Abstract
The quantification of P-glycoprotein [P-gp, ABCB1, multidrug resistance 1 (MDR1)] protein in biological matrices is considered a key factor missing for useful translation of in vitro functional data to the in vivo situation and for comparison of transporter data among different in vitro models. In the present study a liquid chromatography (LC)-mass spectrometry method was developed to quantify P-gp membrane protein levels in different biological matrices. The amount of P-gp transporter protein was measured in Caco-2 cell monolayers and in inside-out human embryonic kidney (HEK)-MDR1 vesicles. From both in vitro systems, two preparations with different functionality were used. Transporter function was determined as digoxin efflux in Caco-2 cell monolayers and N-methylquinidine (NMQ) uptake in membrane vesicles, and, in addition, mRNA expression in the Caco-2 monolayers was measured. The results showed an excellent relationship between NMQ uptake functionality in inside-out HEK-MDR1 vesicles and protein contents. Similar concordance between the digoxin efflux and P-gp content in different Caco-2 cell cultures was observed, whereas mRNA levels are indicative of increased P-gp content and activity in older Caco-2 cultures, however, not yielding the same quantitative relationship. The results from both Caco-2 and HEK-MDR1 membrane vesicles confirm that the protein content is directly related to the level of activity in the respective system. The method presented here to quantify P-gp protein by LC-multiple reaction monitoring will facilitate the development of future methodologies to bridge between expression systems and cell/tissue models and to scale from in vitro models to whole organs.
Introduction
Drug transporters are increasingly recognized as important for drug disposition and drug-drug interactions. P-glycoprotein (P-gp, ABCB1, MDR1) is a key transporter active on many drugs, and there is a large body of clinical evidence for its critical role in drug absorption, distribution, and elimination. P-gp is highly expressed in several organs known to be important for drug disposition. The localization of P-gp has been shown on the apical membrane of columnar epithelial cells of intestine and proximal tubules in the kidney. In the liver, P-gp is localized to the canalicular side of hepatocytes. The orientation of P-gp makes this transporter protein important to protect the body against a possible toxic response by excreting compounds into the intestinal lumen, bile, and urine. Inhibition of the P-gp transporter may also be the basis for drug-drug interactions. The classic example is the increase in oral bioavailability of digoxin when coadministered with the P-gp inhibitor quinidine (Drescher et al., 2003; Igel et al., 2007).
Several in vitro models have been used to investigate transporters and their role in drug uptake and efflux in cells, revealing mechanistic understanding of drug transporter affinities and kinetics. However, the results from in vitro systems have been difficult to fully exploit in in vitro-to-in vivo scaling exercises. A key factor missing for useful in vitro-to-in vivo translation of transporter data is information on the concentrations of the transporter protein both in the in vitro system and in the organs handling the drug. Thus, a method enabling the specific quantitative measurement of the membrane transporter concentration would facilitate prediction of the relevance of individual transporters on human in vivo pharmacokinetics of drugs and drug candidates when using in vitro model systems.
The enzyme-linked immunosorbent assay has been the predominant method used for targeted quantification of a protein, providing good sensitivity and throughput. However, the lack of antibodies with high specificity limits the use of immunological techniques. Moreover, the development of a high-quality enzyme-linked immunosorbent assay requires a significant investment in time and resources. Alternative methods for targeted protein quantification using mass spectrometry-based strategies have been developed to address these issues. The core mass spectrometric technology that has emerged for protein quantification is based on the concept of stable isotope dilution combined with multiple reaction monitoring (MRM) (Gerber et al., 2003; Anderson et al., 2004; Carr and Anderson, 2008). The use of MRM mass spectrometry is based on the measurement of a “proteotypic” tryptic peptide(s) that uniquely and stoichiometrically represents the protein target of interest. Hence, a synthetic stable isotope-labeled version of a proteotypic peptide is used as an internal standard, enabling the protein concentration to be measured by comparing the signals from the isotope-labeled standard peptide to the endogenous peptide in the sample. During the past few years, liquid chromatography (LC) coupled to MRM mass spectrometry has been widely used for protein quantification in biological and clinical samples (Pan et al., 2009). In addition, the quantification of specific membrane transporters has been reported (Kamiie et al., 2008; Li et al., 2008, 2009a,b). Li et al., 2008 demonstrated that the sensitivity of an LC-MRM method for quantification of multidrug resistance-associated protein 2 (ABCC2) exceeded the sensitivity of an immunoblotting assay. In another investigation, Li et al. (2009a) determined the absolute differences for breast cancer resistance protein and bile salt export pump in livers and isolated hepatocytes across species. Li et al. (2009b) have also found that the protein levels of the hepatobiliary transporter multidrug resistance associated protein 2 were significantly different among species in freshly isolated and cryopreserved hepatocytes and snap-frozen liver tissues from human, rat, monkey, and dog. Kamiie et al., 2008 developed an LC-MRM method for simultaneous quantification of 36 membrane proteins (including some transporter proteins) in brain capillary endothelial cells and liver and kidney of the mouse, and Kawakami et al. (2011) reported the simultaneous determination of 11 human cytochrome P450 enzymes with LC-MRM.
