BCRP, breast cancer resistance protein; CV, coefficient of variation; CYP, cytochrome P450; γ-gtp, gamma-glutamyl transpeptidase; LC–MS/MS, liquid chromatography–tandem mass spectrometry; LLOQ, lower limit of quantification; MDR, multidrug resistance protein; MRP, multidrug resistance-associated protein; MRM, multiple reaction monitoring; NTCP, Na+/taurocholate cotransporting polypeptide; OATP, organic anion transporting polypeptide; OCT, organic cation transporter; QC, quality control; UGTs, UDP-glucuronosyltransferases
Mass spectrometry (MS)-based multiplexed multiple reaction monitoring quantification of proteins has recently evolved as a versatile tool for accurate, absolute quantification of proteins. The purpose of this study was to examine the validity of the present method with regard to standard bioanalytical criteria for drug transporters, cytochrome P450 (CYP) enzymes and uridine 5′-diphospho-glucuronosyltransferases (UGTs). Membrane preparations from human liver tissue were used for target protein quantification. As a result, the determination coefficients (r2) of all targets were greater than 0.986. In the absence of matrix, inaccuracy values (expressed as % deviation) were −8.1% to 20.3%, whereas imprecision values (expressed as % coefficient of variation) were within 15.9%. In the presence of matrix, which consisted of digested plasma membrane fraction for transporters and digested microsomal membrane fraction for CYP enzymes and UGTs, respectively, the inaccuracy was −15.3%–8.1%, and the imprecision were within 18.9%. Sufficient sample stability of membrane fraction was shown for three freeze–thaw cycles, 32 days at −20°C, and in processed samples for 7 days at 10°C. In conclusion, this study demonstrated, for the first time, that the MS-based assay with nano-liquid chromatography provides adequate reliability and robustness for the quantification of selected drug transporters, P450 enzymes and UGTs. © 2011 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 100:4037–4043, 2011
Drug transporters, cytochrome P450 (CYP) enzymes and uridine 5′-diphospho-glucuronosyltransferase (UGT) enzymes, play an important role in the absorption, distribution, metabolism, and excretion (ADME) profile of most drugs, xenobiotics, and endogenous compounds.1–5 Absolute quantitative information on the expression levels of these proteins in various tissues and organs (e.g., liver and small intestine), in humans and in animal models, and in vitro systems can help to improve understanding and prediction of the pharmacokinetic and safety profile of new drugs.
Quantitative studies on these proteins applying liquid chromatography–mass spectrometry (LC–MS) technology and regulatory quality standards for bioanalytical assays have not been performed in the past, despite the significant progress in the field of drug transporter and drug metabolism research. Instead, relative mRNA expression has been widely used as a surrogate for protein expression level. Several studies, however, demonstrated only poor correlation between mRNA and protein expression levels.6–8 On the contrary, quantitative immunological approaches such as enzyme-linked immunosorbent assay (ELISA) are capable of highly sensitive quantification, but development of specific antibodies can be time consuming (>1 year) or may fail, for example, for protein classes showing high homology in amino acid sequence.9,10 Recently, several quantitative proteomics methods based on MS have been developed as alternatives to the conventional immuno-based methods.11–13 Kamiie et al.14 first described a highly sensitive and simultaneous quantification method for drug transporters by LC coupled to tandem MS (LC–MS/MS). Kawakami et al.15 reported a quantification method for CYP enzymes and simultaneously quantified expression of several CYP enzymes in human liver microsomes. Ito et al.16 and Uchida et al.17 quantified drug transporters, receptors, and junctional proteins in monkey and human blood–brain barrier.
The principle of this method is based on the quantification of unique peptide fragments by means of multiple reaction monitoring (MRM), which are released from the target proteins by previous trypsin digestion. Synthetic peptides and stable isotope-labeled peptides of identical amino acid sequence serve as calibrators and as internal standard, respectively. Nano-flow LC (nano-LC) combined with MRM analysis allows for the sensitive and selective quantification of multiple target peptides within the same sample and a single chromatographic run.
