MDR1 G1199A polymorphism alters permeability of HIV protease inhibitors across P-glycoprotein-expressing epithelial cells

Objective: To evaluate the impact of the human multidrug resistance gene (MDR1) G1199A polymorphism (amino acid change Ser400Asn) on P-glycoprotein (P-gp)-dependent transepithelial permeability and uptake kinetics of HIV protease inhibitors (PI), by using recombinant epithelial cells expressing wild-type MDR1 (MDR1_wt) or the G1199A variant (MDR1_1199A).

Methods: Using a recombinant expression system developed previously, the transepithelial permeability and uptake kinetic parameters of five PI, amprenavir, indinavir, lopinavir, ritonavir, and saquinavir were estimated across polarized epithelial cells.

Results: For all PI, the transepithelial permeability ratio (basolateral-to-apical transport divided by apical-to-basolateral transport) was significantly greater in MDR1_1199A cells than MDR1_wt cells: amprenavir (1.7-fold), indinavir (1.8-fold), lopinavir (1.5-fold), ritonavir (2.8-fold), and saquinavir (2.1-fold). However, the impact of G1199A on P-gp activity appeared to primarily influence drug permeability in the apical-to-basolateral direction. Kinetic analysis of ritonavir and saquinavir uptake by MDR1_wt- and MDR1_1199A-expressing cells showed that V_max was similar, while uptake K_m was significantly higher in cells expressing the G1199A variant suggesting that alterations in P-gp-dependent efflux mediated by G1199A were due to changes in transporter affinity.

Conclusions: Alterations in transepithelial permeability of HIV PI due to the G1199A polymorphism may impact oral bioavailability of PI and penetration into cells and tissues of the lymphoid and central nervous systems.

Introduction

Although highly active antiretroviral therapy (HAART) is effective in containing plasma HIV loads, eventual disease progression to AIDS occurs even in individuals receiving intensive antiretroviral therapy. HAART combines the use of reverse transcriptase inhibitors and protease inhibitors (PI) each targeting different mechanisms of the life cycle of HIV. It is widely believed that viral sanctuary sites, with limited access to chemotherapy, exist in the body that ultimately lead to disease progression. Potential viral sanctuary sites that have been proposed include lymph nodes and lymphoid tissues and the central nervous system (CNS) ^[1-4]. Cellular and tissue dependent mechanisms that reduce local drug concentrations may compromise effectiveness of antiretroviral therapy and promote viral drug resistance.

P-glycoprotein (P-gp), the gene product of the human multidrug resistance gene (MDR1), has been shown to mediate the transport of HIV PI in both in vitro and in vivo models ^[5-9]. The oral bioavailability of PI is poor and variable, and while solubility and contribution of cytochrome P450-mediated first-pass metabolism also limit bioavailability, the highly efficient efflux by P-gp of PI in the intestine has been proposed to play a major role in limiting their bioavailability ^[5,10,11]. Furthermore, expression of P-gp may play a role in limiting effective delivery of HIV PI into viral sanctuary sites. P-gp expression and activity have also been documented in various subtypes of lymphocytes, as well as in endothelial cells at the blood-brain barrier ^[12-14]. The P-gp-dependent efflux of PI limits intracellular drug accumulation in lymphocytes ^[15-18], and reduces brain permeability at the blood-brain brain barrier ^[19,20]. Therefore, P-gp constitutes a significant obstacle for the access of PI to HIV sanctuary sites, and ultimately reduces the effectiveness of antiretroviral therapy in these tissues.

Researchers have evaluated the effect of P-gp expression on HIV disease progression and PI pharmacokinetics. In HIV patients, data suggest that expression of functionally defective P-gp increases with disease progression ^[21]. In addition, overexpression of MDR1 in HIV positive patients was shown to correlate with decreased systemic trough concentrations nelfinavir, indinavir, amprenavir, ritonavir, and saquinavir as well as decreased accumulation in lymphocytes ^[22]. Studies have also shown that saquinavir, ritonavir, nelfinavir, indinavir, amprenavir and lopinavir induce the expression of P-gp in human lymphocytes, which could act to further decrease intracellular accumulation of PI ^[23,24].