Here, we describe the development of a sensitive LC-MRM method for quantifying human P-gp. The P-gp levels were measured in both HEK293 membrane vesicles expressing human P-gp (Karlsson et al., 2010) and Caco-2 cells of different time in culture. Furthermore, the P-gp protein expression level was correlated with mRNA expression level and functional analysis. The generation of such quantitative data of transporters together with functional data will be a key tool for the translation of results from various in vitro model systems to in vivo.
Materials and Methods
Chemicals and Reagents.
Acetonitrile and water (LC-MS grade) were purchased from Thermo Fisher Scientific (Leicestershire, UK). Phosphate-buffered saline, pH 7.4, a Colloidal Blue Staining kit, NuPAGE 4 to 12% bis-Tris gel, NuPAGE CDS sample buffer 4×, NuPAGE MES SDS running buffer 20×, NuPAGE reducing agent, NuPAGE antioxidant, and Hanks' balanced salt solution were all purchased from Invitrogen (Carlsbad, CA). NaHCO3, Na2CO3, and formic acid were obtained from Fluka (Steinheim, Germany). Precision Plus Protein standards (Kaleidoscope) were obtained from Bio-Rad Laboratories (Hercules, CA). Bovine serum albumin, ammonium bicarbonate (AMBIC), HEPES, iodoacetamide (IAA), and Tris-HCl were purchased from Sigma-Aldrich (Steinheim, Germany). The protease inhibitor cocktail (complete Mini) was obtained from Roche Applied Science (Mannheim, Germany), dithiothreitol (DTT) was obtained from Genomic Solutions Inc. (Ann Arbor, MI), and the PPS Silent Surfactant was obtained from Protein Discovery Inc. (Knoxville, TN). Sequencing grade-modified trypsin was a product of Promega (Madison, WI). The synthetic proteotypic peptide and its corresponding stable isotope-labeled (SIL) variant were purchased from Thermo Fisher Scientific (Ulm, Germany). Tritiated N-methylquinidine and digoxin were purchased from RC Tritec Ltd. (Teufen, Switzerland) and PerkinElmer Life and Analytical Sciences (Waltham, MA), respectively. All other chemicals were of at least analytical grade and were obtained from commercial sources.
HEK-MDR1 Vesicle Preparation and Functional Assay.
HEK293-Epstein-Barr virus nuclear antigen cells were cultured and transiently transfected with human P-gp as described earlier (Karlsson et al., 2010). Purified P-gp membrane vesicles from transfected HEK293-Epstein-Barr virus nuclear antigen cells were prepared as described previously (Karlsson et al., 2010). Membrane protein content was determined using a BCA Protein Assay Kit (Thermo Fisher Scientific).
The vesicular transport activity was measured using a rapid filtration technique on 96-well filter plates (MultiScreenHTS-FB Plate; Millipore Corporation, Billerica, MA). After a preincubation period of 5 min, membrane vesicles (50 μg of protein/75-μl reaction volume) were incubated at 37°C in the presence or absence of 4 mM ATP in assay buffer (250 mM sucrose, 10 mM MgCl2, and 10 mM Tris-HCl, pH 7.0) containing radiolabeled probe substrate [3H]N-methylquinidine (NMQ) (1 μM, 3 μCi/ml) for 2 min. The uptake reaction was stopped by adding ice-cold washing buffer, immediately transferring the vesicles to the filter plate, and washing the filter with washing buffer (250 mM sucrose, 100 mM NaCl, and 10 mM Tris-HCl, pH 7.0). The radioactivity in the membrane vesicles retained on the filter was measured with a TopCount scintillation counter (TopCount NXT; Thermo Fisher Scientific). ATP-dependent transport (picomoles per minute per milligram of protein) was calculated as the difference between the values obtained in the absence of ATP from those in the presence of ATP. Assays were run at least in triplicate.
Caco-2 Cell Monolayer Culture and Transport Assay.
Caco-2 cells were purchased from American Type Culture Collection (Rockville, MD) at passage 18 and maintained in Dulbecco's modified Eagle's medium containing 10% heat-inactivated fetal calf bovine serum, 1% nonessential amino acids, and 1.5% l-glutamine in an atmosphere of 95% air and 5% CO2 at 37°C. All tissue culture media were obtained from Invitrogen (Paisley, Scotland). For functional studies and P-gp quantitation, monolayer cultures were grown on polycarbonate culture inserts in medium also containing antibiotics (100 U/ml penicillin and 100 μg/ml streptomycin). The cells were seeded at an initial density of 2.5 × 105 cells/1.13 cm2 filter (pore size 0.4 μm, polycarbonate membrane Transwell filters; Corning Life Sciences, Lowell, MA). The medium was changed every 2nd day. Cells for protein quantification were harvested by trypsinization after 10 and 29 days in culture on filters and pelleted by centrifugation at 300g for 5 min.
