Article pubs.acs.org/ac
Online 2D-LC-MS/MS Assay To Quantify Therapeutic Protein in Human Serum in the Presence of Pre-existing Antidrug Antibodies Yinghua Shen,* Guodong Zhang, Jinsong Yang, Yongchang Qiu, Thomas McCauley, Luying Pan, and Jiang Wu* Bioanalytical and Biomarker Development, Research and Nonclinical Development, Shire, Lexington, Massachusetts 02421, United States ABSTRACT: The formation of antidrug antibodies (ADA) can interfere with the accurate quantitation of therapeutic proteins, leading to significantly underestimated drug concentrations and confounded pharmacokinetic (PK) data interpretation. Although highly desirable, development of ADAtolerant bioanalytical methods enabling unbiased measurement of both free and ADA-bound drug presents a considerable challenge. We report herein the development and validation of a robust LC-MS assay capable of quantifying therapeutic protein immunoglobulin A1 protease (IgAP) in human serum in the presence of pre-existing anti-IgAP antibodies. The procedure included sodium dodecyl sulfate (SDS) denaturation and chemical reduction of serum proteins to dissociate ADA-drug bindings, followed by tryptic digestion of protein pellets and subsequent LC-MS analysis of the surrogate IgAP peptide using stable isotope labeled peptide internal standard. Substantial enhancements in the sensitivity and selectivity were achieved by the combination of online two-dimensional reversed-phase LC (2D-LC) operated in high and low pH buffers, respectively, for efficient enrichment and quantitation of the surrogate peptide by multiple-reaction monitoring (MRM) mass spectrometry. Unlike ligand-binding assay, our method is not prone to interferences from ADA, allowing accurate and precise measurement of the IgAP in the range of 0.05 to 10 μg/mL in 25 μL of human serum with a wide range of anti-IgAP antibody levels. The intra- and inter-run precision (coefficient of variation (CV%)) was within 11.5% and 10.5%, respectively, and the bias was within ±7.1% for all quality control (QC) concentrations. With little modification, the described method can readily be applicable to the quantitation of other biotherapeutic proteins in the ADA-positive clinical matrices.
T
likely due to the prior exposure to the therapeutic or structurally similar agents.5,6 The ADA response has a hostspecific polyclonal nature, with characteristics of diverse ADA concentrations, affinity, isotypes, and binding epitopes. As a result, the ADA can modulate and neutralize the therapeutic effects and potentially impact dose-exposure relation, bioavailability, pharmacokinetics (PK), pharmacodynamics (PD), and ultimately efficacy and safety profiles of the drug. To that end, risk assessment of immunogenicity is required by regulatory agencies to address major safety concerns during drug development and postmarketing surveillance.8 While a number of recent publications and regulatory documents offer valuable insights and guidelines to assess the presence and potential risks of the ADA, there is a lack of adequate bioanalytical methods to measure how the ADA would impact the preclinical and clinical PK profiles.5,6,9 As for all drugs, a reliable bioanalytical method possessing high selectivity, accuracy, and precision is essential for
he advancement of biotechnology during the past two decades has led to a steady increase of protein drugs that offer high therapeutic potential for a range of diseases. The pipeline and modalities of therapeutic protein candidates under preclinical and clinical development continue to rapidly grow in the pharmaceutical industry. These include recombinant monoclonal antibodies (mAbs), unmodified proteins, fusion proteins, and cytotoxins conjugated to antibodies.1−3 Despite the diversity of molecular structure, all biotherapeutics are designed to functionally modulate specific endogenous targets involved in signaling pathways, to replace dysfunctional enzymes, or to function as agents delivering cytotoxins to cells, in order to control or stop disease progression. In spite of the numerous advantages such as high specificity and low toxicity, administration of a biotherapeutic can induce undesirable immune responses resulting in the formation of antidrug antibodies (ADA) in both preclinical animals and clinical subjects.4−6 Generation of the immunogenicity depends on multiple factors such as nature of the protein therapeutic, its relationship to endogenous protein, patient population, treatment regimen, formulation, etc.7 In addition, untreated animals and human subjects sometimes exhibit cross-reactivity with a therapeutic protein implying the presence of pre-existing ADA, © XXXX American Chemical Society
Received: June 17, 2015 Accepted: August 3, 2015
A
DOI: 10.1021/acs.analchem.5b02293 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
to limited case studies such as PEGylated antibody where the specific biotherapeutic proteins can readily be isolated from other serum proteins and nonclinical PK studies where assay sensitivity does not present a significant issue.11,35 There is a substantial demand to implement a highly sensitive and generally applicable bioanalytical platform to overcome the issue. We present herein a sensitive and robust online twodimensional reversed-phase LC-based mass spectrometric assay (2D-LC-MS/MS) to accurately measure IgAP in human serum in the presence of pre-existing anti-IgAP antibodies. Conditions were optimized for sample preparation, digestion, selection of surrogate peptides, and online peptide fractionation to achieve reproducible sample preparation and enhance enrichment of the surrogate peptide. Implementation of the online 2D-LC utilizing reversed-phase columns operated at high pH (pH 8) and low pH (pH 3), respectively, exhibits remarkable selectivity to enrich surrogate peptide(s), leading to the improvement of the detection limit by 40-fold (lower limit of quantitation, LLOQ: 0.05 μg/mL). The assay performance has been validated following the regulatory guideline in order to support clinical sample analysis.