So far, regulatory authorities such as US Food and Drug Administration (FDA) and European Medicines Agency provide only guidance on acceptance criteria for the quantification of either small molecules by LC–MS/MS or for the quantification of proteins by ligand-binding assays.18,19 Currently, no guidance is given for the quantification of proteins by LC–MS/MS. Moreover, there are only sparse reports available in the literature evaluating assay performances of large molecules, such as proteins, by LC–MS/MS with regards to reliability and robustness. In the present study, we demonstrate reliability, such as linearity, reproducibility, and accuracy, of the simultaneous determination of eight drug transporters, eight CYP enzymes, three UGT enzymes, and two marker membrane proteins by nano-LC–MS/MS.
MATERIALS AND METHODS
All peptides were chosen for synthesis based on the in silico selection criteria as described previously15 and purchased from Thermo Fisher Scientific (Ulm, Germany). The amino acid sequences and the conditions for LC–MS/MS detection were listed in the Supplemental Table 1. Peptide purity (>95%) was provided by the manufacturer, using reversed-phase high-performance liquid chromatography (HPLC) with ultraviolet detection (with a detection wavelength of 215 nm) and matrix-assisted laser desorption/ionization–time-of-flight MS analyses. Other chemicals were commercial products of analytical grade.
|Inaccuracy [Deviation (%)]||Imprecision [CV (%)]|
Mass Spectrometric Analysis
All samples were analyzed by the nano-LC system (Ultimate 3000; Dyonex, Amsterdam, the Netherlands), which was connected to an ESI-triple quadrupole mass spectrometer (QTRAP5500; AB Sciex, Foster City, California). The nano-LC system consisted of a trapping column (Pep Map C18, 5 μm particles, 300 μm inner diameter × 5 mm length) and a separation column (Nano HPLC Capillary Column, 3 μm particles, 100 μm inner diameter; Nikkyo Technos, Tokyo, Japan). Linear gradient was 0%–45% acetonitrile in 0.1% formic acid and was applied to elute the peptides at a flow rate of 200 nL/min. The spectrometer was set up to run MRM analysis for peptide detection with 10 ms dwell time per channel. As shown in Supplemental Table 1, four MRM transitions were set for each target protein and its internal standard peptide due to the possibility that only a single transition might have been hidden by the nontarget components. The ion counts in the chromatograms were determined by data acquisition procedures in Analyst software 1.5 (AB Sciex).
Preparation of Standard and Quality Control/Method Validation Samples
The calibration curve consisted of a blank sample, a zero sample with internal standards, and six nonzero samples spiked with internal standards. With the serial dilutions with 0.1% formic acid, the following standards were prepared: 1, 2, 5, 10, 20, and 50 fmol [CYP1A2, CYP2B6, CYP2C9, CYP2C19, CYP2D6, CYP3A4, CYP3A43, breast cancer resistance protein (BCRP), multidrug resistance protein (MDR)1, organic anion transporting polypeptide (OATP)1B1, OATP1B3, OATP2B1, Na+/taurocholate cotransporting polypeptide (NTCP), and Na+/K+ ATPase] or 2, 4, 10, 20, 40, and 100 fmol (CYP2C8, MRP2, OCT1, UGT1A1, UGT1A1_variant, UGT2B7, and gamma-glutamyl transpeptidase (γ-gtp)]. Quality control (QC) samples for batch acceptance were prepared as 2.5, 10, and 40 fmol or 5, 20, and 80 fmol.
