Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

The lungs have been shown to affect the arterial concentration history of drugs given by rapid i.v. infusion as a result of either metabolism or, more commonly, extensive partitioning into pulmonary tissue (Roerig et al., 1994). The latter mechanism results in an increased pulmonary mean transit time for the drug on first-pass, decreased peak arterial drug concentrations, and slow release from lung tissue as blood concentrations fall below those in the lung. Lipophilic basic amines such as lidocaine (Post, 1979; Krejcie et al., 1997), propranolol (Howell and Lanken, 1992), fentanyl (Roerig et al., 1987; Taeger et al., 1988), and sufentanil (Boer et al., 1996) are examples of drugs that partition extensively into pulmonary tissue.

Pulmonary drug uptake has been studied in vivo by means of double or multiple indicator dilution techniques. The pulmonary uptake has usually been expressed either as the percentage of injected dose not reaching the arterial sampling site after passage of 95% of an intravascular marker (Post, 1979; Roerig et al., 1987, 1994; Taeger et al., 1988), but more recently it has been described as an apparent distribution volume derived from functions describing the right-skewed distribution of transit times across the central (pulmonary) circulation and pulmonary blood flow (Boer et al., 1996; Krejcie et al., 1996a; 1997). From such studies, it has been suggested that the pulmonary uptake of lidocaine is not first order (Jorfeldt et al., 1979), that the disposition of fentanyl in the lung is multicompartmental (Taeger et al., 1988), and that propranolol inhibits the pulmonary uptake of fentanyl (Roerig et al., 1989). These questions have been difficult to address definitively in vivo because it is not possible to characterize precisely the delayed, right-skewed, first-pass, arterial disposition curve of drugs with significant pulmonary uptake because the latter portions of the pulmonary disposition curve are obscured by recirculation of the drug (Boer et al., 1996; Krejcie et al., 1997).

We wished to examine the pulmonary uptake of fentanyl using multiple indicator dilution methodology in an in vitro system. We used a flow-through system consisting of monolayers of bovine pulmonary artery endothelial (BPAE) cells on solid microcarrier beads placed in a chromatography column to determine whether endothelial uptake alone could account for the in vivo pulmonary pharmacokinetic findings.

The pulmonary tissue partitioning of these drugs as been thought to involve only passive processes such as lipid partitioning and electrostatic binding (Roerig et al., 1994). Conversely, this partitioning may be an active process mediated by transporters in the endothelium that function to establish large concentration gradients for these lipophilic substrates. We studied the uptake of fentanyl by monolayers of pulmonary artery endothelial cells cultured in wells and tested the hypothesis that drug uptake is both passive and active by fitting the data to a model that included both a diffusional pathway and a saturable active uptake pathway.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Cell Culture Methods. The cell culture methodology has been described previously (Waters, 1996). Briefly, BPAE cells were obtained from the American Type Culture Collection (CCL-209; Rockville, MD). Cell monolayers were grown in minimal essential medium containing Earle's salts, nonessential amino acids, 10% fetal bovine serum, and gentamicin, and were subcultured using 0.1% trypsin-EDTA. Cells between passages 16 and 21 were grown on gelatin-coated plastic microcarrier beads (Sigma, St. Louis, MO). Cells were seeded at a density of 2 × 10⁴ cells/cm³ and stirred intermittently overnight to promote attachment. Microcarrier cultures were then maintained at 60 rpm continuously, fed three times a week, and used for assays between 7 and 15 days postseeding. Cells were grown to confluence, as verified by phase-contrast microscopy.