quantitative measurement of the therapeutic protein concentration in plasma or serum in support of preclinical and clinical PK studies. Ligand-binding assays (LBA) or immunoassays are currently the preferred method of choice owing to their ease of use, high sensitivity, and analytical throughput. However, accurate measurement of protein concentration by LBA in the presence of ADA in the matrix typically requires two different antibodies that bind to discrete epitopes of therapeutics that are distinctive from the ADA epitopes. Because of the host-specific and highly heterogenic nature of ADA responses, it would be challenging, if not impossible, to find adequate antibody pairs for this purpose. Data from our and other laboratories demonstrated that the presence of the ADA interferes in LBA-based PK measurement, leading to significantly underestimated drug concentrations depending on the ADA levels and choice of LBA reagents and assay formats.10 There are instances where the LBA fails to quantify concentrations of therapeutic proteins administrated at low doses due to competitive ligand binding by the ADA.11 Immunoglobulin A nephropathy (IgAN) is a chronic kidney disease characterized by predominant deposition of IgA1containing immune complexes in glomerular mesangium, which triggers local inflammatory responses and glomerulonephritis. 12−15 Precise mechanisms leading to the mesangial deposition of IgA1 complexes are unknown. Immunoglobulin A1 protease (IgAP) is a proteolytic enzyme produced by several species of pathogenic bacteria such as H. Inf luenzae. IgAP specifically cleaves the hinge region of human IgA1, thereby removing the deposition of mesangial IgA1 complexes in glomeruli. This remarkable characteristic makes IgAP a promising therapeutic agent for the treatment of IGAN.15 Despite this attractive feature, pre-existing immunogenicity due to prior exposure to the bacteria was found previously16,17 and in our own study of a number of individual sera. The presence of these ADA represents a major analytical challenge to accurately measure IgAP concentration in clinical studies by the LBA. Propelled by the advances in the proteomic techniques, liquid chromatography−tandem mass spectrometry (LC−MS/ MS) has become a powerful technique to quantify endogenous biomarkers in complex biological matrices such as serum and plasma.18−21 The LC-MS/MS approach quantifies proteins by targeted monitoring of the appropriate surrogate peptides originating from the proteins using multiple-reaction monitoring (MRM) mode.22−28 A great deal of effort has been made to improve its specificity and sensitivity by reducing sample complexity using immunoaffinity capture or multidimensional chromatographic fractionation at protein and/or peptide levels.29,30 In recent years, the LC-MS/MS assays have proliferated to absolute quantitation of therapeutic proteins to support nonclinical PK/TK studies where drug doses typically exceed mg/kg and circulating drug concentrations are mostly above the μg/mL range.11,22−28,31−33 Measurement of clinical PK profiles of therapeutic proteins at ng/mL concentrations needs considerable efforts to enrich low abundance of surrogate peptides from the vast amount of other peptides in the matrix and has been demonstrated by several groups.11,26,27,34,35 In contrast to the LBA that detects the protein analyte specifically bound to reagent antibodies in the matrix (free drug), the LC-MS/MS, in principle, is able to measure surrogate peptides derived from the total protein (free and ADA-bound drug), thereby circumventing the pitfalls of the LBA.11 However, the method so far has only been applied
■
MATERIALS AND METHODS Chemicals and Reagents. Recombinant IgAP (protein accession number: P44969; molecular weight ∼109 kDa) expressed in E. coli was obtained from Shire. It is the mature form (protease domain spanning amino acid 26 to 1014) of serine type IgA1 proteases from H. inf luenza. The protein was obtained as a stock solution in PBS buffer (1.5 mg/mL), and the purity of the protein is higher than 95%. Mouse anti-IgAP mAb 8-1 and HRP-rabbit anti-IgAP pAb reagents were developed in house. TCPK treated trypsin from bovine pancreas, tris(2-carboxyethyl)phosphine hydrochloride (TCEP), iodoacetamide (IAA), ammonium bicarbonate (NH4HCO3), urea, formic acid, HPLC grade methanol (CH3OH), and phosphate-buffered saline (PBS) were all purchased from Sigma-Aldrich (St. Louis, MO, USA). Mass spec grade trypsin/lys-C mixture was a product of Promega (Madison, WI, USA). The uHLB 96-well plate was obtained from Waters (Milford, MA, USA). The unlabeled surrogate peptide, FSVGATNVEVR, and its 13C-labeled internal standard (IS), FSVGATNVEV[13C5]R[13C6], were custom-synthesized at New England Peptide (Gardner, MA, USA). All the control human serum was purchased from Bioreclamation Inc. (Westbury, NY). 2 mL of 96-well SiliGuard glass coated plates were obtained from Analytical Sales and Services, Inc. (Pompton Plains, NJ). Surrogate Peptide Selection. To choose the appropriate surrogate peptide for IgAP quantification, the recombinant IgAP was denatured in 8 M urea, reduced with 5 mM TCEP at 60 °C for 30 min, alkylated with 5 mM IAA, and subsequently digested overnight at 37 °C by trypsin/lys-C mixture (1:50 E/ S). The resulting peptide mixture was analyzed on an Orbitrap Velos mass spectrometer (Thermo Fisher, Waltham, MA, USA). The instrument was operated in the data-dependent acquisition mode to collect MS and MS/MS spectra of the peptides with ion intensities exceeding the predefined threshold. The mass spectra were searched against the customer-built IgAP sequence database using Mascot search engine (Matrix Sciences, London, UK) to identify peptides and calculate their relative ion intensities. B