At both culturing times (10 and 29 days), transport experiments were performed in Hanks' balanced salt solution buffered with 25 mM HEPES at pH 7.4. The Papp value of the P-gp substrate [3H]digoxin (28 nM) was determined in the apical-to-basolateral (A-B) and basolateral-to-apical (B-A) direction in the absence and presence of verapamil (100 μM), a well characterized inhibitor of P-gp. The apparent permeability values (Papp) were calculated in all experiments according to the equation Papp = (dQ/dt)/(A · C0), where dQ/dt is the slope of the cumulative amount transported during the time course of the period studied, A is the monolayer culture area, and C0 is the starting concentration.
Functionality of P-gp protein is expressed as the bidirectional transport ratio of the specific probe substrate digoxin: efflux ratio = Papp BA/Papp AB. Inhibition of digoxin transport by verapamil in the Caco-2 cell system is interpreted as functional evidence of P-gp-mediated transport.
mRNA Isolation and Quantification (Reverse Transcription-Polymerase Chain Reaction).
mRNA of six filters with Caco-2 cells was extracted on day 10 and day 29, respectively, using a phenol-chloroform extraction method (RNA-Stat) as described previously (Seithel et al., 2006). mRNA quality was confirmed with integrity of 18S and 28S RNA bands in agarose gel electrophoresis, and RNA was quantified with a NanoDrop spectrophotometer. cDNA was prepared according to the manufacturer's protocol using an Invitrogen SuperScript II Kit. Reverse transcription-polymerase chain reaction was performed using the Applied Biosystems Assays-on-Demand for ABCB1 (Hs00184500_m1) and PPIA (Hs99999904_m1) on a 7500 instrument and analyzed using SDS2.3 software.
mRNA expression of ABCB1 was calculated using the ΔCT method against PPIA as a reference gene: ΔCT = CT PPIA − CT ABCB1.
Relative gene expression (REL) was then calculated as follows: 2−ΔCT.
Extraction of Membrane Fraction.
An overview of the extraction procedure is depicted in Fig. 1. The cell pellets were resuspended in 10 volumes of ice-cold 10 mM NaHCO3, pH 8, followed by addition of protease inhibitor cocktail according to the instructions of the manufacturer (Roche Applied Science). The cells were allowed to swell for 10 min and were subjected to lysis using a glass Dounce homogenizer (25 strokes) on ice. Nuclei, unbroken cells, and mitochondria were spun down (10,000g for 10 min at 4°C). The postnuclear supernatant containing the membranes of interest was adjusted to 100 mM Na2CO3 and sonicated for 20 min to disrupt electrostatic interactions between the membranes and the membrane-associated proteins. After the carbonate wash, the membrane material was collected by ultracentrifugation (120,000g for 1.5 h at 4°C). The membrane pellet was resuspended in 50 mM AMBIC, using tip sonication to properly disperse the sample. The membrane samples were stored at −80°C until future analysis. Total protein concentrations of the various membrane fractions were determined with a DC Protein Assay Kit (Bio-Rad Laboratories).
Schematic overview of the membrane preparation procedure and identification of proteotypic peptide (left) and sample workup for LC-MS quantification of specific membrane protein (right).
SDS-PAGE Separation.
The vesicle preparation of the HEK293 cells overexpressing human P-gp was diluted 10 times, and 40 μg of total protein was loaded onto gel and separated under reducing conditions on a precast bis-Tris SDS-PAGE gel. The protein bands were visualized by using a Colloidal Blue Staining kit. The protein band corresponding to P-gp was cut out and dissected into 2-mm pieces that were transferred to a 1.5-ml Eppendorf microtube and subjected to in-gel digestion. In brief, the gel pieces were destained using 25 mM AMBIC in 70% acetonitrile and dried by vacuum centrifugation for 10 min, reduced with 10 mM DTT at 56°C for 45 min, and subsequently IAA-alkylated for 30 min. Finally, the gel pieces were dried by vacuum centrifugation for 15 min and then incubated with trypsin (0.15 μg) at 37°C overnight. The trypsin-digested peptides in the gel pieces were extracted with a solution containing 1% formic acid in 5% acetonitrile in water. The peptide extract was stored at −80°C before analysis by nano-LC-Chip-ESI-QTOF MS for proteotypic peptide identification.
Identification of the Proteotypic Peptide.