Cell-Column Chromatography Methods. The cell-column chromatography method described previously (Waters, 1996) was used to evaluate single-pass, flow-through pulmonary endothelial uptake of drugs. Cell-covered microcarrier beads were poured into a water-jacketed glass column (0.65 cm diameter; Rainin, Woburn, MA) and the column was perfused using a Gilson peristaltic pump (Gilson Medical Electronics, Villiers Le Bel, France) with flow rates ranging from 0.9 to 1.1 ml/min. In two experiments, the column was loaded with beads without cells. Three drugs with differing pharmacokinetic properties were injected into the perfusate at the top of the column. Blue dextran (10 mg/ml, mw 2 × 10⁶, Sigma) was used as a flow (or reference) tracer, because it is a large hydrophilic molecule that does not enter the endothelial cells. [¹⁴C]Antipyrine (mw 188, New England Nuclear, Cambridge, MA) is a lipophilic drug frequently used as a marker of flow-limited tissue distribution in many tissues, and tissue water space as its volume of distribution (V_d) is nearly that of true water space markers such as deuterium. [³H]Fentanyl (mw 336, Janssen Pharmaceutica, Beerse, Belgium) was used as a typical lipophilic basic amine that has been shown to have extensive first-pass pulmonary uptake. The comparison of the mean transit times (MTTs) across the column were used to assess the partitioning into the pulmonary endothelium.

Procedure for Equilibrium Kinetics. Cells between passages 16 and 21 were grown to confluence in polystyrene wells (3.8 cm²) in minimal essential medium. On the day of the experiment the media in the polystyrene wells was replaced with 1 ml of HBSS with 0.5% bovine serum albumin that contained 0 to 500 µg/ml (0-946 µmol) of unlabeled fentanyl citrate and 0.14 µmol [³H]fentanyl. Cells were incubated with these solutions for 10 min at 37°C. The supernatant was removed and the cells, after two washes with HBSS that contained excess unlabeled fentanyl to prevent back diffusion of cell-associated [³H]fentanyl during the wash, were solubilized in 1 N NaOH. The [³H]fentanyl was then counted by liquid scintillation as described previously (Bowsher et al., 1985). The number of cells in a representative well from each plate was counted using a Coulter Counter (Coulter Electronics, Hialeah, FL). The experiments were performed in triplicate.

Pharmacokinetic Analyses. Single-pass indicator dilution analysis of the outflow concentration histories as the sum of parallel -distribution functions was used to estimate the apparent pulmonary endothelial distribution volumes as described previously (Krejcie et al., 1996a). Concentration versus time data were fitted to a single -distribution and to the sum of two or three -distributions using TableCurve2D, version 3.0 (Jandel Scientific, San Rafael, CA) on a Pentium-based personal computer using constant standard deviation weighting. Model selection among one, two, or three -function models was made on the bases of visual inspection of fit quality, adjusted r², and Akaike information criteria. In general, models that describe drug concentration histories measured at an organ, or in this case, column outflow, following impulse inflow administration, are unimodal, asymmetric, and right-skewed and can be characterized by the sum of -distribution functions:

(1)

These functions describe a drug or indicator concentration history C(t)

(2)

where A is the area under the drug concentration versus time relationship [area under the concentration curve (AUC) or zeroth moment]. Higher order moments describe the characteristics of the concentration versus time profile. The first moment estimates the MTT, the second central moment estimates the variance (²), and the third central moment estimates the skewness (). Using the  function, which describes right-skewed, lagged distributions (e.g., single-pass drug concentration history), the higher order moments are easily calculated (using n and k from eq. 1) as follows:

(3)

(4)

(5)

Assuming no drug is lost, flow (Q) can be estimated from each of the outflow curves

(6)

The apparent V_d between the injection port and the elution site can be estimated from:

(7)

For blue dextran, this V_d is an estimate of the sum of the perfusate volumes in the inlet tubing, outlet tubing, and column exclusive of cells and beads. The difference between the V_d of blue dextran and antipyrine and between blue dextran and fentanyl are the endothelial cell water volume and apparent fentanyl endothelial cell volume, respectively.

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

(16)

View larger version (19K): [in this window] [in a new window]   Fig. 1.   The model of drug uptake from the supernatant (CS) into pulmonary vascular endothelial cells by diffusion (CD) and by active transport (CT). Diffusional equilibrium is characterized by a partition coefficient (H), whereas active transport is characterized by a rate constant for binding to a specific cellular binding site (kt) and for dissociation from that site (ko).