Preparation of Samples in the Presence of Microsomal and Plasma Membrane Fractions for the Matrix Effects and Stability
Frozen human liver tissues were obtained from Human Tissue Bank (Oxford, UK). Tissue pieces of 1 g were dissected and homogenized using Potter–Elvehjem homogenizer in buffer A [0.1 M KCl–phosphate buffer (pH7.4) containing a protease inhibitor cocktail]. The homogenates were centrifuged at 10,800 × g for 20 min at 4°C and the supernatants were collected and ultracentrifuged at 100,000 × g for 60 min at 4°C. The microsomal pellet was suspended in buffer A and ultracentrifuged at 100,000 × g for 60 min at 4°C. The resulting pellet was suspended in buffer B [20 mM Tris–HCl buffer containing 0.25 M saccharose and 5.4 mM ethylenediaminetetraacetic acid (EDTA)] and part of the solution was stored as microsomal fraction at −80°C. The remainings were layered on top of a 38% (w/v) sucrose solution and centrifuged at 100,000 × g for 30 min at 4°C. The turbid layer at the interface was recovered, suspended in buffer B, and centrifuged at 100,000 × g for 30 min at 4°C. Plasma membrane fraction was obtained from the resulting pellet, which was suspended in buffer B. Protein concentration was measured by the Lowry method using the DC protein assay reagent. Protein samples were suspended in suspension buffer containing 7 M guanidine hydrochloride, 10 mM EDTA, and the proteins were S-carbamoylmethylated. The alkylated proteins were precipitated with a mixture of methanol and chloroform. The precipitates were dissolved in 6 M urea, diluted with 100 mM Tris–HCl (pH 8.0), and digested with TPCK (L-(tosylamido-2-phenyl) ethyl chloromethyl ketone)-treated trypsin sequence grade-modified trypsin (Promega, Madison, Wisconsin) at an enzyme/substrate ratio of 1:100 at 37°C for 16 h.
Evaluation of Accuracy and Precision of the Analytical Method
In order to evaluate accuracy and precision of the analytical method, the following parameters were defined and calculated for each analysis.
|Inaccuracy [Deviation (%)]||Imprecision [CV (%)]|
Accuracy and precision with trypsinized microsomal or plasma membrane fractions (Table 3) were calculated as follows:
where MF is membrane fraction (trypsinized microsomal or plasma membrane fraction), QCMF is amount of analytes after spiking standard solution to membrane fraction, matrix is amount of analytes originally contained in membrane fraction, spiked is amount of analytes contained in standard solution, and SD is standard deviation.
|Protein||Amount in Membrane Fractiona (fmol/sample)||Spiked Amount of Peptide (fmol/sample)||Amount in Spiked Sample (fmol/sample)||Calculated Amount of Spiked Peptideb (fmol/sample)||Inaccuracy [Deviation (%)]||Imprecision [CV (%)]|
Stability of peptides in trypsinized microsomal or plasma membrane fraction and in processed samples in the autosampler (Table 4) were calculated as follows:
|Difference from Initial (%)|
|Protein||−20°C for 32 Days||Three Cycles of Freeze–Thaw||10°C for 7 Days|
Linearity of Calibration Curves for Peptide Quantification by LC–MS/MS
Calibration samples, containing seven nonzero standards ranging from 1 to 50 fmol or from 2 to 100 fmol, respectively, were spiked with 40 or 80 fmol of internal standard peptide and analyzed for generation of calibration curves. A weighting factor 1/x was applied. Calibration curves obtained for each target peptide are shown in Supplemental Figure 1. As shown in Table 5, the determination coefficients (r2) of all targets within three runs were greater than 0.986. Therefore, lower limit of quantification (LLOQ) was set 1 fmol (CYP1A2, CYP2B6, CYP2C9, CYP2C19, CYP2D6, CYP3A4, CYP3A43, BCRP, MDR1, OATP1B1, OATP1B3, OATP2B1, NTCP, and Na+/K+ ATPase) or 2 fmol (CYP2C8, MRP2, OCT1, UGT1A1, UGT1A1_variant, UGT2B7, and γ-gtp).
Accuracy and Precision of Quantification in Absence of Trypsinized Microsomal or Plasma Membrane Fraction
Accuracy and precision of quantification were assessed by analyzing standard samples in three different amounts. Percentage of inaccuracy was determined as the percentage of deviations of the quantified amounts from their nominal values, imprecision was determined as the percentage of the coefficient of variation. Intra-assay precision was determined by six technical replicates from a single sample preparation (Table 2). Interassay precision was determined by 10 replicates resulting from three preparations (Table 3). In both intra- and interassay precision determinations, inaccuracy in both high and middle amount samples (HQC and MQC) were below 8.1% among all peptides. Inaccuracy in the low amount sample (LQC) was 20.3% at maximum. Imprecision values were below 13.1% in HQC and MQC, and increased in LQC to 15.1% at maximum.