A 1-μl aliquot of the tryptic peptide extract was injected onto an LC-MS system consisting of a 1200 series system, an LC-Chip Cube MS interface, and a 6520 ESI-QTOF mass spectrometer (Agilent Technologies, Santa Clara, CA). Chromatography was performed on an LC-Chip (Agilent Technologies) that incorporated a 40-nl enrichment column and a 150 mm × 75 μm analytical column packed with ZORBAX 300SB-C18, 5-μm particles. The tryptic peptides were loaded onto the enrichment column with 97% solvent A (2.5% acetonitrile and 0.1% formic acid) and 3% solvent B (95% acetonitrile and 0.1% formic acid) at a flow rate of 4 μl/min. Elution was performed using an increasing gradient from 3% B to 40% B in 50 min, at a flow rate of 300 nl/min. The following QTOF conditions were used: drying gas, 5 l/min (350°C); fragmentor, 175 V; skimmer, 60 V; capillary voltage, 1800 V; acquisition rate and time, 4 spectra/s (threshold 200 absolute, 0.01% relative) and 250 ms/spectrum; MS scan range and rate, 296 to 2500 at 4 Hz; MS/MS acquisition rate and time, 3 spectra/s (threshold 5 absolute, 0.01% relative) and 333.3 ms/spectrum; MS/MS scan range and rate, 50 to 2500 at 3 Hz; collision energy slope, 3.3 V; offset, 2.5 V; auto MS/MS, 5 precursor; active exclusion on with 2 repeat and with release after 0.1 min; preferred charge state, 2, 3, >3, unknown; and internal reference mass correction enabled. The raw data files were exported to a Mascot Generic Format by MassHunter (version B.02.00; Agilent Technologies) to create peak lists on the basis of the recorded fragmentation spectra. Peptides and proteins were identified by Mascot version 2.2.0 (Matrix Science, London, UK) against the SwissProt database with a precursor mass tolerance of 10 ppm, a fragment ion mass tolerance of ±0.2 Da, and trypsin specificity allowing for up to one missed cleavage. Carbamido methylation of cysteine was set as a fixed modification, and methionine oxidation was allowed as a variable modification. Peptides were rejected if the Mascot score was less than the 95% confidence limit, based on the identity score of each peptide.
Tryptic Digestion Protocol and Preparation of Calibration Curve.
An aliquot of 40 μl of each membrane sample was diluted 1:1 with PPS stock solution (0.2% PPS and 10% acetonitrile in 50 mM AMBIC) in a 1.5-ml Eppendorf tube. The PPS enhanced the extraction and solubilization of hydrophobic membrane proteins and improved the in-solution tryptic digestion of membrane proteins. The membrane samples were then reduced (10 mM DTT at 50°C for 30 min) and alkylated (15 mM IAA in the dark at room temperature for 45 min). After addition of internal standard, 25 fmol of SIL P-gp proteotypic peptide (AGAVAEEV[13C615N1]LAAIR), the samples (2–50 μg of protein) were digested by trypsin in a final volume of 100 μl at 37°C for 6 h. The ratio of trypsin and protein was 1:20. To avoid detergent interference in the subsequent mass spectrometric analysis, the PPS was hydrolyzed by adding 2.5 μl of undiluted formic acid for 1 h. The samples were centrifuged at 16,000g for 10 min before analysis by LC-ESI-MS/MS. The external calibration curve was prepared by spiking the synthetic proteotypic peptide AGAVAEEVLAAIR in the digestion matrix with the addition of a fixed amount at 2500 pM of the SIL internal standard peptide. Data were processed by integrating the appropriate peak areas generated from the extracted ion chromatograms for the 13-mer analyte peptide and the SIL internal standard peptide by MassHunter version B.01.04. The ratio of the peak area of the proteotypic peptide to the SIL peptide was plotted against the concentration of the synthetic native peptide for constructing the regression analysis.
LC-MS/MS Quantification of P-gp.
The selected proteotypic peptide was subjected to positive ion LC-MRM analysis using an LC-MS system comprising an ultra-high-pressure liquid chromatography system (1290 Infinity binary pump, 1290 Infinity autosampler with thermostat, and 1290 Infinity thermostated column compartment; Agilent Technologies, Waldbronn, Germany) coupled to a triple quadrupole mass spectrometer (6460; Agilent Technologies). Chromatography was performed on a 1.0 × 50 mm C18 column (Acquity UPLC BEH C18, 1.7-μm size beads, 130-Å pore size; Waters, Milford, MA). Mobile phase A consisted of 2.5% acetonitrile and 0.1% formic acid, and mobile phase B consisted of 95% acetonitrile and 0.1% formic acid. The tryptic peptides were separated and eluted using a linear gradient starting from 5% B that progressed to 60% B in 10 min. A sample volume of 10 μl was injected onto the column at a flow rate of 0.2 ml/min. The MRM transitions for the proteotypic peptide monitored represented the doubly charged precursor ion (AGAVAEEVLAAIR)2H+ (m/z 635.4) to the singly charged product ions y9, y8, and y7 with m/z 971.5, 900.5, and 771.5, respectively. Likewise, the MRM transitions for the SIL internal standard peptide were the corresponding doubly charged precursor ion with m/z 638.9 to the singly charged product ions y9, y8, and y7 with m/z 978.5, 907.5, and 778.5, respectively. The instrument settings of the 6460 triple quadrupole mass spectrometer were as follows: drying gas temperature, 300°C; drying gas flow, 6 l/min; nebulizer pressure, 40 psi; capillary voltage, 4500 V, sheath gas temperature, 350°C; sheath gas flow, 11 l/min, fragmentor voltage, 150 V; dwell time, 50 ms; and collision energy for y9, y8, and y7 ions, 22, 18, and 18 eV, respectively.