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Fentanyl Uptake in Cell Column. Fentanyl and antipyrine were injected simultaneously onto the cell columns to allow us to compare the endothelial cell uptake of fentanyl with that of a lipophilic, flow-limited tracer of similar size. Figure 2A shows the outflow concentration histories for blue dextran, antipyrine, and fentanyl from a typical cell-column chromatography experiment with the symbols representing the actual measured fractional dose collected during each collection interval and the lines representing the best fit by sums of -distribution functions. The curves for the impermeable blue dextran and the permeable antipyrine are very similar, with the MTT of antipyrine being only marginally longer than that of blue dextran. However, the outflow fentanyl concentration history had a much longer MTT and more rightward skewing than those of the other two indicators. This is reflected in the  parameters for these curves, shown in Table 1; because of the exaggerated rightward skewing of fentanyl, a sum of three -distributions is required to fit the fentanyl curve, whereas a sum of two -distributions is sufficient to fit both the antipyrine and blue dextran curves. Figure 2B shows that when solutions were collected from a column of beads without cells, there was little difference among the curves, demonstrating that the cells, and not nonspecific binding, were responsible for the increased MTTs shown in Fig. 2A.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Drug concentration histories for blue dextran (), [14C]antipyrine (), and [3H]fentanyl () of effluent collected from a column with solid microcarrier beads with (A) and without (B) monolayers of bovine pulmonary artery endothelial cells. Data are expressed as the fraction of the dose collected by the end of each collection interval. The solid, dashed, and dotted lines are the fitted sum of -distribution functions for the respective observed data.

View this table: [in this window] [in a new window]   TABLE 1 Parameters for sum of -distribution functions for data in Fig. 2

View larger version (13K): [in this window] [in a new window]   Fig. 3.   Results of a linear regression analysis of the relationship between the MTTs (dependent variable) of the drugs and the heights of the cell columns (independent variable) for different experiments all perfused with HBSS. Blue dextran (), [14C]antipyrine (), and [3H]fentanyl () are plotted as the mean MTT for each column.

Fentanyl Equilibrium and Model Evaluation. To test our hypothesis that fentanyl uptake by endothelial cells occurs by two processes, diffusion into the membrane and specific uptake by a binding site, we incubated cells for 10 min with increasing doses of fentanyl. If fentanyl uptake occurred by a simple diffusion mechanism, then the ratio of cell-associated fentanyl to free fentanyl (K_EQ) would be constant at all doses. As shown in Fig. 4, K_EQ was significantly higher at low doses of fentanyl than it was at high doses. The line in Fig. 4 is the best fit of eq. 16 to our data; this sigmoid relationship was well predicted by our model (adjusted r² = 0.789). From this fit, we determined H to be 1.36 × 10^-8 ml/cell, R_max to be 2.06 × 10^-7 nmol/cell, and k_o/k_t or C_S-50% to be 2.87 µM.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   The ratio of total (labeled plus unlabeled) cell-associated fentanyl concentrations (nanomoles per cell) to supernatant fentanyl concentrations (nanomoles per milliliter) versus supernatant fentanyl concentrations (nanomoles per milliliter, micromolar) of each well. , observed values; line, best fit of these data to eq. 16. Note the sigmoid shape of this line with the upper plateau delineating the combined effects of diffusional equilibrium (H) and active transport, the lower plateau representing H alone, and the ko/kt at the midpoint of the vertical portion.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

There are numerous in vivo reports of extensive first-pass pulmonary uptake of fentanyl following i.v. administration (Roerig et al., 1987; Taeger et al., 1988; Roerig et al., 1989, 1994; Niemann et al., 1996). Fentanyl distribution into the lung has been described as rapid uptake, followed by slow release that makes a fentanyl i.v. bolus dose resemble a slower i.v. infusion of drugs not having extensive pulmonary uptake. Drug metabolism is not a factor in pulmonary drug uptake (Roerig et al., 1994). In pharmacokinetic terms, pulmonary uptake is a partitioning of drug into lung tissue and should be expressed as an apparent V_d.

The lung is unique among tissues in the large percentage of its total cellular composition that is endothelium; almost half of the body's endothelium is in the lung (Simionescu, 1991). Therefore, we sought to determine whether an in vitro perfused system, composed entirely of pulmonary endothelial cells, would produce pharmacokinetic data (i.e., V_ds) that would demonstrate the endothelium is the site of extensive fentanyl pulmonary uptake. We quantified this uptake using time density analysis and multiple indicator dilution methodologies.