Accuracy and Precision with Trypsinized Microsomal or Plasma Membrane Fraction
To determine the effect of matrix, samples were prepared by spiking target peptide standards into trypsinized membrane fraction prepared from liver tissues. Microsomal fraction was used as matrix for the analysis of CYP enzymes and UGT enzymes, whereas plasma membrane fraction was used for analysis of drug transporters to reflect differences in matrix composition due to differing sample preparations. Quantification of spiked samples was performed by subtracting the amount found in blank matrix sample from that in the spiked peptides. Inaccuracy and imprecision were determined in three analytical runs. Inaccuracy (% deviation) ranged from −15.3% to 8.1%, and imprecision (% CV) ranged from 3.9% to 18.9%. These values were in a similar range to those obtained for intra- and interassay precision without matrix.
Stability of Peptides in the Membrane Fraction
Stability of peptide analytes in matrix was assessed following three freeze–thaw cycles for 32 days at −20°C. Autosampler stability was determined for 7 days at 10°C. Results are summarized in Table 5. Acceptable stability was shown for all three conditions. At −20°C for 32 days, quantified amounts decreased by 13.5% at most. Following three freeze–thaw cycles, values varied from −3.4% to 7.2%. After 7 days under autosampler conditions, the values of all proteins were found within 19.6%, except for CYP2D6.
Reliability and robustness of a LC–MS/MS-based assay for simultaneous quantification of multiple drug transporters, CYP enzymes and UGT enzymes, were demonstrated in this study. Absolute quantitative data on protein expression of drug metabolizing enzymes and drug transporters clearly are of high value for scientists involved in ADME/Tox research and development. It provides an essential set of information for an accurate interpretation and prediction of the PK profile of new drugs. Although validation studies on these target proteins have been conducted earlier,20,21 the herein reported assessment of a simultaneous assay for multiple drug transporters, CYP enzymes and UGT enzymes, was performed for the first time. The present study comprised essential bioanalytical parameters including linearity, reproducibility, and stability.
According to the current FDA guidance18 on chromatographic bioanalytical assay, inaccuracy and imprecision of a quantitative assay should both be within ±15%, except for the LLOQ (20%). The present study demonstrated that inaccuracy of standard peptide was within ±20.3% in the absence of plasma membrane protein-digested matrix, and imprecision of standard peptide was not greater than 15.9% in the absence of plasma membrane protein-digested matrix. Therefore, the presented method did not meet the FDA criteria for a chromatographic bioanalytical assay.
The criteria for chromatographic assays were initially targeted for quantification of small molecules rather than macromolecules such as proteins or peptides. The latter are unlike more complex with respect to sample preparation and analytical technique. For example, nano-flow LC rather than conventional LC applied for chromatographic separation was required to achieve sufficient analytical sensitivity for quantification of low abundant proteins such as some drug transporters. Because nano-LC–MS/MS has been mainly applied to comprehensive and qualitative global proteomics or relative quantitative proteomics, but not to absolute quantification, respective quality criteria for nano-LC–MS/MS-based quantification assays are not given to date.
In their workshop report, the FDA also referred to criteria for ligand-binding assay such as ELISA.22 Acceptance criteria for inaccuracy and imprecision were defined as within ±20%, except for LLOQ and upper limit of quantification, where within ±25% is considered acceptable. Our assay clearly meets these criteria for ligand-binding assays and is therefore fully applicable from the view point of a protein assay. Inaccuracy and imprecision in presence of matrix were assessed and likewise met the criteria.
Sample stability of protein analytes usually is a critical parameter affecting the quality of an analytical method. Even under freezer (−20°C) conditions, proteins are prone to enzymatic degradation. The present study showed acceptable peptide analyte stability after tryptic digest for 30 days at −20°C and for 7 days under autosampler conditions at 10°C, which is an important requirement for most analytical laboratories within the field of pharmaceutical research and development. Our results also suggest that samples are more stable once treated by trypsin digestion as compared with the nondigested protein samples.
Conventional ELISA methods still represent the “gold standard” for protein quantification in a standardized and large-scale setting such as clinical studies. Sample throughput, assay run time, and sensitivity are the main advantages compared with MS-based assays. On the contrary, the increasing demand for multiplexed analyses at preclinical (e.g., ADME/Tox) and early clinical stages, the relatively short assay development times, and its applicability to virtually any target protein justify the use of MS in the bioanalysis of large molecules and will further promote its application to pharmaceutical research and development in the future.