Data Analysis.
The functional data in HEK-P-gp vesicles were acquired in a minimum of two experiments performed on different days. The comparison of the P-gp protein amount among different vesicle preparations and 10- and 29-day-old Caco-2 cultures was statistically analyzed using Student's t test. p < 0.05 was regarded as statistically significant.
Results
Selection of Proteotypic Peptide.
The membrane protein preparation was separated by SDS-PAGE followed by Coomassie Blue staining. The P-gp protein band (141 kDa) was excised for in-gel tryptic digestion and the extracted tryptic peptides were analyzed by a nano-LC-ESI-QTOF mass spectrometer. The acquired MS/MS data were then subjected to database searching using the Mascot search engine. On the basis of the intensity of the fragment ions and reproducibility, the 13-mer peptide AGAVAEEVLAAIR, which was produced by tryptic digestion of human P-gp protein was selected as the proteotypic peptide with a Mascot score of 109. The MS spectrum and the MS/MS spectrum of the selected proteotypic peptide, represented by H-AGAVAEEVLAAIR-OH, are illustrated in Fig. 2. The predominant fragment ions (C-terminal y ions) are corroborated with the sequence of the tryptic fragment.
MS and MS/MS spectra of the proteotypic peptide for P-gp identified on a QTOF instrument. A, the MS spectrum of H-AGAVAEEVLAAIR-OH (doubly charged, m/z 635.36), the precursor ion of the proteotypic peptide. B, product ion mass spectrum from the collision-induced dissociation of m/z 635.36 precursor ion (AGAVAEEVLAAIR)2H+ to singly charged products; predominant fragment ions (C-terminal y ions) are highlighted.
P-gp is a member of the superfamily of ABC transporters, which is one of the largest families of transmembrane proteins. More specifically, P-gp belongs to ABC transporter subfamily B, which is composed of several protein members as outlined in Table 1. A BLAST search was performed to ensure that no exact amino acid sequence matches exist from any other ABC family member (Table 1). The sequence alignment demonstrated that the AGAVAEEVLAAIR peptide is conserved in P-gp across species of human, rat, mouse, and dog. However, the selected proteotypic peptide could easily be distinguished from the human ABCB subfamily transporters listed in Table 1. Therefore, by taking the peptide sequence analysis based on the LC-MS/MS data along with the confidence of a selective tryptic fragment for P-gp, this 13-mer tryptic peptide (AGAVAEEVLAAIR) was selected as the surrogate marker for P-gp protein that can be used for the development of an LC-MRM method.
Sequence specificity ABC transporter subfamily B
The selected proteotypic peptide, AGAVAEEVLAAIR, for human P-gp is unique and distinct from other human transporters from the ABCB subfamily; different amino acids are depicted in bold italic type. The sequence is specific for an ABCB1 region that is highly conserved between species and found to be identical in abcb1a and abcb1b isoforms in mouse and rat and abcb1 in dog and human.
LC-MRM Method Development for Quantification of P-gp.
A synthetic variant of the selected proteotypic peptide (AGAVAEEVLAAIR) was purchased along with its SIL variant (AGAVAEEVL*AAIR). The SIL peptide, labeled (13C6,15N1) at the position of leucine, was used as an internal standard to normalize the acquired data. The product ion spectra of P-gp and the corresponding SIL peptide are shown in Fig. 3. Three of the most intense transition ions, y7, y8, y9 and y′7, y′8, y′9, were selected, respectively, as the unique MRM signature for the development of a specific LC-MRM method for P-gp quantification. The peak areas of all monitored parent to product ion transitions of the synthetic proteotypic peptide were normalized by the peak area of the corresponding MRM transitions of the SIL internal standard. The analytical procedure included a standard curve placed at the beginning and end of the run to bracket the unknowns, which resulted in a total of two measurements for each standard calibration level. The calibration curve was linear over the concentration range of 10 to 5000 pM for P-gp and is illustrated in Fig. 4. The correlation coefficient (r2) was greater than 0.99. The limit of quantification for P-gp was determined at 10 pM, as defined by a signal/noise ratio response of 3 and with accuracy within ± 10% and precision better than 10%. The response of a blank sample analyzed immediately after the analysis of a sample at the upper limit of quantification was less than 1% of the response of the limit of quantification.
Top, y ions formed from collision-induced transition from the proteotypic peptide. The three most intense transition ions, the y ions of m/z 971.5, 900.5, and 771.5, were selected to develop the LC-MRM method. Bottom, corresponding y ions from the heavy peptide (internal standard) are shown; for all, a mass shift of +7 Da adds to the corresponding transitions of the SIL peptide due to the isotope labeling.
Representative calibration curve over the concentration range 10 to 5000 pM. The concentration of SIL internal standard was fixed at 2500 pM. Each calibration standard was measured in duplicate before and after each analytical run to bracket the unknowns. Inset, extracted ion chromatogram at the lowest point in the standard curve (10 pM), showing a clear signal against baseline.
