a chromatography column with a nonrecirculating perfusate. The perfusate was sampled at frequent intervals after bolus doses of fentanyl and antipyrine, and a nondiffusible marker was injected at the inlet of the columns. These results suggest that the drug transport model 10is more appropriate because drug trapping creates a slowly effluxable drug pool. 36,37 

The k^o/k^tvalue, which is the concentration of fentanyl leading to 50% occupancy of transporters, for isolated cells was 1.84 μm or 619 ng/ml and is well above the clinical range of arterial fentanyl concentrations. 38This suggests that fentanyl, at normally observed clinical concentrations in humans, should be subject to facilitated pulmonary uptake.

Because verapamil is used regularly as a calcium-channel blocker, the question arises as to whether patients with therapeutic verapamil concentrations are at risk for complications when fentanyl is administered because of increased peak fentanyl concentrations resulting from decreased pulmonary fentanyl uptake. Our results suggest that verapamil would have little effect on the pulmonary uptake of fentanyl at blood concentrations less than 1 μm (454.6 ng/ml), which is more than 3 times typical therapeutic verapamil concentrations (120 ± 20 ng/ml). 39 

In a previous report, 10we demonstrated facilitated transport of fentanyl by bovine pulmonary artery endothelial cells. In the current study, we compared the uptake of fentanyl and alfentanil to that of antipyrine, a marker of extravascular lung water, 11,12by HMVE-L cells. Alfentanil uptake by HMVE-L cells was indistinguishable from that of antipyrine for the concentration ranges studied. In contrast, at low concentrations, fentanyl sequestration into HMVE-L cells was substantially greater than that of antipyrine, suggesting fentanyl uptake by HMVE-L cells is a result of diffusion and facilitated transport. A drug transporter located in the pulmonary endothelium that functions to temporarily remove xenobiotics from the circulation would be uniquely situated to serve as a control mechanism for drug disposition. Our results show for the first time that facilitated uptake of fentanyl occurs in human lung endothelial cells. Identification and characterization of the transporter may lead to better understanding of factors that influence pulmonary drug uptake and interindividual differences in early arterial blood drug concentrations.

The authors thank Jia Zhao, B.S., Zhao Wang, M.D., and Ushma Savla, Ph.D., for technical assistance.