DOI: 10.1021/acs.analchem.5b02293 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry Preparation of Calibration Standards, IS, and Quality Control (QC) Samples. Calibration standards (0.05, 0.10, 0.20, 0.50, 1.0, 2.0, 5.0, and 10.0 μg/mL) and QCs (0.05 μg/ mL (LLOQ), 0.15 μg/mL (LQC), 2.0 μg/mL (MQC), and 8.0 μg/mL (HQC)) were prepared by an intermediate dilution of the IgAP stock solution into normal human serum (pool of 25 female and male) followed by independent serial dilution with two different lots of pooled human serum to obtain the desired final concentrations, respectively. The calibration standards were prepared in duplicates, whereas QC samples were prepared in six replicates. To examine the effect of anti-IgAP antibody titer levels on the accuracy of the assays, 18 individual human serum samples with different anti-IgAP levels were spiked with IgAP stock solution to final concentrations of 0.15 μg/mL (LQC), 2.0 μg/mL (MQC), and 8.0 μg/mL (HQC), respectively. The IS stock solution was prepared in CH3OH/ H2O (1:1, v/v) at a concentration of 1 nmol/mL, which is equivalent to a nominal concentration of approximately 109 μg/mL IgAP. Sample Preparation. Aliquots of 25 μL serum samples (calibration standards, QCs, blanks, or study samples) were transferred into a low-binding 96-well glass coated plate (2 mL capacity). The proteins were denatured by adding 2.5 μL of 10% sodium dodecyl sulfate (SDS), followed by vigorous vortexing for 5 min on an Eppendorf Thermomixer. Protein reduction was carried out by adding 5.0 μL of 500 mM TECP and subsequent incubation for 50 min at 60 °C in a preheated thermomixer at 600 rpm. Total serum proteins were precipitated by adding 1.0 mL of CH3OH to each well followed by vigorous vortexing for 5 min. After centrifugation at 2000g for 10 min, the methanol supernatant (0.9 mL) was removed and discarded by Tomtec liquid handler, and the plate was placed in a nitrogen blower to remove the remaining methanol. The protein pellets were resuspended in 500 μL of 100 mM NH4HCO3 digestion buffer containing 5 mM of IAA and 0.115 pmoL of IS (equivalent to 0.5 μg/mL of IgAP) by vigorous vortexing in the dark for 5 min and further incubation in the dark for 35 min at 25 °C, 600 rpm to promote alkylation. To the protein suspension was added 30 μL of 8.0 mg/mL freshly prepared trypsin solution in 100 mM NH4HCO3. The proteolytic digestion was carried out at 37 °C, 600 rpm for overnight in the dark. The digested samples were acidified by 50 μL of 10% formic acid, followed by vigorous vortexing for 5 min. After centrifugation at 4000g for 10 min, 400 μL of the supernatants was transferred to a Waters uHLB plate (2 mg capacity) preconditioned sequentially with 1 mL of CH3CN and 1 mL of 0.1% formic acid (vacuum was applied to remove solvent completely). After washing with 500 μL of 0.1% formic acid, the peptides were eluted by 100 μL of 80:20 (v/v) CH3CN/H2O and collected in a clean 96-well collection plate. The samples were dried in a speedvac and reconstituted in 80 μL of 25 mM NH4HCO3 solution containing 10% CH3CN prior to LC-MS/MS analysis. Enrichment of Surrogate Peptide by Two-Dimensional Reversed-Phase LC. An automated two-dimensional reversed-phase LC system (2D-LC) operated at high and low pH, respectively, was employed for enhanced separation of the surrogate peptide from matrix interference. The 2D-LC configuration (Figure 1) consisted of a Waters UPLC as the first dimension (Milford, MA) connected to a Shimadzu Nexera UHPLC as the second dimension (Columbia, MD, USA) through a six-port, two-position valve (Valco Inc., Austin, TX).
Figure 1. Schematic description of an automated standard flow 2D-LC platform and the workflow: (A) (valve at position 1) The auto sampler loads the sample, and the peptides are first separated in high-pH eluents on the BEH C18 column, while the second dimension low-pH eluents are delivered to condition the XBridge C8 column; (B) (valve at position 2) the valve connects the BEH C18 column with the XBridge C8 column; the fraction containing the surrogate peptide and IS was eluted from the C18 column and trapped on the C8 column; (C) (valve at position 1) low-pH gradient is delivered to elute the surrogate peptides from the C8 column for MS/MS analysis.
The peptide sample (10 μL) was loaded onto an Acquity UPLC BEH C18 column (1.7 μm, 2.1 × 100 mm, Waters Corporation, Milford, MA, USA) maintained at 55 °C, and chromatographic separation was performed in high pH mobile phase at a flow rate of 250 μL/min using the following gradients: 0.0−0.2 min, 10% B; 0.2−12 min, 10−13% B; 12− 12.1 min, 13−80% B; 12.1−16.1 min, 80% B; 16.1−16.5 min, 80−10% B; 16.5−20 min, 10% B, where mobile phases A and B were 25 mM aqueous NH4HCO3 (pH 8.0) and CH3CN, respectively. Column switch was achieved by the six-port valve, which was placed initially in position 1 during the sample injection and gradient elution of the first LC at which the column eluent was diverted to waste (Figure 1). The LC fraction containing the surrogate peptide and IS, eluting between 10.1 and 10.6 min from the first column, was directed to the second column by switching the valve to position 2 at 10.1 min. Shortly after complete elution of the analyte from the first column and transfer to the second column, the valve was switched back to position 1 at 10.6 min, and gradient elution of the trapped peptides for the second column began. The low pH separation was achieved on an Acquity UPLC XBridge C8 column operated at 300 μL/min and maintained at 55 °C (2.5 μm, 2.1 × 100 mm, Waters Corporation, Milford, MA, USA). Mobile phases consisted of 0.5% formic acid (A) and 0.5% formic acid in CH3CN (B). The column gradient of the second dimensional LC was programed synchronically with the first dimensional separation as follows: 0.0−11.1 min, 2% B; 11.1− 17 min, 2−30% B; 17−17.1 min, 30−80% B; 17.1−19.4 min, 80% B; 19.4−19.5 min, 2% B; 20 min, stop. The total analytical time for the entire 2D-LC run was 20 min. MRM Data Acquisition and Processing. The 2D-LC system was coupled to an AB Sciex API 5500 Qtrap mass spectrometer equipped with a TurboIonspray source (Concord, Ontario, Canada) for peptide quantitation. The column eluent was ionized in the positive ionization and analyzed by the Qtrap analyzers in the MRM mode. Mass spectrometer parameters C
DOI: 10.1021/acs.analchem.5b02293 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
was spiked into the human serum samples to mimic the real study samples. It was thus selected as a surrogate for the quantitation of IgAP. The peptide sequence was further blasted against the protein database to ensure it has no overlap with the hypothetical tryptic peptides derived from the human proteins. A heavy peptide IS containing 13C-uniformly labeled valine and arginine at C-terminus was synthesized to eliminate the potential contribution of the impurity in the singly labeled IS to surrogate peptide MRM channel. The specificity of the surrogate peptide and their transition ions were further examined by running pooled human serum samples with and without the spiked IgAP following the experimental procedures to ensure the lack of interference peaks. Finally, the MRM parameters were optimized to further maximize the sensitivity of the peptide. Tryptic Digestion with/without SDS and TECP Treatment. Efficient and reproducible digestion by trypsin is a critical step to ensure the accuracy and precision of the assay. We aimed to develop a general serum pretreatment method that can be