Improvement of Detection Sensitivity by Minimizing Analyte Adsorption.
A significant improvement in the detection sensitivity can be obtained for specific tryptic peptides by a simple addition of 10% acetonitrile in the digestion matrix. Depending on the physicochemical properties, a peptide can be more or less amenable for adsorption to plastic surfaces (pipette tips, sample vials, and tubing). One option to avoid such binding is to use siliconized or other low-binding surfaces that may be readily obtainable for tips and vials and less straightforward for tubings and capillaries in the LC system. Another generic approach is adding a small percentage of organic solvent to the buffer to prevent binding to surfaces. It has previously been demonstrated that tryptic digestion in mixed organic (e.g., acetonitrile)-aqueous solvent systems is highly efficient in terms of both rate of digestion and amino acid sequence coverage (Russell et al., 2001). Russell et al. (2001) used as much as 80% acetonitrile during trypsination. In this study we were able to obtain reproducible P-gp quantifications when performing the digest with and without the presence of 10% acetonitrile. In the case of the AGAVAEEVLAAIR peptide, we observed almost a factor of 10 in improvement of the detection sensitivity as illustrated in Fig. 5, indicating that this peptide has a tendency to adsorb that can be minimized by the addition of a small amount of an organic modifier.
Detection sensitivity was significantly improved after addition of 10% acetonitrile (ACN) to the digestion matrix.
Validation of LC-MRM Quantification for P-gp.
To ensure appropriate performance of the established LC-MRM method, quality control samples were prepared by spiking the synthetic proteotypic peptide in the membrane fraction of a membrane vesicle preparation originating from HEK293 cells at concentrations of 100 and 1000 pM subsequently followed by tryptic digestion overnight according to the protocol previously described under Materials and Methods. The SIL peptide was also spiked into each sample as an internal standard before trypsin digestion. The amount of quality control peptide was determined by subtracting the baseline of P-gp protein in the biological matrix. The percentage relative error of the prepared validation samples (100 and 1000 pM) at three independent days demonstrated the accuracy, whereas the coefficient of variation represented the precision of the LC-MRM method. The validation results of the synthetic proteotypic peptide are shown in Table 2. Reproducibility of the tryptic digestion of the membrane preparation was evaluated by three independent digestions and LC-MRM measurements of the P-gp concentration and resulted in a coefficient of variation of less than 5% (data not shown), clearly indicating that the trypsination process was reproducible.
Intraday and interday validation of P-gp synthetic proteotypic peptide quantification
The synthetic proteotypic peptide was spiked to the membrane fraction of a membrane vesicle preparation originating from untransfected HEK293 cells (matrix) at two concentrations (100 and 1000 pmol/l). Preparations were performed at three independent occasions.
P-gp Mediated NMQ Uptake in Two Batches of HEK-MDR1 Vesicles.
The ATP-dependent uptake of the P-gp substrate NMQ was determined in two different vesicle batches, numbers 3 and 6. The protein concentration of both preparations were very similar (13 and 10 mg/ml, respectively); however, the activity measurements differed significantly: 61 ± 11 pmol · min−1 · mg protein−1 and a +ATP/−ATP ratio of 4.4 in batch 3, whereas in batch 6 the active uptake was 220 ± 30 pmol · min−1 · mg protein−1 and the +ATP/−ATP ratio was 9.3 (Table 3).
P-gp content as determined by LC-MRM-MS and functionality in two batches of HEK-MDR1 vesicles and in 10- and 29-day old Caco-2 cell monolayers
P-gp activity is determined as [3H]NMQ uptake in inside-out HEK-MDR1 vesicles and the [3H]digoxin efflux ratio, as described under Materials and Methods. Values are given as mean ± S.D.; n = 3–4.
P-gp Expression and Transporter Functionality in Caco-2 Cell Monolayers Grown for 10 and 29 Days.
In accordance with previous publications, P-gp mRNA levels increase in Caco-2 cell monolayers with time in culture. In this study, a 3.4-fold increase in normalized mRNA levels was observed between day 10 and day 29 (data not shown).
The bidirectional transport of the probe substrate digoxin was clearly direction-dependent, with more than 4.2-fold higher basal to apical transport on day 10 that increased to more than 10-fold on day 29. The calculated efflux ratio increased 2.2-fold between day 10 and day 29 (Table 3). The efflux was inhibitable in presence of 100 μM verapamil (data not shown).
LC-MRM Quantification of P-gp in HEK-MDR1 Vesicles and Caco-2 Cell Monolayers Grown for 10 and 29 Days.