Genetic polymorphisms in MDR1 are thought to play a role in P-gp expression and function, and may influence HIV PI levels and clinical response to antiretroviral drug therapy. In particular, the MDR1 C3435T polymorphism has been associated with alterations in P-gp; however, it is difficult to determine whether the C3435T polymorphism acts directly at the molecular level since it does not code for an amino acid modification. Therefore, the influence of MDR1 polymorphisms that lead to an amino acid modification must be evaluated at the cellular level. To accomplish this, we have developed a tool to probe the penetration and uptake of PI. A recombinant cell system expressing a non-synonymous MDR1 polymorphism, a G→A transition at nucleotide 1199 (G1199A) was developed previously by us ^[25]. G1199A results in a serine to asparagine transition at amino acid 400 (Ser400Asn) in a cytoplasmic domain of P-gp, with an allelic frequency of approximately 5.5% in Caucasians ^[26-28]. In our previous studies, we confirmed that cells expressing either wild-type MDR1 (MDR1_wt) or the G1199A polymorphism (MDR1_1199A) expressed comparable levels of P-gp on the apical membrane of epithelial cells by real-time reverse transcriptase (RT)-PCR, immunoblot, cell surface expression analysis, and deconvolution immunofluorescent microscopy ^[25].

In this report, functional significance of the MDR1 G1199A polymorphism was evaluated with the recombinant cell expression system, to assess the influence of G1199A on the transport and permeability of HIV PI. Our results suggest that altered transepithelial permeability of HIV PI due to MDR1 polymorphisms may affect absorption of PI from the intestine as well as penetration into potential HIV sanctuary sites such as lymphoid tissues and the CNS.

Methods

Cell culture

LLC-PK1 control and recombinant MDR1-expressing cells were grown in culture media and transport studies performed as described previously ^[25].

Transepithelial permeability assay

Transepithelial permeability of PI was evaluated by the protocol established previously ^[25] with radiolabeled ³H-amprenavir, ³H-lopinavir, ³H-ritonavir, and ³H-saquinavir (Moravek Biochemical, Brea, California, USA) at 5 μM (1 μCi/well) in Opti-MEM medium (Invitrogen, Carlsbad, Ccalifornia, USA) in quadruplicate. Radioactivity was measured with a 1600 TR liquid scintillation analyzer (Canberra Packard, Zellik, Belgium). Indinavir was performed at 5 μM and aliquots diluted 1: 1 in methanol for liquid chromatography-coupled mass spectroscopy (LC/MS) using a protocol established previously ^[1]. Inhibition of transport was performed at 1 μM GF120918 [N-(4-[2-(1,2,3,4-tetrahydro-6,7-dimethoxy-2-isoquinolinyl)ethyl]-phenyl)-9,10-dihydro-5-methoxy-9-oxo-4-acridine carboxamide] kindly provided by GlaxoSmithKline (Research Triangle Park, North Carolina, USA). All PI experiments were duplicated. Percent transport was calculated as amount of drug in the recipient compartment divided by initial amount in the donor compartment. Apparent permeability (P_app) was calculated as P_app = [1/(A × C₀)] × [dQ/dt], where A is the surface area of permeable support, C₀ is the initial concentration in the donor compartment, and dQ/dt is the initial rate of transfer of compound into the acceptor compartment, and was estimated in the apical-to-basolateral direction (P_{app A→B}) and in the basolateral-to-apical direction (P_{app B→A}) ^[29]. The ratio of P_appB→A/P_appA→B was also estimated to evaluate net directional flux.

Uptake kinetics of PI

To evaluate influence of G1199A on P-gp efflux activity, the recombinant cells were exposed to 5 μM (0.5 μCi/well) ³H-ritonavir or ³H-saquinavir to optimize time-course of intracellular uptake in 0.25 × 10⁶ cells seeded overnight in 12-well tissue culture plates (Corning, Corning, New York, USA). The uptake time-course was linear up to 60 min. Therefore, concentration-dependent drug accumulation was performed at 45 min over the concentration range. Cells were solubilized with 0.1 N NaOH and radioactivity was measured. The V_max and K_m values of drug uptake were analyzed based on the Michaelis-Menten model. Kinetic analyses were repeated three times and WinCurveFit software (Kevin Raner Software, Mt. Waverley, Australia) was used to model the data.

Statistical analysis

Student's two-sided t test was used to evaluate differences between two sets of data. P values < 0.05 were considered statistically significant.

Results

Directional transepithelial permeability

The rate of transfer of protease inhibitors across an epithelial cell barrier was measured with alone (apical-to-basolateral, Fig. 1a-c; basolateral-to-apical, Fig. 2a-c) and in the presence of a P-gp-specific inhibitor GF120918 (apical-to-basolateral, Fig. 1d-f; basolateral-to-apical, Fig. 2d-f). Drug was placed in either the apical or basolateral compartment and percent transport was estimated at each time point.

Fig. 1

Fig. 2

Apical-to-basolateral flux, in the absence of GF120918, was greatly reduced by the expression of either MDR1_wt or MDR1_1199A compared to control cells, indicating expression of functional P-gp (Fig. 1a-c). In control cells, movement of drug across a cell membrane is assumed to be mediated primarily by non-P-gp mechanisms, either diffusion or contribution of other transporters; notably, the PI under study did not cross the epithelial membrane all at the same rate, indicating rates of transfer is likely a function of their physiochemical properties. The rank order of transfer of PI corresponded between cell types, with amprenavir displaying the highest apical-to-basolateral flux and saquinavir the lowest.

Next, basolateral-to-apical flux, in the absence of GF120918, was evaluated in the three cell types (Fig. 2a-c). Expression of either MDR1_wt or MDR1_1199A enhanced basolateral-to-apical flux compared to control cells, confirming expression of P-gp. Again, flux in the control cells was not uniform across the PI, but rank order corresponded between the cells with lopinavir displaying the highest basolateral-to-apical flux and indinavir the lowest.

Differences in the activity of P-gp between cells expressing MDR1_wt or MDR1_1199A were observed for all PI. The transepithelial data were evaluated in each direction and the data are presented in Table 1. A significantly lower apical-to-basolateral flux was observed in MDR1_1199A cells for all PI compared to MDR1_wt cells [amprenavir: 1.7-fold (P < 0.0001); indinavir: 1.9-fold (P < 0.001); lopinavir: 1.6-fold (P < 0.0001); ritonavir: 2.3-fold (P < 0.0001); and saquinavir 2.3-fold (P < 0.0001)]. These data indicate that P-gp expressed in MDR1_1199A cells was more efficient at limiting absorptive transport across the epithelial barrier than was P-gp expressed in MDR1_wt cells. However, no differences were observed between MDR1_wt and MDR1_1199A cells in the basolateral-to-apical direction, with the exception of a modest increase in efflux of ritonavir (1.2-fold; P < 0.05) in MDR1_1199A cells. Therefore, although a significant amount of drug was secreted across the epithelial cells in the basolateral-to-apical direction in both MDR1 cells, there appears to be little G1199A genotypic differences. Such a rapid basolateral-to-apical transfer was observed with lopinavir, the rate of accumulation in the apical compartment appeared to diminish over time in both MDR1-expressing cells (Fig. 2a-c); therefore, transfer kinetics may not be under sink conditions at later time points. These data suggest that although both MDR1_wt- and MDR1_1199A-expressing cells display efficient efflux of PI, the influence of the G1199A polymorphism may have the largest impact on efflux transport in the absorptive or apical-to-basolateral direction; i.e., the impact of P-gp on transepithelial permeability is asymmetric.

Table 1

The P-gp-specific inhibitor GF120918 was used to confirm that efflux of PI was mediated by P-gp in the apical-to-basolateral (Fig. 1d-f) and basolateral-to-apical (Fig. 2d-f) directions. GF120918 had little effect on the efflux of PI across control cells in either direction. Inhibition with GF120918 effectively blocked P-gp activity in both MDR1_wt and MDR1_1199A cells, and brought flux rates in both directions near to the values observed in control cells. These data confirm that the efflux of PI in these cells is a P-gp-specific effect, and that the potency of GF120918 inhibition was unaffected by G1199A polymorphism.

Collectively, these data indicate that P-gp is the primary contributor of PI efflux across the recombinant MDR1 cell system. Significant differences in P-gp efflux between MDR1_wt and MDR1_1199A cells were observed for all PI in the apical-to-basolateral direction, but negligible genotypic differences existed in efflux in the basolateral-to-apical direction. G1199A expression had the largest impact on the apical-to-basolateral efflux of ritonavir and saquinavir, followed in order by indinavir, amprenavir, and lopinavir. This highly significant decrease in apical-to-basolateral permeability in MDR1_1199A cells was abolished in the presence of GF120918, indicating that the observed G1199A effect was entirely attributable to altered P-gp function.

Apparent transepithelial permeability

After evaluating P-gp-mediated efflux of PI in the individual directions, the next step was to estimate permeability ratios (P_appB→A/P_appA→B) that represent the net active steady-state flux of drug across an epithelial membrane (Table 2). If cells display directional efflux mediated by P-gp, a permeability ratio greater than 1.0 is expected, and in the presence of the P-gp-specific inhibitor GF120918, the ratio should be reduced to about unity. In the control cells, the permeability ratios were close to unity for amprenavir and indinavir, indicating no directional flux; however, the permeability ratios for lopinavir, ritonavir, and saquinavir were greater than unity, but the ratios were unaffected by GF120918 indicating that the net flux of these compounds was mediated by transporters other than P-gp. Cells expressing MDR1_1199A had significantly higher permeability ratios of all PI compared to MDR1_wt [amprenavir: 1.7-fold (P < 0.0005); indinavir: 1.8-fold (P < 0.05); lopinavir: 1.5-fold (P < 0.0005); ritonavir: 2.8-fold (P < 0.0005); and saquinavir: 2.1-fold (P < 0.0005)]. Permeability ratios of saquinavir in both MDR1 cell types were substantially higher than the ratios for other PI, indicating that saquinavir is very efficiently effluxed by P-gp. In the presence of GF120918, permeability ratios were dramatically reduced in both MDR1_wt and MDR1_1199A cells, confirming that efflux was mediated by P-gp. Although the permeability ratios were not reduced to that of the control cells in the presence of GF120918, the numbers were similar between MDR1_wt and MDR1_1199A. These data suggest that G1199A enhances the transepithelial permeability ratio of PI by promoting P-gp-mediated efflux.

Table 2

Uptake kinetics of HIV PI

Traditionally, P-gp transport kinetic studies were performed with plasma membrane vesicles to enrich P-gp activity. Our recombinant MDR1 cells consistently express high levels of P-gp, which allowed for the estimation of uptake kinetic parameters for ritonavir and saquinavir. The preceding analyses suggest that the effect of G1199A may be most significant for ritonavir and saquinavir. Uptake kinetic parameters were estimated from intracellular drug accumulation rate versus medium concentration plots; namely, the Michaelis-Menten constants V_max and K_m (Table 3). V_max, the maximal rate of uptake, is expected to diminish as P-gp efflux transport increases, and K_m, the drug concentration at half-maximal uptake transport, is expected to increase as the affinity of P-gp for the drug increases. For ritonavir, V_max values were similar between MDR1_wt- and MDR1_1199A-expressing cells; however, the K_m value was more than twofold greater in MDR1_1199A-expressing cells than MDR1_wt cells. In the case of saquinavir, V_max was slightly decreased in MDR1₁₁₉₉-expressing cells, while K_m was also greater in MDR1_1199A-expressing cells. Collectively, these data suggest that G1199A impacts apparent K_m of P-gp with minimum impact on V_max for both ritonavir and saquinavir.

Table 3

Discussion

The ability of HIV PI to cross a cellular barrier is important not only for effective absorption from the intestine but also for their penetration into potential HIV sanctuary sites, such as lymphoid tissues and the CNS. We have previously shown that the recombinant cell system expressing either MDR1_wt or MDR1_1199A is a highly reliable and consistent method with which to evaluate functional changes in P-gp due to the G1199A polymorphism ^[25]. Using this system, expression of G1199A was shown to result in variable cellular sensitivity and permeability of P-gp substrates ^[25]. This system has now been extended to study the phenotypic penetrance of the G1199A polymorphism in transepithelial permeability of HIV PI.

Evaluating transport of compounds across an epithelial cell barrier provides the ability to predict how drugs will move across physiologic cell barriers in the body, including the intestine, kidney, and blood-brain barrier. For a physiologic comparison, the apical compartment represents the intestine lumen, the lumen of kidney proximal tubules, and the capillary lumen at the blood-brain barrier. In contrast, the basolateral compartment represents the blood relative to the intestine and kidney and the brain interstitium relative to the blood-brain barrier.

Our apical-to-basolateral transport data indicate that MDR1 expression and, therefore, P-gp efflux dramatically reduces absorption of PI from the apical to the basolateral compartment. This may reflect the role of P-gp in limiting absorption of PI from the intestine after oral administration, and limiting brain penetration at the blood-brain barrier. The basolateral-to-apical data, which represent the contribution of P-gp to systemic clearance when drug in arterial blood is presented to the basolateral cell membrane (i.e., after absorption of orally administered drug or after intravenous administration), suggest that MDR1 expression enhances efflux of PI for exsorption and elimination.

The rank order of PI transfer in the apical-to-basolateral direction did not correspond to the rank order for the basolateral-to-apical direction; however, the two rank orders were similar between cell types. These data suggest that differences in membrane structure or contribution from non-P-gp transporters between the apical and basolateral membranes of the polarized epithelial cells become evident as the physiochemical properties of the drug permeants are varied. The observed rank orders in the recombinant cell system are similar to previously reported studies with MDR1_wt. Apparent permeability ratios estimated in MDR1-MDCKII cells were the largest for saquinavir followed by ritonavir and amprenavir ^[29]. Transport studies in MDR1-MDCKII and Caco-2 cells found that saquinavir had lower apical-to-basolateral flux than ritonavir, corresponding to larger apparent permeability of saquinavir compared to ritonavir ^[30]. Another report in Caco-2 cells observed that apical-to-basolateral permeability was less for saquinavir than indinavir and that the apparent permeability ratio was greater for saquinavir than indinavir ^[31]. Therefore, the LLC-PK1 recombinant expression cell system appears to display transport characteristics of HIV PI similar to other in vitro cell systems.

Differences in P-gp-mediated efflux of PI were observed between MDR1_wt and MDR1_1199A cells. Permeability ratios were significantly higher in MDR1_1199A cells compared to MDR1_wt cells for all PI, demonstrating that MDR1_1199A-expressing cells display a more efficient net flux of amprenavir, indinavir, lopinavir, ritonavir, and saquinavir across an epithelial membrane. However, this net flow analysis derived from permeability ratios does not reveal the exact impact of the G1199A polymorphism on the modulating effects of P-gp upon barrier permeability. Characterization of the influence of the G1199A polymorphism on P-gp activity requires independent evaluation of both absorptive (apical-to-basolateral) and secretory (basolateral-to-apical) processes. Our detailed analysis of the unidirectional flux data clearly showed that the observed changes in permeability ratios were due primarily to changes in apical-to-basolateral efflux between MDR1_wt and MDR1_1199A cells. All drugs displayed decreased apical-to-basolateral transepithelial permeability in MDR1_1199A cells compared to MDR1_wt cells, indicating that the G1199A polymorphism led to greater P-gp-mediated efflux, with the greatest effect observed in efflux of ritonavir and saquinavir, followed by indinavir, amprenavir, and lopinavir. The basolateral-to-apical transepithelial permeability of PI was similar between MDR1_wt and MDR1_1199A cells, with the exception of a modest increase in basolateral-to-apical flux of ritonavir in MDR1_1199A cells. Therefore, the effect of the G1199A polymorphism appears to be most significant when drug is delivered first to the apical membrane of epithelial cells (i.e., the first step in intestinal absorption and penetration into the blood-brain barrier) than when it is presented to the basolateral membrane (i.e., requisite step in secretion and elimination from blood by the kidney and intestine).

The P-gp-mediated efflux of PI in MDR1_wt and MDR1_1199A cells was almost completely inhibited in the presence of GF120918, that is, to the level of the control cells. These data indicate that the asymmetric effect of G1199A expression on transepithelial permeability of PI fully reflects the alteration in P-gp function. However, there was evidence in the control cells, that other transporters may play a role in the permeability of lopinavir, ritonavir, and saquinavir, but not of amprenavir and indinavir.

Since our recombinant cell system expresses high levels of wild-type and variant P-gp, it allowed for a more detailed analysis of the uptake kinetics between MDR1_wt- and MDR1_1199A-expressing cells. Historically, kinetics analysis of P-gp activity has been performed using membrane vesicles, but the high level of expression of P-gp in these recombinant cells allows for evaluation of kinetics in intact cells. The preliminary kinetic analysis of ritonavir and saquinavir indicates that the changes in P-gp-dependent efflux due to G1199A are mediated by an apparent change in affinity, observed in K_m values, for the substrates.

The mechanisms by which G1199A alters P-gp activity are not yet known. The G1199A polymorphism causes a Ser400Asn, which lies adjacent to the first ATP-binding domain of P-gp and to transmembrane domain 6, one of the transmembrane domains important in substrate binding ^[32,33]. However, because a crystal structure for P-gp is not yet available, the exact orientation of Ser400Asn relative to the substrate-binding regions or ATP-binding domain of P-gp is not known. The G1199A polymorphism may affect ATPase activity of P-gp, altering ATP hydrolysis necessary for substrate efflux. Because the impact of G1199A was not equal across the PI evaluated, expression of G1199A may also affect P-gp drug binding, affinity, or transport. Multiple binding sites of P-gp have been proposed, so it is probable that amino acid modification due to MDR1 polymorphisms may differentially modulate efflux of P-gp substrates (reviewed in ^[34]).

The recombinant expression system we have developed will be useful in clarifying the contradictory results in estimating clinical impact of MDR1 polymorphisms on antiretroviral therapy. MDR1 C3435T polymorphism, which does not code for an amino acid modification, has been suggested to influence plasma HIV PI levels and virological and immunological responses to antiretroviral drug therapy, however the data remains inconclusive ^[35-41]. As we have successfully established and characterized a recombinant cell system expressing G1199A, future studies are under development to evaluate the influence of C3435T, alone and in linkage with C1236T and G2677T, on PI uptake and transepithelial permeability, which may help us to understand the significance of C3435T at the cellular level.

In summary, the transepithelial efflux and permeability of amprenavir, indinavir, lopinavir, ritonavir, and saquinavir in the stable recombinant cell system was evaluated and showed that these drugs are excellent substrates of P-gp. Expression of MDR1_wt or MDR1_1199A in LLC-PK1 cells enhances the P-gp-mediated efflux transport of PI across polarized epithelial cells, which can be reversed with the specific P-gp inhibitor GF120918. A significant impact of the G1199A polymorphism on efflux and permeability was observed. Most importantly, the magnitude of a G1199A effect is drug specific and has a preferential effect on drug permeability in the apical-to-basolateral direction. Therefore, the G1199A polymorphism may impact intestinal absorption of PI and penetration into cells and tissues of the lymphoid and central nervous systems. Further phenotype-genotype association studies are needed to confirm these predictions.

Acknowledgement

The authors thank Loren Kinman for help with indinavir analysis.

Sponsorship: Supported in part by NIH grants GM62883, AI52663, NS39178, AI31854, ES07033 and HL56548. E.L.W. is a recipient of the NIH Pharmaceutical Sciences Training Grant (GM07750), and the William E. Bradley Fellowship in Pharmaceutics.