The first step in the pharmacokinetic analysis was to derive the MTT from the delayed, right-skewed drug concentration versus time profiles obtained by sampling column outflow. In our previous in vivo work (Krejcie et al., 1996a; 1997), we found that the sum of -distributions (eq. 1) does this quite well. Figure 2 illustrates typical fits of the sum of -distribution functions to marker concentration histories such as those of the present study. Equation 2 shows that when the data consists of measurements (e.g., concentrations) rather than probabilities, the coefficient (AUC) is not necessarily unity. Thus, with the injected drug dose, eq. 6 yields the flow for the system that was always within 4% of the pump calibration (1.0 ml/min). Finally, eq. 7 states that the product of flow and MTT is the apparent indicator V_d. Thus, in these experiments in which the flow was nominally 1.0 ml/min, the MTT (minutes) is the same, numerically, as the V_d (milliliters).

Blue dextran is a high molecular weight hydrophilic dye that does not cross cellular membranes. The pharmacokinetics of blue dextran was used to estimate the extracellular fluid space in the cell-column chromatography system. Because the MTT of blue dextran in a 1-cm column was 0.83 min and the column flow rate was 1.0 ml/min, by eq. 7 the extracellular volume of the system is estimated to be 0.83 ml. We can break this volume down using results from the experiments with different cell-column heights (0.9-1.8 cm) (Fig. 3). From the y-intercept of this regression of column height versus MTT, the majority of the extracellular volume (0.72 ml) is in the tubing leading into and out of the cell column. Thus the nonbead, noncell (void) volume in a 1.0-cm column is only 0.11 ml, or 33% of the 0.33-ml physical volume. We (Waters, 1996) estimated the void fraction in the cell column, based on the pressure-flow relationship and measurements of perfusate viscosity, bead diameter, and column dimensions. We consistently found the void fraction to be 0.3, which is nominally the value obtained using the current pharmacokinetic techniques.

In this cell-column chromatography system, antipyrine and fentanyl tissue distribution volumes were derived by subtracting the MTT of blue dextran from those of antipyrine and fentanyl. This leaves the MTTs of the interactions of antipyrine and fentanyl with the pulmonary endothelium and the gelatin coating of the plastic microcarrier beads. To assess antipyrine and fentanyl uptake into the microcarrier beads' gelatin matrix, two injections were made into a bead-filled column, devoid of cells (Fig. 2B). Antipyrine had a bead gelatin V_d of 0.038 ml/cm column height, whereas fentanyl had a bead gelatin V_d of 0.14 ml/cm column height. After making the correction for the respective microcarrier bead MTTs, the MTTs and apparent antipyrine and fentanyl cellular V_ds can then be determined.

Antipyrine is a lipophilic drug that has been used as a marker of both total tissue water and flow-limited tissue distribution (Krejcie et al., 1996b). Its distribution space is approximately 66% of tissue weight (Soberman et al., 1949). In a cell column of 1.0 cm height, the apparent nonvoid, nontubing antipyrine distribution volume was estimated to be 0.056 ml (Table 2), or 17% of the 0.33 ml physical volume. After making the correction for that portion of the antipyrine MTT that is due to the microcarrier bead gelatin layer, the MTT and apparent cellular V_d of antipyrine was 0.018 ml/cm column height (Table 2). Assuming a cell-covered bead surface area of 82.1 cm²/cm column height (Waters, 1996), the apparent endothelial cell thickness is 2.2 µm or, using a specific gravity of 1.0 and the 66% tissue weight correction, a physical thickness of 3.3 µm.

In the current paradigm of a cell column composed of pulmonary endothelial cells, the outflow fentanyl concentration history was similar to that reported in vivo (Roerig et al., 1987; Taeger et al., 1988; Niemann et al., 1996). The MTT of fentanyl was more than twice that of either blue dextran or antipyrine (Table 2). Unlike antipyrine, only a small percentage of the nonvoid, nontubing fentanyl MTT can be attributed to the microcarrier bead gelatin layer (0.14 min or 11% for fentanyl versus 0.038 min or 68% for antipyrine). After correcting for the MTT of the microcarrier bead gelatin layer, the MTT and apparent cellular V_d of fentanyl was more than 60 times that of antipyrine (Table 2).

Animal studies and our initial experiments indicate that pulmonary drug uptake is saturable (Anderson et al., 1974; Eling et al., 1975; Roerig et al., 1983), suggesting an active transport mechanism is responsible. Therefore, we developed a model of fentanyl uptake by pulmonary endothelium (eq. 16, Fig. 1) and designed experiments to validate it. This model includes terms for first order diffusional uptake (characterized by H) and for a saturable uptake (characterized by R_max and k_o/k_t or C_S-50%) into the cells. Because fentanyl is a small (mw 336) lipophilic compound, it should diffuse freely into cells and establish a plasma (or incubation medium) to cell partitioning that reflects the physicochemical characteristics of the respective environments and the drug. At high fentanyl concentrations, the fraction of the dose in the cells relative to the dose is essentially constant (Fig. 4), suggesting that simple drug partitioning dominates uptake at high concentrations. At low concentrations, the fractional fentanyl uptake was not constant (Fig. 4) and the increase in relative uptake with decreasing fentanyl doses suggests a specific uptake mechanism. Our model fit these data well and predicts a diffusional equlibrium constant, H, (eq. 9) of 1.36 × 10⁸ ml/cell, a total transport capacity, R_max, of 2.06 × 10⁷ nmol/cell, and the free drug concentration (C_S) which leads to 50% occupancy of the transporters, k_o/k_t or C_S-50%, of 2.87 µM. These findings suggest that the large tissue to plasma partition gradient seen in the lung at clinical fentanyl concentrations may be due to a substrate-specific transporter.

A membrane protein that transports a wide range of lipophilic drugs is p-glycoprotein (Garrigos et al., 1997; Stratmann et al., 1997). p-Glycoprotein maintains concentration gradients for freely diffusible, lipophilic drugs in various tissues (Leu and Haung, 1995; Schinkel et al., 1996), including lung (Bagrij et al., 1998). The fentanyl k_o/k_t or C_S-50% in the present study, 2.87 µM,_, is similar to the half-maximal concentrations for the p-glycoprotein interaction of verapamil, progesterone, and daunomycin, which are 1.5 µM, 25 µM, and 26.8 ± 13.4 µM, respectively (Kwon et al., 1996; Garrigos et al., 1997). Consistent with this is the observation that the R_max in the present study, 2.06 × 10⁷ nmol/cell or 1.24 × 10⁸/cell, is similar to the number of vascular endothelial albumin receptors, 4.2 × 10⁷/cell (Siflinger-Birnboim et al., 1991) Thus, although we did not attempt to identify p-glycoprotein as the binding site in the present study, our results are consistent with a p-glycoprotein-like transporter facilitation of fentanyl uptake into the pulmonary vascular endothelium.

In conclusion, single-pass pharmacokinetics using cell-column chromatography yield parameter estimates consistent with the physical dimensions of the apparatus, including the injection and collecting system and the void column fraction. Fentanyl uptake in BPAE cells was more than 60 times that of antipyrine, a flow-limited cellular volume marker in this in vitro system. Saturation kinetics performed in static wells demonstrated that pulmonary endothelial cells contained a higher fentanyl concentration at lower fentanyl supernatant concentrations than would be expected if uptake occurred by diffusion alone. These observations can be explained with a model of fentanyl uptake that includes both passive diffusion and saturable active uptake. This suggests the extensive first-pass fentanyl pulmonary uptake observed in vivo is due to drug uptake into the vascular endothelium by both a passive and a saturable active uptake process. Further work is needed to establish whether or not a drug transporter, such as p-glycoprotein, is responsible for establishing the large fentanyl concentration gradient observed both in vivo and in vitro. Further studies are also needed to determine whether fentanyl uptake by human lung endothelial cells is similar to uptake by BPAE cells.