The LC-MRM method developed above was applied for the determination of P-gp protein levels in the membrane fraction extracted from two HEK293-MDR1 vesicle batches and Caco-2 cells that were cultured for 10 and 29 days, respectively. The specific MRM transitions based on the chromatographic retention time and MS spectrum were used to detect and quantify the endogenous P-gp proteotypic peptide by LC-MS/MS in MRM mode. Figure 6 shows the reconstituted ion chromatograms at different lengths of culture period of the tryptic proteotypic P-gp fragment (AGAVAEEVLAAIR) released from trypsination of Caco-2 cells. The human P-gp protein was determined at 4.0 fmol/μg protein (10 days) and 7.9 fmol/μg protein (29 days). The amount of P-gp in the cell sample grown for 29 days was approximately 2-fold higher than that in the sample that was grown for 10 days. HEK-MDR1 vesicle batches differed nearly 4-fold in their P-gp protein contents: 8.2 and 32.0 fmol/μg protein in batch 3 and batch 6, respectively.
Representative reconstructed mass chromatograms of P-gp proteotypic peptide (transition of m/z 635.4 → m/z 971.5) from trypsinated membrane preparations of Caco-2 cells cultured for 10 days (top) and 29 days (bottom), respectively.
Discussion
P-gp is abundant in all organs important for the disposition of many drugs in the body (Lin and Yamazaki, 2003; Sun et al., 2004; Lee et al., 2010). The combination of the wide distribution in the body and the broad substrate spectrum for P-gp implies that mechanistic and quantitative models are needed to understand how this transporter affects the pharmacokinetics of drugs in the body. Caco-2 cells and Madin-Darby canine kidney-MDR1 and human P-gp-transfected vesicles are currently common in vitro systems used to address the mechanistic relevance of substances as substrates or inhibitors of P-gp (Guidance for Industry: Drug Interaction Studies—Study Design, Data Analysis, and Implications for Dosing and Labeling, 2006, http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm072101.pdf). In particular, interpretation of inhibition data from in vitro models are increasingly well understood and are in many cases successfully translated to relevant information for the in vivo situation (Fenner et al., 2009; Cook et al., 2010). In contrast, extrapolation of substrate kinetic data from in vitro models to in vivo pharmacokinetics remains more challenging. Even though P-gp substrates can with good confidence be identified using in vitro systems, the quantitative relevance of the active transport process for the pharmacokinetics in vivo is still difficult to predict from in vitro data.
In comparison, successful in vitro-in vivo extrapolation for hepatic metabolic clearance has been applied for cytochrome P450-mediated metabolism. A key factor for the well accepted in vivo prediction of hepatic clearance is the possibility of measuring the specific cytochrome P450 content in liver tissue and in the various in vitro systems used. The amount of active protein in the in vitro system can thus be directly contrasted and related to the amount of active protein in the tissue (Barter et al., 2007). Today, similar scaling factors and extrapolation equations for membrane transporters are scarcely available. Transporter substrate kinetics can be compared and ranked in affinities in the same in vitro system. However, for a useful comparison between different in vitro systems and prediction to in vivo, the absolute amount of functional protein is needed. The protein atlas project (Nilsson et al., 2005) is to some extent filling this gap with semiquantitative information from protein data (Western blot and immunohistochemistry) and mRNA expression. However, it is highly warranted to be able to use specific scaling factors that describe the amount of transporter protein per cell unit or gram of tissue. In a recent publication, Bolger et al. (2009) applied relative expression scores normalized to ileum to account for varying P-gp levels in different intestinal regions and simulated pharmacokinetic and absorption profiles after different doses using an advanced compartmental absorption and transit model in GastroPlus simulations. The relative expression profiles from literature sources were empirically compared in the simulation model and built into physiological characteristics of the small and large intestines for the simulation. The authors proposed an advanced compartmental absorption and transit model that was able to accurately account for the regional and temporal changes in concentration and carrier-mediated transport and reproduced the nonlinear dose dependence observed in vivo.
