Quinidine was sourced from ThermoFisher Scientific (Waltham, MA), while methadone hydrochloride came from the U. S. Pharmacopeia (Rockville, MD). Rhodamine B, desipramine hydrochloride, and ADP sodium were procured from Sigma-Aldrich (St. Louis, MO). Cholesterol, tris-HCl, and disodium ATP were obtained from Amresco (Solon, OH). Dithiothreitol (DTT) was purchased from Gold Biotechnology (Olivette, MO), and ethylene glycol tetraacetic acid (EGTA) and imidazole were sourced from Alfa Aesar (Tewksbury, MA). Escherichia (E.) coli total extract lipid and E. coli polar extract lipid were acquired from Avanti Polar Lipids Inc. (Alabaster, AL). The detergent n-dodecyl-β-D-maltoside (DDM) was procured from MilliporeSigma (Formerly, EMD Millipore Corporation) (Burlington, MA), and 4-(2-hydroxyethyl)−1−piperazineethanesulfonic acid (HEPES) and acrylamide from MilliporeSigma (Formerly, Calbiochem) (Burlington, MA). Sodium orthovanadate (Na₃VO₄) was sourced from Enzo Life Sciences (Farmingdale, NY). All other chemicals were acquired from ThermoFisher Scientific (Waltham, MA).

The his-tagged wild-type mouse Pgp (Abcb1a, MDR3) was purified from Pichia (P.) pastoris using affinity chromatography with nickel-nitrilotriacetic acid (Ni-NTA) (Thermo Fisher Scientific) followed by diethylaminoethyl cellulose (DEAE) resin (Thermo Fisher Scientific) as previously described (Lerner-Marmarosh et al., 1999; Bai et al., 2011). Briefly, Pgp solubilized by DDM was reconstituted into 200 nm liposomes using a previously established procedure so that the orientation of the transporter was inside-out with nucleotide-binding domains projected outside of the liposome (Ledwitch et al., 2016; Nguyen et al., 2020). Liposomes were prepared using 80% wt/vol E. coli Polar Lipid Extract (Avanti Polar Lipids) and 20% wt/vol cholesterol. A thin lipid film was created by evaporating the mixture of lipid extract and cholesterol dissolved in chloroform using a rotorvap system for 1 h. Rehydration of the dried lipid film in an aqueous phase (0.1 mM EGTA and 50 mM Tris-HCl) was followed by ten cycles of freeze-thaw in liquid nitrogen, generating liposomes of various sizes. The unilamellar liposomes with an average size of 200 nm were created by extruding the multilamellar liposome eleven times through a 400 nm filter with a LIPEX extruder. Bigger liposome size might lower SPR signal and decrease the signal sensitivity because analytes used are relatively small compared to ligand. Detergent-solubilized Pgp was dialyzed in HEPES buffer (20 mM HEPES, 100 mM NaCl, 5 mM MgCl₂, 2 mM DTT, pH 7.4) for 2 h to remove DDM and glycerol. Dialyzed-protein was incubated with the extruded liposome with a lipid-to-protein ratio of 6.25 μM Pgp (mg ml⁻¹ lipid)⁻¹ for an hour and dialyzed for another 2 h to make proteoliposomes. The concentration of protein reconstituted in the liposomes was determined using a DC Protein Assay Kit II (Bio-Rad, Hercules, CA).

The ability of Pgp-embedded liposomes to transport pharmaceutical agents is conducted using a dialysis tube in Chifflet buffer (150 mM NH₄Cl, 5 mM MgSO₄, 0.02% wt/vol NaN₃, 50 mM Tris-HCl, pH 7.4) (Cui et al., 2022). Briefly, 1 mg/mL of liposome (control) or proteoliposome suspension in Chifflet buffer was transferred into narrow dialysis tubing (3.5 kDa MWCO), which was then placed into 50 mL centrifuge tubes containing 20 mL of Chifflet buffer added with 2 mg of rhodamine B and with or without 5 mM of ATP. Then, the tubes were incubated at 37 °C for up to 2 h. At time points 0, 30 min, 60 min, and 120 min, 200 μL of the buffer outside the dialysis tubing was collected, and 200 μL of fresh buffer was added to the tubes to maintain a total of 20 mL release media. The collected samples had their absorbances tested at 550 nm using UV-vis spectroscopy. The study was carried out in triplicate, and samples were evaluated through a standard curve created successive 2-fold dilutions of free rhodamine B (R² = 0.9987) to calculate the ratio of absorbance to rhodamine B concentration.

After calculating rhodamine B concentrations using absorbance through the standard curve, the transport efficiency was calculated using the following equation. Briefly, concentration (I) is the initial concentration of rhodamine B in the buffer, while concentration (S) is the concentration of rhodamine B in the buffer at each time point. Reduced values indicate the drug being transported into the proteoliposome. T r a n s p o r t   e f f i c i e n c y   % = R h o B   c o n c e n t r a t i o n   I − S R h o B   c o n c e n t r a t i o n   I × 100 (1)

ATPase assays were carried out to measure the ATP hydrolysis of Pgp in the presence of three different substrates, including desipramine, methadone, and quinidine. The Chifflet method, which detects the formation of inorganic phosphate following ATP hydrolysis, was used as described previously (Ledwitch et al., 2016; Gibbs et al., 2018; Nguyen et al., 2020). The ATPase activity of the three drugs was measured in the presence of 50 nM Pgp reconstituted in liposomes in Chifflet buffer.

The ATPase activity of desipramine, methadone, and quinidine were fit by nonlinear regression using Igor Pro 6.2 software (Wavemetrics, Tigard, OK) as described previously (Ledwitch et al., 2016; Wilt et al., 2017; Nguyen et al., 2020). The ATP hydrolysis kinetics of desipramine and methadone fit with the monophasic equation derived from the Michaelis-Menten equation, where v is the ATP hydrolysis rate, V _max is the maximum ATP hydrolysis rate at saturating drug, [L] is the ligand concentration, K _m is the Michaelis-Menten constant, and v _basal is the basal ATPase activity in the absence of the drug.

v = v max L K m + L + v b a s a l (2)

Unlike the ATP hydrolysis kinetics of methadone and desipramine, quinidine established a biphasic ATPase activity profile. Substrate inhibition was used for biphasic kinetics, where K I is the inhibitory constant. v = v max 1 + K m L + L K I + v b a s a l (3)

Surface plasmon resonance is typically used to measure the real-time kinetics of the interaction between two molecules. The classic SPR monitors kinetic events by measuring the change in refractive angle of the incident light upon changes on the sensor surface (Myszka, 1997). The sensor in the classic SPR system is normally made of gold, which exhibits a broad plasmonic field. The SPR system we used in this study is a localized SPR (LSPR) from Nicoya Lifescience (Canada) instead of a classic SPR. Localized SPR monitors the change in absorbance wavelength at the local point where binding events occur instead of the change in the refractive angle (Hutter and Fendler, 2004; Willets and Van Duyne, 2007). The sensor of the LSPR system is made of nanogold, which has a smaller plasmonic field than a gold film and is therefore sensitive to localized changes instead of bulk changes. LSPR is known to have similar sensitivity for biomolecular binding events to SPR but is less susceptible to drift due to bulk changes (Willets and Van Duyne, 2007).

The SPR experiment was carried out with the three drugs, desipramine, methadone, and quinidine, in the presence of 5 mM ATP, 5 mM ADP, or no nucleotide as analytes shown in Figure 1. Analyte concentrations used were 500, 250, 125, 62.5, and 31.25  μ M . Pgp reconstituted in liposomes was used as the ligand and were immobilized onto the liposome or NTA sensor at a saturated level, which typically takes two injections of the ligand at 20 μM of the protein. The Liposome sensor has a hydrophobic chain to capture the bilayer portion of the proteoliposome, while the NTA sensor captures the His tag on Pgp. The running buffer contained 100 mM Tris, 150 mM NaCl, 5 mM MgCl₂, 0.5% DMSO, and pH 7.4. The running buffer contained the same amount of nucleotide (ATP, ADP, or no nucleotide) as in the analyte solution. The association phase was set to 1,200 s and the dissociation phase was set to 800 s. All the experiments were performed at a speed of 5  μ l / ⁡ min . Regeneration of the liposome sensor was carried out using 20 mM CHAPs, while the regeneration of the NTA sensor was carried out using 300 mM EDTA and 250 mM imidazole. Double referencing was carried out and subtracted from the data before further analysis.

COPASI is the software used previously by other studies to fit complex kinetics data. SPR data with multiple reactions and compartments were also successfully modeled using COPASI (Collier, 2012; Kim et al., 2012). Therefore, we used this software to fit the transport data obtained from the SPR experiment. The data were fit with two models, the vacuum cleaner and the flippase model. Each model consists of five different compartments, including bulk solvent, the sensor’s surface, the liposome’s outer leaflet, the liposome’s inner leaflet, and the liposome’s inner space. There are six reactions used in each model to represent the whole transport process. Specifically, in the vacuum cleaner model, the following reaction was used: (Brum, 2020): analyte diffused from the bulk to the surface of the sensor, (Ambudkar et al., 1998), analyte diffused from outside to outer leaflet of the proteoliposome, (Gameiro et al., 2017), analyte bound to Pgp, (Glavinas et al., 2008), Pgp transported analyte into the inner space of the proteoliposome, (Sharom et al., 1999), analyte diffused from the outer leaflet to the inner leaflet of the proteoliposome, (Toyoda et al., 2019), analyte diffused from the inner leaflet to the inner space of the proteoliposome. The flippase model used similar reactions except for reaction (Glavinas et al., 2008), in which Pgp transported the analyte to the inner leaflet instead of the inner space of the proteoliposome.

The global quantities were defined as the SPR output signal, represented by the following equation: R U = k SIN t + S L O t + S P G P t + S L I t + C (4)

Where R U is the relative unit in SPR sensorgram, k is the overall scaling, C is the baseline constant, SIN t , S L O t , S P G P t , and S L I t represent the time-dependent evolution of the analytes incorporated into the inner space of the proteoliposome, the outer leaflet of the proteoliposome, the Pgp, and the inner leaflet of the proteoliposome, accordingly. A time course kinetics simulation of the SPR signal was carried out with intervals of 50,000 and an interval size of 0.06 s for a duration of 3,000 s using the deterministic method. Following the simulation, parameter estimation was used to fit the experimental SPR data. All available convergence methods were implemented, but the best result was obtained with an evolutionary programming algorithm. Data with different concentrations of the analytes were fit globally with the kinetics for each part of Equation 4, varied freely. The scaling factor k and baseline constant C were fitted locally for each concentration. As indicated in the previous AFM study, the amount of Pgp reconstituted into the liposome has 80% inside-out orientation (Sigdel et al., 2018). Only Pgp with the nucleotide binding domains projecting out of the proteoliposome has access to ATP and can efflux substrates. Therefore, a restraint was applied to the number of Pgp involved in active transport during simulation.

Utilizing rhodamine B, a well-established Pgp substrate (Sarver et al., 2002; Loetchutinat et al., 2003), the vesicle transport assay was conducted. The transport efficiency of rhodamine B was depicted in Figure 2, illustrating rates at 10, 30, 60, and 120 min into the assay. Rhodamine B is recognized for its Pgp-mediated membrane transport facilitated by ATP (Loetchutinat et al., 2003). Our findings demonstrated a substantial increase in rhodamine B accumulation within proteoliposome in the presence of ATP compared to its absence. Notably, within 10 min, rhodamine B uptake without ATP was around 5%, whereas with 5 mM ATP, it escalated to 10%, with statistical significance. Over 120 min, ATP-assisted transport raised vesicular rhodamine B content to 17%, while this remained at 12% without ATP. Rhodamine uptakes in proteoliposome in the absence of ATP is due to passive diffusion. These outcomes highlight successful Pgp-mediated transport of rhodamine B within proteoliposome with ATP support, validating the functionality of the employed inside-out vesicle system. ATPase of methadone, quinidine, and desipramine confirms the activity of the transporter.

To further assess proteoliposome activity, an ATPase assay was conducted using three additional Pgp substrates: methadone, quinidine, and desipramine. Figure 3 depicts the ATP hydrolysis kinetics curve for each substrate across varying concentrations. The ATPase data reveals that, in the presence of these drugs, Pgp’s basal ATP hydrolysis rate measured 482 ± 50 nmolmin⁻¹mg⁻¹, consistent with prior investigations (Gibbs et al., 2018; Nguyen et al., 2020). The presence of sodium orthovanadate, a non-competitive ATP inhibitor, resulted in a decline of Pgp’s ATPase activity to approximately 100 nmolmin⁻¹mg⁻¹, corroborating findings from previous studies (Ledwitch et al., 2016; Wilt et al., 2017).

Methadone incited monophasic ATP hydrolysis activation in Pgp, revealing kinetics with a peak of approximately 800 nmolmin⁻¹mg⁻¹ at saturating methadone levels. Curve fitting yielded v max and K m values of 849 ± 45 nmolmin⁻¹mg⁻¹ and 28 ± 10 μM, respectively, consistent with a prior methadone-Pgp study (Gibbs et al., 2018). Similarly, desipramine mirrored methadone’s behavior, eliciting monophasic ATPase activity and implying a solitary binding site on the transporter for both compounds. In the presence of desipramine, Pgp’s ATPase kinetics revealed v max and K m values of 904 ± 65 nmolmin⁻¹mg⁻¹ and 12 ± 4 μM, respectively. Notably, this study uniquely unveils the ATP hydrolysis kinetics of Pgp influenced by desipramine, as no prior research reported v max and K m values for this interaction.

In contrast to methadone and desipramine, quinidine induced Pgp’s ATP hydrolysis in a biphasic manner, implying the presence of at least two quinidine binding sites on Pgp, as observed with other substrates like verapamil, lidocaine, and loperamide. The kinetics demonstrated a peak at approximately 900 nmolmin⁻¹mg⁻¹ at a quinidine concentration of 15.6 μM. Fitting the biphasic kinetics curve to Equation 3 yielded v max , K m , and K I values of 626 ± 44 nmolmin⁻¹mg⁻¹, 2.9 ± 1 μM, and 131 ± 34 μM, respectively. Notably, the K m and v max values obtained in this study were consistent with those reported in a prior investigation involving quinidine and Pgp (Matsunaga et al., 2006).

The SPR signal amplitude corresponds to alterations in mass on the sensor surface (Myszka, 1997). Consequently, the signal amplitude can serve to monitor changes in mass within the proteoliposome. Should the analyte occupy the proteoliposome, this would elevate weight, translating into an increased signal amplitude. Leveraging this SPR attribute, we can gauge the relative transport rate of the Pgp substrate by observing fluctuations in signal amplitude. Active Pgp-mediated transport, facilitated by ATP, is expected to yield greater substrate accumulation within the proteoliposome than in the absence of ATP. This heightened substrate accumulation within the proteoliposome would generate a more pronounced SPR signal.

Experiments involving three distinct Pgp substrates consistently demonstrated that the SPR signal amplitude in the presence of 5 mM ATP significantly surpassed the amplitude in the absence of nucleotides or with ADP (Figure 4). The ratio between SPR amplitude in the presence of ATP and that without nucleotides is an indicator of each substrate’s relative transport rate. Table 1 presents the ratios for quinidine, methadone, and desipramine as 1.64 ± 0.12, 1.31 ± 0.1, and 1.30 ± 0.08, respectively. These ratios align with prior investigations that assessed Pgp transport rates of these three compounds using a cell-based system. Mahar et al. and Tournier et al. reported Pgp efflux rates of 2 for desipramine and methadone (Doan et al., 2002; Tournier et al., 2011). However, quinidine’s reported Pgp efflux rate was 10 (Patil et al., 2011). This concurs with the SPR amplitude findings, indicating that quinidine exhibits significantly higher transport rate than methadone and desipramine, while the latter two share the same transport rate. The larger difference in sensitivity between in vitro cell-based assay and SPR assay is because the design of the SPR assay involved large molecular weight ligand and small molecular weight analyte, which decreased the overall signal sensitivity. However, SPR assay could detect the transport of weak substrates like methadone and desipramine suggesting that the assay is sufficient for screening Pgp efflux. Hence, the SPR signal amplitude proves valuable in monitoring the relative Pgp transport rate across diverse compounds.

Recognized for its capability to track real-time binding kinetics, SPR serves as a means to not only gauge Pgp’s relative transport rate across diverse compounds but also to delve into the transport kinetics of each compound (O’Shannessy, 1994). Presently, two prominent models for Pgp transport, the vacuum cleaner and flippase models have gained acceptance (Romsicki and Sharom, 2001; Sharom, 2014; Kim and Chen, 2018). The SPR data were subjected to fitting using both models. Figure 5 portrays the fittings and residuals for the desipramine dataset, revealing a distinct preference for the vacuum cleaner model over the flippase model. This outcome signifies that the SPR-derived desipramine transport data align more favorably with the vacuum cleaner model, thus suggesting its support for the vacuum cleaner transport mechanism.

The kinetics of each reaction encompassed in the transport process of the three drugs were also assessed, as outlined in Table 2. The values of k_on and k_off for the reaction involving substrate diffusion from the exterior to the outer leaflet exhibited remarkable similarity for desipramine (within the 10⁻⁶ M⁻¹s⁻¹ and 10⁻⁶ s⁻¹ range), suggesting a relatively balanced rate for this bidirectional process in desipramine. Conversely, in the case of methadone, the on-rate (5.07 x 10⁻⁵ M⁻¹s⁻¹) for diffusion from the exterior into the outer leaflet slightly exceeded the off-rate (8.34 x 10⁻⁶ s⁻¹), implying a slight bias against methadone’s diffusion into the outer leaflet. For quinidine, the k_on for diffusion into the outer leaflet of the proteoliposome was 4.62 x 10⁻⁸ M⁻¹s⁻¹, while the k_off was 1.20 x 10⁻⁵ s⁻¹, indicating a greater likelihood for quinidine to stay in the exterior than to diffuse into the outer leaflet of the bilayer.

In the reactions involving desipramine’s diffusion from the outer leaflet to the inner leaflet and the departure of substrate from the outer leaflet to bind to Pgp, the k_on was significantly lower than the k_off. This observation suggests a preference for the desipramine-outer leaflet complex over the desipramine-inner leaflet and desipramine-Pgp complex. Similar trends were observed with both methadone and quinidine, indicating that all three drugs displayed a preference for complex formation with the outer leaflet rather than with Pgp or the inner leaflet of the bilayer.

Remarkably, the reactions in which Pgp facilitated the transport of substrates inside the vesicle exhibited significantly smaller k_off values (6.44 x 10⁻⁸ s⁻¹ for desipramine, 1.89 x 10⁻⁴ s⁻¹ for methadone, and 7.91 x 10⁻⁸ for quinidine) compared to the corresponding k_on values (3.57 x 10⁻⁵ M⁻¹s⁻¹ for desipramine, 2.75 x 10⁻⁴ M⁻¹s⁻¹ for methadone, and 1.6 x 10⁻³ M⁻¹s⁻¹ for quinidine), indicating a pronounced preference for substrate transportation into the proteoliposome interior. In the final reaction, involving substrates diffusing from the inner leaflet to the inner space of the proteoliposome, the k _on rates were significantly lower than the k _off rates for all three drugs. This observation suggests that all three substrates preferred the hydrophobic inner leaflet over the hydrophilic core of the proteoliposome.