applied, with little modification, to multiple protein therapeutics. The effect of SDS (for protein denaturation) and TECP (for protein reduction) on digestion yield was investigated. For this purpose, QC samples were spiked with 2.0 and 10.0 μg/mL of IgAP, respectively. The serum proteins were precipitated by CH 3 OH with or without prior denaturation/reduction. The precipitated protein samples thus obtained were processed in parallel and loaded to the 2D-LC-MS/MS system. As justified from the relative peak areas of the surrogate peptide/IS, the recoveries of IgAP from precipitated serum samples without prior protein denaturation and reduction were ∼46% (at 2 μg/mL) and ∼10% (at 10 μg/ mL) lower compared to denatured/reduced protein samples (data not shown). It is predicted that the decline in digestion yield would be more substantial at lower IgAP concentration. Obviously, the prior protein denaturation and reduction facilitated the dissociation of IgAP from ADA and promoted the tryptic digestion efficiency of the IgAP by increased exposure to the enzyme. Different types of denature agents were explored such as 8 M urea, 0.1% RapiGest SF, heating at boiling, etc., and SDS was selected in our final protocol as it effectively denatured all proteins, was cost-effective, and was compatible with the downstream protein precipitation and other subsequent steps. We found vigorous vortexing was essential to ensure protein denaturation and reconstitute the protein pellets into the uniform protein slurry for efficient tryptic digestion and improved assay accuracy and precision. With the current setup, the time course study to monitor the release and digestion efficiency of the tryptic peptides in the serum mixture is hard, because the incomplete digestion of serum proteins tended to change the chromatographic profile of the 2D-LC separation. In our experiments, overnight digestion at 37 °C with a trypsin/protein ratio of 1:10 gave rise to reproducible and consistent results and was utilized throughout the experiments. Recently, pellet digestion of serum proteins by organic solvents has been described in a number of papers with great success.26,27,36−38 A variety of different conditions have been reported to shorten the digestion time of protein pellets yet not to compromise the digestion efficiency.26,27,36−38 The reported conditions included higher temperature, excess amount of trypsin, and/or elimination of denature/reduction/alkylation steps, aiming to maximize throughput of the assay. We found
employed for the detection of surrogate peptide and its IS were as follows: ionspray voltage (IS) +5000 V, temperature 600 °C, nebulizer gas (GS1) 55 units, TurboIonSpray gas (GS2) 55 units, collision-activated dissociation gas setting medium, curtain gas 25, dwell time 100 ms, and entrance potential 10 V; declustering potential +100 V, collision energy +29 V, and collision exit potential +14 V. The doubly charged precursor ions at m/z 589.8 and 595.4 were selected for collision-induced fragmentation, and their respective product ions at m/z values of 717.6 and 728.5 were monitored for the surrogate peptide and its IS, respectively. Analyst software (Version 1.5.2) was used for raw MRM data acquisition and chromatogram processing, as well as data regression using peak area ratios of the surrogate peptide to the IS. Calibration curves were constructed using peak area ratios of the calibration standards by applying a linear, 1/x weighted, least-squares regression algorithm. All calibration standards and QCs were then back calculated from their peak area ratios against the calibration line. Mean accuracy and precision statistics for QC samples were calculated using Microsoft Excel. Determination of Anti-IgAP Antibodies in Serum. Anti-IgAP antibodies in human serum were determined by a bridging electrochemiluminescent immunoassay using the Meso Scale Discovery (MSD) technology platform (Meso Scale Diagnostics, LLC, Rockville, MD). Immobilized IgAP was used as the capture reagent, and sulfo-tagged IgAP was used as the detection reagent. The anti-IgAP antibody levels were expressed as mean ECL values. Quantification of IgAP in Human Serum by LBA Assay. IgAP in 18 individual human serum samples with a variety of different anti-IgAP antibody levels was measured by an enzyme-linked immunosorbent assay (ELISA) using paired anti-IgA protease Abs (mouse anti-IgAP mAb 8-1, HRP-rabbit anti-IgAP pAb), with the concentrations back calculated from the IgAP calibration curve in the same assay. Assay Validation in Human Serum with and without Anti-IgAP Antibodies. The performance of the LC-MS/MS assay was validated in accordance with current regulatory guidelines by assessing precision, accuracy, selectivity, and matrix effect using various QC samples prepared in pooled human serum. In addition, the same individual human serum samples used for the LBA assay were subjected to analysis by the 2D-LC-MS/MS. The results obtained from the LBA and 2D-LC-MS/MS were compared to evaluate the effect of antiIgAP antibody level on accurate determination of IgAP concentration.
■
RESULTS AND DISCUSSION Selection of Surrogate Peptide. IgAP contains 995 amino acids and a molecular weight of 109 kDa; two cysteine residues are linked to form an intradisulfide bond. Tryptic digestion of the IgAP solution under the denatured conditions identified 64 peptides ranging from 750 to 2700 Da; most of the peptides were from specific tryptic cleavage at both sides. Four peptides with desired sequence specificity and reasonable size were selected for further method development: FSVGATNVEVR and IATLINPQYVVGVK from the N-terminus and GTLIVEGK and VGDGTVILK from middle of the sequence. All peptides demonstrated the complete digestion pattern at both ends with minimal nonspecific digestion product, and most importantly, all possessed the excellent MS responses. The peptide, FSVGATNVEVR, however, maintained the superior performance with little optimization when the IgAP D
DOI: 10.1021/acs.analchem.5b02293 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 2. (a) MRM ion chromatogram of the spike-in surrogate peptide in serum tryptic peptides (equivalent to 2.0 μg/mL IgAP) separated by the first dimension RPLC system; (b) MRM ion chromatogram of the surrogate peptide at the same concentration following separation by the 2D-LC system (the highlighted fraction in (a) was transferred to the second RP column); (c) MRM ion chromatogram of the spike-in surrogate peptide at 0.05 μg/mL following separation by the 2D-LC system; (d) surrogate peptide MRM ion chromatogram of the blank matrix following separation by the 2D-LC system.
complete denaturation followed by reduction is essential for IgAP quantitation. Without denaturation and protein reduction, the accuracy of the IgAP was lower at 2 μg/mL (46%) and would be significantly lower at lower concentrations (data not shown). The discrepancy between our observation and published data was likely due to the specific property of the target protein. Indeed, some proteins may be prone, but others may be more resistant, to proteolytic digestion without disrupting the tertiary structure of the protein.26,27,38 One unique advantage about our protocol is that the SDS, reducing agent, together with salts, phospholipids, and other soluble species can be easily removed from the proteins as they remain in solution after protein precipitation. Immunocapture of IgAP To Improve Assay Sensitivity. In order to support the human PK study, an assay with sensitivity of low ng/mL is usually needed. Human serum is perhaps the most complex matrix. It contains thousands of serum proteins with a huge dynamic range of expression. Upon tryptic digestion, the serum proteins were chopped into hundreds of thousands of peptides, many of which would coelute and interfere with the analyte measurement. We explored a number of different approaches attempting to simplify the protein mixture and thus improve the sensitivities of the LC-MS/MS assay. The simple approach was to enrich the target protein by immunocapture extraction using the antiIgAP antibodies to reduce the sample complexity. This approach yielded lower recovery for study samples with ADA, as the endogenous ADA competed with the immunocapture antibodies and significantly interfered with the accurate measurement. Another promising approach was immunocapture by protein G/A, which captures different subtypes of IgG and indirectly pulls down the ADA-bound IgAP. To capture free IgAP in the serum, we spiked an excess amount of anti-
IgAP antibodies in the test samples as bait. We found however the accuracy of this approach was typically low and unreproducible, presumably because the IgAP protein might bind to other antibodies (e.g., such as IgA, IgM, IgE) or proteins, which compromised its affinity binding to protein A/ G. Performance of the 2D-LC Platform. The utilization of multiple-dimensional separations at protein and/or peptide levels enables much improved proteome coverage and has been routinely used in proteomics-based drug target and biomarker discovery. A variety of “orthogonal” separation schemes have readily been available such as online or offline strong-cation exchange (SCX), anion exchange (SAX), reversed-phase (RP), and hydrophilic interaction (HILIC) in addition to reversed phase (RP) as a second dimension.39−44 In contrast to global proteomic identification aiming to identify as many peptides as possible, the targeted quantitation centers on quantitation of surrogate peptide only. We found the reversed-phase separations at high and low pH were ideal for this purpose as they offered the highest resolving power in the first dimension and the easiest conjunction with the second dimension. After desalting with the uHLB plate, the peptide mixture was first fractionated by online RP-LC at high pH (pH 8). A very shallow gradient elution (10−13% B over 11.8 min) was applied, which proved sufficiently selective to enrich the analyte from other peptides released by trypsin digestion of the serum proteome. Next, the very narrow fraction (0.5 min) containing the analyte was transferred to the second column, which was operated in acidic pH (3.0) to offer further separation of the trapped peptides. Instead of eluting peptides from the first column onto a trap column as typically utilized previously, our system eliminated the use of trap column and the peptides were directly transferred to the analytical column. We concluded that E
DOI: 10.1021/acs.analchem.5b02293 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
expressed as bias% from the respective nominal concentration and coefficient of variation (CV%) of the multiple measurements at each concentration level, respectively, should be within the acceptance limits of 20% (25% at the LLOQ). For each run to be accepted, the calculated concentrations for 75% of the calibration standards and 67% of the QC samples must meet the acceptance criteria. All calibration standards passed the above acceptance criteria. Four QC levels (LLOQ, LQC, MQC, and HQC), each in six replicates, were evaluated in three consecutive runs to assess intra-run and inter-run accuracy and precision. As demonstrated in Table 1, all three validation runs passed the acceptance criteria for accuracy and precision. The overall accuracy (bias%), intraday precision (CV %), and interday precision (CV%) were 7.1%, 11.8%, and 10.5%, respectively. The excellent accuracy and precision implied that consistent sample processing and digestion efficiency were achieved across samples of different concentrations, including LLOQ samples. Selectivity. From the double blank samples, there were no significant interferences from human serum for either surrogate peptide or IS detection, suggesting the assay was selective (data not shown). Furthermore, the use of two isotope-labeled amino acids as IS eliminated possible interference to the surrogate peptide MRM channel. Most noticeably, the presence of ADA in the serum did not interfere with the assay or its LLOQ, as demonstrated below in the accuracy and precision study in serum samples with different anti-IgAP antibody levels. Stabilities of IgAP in Serum and Processed Samples. Stabilities of the IgAP in human serum and the processed peptide samples were evaluated using six replicates of LQC and HQC samples. The frozen QCs (−80 °C) were subjected to three freeze−thaw cycles, as well as exposure for 6 h at room temperature. The stressed QC samples were processed and IgAP quantified against freshly prepared calibration curves. As shown in Table 2, the bias (%) of the stability test results ranged from −3.3% to 10.6%, which were well within the 20% allowed bias criterion. Therefore, the stabilities for these stress conditions were established. Comparison of the LBA and 2D-LC-MS/MS for ADAPositive Serum Samples. A bridging immunoassay using the MSD platform was utilized to screen pre-existing anti-IgAP antibodies in normal human serum. The results indicated that the majority of the healthy humans were anti-IgAP antibody positive, most likely due to prior bacterial infections (data not shown). We randomly selected 18 human serum samples with different levels of anti-IgAP antibodies to evaluate the effect of the two analytical platforms (LBA and 2D-LC-MS/MS) on IgAP quantification. For this purpose, each sample was spiked with IgAP stock solution to final concentrations of LQC (0.15 μg/mL), MQC (2.0 μg/mL), and HQC (8.0 μg/mL), respectively, and was subjected to analysis by both LBA and 2D-LC-MS/MS platforms. The accuracy of the IgAP
the simplified configuration at least partially attributed to the superb reproducibility of the assay. In order to estimate the gain in sensitivity provided by the 2D-LC separation compared to the traditional 1D-LC separation, the same samples were analyzed by both methods (Figure 2). In 1D-LC, the ion chromatogram of the spike-in surrogate peptide suffered from high background and interference peaks from other peptides with the same precursor and transition masses (Figure 2a). The analyte peak (RT ∼ 10.4) was buried in the background and barely discernible. The much higher noise limited the LLOQ of the detection to ∼2 μg/mL. In comparison, Figure 2b clearly shows the significant background reduction of the analyte ion chromatogram, after the highlighted fraction in Figure 2a was subjected to the second dimensional separation. Noticeably, baseline separation from the interfering peptides and decent signal-to-noise ratio were achieved, leading to reduction in matrix effect. Figure 2c displays an ion chromatogram of the analyte at LLOQ (0.05 μg/mL, signal-to-noise ratio ∼10) using this system, whereas Figure 2d demonstrates the background signal from the blank matrix sample is minimal. Overall, our system offered more than 40-fold enhancement in the sensitivity. Assay Validation. Calibration Curve. The calibration curve exhibited a linear range from 0.05 to 10 μg/mL in human serum (Figure 3). A linear regression model with 1/x weighting
Figure 3. Representative calibration curve prepared using pooled human serum.
was selected as the best fit. To avoid the potential carryover of the analyte, the higher concentration was not pursued. The assay sensitivity should be able to adequately cover the real study sample concentrations in human PK study. Accuracy and Precision. Data from three consecutive validation runs were used to assess the accuracy and precision of calibration curves and QCs. The accuracy and precision,
Table 1. Intra-run (n = 6) and Inter-run (n = 18) Precision and Accuracy of the 2D-LC-MS/MS Assay in Human Serum intra-run (n = 6) IgAP QC
nominal conc (μg/mL)
LLOQ LQC MQC HQC
0.050 0.15 2.0 8.0
inter-run (n = 18)
measured conc ± SD (μg/mL)
bias (%)
CV (%)
± ± ± ±
−4.5 0.9 0.4 7.1
8.9 5.9 11.8 4.9
0.0478 0.151 2.01 8.57
0.0043 0.0090 0.24 0.42
F
measured conc ± SD (μg/mL)
bias (%)
CV (%)
± ± ± ±
1.2 3.6 0.1 3.9
10.5 8.1 10.3 6.7
0.0506 0.155 2.00 8.31
0.0053 0.013 0.21 0.55
DOI: 10.1021/acs.analchem.5b02293 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry Table 2. Stability of IgAP in Human Serum by 2D-LC-MS/MS Assay LQC (nominal conc = 0.15 μg/mL)
HQC (nominal conc = 8.0 μg/mL)
stability
measured conc ± SD (μg/mL)
bias (%)
CV (%)
measured conc ± SD (μg/mL)
bias (%)
CV (%)
freeze−thaw bench-top processed sample
0.159 ± 0.015 0.152 ± 0.010 0.152 ± 0.023
5.7 1.1 1.4
9.3 6.6 14.8
7.74 ± 0.57 7.95 ± 0.58 8.85 ± 0.40
−3.3 −0.7 10.6
7.4 7.3 4.5
Table 3. Accuracy of the IgAP Quantification in 18 Individual Human Serum Samples with Different Anti-IgAP Antibody Levels as Measured by the LBA and 2D-LC-MS/MS Assays LQC (norm. conc = 0.15 μg/mL) LBA sample lot #
ADA level (ECL value)
BRH863302 BRH863305 BRH863307 BRH863291 BRH863311 BRH863279 BRH863289 BRH863290 BRH863299 BRH863282 BRH863284 BRH863298 BRH863286 BRH863313 BRH863306 BRH863294 BRH863280 BRH863293
526 1490 2688 3317 3990 5074 5740 5797 6285 6855 8518 9436 10 546 12 336 14 340 16 885 20 782 30 399
MQC (norm. conc = 2.0 μg/mL)
2D-LC-MS/MS
mean
accuracy (%)
0.0648 0.0629 BLQ BLQ BLQ BLQ BLQ BLQ BLQ BLQ BLQ BLQ BLQ BLQ BLQ BLQ BLQ BLQ
43.2 41.9 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
mean
accuracy (%)
0.140 0.171 0.162 0.179 0.142 0.158 0.139 0.173 0.145 0.169 0.143 0.153 0.170 0.142 0.164 0.168 0.144 0.148
93.3 113.7 107.7 119.3 94.3 105.3 92.3 115.3 96.3 112.7 95.3 101.7 113.3 94.7 109.3 111.7 95.7 98.7
LBA
HQC (norm. conc = 8.0 μg/mL)
2D-LC-MS/MS
mean
accuracy (%)
1.814 1.665 0.472 0.529 0.621 0.220 BLQ BLQ 0.668 0.224 BLQ 0.075 BLQ 0.051 BLQ 0.209 BLQ BLQ
90.7 83.2 23.6 26.4 31.1 11.0 0 0 33.4 11.2 0 3.7 0 2.6 0 10.4 0 0
mean
accuracy (%)
1.83 2.21 1.99 2.19 2.16 1.87 1.96 1.88 2.02 2.04 2.10 1.92 1.89 1.90 2.02 2.26 1.88 1.67
91.5 110.5 99.2 109.5 108.0 93.5 97.7 93.7 101.0 101.8 105.0 95.7 94.2 94.7 100.8 113.0 93.7 83.2
LBA
2D-LC-MS/MS
mean
accuracy (%)
mean
accuracy (%)
8.390 8.498 6.065 6.884 6.336 5.996 4.455 4.118 3.717 5.653 4.878 3.314 5.073 3.298 0.997 2.768 0.319 0.039
104.9 106.2 75.8 86.0 79.2 75.0 55.7 51.5 46.5 70.7 61.0 41.4 63.4 41.2 12.5 34.6 4.0 0.5
7.190 8.210 8.460 8.690 8.380 7.000 7.515 8.315 8.115 8.415 8.375 8.590 7.630 8.440 8.435 9.340 9.110 7.130
89.9 102.6 105.8 108.6 104.8 87.5 93.9 103.9 101.4 105.2 104.7 107.4 95.4 105.5 105.4 116.8 113.9 89.1
levels as well as spike-in concentration of the IgAP. As the ADA level increased, accuracy of the measurement deteriorated. For HQC (8.0 μg/mL), 3 out of 18 samples exhibited the acceptable accuracy (bias% ≤ 20%), but the sample with the highest ADA levels (ECL 30399) only gave 0.5% of the nominal concentration. In contrast, only 2 MQC samples with the lowest ADA levels met the acceptance criteria, whereas no LQC sample met the same criteria. In fact, the measured concentration of all the LQC samples was below 50% of the nominal concentration value. As the ADA levels further elevated, the IgAP in LQC samples was no longer detectable by the LBA (16/18 samples were BLQ) (Table 3). It is worthy to mention that, due to the heterogeneous nature of the ADA, there seems to be no linear reverse correlation between recoveries and ADA levels, and it is not always possible to predict how the ADA levels would interfere with the LBA assay. Nevertheless, it is evident from the current data that the presence of ADA affects more significantly the accurate measurement of proteins at lower concentrations. The same QC samples evaluated by the LBA were subjected to analysis by the 2D-LC-MS/MS. The serum proteins were extracted in duplicate following the described procedure, and IgAP concentration in each sample was quantified against the calibration curve. As summarized in Table 3, all the samples met the assay acceptance criteria and there is no noticeable bias against serum samples with the high ADA levels. Figure 5 demonstrates that the bias% between the measured and nominal concentrations is within ±20% for all the LQC, MQC, and HQC samples. In contrast to the LBA assay, which
determination is summarized in Table 3 and is plotted against the level of endogenous anti-IgAP antibodies (measured as mean ECL value) for each sample (Figure 4). For simplicity, the IgAP concentration below the LBA LLOQ is replaced by zero. As shown in Figure 4, the immunoassay was susceptible to interference by the pre-existing ADA. The declined accuracy of the IgAP measurement seemed to be correlated with the ADA
Figure 4. Correlation of anti-IgAP antibody levels (expressed by mean ECL values) and accuracy of IgAP measurement by the LBA assay in the 18 individual human serum samples spiked with various concentrations of IgAP. For simplicity, the concentration values below the assay LLOQ were replaced by zero. G
DOI: 10.1021/acs.analchem.5b02293 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry Notes
The authors declare no competing financial interest.
■
Figure 5. Correlation of anti-IgAP antibody levels and accuracy of IgAP measurement by 2D-LC-MS/MS assay in the 18 individual human serum samples spiked with various concentrations of IgAP.
measured only the free IgAP binding to the capture/detection antibodies, the LC-MS/MS assay was able to measure both free IgAP and ADA-bound IgAP with incredible accuracy.
■
CONCLUSION There is an increasing interest in developing novel bioanalytical techniques for accurate measurement of protein therapeutic in ADA-positive matrix to support nonclinical and clinical PK/PD studies. The ADAs form an immune complex with the drug thus preventing the binding of the drug to the capture/ detection antibodies used in the immunecapture. Therefore, LBA, the current gold standard for PK measurement of protein therapeutics, often yields significantly lower concentrations in ADA-positive samples, especially at low drug concentration zone, as the method only measures “free” drug. LC-MS platform, in contrast, is able to quantify “total” drug by dissociating the ADA bound drug through denaturation and subsequent tryptic digestion of both free and ADA bound proteins. To achieve the sensitivity suitable for clinical PK study, we implemented an online orthogonal 2D-LC system operated in high pH for serum peptide fractionation and low pH for selected separation and MS analysis of the surrogate peptide. The sample preparation protocol and orthogonal separation described herein is capable of determining 0.05 μg/ mL IgAP with high accuracy and precision. The bottleneck of the approach described here is the analytical throughput, which is largely limited by the preparation of tryptic peptides (1 day) and 2D-LC analytical run time (20 min). Expedited tryptic digestion protocol was recently reported, holding promise to shorten the sample process time.27,38 Furthermore, we are exploring the use of miniature columns to improve the 2D-LC output. Overall, the described assay is easy to implement and needs minimal optimization to be adapted to other therapeutic proteins. Most importantly, it does not need the timeconsuming production and characterization of antibody reagents. Our results offer compelling evidence that the platform is suitable for further application in clinical PK studies conducted in regulatory laboratories.
■
REFERENCES
(1) Beck, A.; Wagner-Rousset, E.; Bussat, M. C.; Lokteff, M.; Klinguer-Hamour, C.; Haeuw, J. F.; Goetsch, L.; Wurch, T.; Van Dorsselaer, A.; Corvaia, N. Curr. Pharm. Biotechnol. 2008, 9, 482. (2) Beck, A.; Wurch, T.; Bailly, C.; Corvaia, N. Nat. Rev. Immunol. 2010, 10, 345. (3) Nelson, A. L.; Dhimolea, E.; Reichert, J. M. Nat. Rev. Drug Discovery 2010, 9, 767. (4) Chirmule, N.; Jawa, V.; Meibohm, B. AAPS J. 2012, 14, 296. (5) Kelley, M.; Ahene, A. B.; Gorovits, B.; Kamerud, J.; King, L. E.; McIntosh, T.; Yang, J. AAPS J. 2013, 15, 646. (6) Thway, T. M.; Magana, I.; Bautista, A.; Jawa, V.; Gu, W.; Ma, M. AAPS J. 2013, 15, 856. (7) Singh, S. K. J. Pharm. Sci. 2011, 100, 354. (8) Parenky, A.; Myler, H.; Amaravadi, L.; Bechtold-Peters, K.; Rosenberg, A.; Kirshner, S.; Quarmby, V. AAPS J. 2014, 16, 499. (9) Sailstad, J. M.; Amaravadi, L.; Clements-Egan, A.; Gorovits, B.; Myler, H. A.; Pillutla, R. C.; Pursuhothama, S.; Putman, M.; Rose, M. K.; Sonehara, K.; Tang, L.; Wustner, J. T.; Global Bioanalysis, C. AAPS J. 2014, 16, 488. (10) Schwickart, M.; Vainshtein, I.; Lee, R.; Schneider, A.; Liang, M. Bioanalysis 2014, 6, 1939. (11) Wang, S. J.; Wu, S. T.; Gokemeijer, J.; Fura, A.; Krishna, M.; Morin, P.; Chen, G.; Price, K.; Wang-Iverson, D.; Olah, T.; Weiner, R.; Tymiak, A.; Jemal, M. Anal. Bioanal. Chem. 2012, 402, 1229. (12) Lai, K. N. Nat. Rev. Nephrol. 2012, 8, 275. (13) Mistry, D.; Stockley, R. A. Int. J. Biochem. Cell Biol. 2006, 38, 1244. (14) Roberts, I. S. Nat. Rev. Nephrol. 2014, 10, 445. (15) Xie, L. S.; Huang, J.; Qin, W.; Fan, J. M. Nephrology 2010, 15, 584. (16) Gilbert, J. V.; Plaut, A. G.; Longmaid, B.; Lamm, M. E. Ann. N. Y. Acad. Sci. 1983, 409, 625. (17) Brooks, G. F.; Lammel, C. J.; Blake, M. S.; Kusecek, B.; Achtman, M. J. Infect. Dis. 1992, 166, 1316. (18) Chappell, D. L.; Lassman, M. E.; McAvoy, T.; Lin, M.; Spellman, D. S.; Laterza, O. F. Bioanalysis 2014, 6, 1843. (19) Huttenhain, R.; Malmstrom, J.; Picotti, P.; Aebersold, R. Curr. Opin. Chem. Biol. 2009, 13, 518. (20) Kuzyk, M. A.; Smith, D.; Yang, J.; Cross, T. J.; Jackson, A. M.; Hardie, D. B.; Anderson, N. L.; Borchers, C. H. Mol. Cell. Proteomics 2009, 8, 1860. (21) Parker, C. E.; Borchers, C. H. Mol. Oncol. 2014, 8, 840. (22) Sano, S.; Tagami, S.; Hashimoto, Y.; Yoshizawa-Kumagaye, K.; Tsunemi, M.; Okochi, M.; Tomonaga, T. J. Proteome Res. 2014, 13, 1012. (23) Shi, T.; Gao, Y.; Quek, S. I.; Fillmore, T. L.; Nicora, C. D.; Su, D.; Zhao, R.; Kagan, J.; Srivastava, S.; Rodland, K. D.; Liu, T.; Smith, R. D.; Chan, D. W.; Camp, D. G., 2nd; Liu, A. Y.; Qian, W. J. J. Proteome Res. 2014, 13, 875. (24) Xu, K.; Liu, L.; Maia, M.; Li, J.; Lowe, J.; Song, A.; Kaur, S. Bioanalysis 2014, 6, 1781. (25) Yang, Z.; Hayes, M.; Fang, X.; Daley, M. P.; Ettenberg, S.; Tse, F. L. Anal. Chem. 2007, 79, 9294. (26) Yuan, L.; Arnold, M. E.; Aubry, A. F.; Ji, Q. C. Bioanalysis 2012, 4, 2887. (27) Yuan, L.; Aubry, A. F.; Arnold, M. E.; Ji, Q. C. Bioanalysis 2013, 5, 2379. (28) Zhang, Q.; Spellman, D. S.; Song, Y.; Choi, B.; Hatcher, N. G.; Tomazela, D.; Beaumont, M.; Tabrizifard, M.; Prabhavalkar, D.; Seghezzi, W.; Harrelson, J.; Bateman, K. P. Anal. Chem. 2014, 86, 8776. (29) Neubert, H.; Muirhead, D.; Kabir, M.; Grace, C.; Cleton, A.; Arends, R. Anal. Chem. 2013, 85, 1719. (30) Krisp, C.; Yang, H.; van Soest, R.; Molloy, M. P. Mol. Cell. Proteomics 2015, 14, 1708.
AUTHOR INFORMATION
Corresponding Authors
*Tel: (781)482-1493. E-mail:
[email protected]. *Tel: (781)482-0410. E-mail:
[email protected]. H
DOI: 10.1021/acs.analchem.5b02293 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry (31) Fernandez Ocana, M.; James, I. T.; Kabir, M.; Grace, C.; Yuan, G.; Martin, S. W.; Neubert, H. Anal. Chem. 2012, 84, 5959. (32) Shi, T.; Sun, X.; Gao, Y.; Fillmore, T. L.; Schepmoes, A. A.; Zhao, R.; He, J.; Moore, R. J.; Kagan, J.; Rodland, K. D.; Liu, T.; Liu, A. Y.; Smith, R. D.; Tang, K.; Camp, D. G., 2nd; Qian, W. J. J. Proteome Res. 2013, 12, 3353. (33) Szapacs, M. E.; Urbanski, J. J.; Kehler, J. R.; Wilson, R.; Boram, S. L.; Hottenstein, C. S.; Citerone, D. R. Bioanalysis 2010, 2, 1597. (34) Bronsema, K. J.; Bischoff, R.; Bouche, M. P.; Mortier, K.; van de Merbel, N. C. Bioanalysis 2015, 7, 53. (35) Gong, C.; Zeng, J.; Akinsanya, B.; Jiang, H.; Mora, J.; Chilewski, S.; Gambardella, J.; Allentoff, A.; Gleason, C.; Aubry, A. F.; DeSilva, B. S.; Arnold, M. E. Bioanalysis 2014, 6, 2371. (36) Furlong, M. T.; Titsch, C.; Xu, W.; Jiang, H.; Jemal, M.; Zeng, J. Bioanalysis 2014, 6, 1747. (37) Jiang, H.; Zeng, J.; Titsch, C.; Voronin, K.; Akinsanya, B.; Luo, L.; Shen, H.; Desai, D. D.; Allentoff, A.; Aubry, A. F.; Desilva, B. S.; Arnold, M. E. Anal. Chem. 2013, 85, 9859. (38) Liu, G.; Ji, Q. C.; Dodge, R.; Sun, H.; Shuster, D.; Zhao, Q.; Arnold, M. J. Chromatogr. A 2013, 1284, 155. (39) Kong, R. P.; Siu, S. O.; Lee, S. S.; Lo, C.; Chu, I. K. J. Chromatogr. A 2011, 1218, 3681. (40) Percy, A. J.; Chambers, A. G.; Yang, J.; Borchers, C. H. Proteomics 2013, 13, 2202. (41) Percy, A. J.; Simon, R.; Chambers, A. G.; Borchers, C. H. J. Proteomics 2014, 106, 113. (42) Zhou, F.; Sikorski, T. W.; Ficarro, S. B.; Webber, J. T.; Marto, J. A. Anal. Chem. 2011, 83, 6996. (43) Dowell, J. A.; Frost, D. C.; Zhang, J.; Li, L. Anal. Chem. 2008, 80, 6715. (44) Gilar, M.; Olivova, P.; Daly, A. E.; Gebler, J. C. J. Sep. Sci. 2005, 28, 1694.
I
DOI: 10.1021/acs.analchem.5b02293 Anal. Chem. XXXX, XXX, XXX−XXX