In the present study, we report the successful development of a LC-MRM method to quantify the transport protein P-gp (ABCB1) in in vitro systems often used for P-gp studies and for the first time relate protein abundance to transporter functionality in these systems. Several critical factors that affect both the sensitivity and specificity of an assay have to be considered in the selection of a proteotypic peptide for development of quantitative protein assays based on a proteolytic peptide surrogate marker (the proteotypic peptide). First, the peptide amino acid sequence has to be unique to the target protein and contain no missed cleavage sites. Second, the peptide must be reproducibly observed in proteolytic digests. Third, the amino acid sequence of the peptide should not contain amino acids that are susceptible to chemical modifications. The tryptic fragment used in this study as a marker peptide is highly specific among proteins from the ABCB transporter subfamily. In contrast to the method published recently by Zhang et al. (2011), in which mouse mdr1b was not detected using a 9-mer peptide as a selective tryptic fragment for P-gp, the present method will quantify all P-gp protein in human and preclinical species, including the rodent forms 1a and 1b. ABCB1 is detected selectively from proteins ABCB4 and ABCB11, which have the closest related protein sequences with 2 and 6 amino acids difference, respectively. Other members of the ABCB family are not showing a homologous sequence. Because protein quantification by LC-MRM methodology is based on the measurement of a tryptic peptide, it is critical that the tryptic digestion should achieve 100% efficiency to accurately quantify the protein of interest in a biological sample. The transmembrane domain should be avoided when a proteotypic peptide is selected. Trypsin typically has limited availability in the transmembrane domain that is situated within the lipid bilayer and hereby “protected.” Moreover, even if trypsination occurs in the transmembrane domain, it may be difficult to extract the peptides from the lipid bilayer. The proteotypic peptide identified for P-gp in this study is not located within a transmembrane domain. In our experience the selection of proteotypic peptide is difficult to predict by in silico experiments, and until more reliable algorithms have been developed we prefer to conduct wet laboratory trypsinations subsequently followed by LC-MS experiments. Ideally, a pure protein is used as a reference standard to confirm the complete digestion of the target protein. However, in the present study, the method was developed without a pure protein standard because of the lack of commercial availability of purified P-gp. This is not an uncommon situation for quantification of membrane proteins. One way of addressing this issue is to synthesize a slightly longer surrogate digestion substrate peptide that contains the amino acid residues of the proteotypic peptide (Li et al., 2008). However, the structure of a large and complex transmembrane protein is significantly different to digest compared with a surrogate digestion substrate peptide. The use of gene-transfected cells overexpressing P-gp enabled us to conduct tryptic mapping and subsequently identify a suitable proteotypic peptide. Precision and accuracy of the LC-MRM method are within 10% and meet quality standards for bioanalytical methods. The method produced reproducible quantification data from different in vitro matrices on different occasions. The relevance of the membrane transporter abundance data is confirmed by a good correlation to transporter functionality in the two different in vitro systems. The amount of P-gp in the Caco-2 cell samples from cells grown for 29 days was 2-fold higher than that in cells grown for 10 days, in good agreement with the functional data on digoxin efflux, which increased 2.2-fold between the two cell cultivating times. Similar concordance between NMQ uptake functionality in inside-out HEK-MDR1 vesicles and protein contents was observed. A 3.6-fold increase in functional activity was reflected by a 3.9-fold increase in protein content. The observed abundance functionality relationships in both membrane vesicle preparations and monolayer grown cells show that the developed LC-MRM method is applicable to quantify the specific protein contents in the in vitro systems.
The LC-MS method to quantify membrane protein levels presented here enables accurate determinations of human P-gp protein abundance in isolated cells. The high analytical sensitivity of the method is promising for its application to different types of human tissues in the future.
On the basis of the fact that the proteotypic peptide used in this study is a homologous sequence in human, rat, dog, and mouse P-gp, the technology is readily transferable to quantitation of nonhuman P-gp and will enable direct interspecies comparison of protein abundance. In addition, comparison of protein levels among species and tissues will be possible. Such knowledge is warranted to guide interpretation of observed interspecies differences in pharmacokinetic profiles. In addition, interindividual differences in P-gp protein levels between donors can be quantified and assist understanding of human variability.
In summary, this method to quantify P-gp by LC-MRM will facilitate future efforts to bridge between expression systems and cell/tissue models and to scale from in vitro models to whole organs. Availability of such scaling factors will be instrumental in successful extrapolation to human to understand in a quantitative manner the impact of P-gp-mediated transport on overall pharmacokinetic profiles and drug distribution. Because the P-gp quantitative method developed in the present study also could be used in animals, the application could be extended to describe levels and variations of P-gp levels in various organs in animals and how these may affect pharmacokinetics in vivo.
Authorship Contributions
Participated in research design: Miliotis, Palm, Andersson, and Hilgendorf.
Conducted experiments: Miliotis, Ali, and Hilgendorf.
Contributed new reagents or analytic tools: Miliotis, Lundquist, and Ahnoff.
Performed data analysis: Miliotis, Ali, Palm, and Hilgendorf.
Wrote or contributed to the writing of the manuscript: Miliotis, Palm, Andersson, and Hilgendorf.
Footnotes
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
doi:10.1124/dmd.111.040774.
-
ABBREVIATIONS:
- P-gp
- P-glycoprotein
- ABC
- ATP-binding cassette
- MDR/mdr
- multidrug resistance
- MRM
- multiple reaction monitoring
- LC
- liquid chromatography
- HEK
- human embryonic kidney
- MS
- mass spectrometry
- bis-Tris
- 2-[bis(2-hydroxyethyl)amino]-2-(hydroxymethyl)propane-1,3-diol
- MES
- 4-morpholineethanesulfonic acid
- AMBIC
- ammonium bicarbonate
- IAA
- iodoacetamide
- DTT
- dithiothreitol
- PPS
- 3-[3-(1,1-bisalkyloxyethyl)pyridin-1-yl]propane-1-sulfonate
- SIL
- stable isotope-labeled
- NMQ
- N-methylquinidine
- PAGE
- polyacrylamide gel electrophoresis
- ESI
- electrospray ionization
- QTOF
- quadripole time of flight.
- Received May 31, 2011.
- Accepted September 23, 2011.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics
