Article pubs.acs.org/ac
In Vivo Biotransformation of the Fusion Protein TetranectinApolipoprotein A1 Analyzed by Ligand-Binding Mass Spectrometry Combined with Quantitation by ELISA Manfred Zell,† Christophe Husser,† Roland F. Staack,‡ Gregor Jordan,‡ Wolfgang F. Richter,† Simone Schadt,*,† and Axel Paḧ ler† †
Roche Pharma Research and Early Development, Roche Innovation Center Basel, F. Hoffmann-La Roche Ltd, Grenzacherstrasse 124, 4070 Basel, Switzerland ‡ Roche Pharma Research and Early Development, Roche Innovation Center Munich, Roche Diagnostics GmbH, Nonnenwald 2, 82377 Penzberg, Germany S Supporting Information *
ABSTRACT: The in vivo biotransformation of a novel fusion protein tetranectin/apolipoprotein A1 (TN-ApoA1) was investigated by ligand-binding mass spectrometry (LB-MS) in support of enzyme-linked immunosorbent assays (ELISA). The main focus was on catabolites formed by proteolysis of the fusion protein in rabbit following intravenous administration of lipidated TN-ApoA1. The drug and its catabolites were isolated from rabbit plasma by immunocapture with a monoclonal antibody (mAb) binding to the fusion region of TN-ApoA1. The captured drug and catabolites were released from the streptavidin-coated magnetic beads, separated by monolithic RP capillary HPLC, and online detected by high-resolution mass spectrometry. In addition, the same extract was digested with LysN to confirm or further narrow down the structure of the found catabolites. Two pharmacologically active catabolites were identified with conserved fusion region. The major catabolite [3-285] was formed by truncation of AP at the Nterminus and the minor catabolite [29-270] by truncations of either side of the TN-ApoA1 sequence. Since the ELISA determined the sum of TN-ApoA1, along with its two main catabolites, the individual PK profiles of all three components could be derived by applying their mass peak composition for each sampling point. Parent drug accounted for 25% of drug-related material, whereas that of the catabolites [3-285] and [29-270] accounted for 66% and 9%, respectively. This result could be obtained without catabolite specific ELISAs or quantitative LC-MS assays. It was also confirmed that all relevant functional molecules of TN-ApoA1 in the plasma samples were quantified by the ELISA, which provided a good relationship for pharmacokinetic/pharmacodynamic evaluations.
O
pharmacodynamics (PK/PD) evaluations of biotherapeutics, sensitive, specific, and accurate quantitative methods are required.8 Ligand-binding assays are still the gold standard for quantitation of protein biotherapeutics in support of PK/PD, owing to their functional properties to bind to the target molecule and their superior sensitivity of detection versus mass spectrometry based assays.9−11 However, for investigating the biotransformation of protein biotherapeutics, mass spectrometry is the technique of choice, especially when high-resolution and high-mass-accuracy instruments are employed.12 In contrast to ligand-binding assays, mass spectrometry can distinguish the therapeutic protein from its biotransformation products, like truncations, deamidation, oxidation, or disulfide bond reduction. If immune affinity extraction and LC-MS are
ver the last decades, protein biotherapeutics such as monoclonal antibodies (mAb), antibody-drug conjugates (ADC), therapeutic peptides, and PEGylated peptides/proteins along with fusion proteins have become important drugs for therapy of a multitude of diseases in patients.1−3 Currently, a myriad of protein based pharmaceutical drugs are in preclinical development as well as many candidates in clinical development in the pharmaceutical industry. These biotherapeutics also undergo biotransformation in vivo, like small molecule drugs but in somewhat different pathways. In contrast to small molecule drugs, where the metabolism is mainly mediated by Cytochrome P450 (CYP) enzymes, the biotransformation of protein based drugs is governed by proteolysis, oxidation, deamidation, isomerization, and particularly in the case of mAbs by disulfide bond formation/disruption, glycosylation, and phosphorylation to form structural modifications.4−7 These modifications to the drug may impact stability and pharmacological function. Thus, in support of pharmacokinetic/ © XXXX American Chemical Society
Received: August 19, 2016 Accepted: November 7, 2016 Published: November 7, 2016 A
DOI: 10.1021/acs.analchem.6b03252 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
7.5), 10 mM methionine, and 140 mM NaCl, to obtain a concentration of 5.44 mg/mL (protein content) used as stock solution. The purity of RO5319615 was >85%. Lipidated TNApoA1 (RO5327819) was prepared as follows: RO5319615 was lipidated with pure POPC/DPPC (3/1; w/w) in a protein to lipid ratio of 1/60. POPC: Palmitoyl oleoylphosphatidylcholine; DPPC: Dipalmitoylphosphatidylcholine. The stock solution of RO5327819 (9.5 mg/mL) was prepared by dissolving the appropriate amount of RO5327819 in the same solvents as used for neat TN-ApoA1. For immunecapture, biotinylated mAb 5.138.259 was used (2.0 mg/mL; Roche Diagnostics GmbH, Penzberg, Germany). Chemicals, Reagents, and Materials. Formic acid, 98− 100% (Suprapur) and ethanol (Lichrosolv) were purchased from Merck (Darmstadt, Germany). Acetonitrile (HPLC grade S) was obtained from Rathburn (Walkerburn, U.K.). Water was purified “in-house” using a Milli-Qplus 185 system from Millipore (Volketswil, Switzerland) to obtain deionized water. Water used for chromatography was Lichrosolv grade from Merck. The 0.1% formic acid solution was prepared by adding 1.0 mL of formic acid (98−100%) to water in a 1.0-L volumetric flask and made up to volume with deionized water. An aqueous solution of Tween 20, 10% (w/v), premixed PBS buffer (pH 7.4) and trypsin sequencing grade was supplied by Roche Diagnostics (Mannheim, Germany). Ammonium bicarbonate puriss p.a. (eluent additive for LC-MS) and glycine hydrochloride solution (100 mM glycine in 0.1 M HCl, analytical standard) were supplied by Fluka (Buchs, Switzerland). Trifluoroacetic acid (purity >99.5%) and Pierce magnetic beads coated with streptavidin (10 mg/mL) were supplied by Thermo Fisher Scientific (Reinach, Switzerland). Rabbit plasma was supplied by the animal care unit of F. Hoffmann-La Roche Ltd. (Basel, Switzerland). Instrumentation. Liquid Chromatography of Intact Protein. A Waters nanoAQUITY UPLC system (Waters, Milford, MA, USA) was used in combination with a Synapt G2 HDMS QTof instrument (Waters, Manchester, UK). The components of the nanoACQUITY system were specifically designed for operating at pressures up to 5000 psi. The isolated protein fraction was analyzed on a monolithic HPLC column ProSwift RP from Thermos Fisher Scientific, 1.0 mm i.d. × 50 mm, at a flow rate of 40 μL/min. The heating and trapping module of the system was circumvented by using an off-line column heater HotDog XL5090 from ProLab (Basel, Switzerland) to maintain the ProSwift column temperature at 70 °C. The system’s sample manager (autosampler) was used to directly inject aliquots of the sample onto the ProSwift column. Liquid Chromatography of Peptides from Protein Digests. The nanoAQUITY UPLC system was used in a columnswitching mode (single pump trapping) as displayed in Figure 2. The trap capillary (Xbridge BEH130, C18, 0.3 mm i.d. × 50 mm, 3.5 μm particle size) was connected via a Nano-Tee device with the analytical capillary (Xbridge BEH130, C18, 0.3 mm i.d.
combined, a powerful tool for the selective enrichment of active catabolites and the identification of catabolites of the therapeutic proteins emerges. An interesting case example is described by Hager et al., who studied the in vivo biotransformation of Fc fusions of fibroblast growth factor 21 by a combination of differential ELISA with ligand-binding MS and used this data for compound optimization.13 In this article, we describe the investigation of the in vivo biotransformation of the therapeutic fusion protein TN-ApoA1 to select the appropriate ligand-binding assay (ELISA) for quantitation of all structural but still functional variants of this fusion protein. This molecule (TN-ApoA1, RO5319615, Figure 1) is composed of a portion of human apolipoprotein A1
Figure 1. Sequence of the fusion protein TN-ApoA1. Amino acids in blue and pink originate from human tetranectin protein fused to the main portion of ApoA1. Amino acids in green and black originate from apolipoprotein A1. The fusion region is highlighted in pink and green, corresponding to the respective tryptic peptide.
(ApoA1) and tetranectin (TN). The fusion protein was produced by exchanging the signal ApoA1(1-18) and propeptide ApoA1(19-24) portion of human apolipoprotein A1(1-267) with a portion of the human tetranectin protein. Lipidated TN-ApoA1 (RO5327819) was intravenously administered to rabbits. The concentrations of the active components were determined by different ELISAs to establish pharmacokinetic profiles. However, one of the ELISAs, using monoclonal capture antibody (mAb) 11/81 (ELISA 2), demonstrated rapid loss of TN-ApoA1 in plasma whereas ELISA 1 with mAb 5.138.259 could measure concentrations of TN-ApoA1 up to 96 h. The pharmacokinetic (PK) profile, established by ELISA 1 was in good keeping with the results of a bioassay, which demonstrated pharmacological activity following administration of TN-ApoA1 to the rabbit. These results suggested there might be modified forms of TN-ApoA1 in plasma generated by biotransformation which are missed by mAb 11/81 but detected by mAb 5.138.259. One of various hypotheses was that truncated but still pharmacologically active forms of TNApoA1 might be produced by catabolism of this fusion protein. For this reason, a methodology was devised to isolate TNApoA1 and potential catabolites from rabbit plasma by immunocapture and mass spectrometry for characterization of isolated proteins or peptides.14−17 Besides the information about the catabolism of TN-ApoA1, these data are required for a detailed understanding of the specificity of the selected mAbs to better interpret the results of the ELISA using this reagent and to provide fully characterized pharmacokinetic data. The approach presented here has the potential to be more widely applied to support the development of therapeutic proteins.
■
EXPERIMENTAL SECTION Reference Compounds. The native fusion protein tetranectin apolipoprotein A1 (TN-ApoA1; RO5319615) was dissolved in an aqueous solution of 250 mM TRIS buffer (pH
Figure 2. LC-MS nanoAcquity capillary column-switching system from Waters. B
DOI: 10.1021/acs.analchem.6b03252 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry × 100 mm, 3.5 μm particle size) from Waters. The system further consisted of two binary high-pressure mixing pumps BSM1 and BSM2 providing a wide dynamic flow range from 0.2 to 100 μL/min. The column-switching device was assembled between the trap capillary (TC) and analytical capillary (AC) using a Rheodyne narrow-bore 6-port columnswitching valve at ambient temperature. The column-switching valve was controlled by the MassLynx software as a timed event. The heating trapping module (HTM) with simplified “forward flush” sample trapping scheme and an integrated column heater for AC were used. A dedicated auxiliary pump for providing a stable flow of enkephalin “lockmass” solution to the reference sprayer of the standard electrospray source was used. The system’s sample manager AMS (autosampler) was used to directly inject aliquots of the sample (8 μL) onto the trap capillary. Mass Spectrometry. Mass spectrometric analysis was performed using a Synapt G2 HDMS QTof mass spectrometer from Waters equipped with a LockSpray electrospray ionization source. Assay Procedure. Details are in the Supporting Information. Chromatography. Chromatographic Separation of Intact Proteins. A 8-μL aliquot of the eluate, following intact protein extraction, was directly injected onto a monolithic HPLC column ProSwift, 1.0 mm i.d. × 50 mm, kept at 70 °C, using high-pressure gradient elution at a flow rate of 40 μL/ min. The elution was performed with a mixture of 0.1% formic acid (E1) and acetonitrile containing 0.1% formic acid (E2). After injection, eluent E2 was kept at 5% for 2 min, and then raised to 25% within 4 min, and further raised to 95% in 12 min. Thereafter, E2 was maintained at 95% for an additional 4 min. While E2 was raised, E1 was decreased complimentarily. After 20 min following injection, E2 was decreased to 5%. The overall run time of analysis was 22 min. Chromatographic Separation of Peptides from Digest with Trypsin and LysN. A 1- to 8-μL aliquot of the eluate, obtained after removal of the substrate from the capture antibody, was loaded onto TC (Xbridge BEH130 C18, 0.3 mm i.d. × 50 mm, particle size 3.5 μm) of an HPLC columnswitching system using the Acquity autosampler. The loading eluent was 0.2% formic acid/acetonitrile (97/3; v/v) (E1) delivered at a flow rate of 10 μL/min for 5.0 min. Thereafter, TC was switched in line with AC (Xbridge BEH130 C18, 0.3 mm i.d. × 100 mm, particle size 3.5 μm), and the retained analytes were transferred to AC by gradient elution at a flow rate of 10 μL/min using the binary sample manager pump (nanoBSM1) of the nanoAcquity HPLC system. The elution was performed by a mixture of 0.2% aqueous formic acid (eluent E2.1) and 0.1% formic acid in acetonitrile (eluent E2.2). At the start of the linear gradient, eluent E2.2 was kept at 3% for 2 min and then raised to 40% within 90 min. Thereafter, E2.2 was increased to 95% and maintained for 3 min, and then decreased to 3% within 5 min. TC and AC were kept at 40 °C during the whole analysis sequence. Prior to the next injection, TC was decoupled from AC. The overall run time of the analysis cycle was 105 min.
(TN-ApoA1) following intravenous administration to rabbits. For this reason, an analytical methodology was devised to isolate and identify preferentially assumed active biotransformation products of TN-ApoA1 in plasma using mass spectrometry. This assumption could be made since the mAb, used for immunoaffinity extraction with the ELISA 1, resulted in concentrations which were in good correlation with the pharmacological activity. A crucial step to achieve this goal was the specific extraction of the therapeutic protein from plasma, along with its catabolic and other biotransformation products. Further, we were solely interested in identifying pharmacologically active biotransformation products of the fusion protein TN-ApoA1. This prerequisite ruled out the use of high-capacity affinity chromatography (protein depletion columns) to remove the most abundant endogenous proteins in plasma followed by classical protein identification to fish out relevant catabolites of TN-ApoA1.18 Therefore, using monoclonal antibodies directed to the ApoA1 portion of TN-ApoA1 for capturing catabolites of TN-ApoA1 was also not possible.19 For all these reasons, we favored immunocapture from plasma for isolating TN-ApoA1 and its biotransformation products using the same monoclonal capture antibody (mAb 5.138.259) as utilized with ELISA 1, which correlated well with the pharmacological activity. This extraction procedure also supported analysis of proteins by direct mass spectrometry or LC-MS due to its selectivity to produce clean extracts. Additionally, it facilitated the analysis of characteristic peptides of biotransformation products after suitable proteolytic enzymatic digest of the isolated proteins.20,21 Immunocapture. Linking of Biotinylated Monoclonal Capture Antibody (mAb) to Streptavidin-Coated Magnetic Beads. For fast and convenient immunocapture of a limited number of samples, magnetic beads were preferred versus streptavidin agarose resins or other high-throughput formats such as ELISA microplates.22,23 The binding capacity of 1 mg of streptavidin-coated beads corresponds to ∼400 pmol in terms of biotinylated rabbit IgG. Since each streptavidin molecule has three binding sites and the mAb provided at least three biotin moieties per molecule, the 2-fold quantity of mAb (800 pmol) was added to the beads for incubation to generate a tight layer of the streptavidin−biotin complex. Immuno-extraction from Plasma. Since the concentrations of TN-ApoA1 in the plasma samples from rabbit were known from results of ELISA 1, the plasma volume of the respective samples was diluted with drug free control plasma to deliver the antigen in the range of 200 to 400 pmol/50 μL to the mAb in the slurry of the beads. The reaction was sufficiently quantitative after incubation for 2 h. The beads were washed with PBS buffer to remove residues of endogenous plasma proteins. Excessive washing was detrimental to a good recovery. The specifically bound TN-ApoA1 and its metabolites with conserved fusion region were released from mAb using glycine buffer or trifluoroacetic acid/trifluoroethanol. This step proved to be critical to recovery. The glycine buffer more gently removed the substrates, but sometimes the recovery was poor for unknown reasons. Therefore, trifluoroacetic acid/trifluoroethanol was used to improve the recovery. Chromatography of Intact Protein. Different HPLC columns were tested with respect to elution properties for TNApoA1 and human ApoA1 and their capabilities to chromatographically separate these proteins. A further requirement was that the eluent used should be fully compatible for subsequent mass spectrometry detection of the proteins. For this reason,
■
RESULTS AND DISCUSSION Analytical Strategy for Identification of Biotransformation Products. The determination of TN-ApoA1 in plasma, using different ELISAs, indicated catabolism or other types of biotransformation of the therapeutic fusion protein C
DOI: 10.1021/acs.analchem.6b03252 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
agreement. Twenty-eight amino acids were split off the Nterminus and 15 amino acids from the C-terminus of the TNApoA1 sequence. There were further mass peaks in the region 10000 to 17000 Da. All of these mass peaks represent 1/2 or 1/ 3 of the mass peaks at 32772 Da, 32604 Da, and 27683 Da. Apparently, these “subharmonic” mass peak artifacts were caused by the deconvolution software MaxEnt1 of MassLynx. The increase of the number of iterations from 12 to 30 reduced the peak height of these “subharmonic” mass peaks to some extent but did not remove them from the spectrum. The additional application of the software BiopharmaLynx from Waters (Manchester, UK) did not generate these “subharmonic” mass peaks. Figure 4A displays the HPLC chromatograms from a predose plasma sample along with plasma samples taken after 5 min, 7
trifluoroacetic acid was not used as a modifier in the elution solvent. The first eluent choice was composed of 0.1% formic acid and acetonitrile to reduce the back-pressure of the HPLC column. The Acquity BEH300 C4 column, 1.7 μm particle size, 300 Å pore size (Waters, Milford, MA, USA) has been successfully applied to the analysis of monoclonal antibodies. Nevertheless, the elution properties for TN-ApoA1 were modest, showing a broad protein peak with significant tailing. In contrast to these findings, the test of an Acquity UPLC HSS T3 column, 1.8 μm particle size, 100 Å pore size (Waters) showed improved peak shape but also some tailing. In addition, a reversed-phase column with monolith technology designed for fast, high-resolution HPLC of proteins was investigated.24,27 The monolithic RP ProSwift column showed improved peak shape with respect to band broadening and tailing. Mass Spectrometry. LC-MS Analysis of Isolated Proteins from Plasma. In a first step, mass spectrometric analysis of intact protein was applied to identify catabolic and other biotransformation products of TN-ApoA1 in plasma from animals.25,26 Following immunocapture, the isolated proteins or peptides were analyzed by direct infusion mass spectrometry. Under electrospray ionization, strong sodium ion adducts and other adducts were produced, which resulted in ionization suppression. The eluate after immunocapture was analyzed on a monolithic reversed phase HPLC column coupled to the electrospray interface of the QTof mass spectrometer. Figure 3 displays the deconvoluted mass spectrum of the major protein peak in the LC total ion chromatogram from a
Figure 4. Total ion chromatograms (A) and the corresponding deconvoluted mass spectra (B) of the plasma samples taken at various sampling times following intravenous administration of TN-ApoA1 to rabbit.
h, and 24 h after i.v. administration of TN-ApoA1 to rabbit. The total ion chromatogram (TIC) of the predose sample indicated the extent of nonspecific binding of endogenous proteins to the mAbs, which was found to be low. Owing to the specificity of the immunocapture, only one abundant peak was found in the TIC, where the parent drug (TN-ApoA1) superimposed its biotransformation products. An excerpt of the pertinent deconvoluted mass spectra is depicted in Figure 4B: The catabolite [3-285] at 32604 Da was produced in small amounts after 5 min in the rabbit, underscoring the speed of AP abstraction in vivo. Already 7 h after i.v. administration of TNApoA1, this major catabolite exceeded the parent drug. The mass peak heights of the proteins [1-285] and [3-285] can be directly compared, since the supposition is that their ionization efficiencies are the same. Already 24 h after administration, catabolite [3-285] was the major circulating component in rabbit and only a small trace of administered TN-ApoA1 was left. LC-MS Analysis of Digested Protein. The results of the intact protein analysis clearly indicated the occurrence of one major and at least one additional catabolite of TN-ApoA1 in rabbit plasma. The cleavage of AP from the N-terminus was found to be the dominating biotransformation pathway. For this reason, we wanted to unambiguously confirm this cleavage by digestion analysis of the isolated proteins after immunocapture. The sequence of TN-ApoA1 contains 26 lysine and 17 arginine amino acids which are distributed over the whole sequence. Therefore, tryptic digestion was the first choice to
Figure 3. Deconvoluted MS spectrum of immunocaptured main components in plasma from rabbit 2 taken 7 h after i.v. administration.
plasma sample taken 7 h after i.v. administration of 20 mg/kg TN-ApoA1 to rabbit. The mass peak at 32772 Da corresponds to unchanged TN-ApoA1. However, the major mass peak in the spectrum at 32604 Da was found to be a truncation product of the fusion protein. The mass difference of 168.4 Da was in acceptable agreement with the residue mass of alanine-proline (AP). Apparently, AP was cleaved of the N-terminus of TNApoA1 by proteolysis to produce the catabolite [3-285]. Notably, water was abstracted from the protein molecules, probably caused by the electrospray process. In addition, a small mass peak at 32624 Da was found, being 20 Da higher than catabolite [3-285]. Presumably, this mass peak was produced by water and sodium adducts as well as oxidation of catabolite [3-285], which were not resolved by the QTof MS. The additional mass peak at 27863.20 Da could be assigned to catabolite [29-270] predicted at 27863.44 Da. The mass difference of 0.24 Da between the measured and predicted average mass from the sequence [29-270] was in good D
DOI: 10.1021/acs.analchem.6b03252 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 5 depicts the respective XICs of the signature peptide APIVNA at different time points. Apparently, no trace of TN-
obtain characteristic peptides of the proteins under investigation. Figure 1 displays the sequence of TN-ApoA1. The fusion region is highlighted by the amino acids in pink and green. The pink ones belong to the tetranectin and the green ones to the ApoA1 sequence. This annotated tryptic peptide of the fusion region could be used to monitor TN-ApoA1 together with all of its catabolic metabolites. A further advantage of this approach was to achieve better sensitivity using mass spectrometry for tryptic peptide detection compared to the highly charged intact proteins. This opens up the opportunity to further identify immuno-extracted proteins which exhibit poor MS response, causing them to be poorly detectable in the total ion chromatogram of intact protein analysis. The first approach applied was to digest the complete eluate after immunocapture of the plasma sample and conduct the separation of the tryptic peptides on the HPLC capillary combined with MSE mass spectrometry detection. Owing to the high sequence coverage of >95%, most of the predicted peptides could be assigned to chromatographic peaks. The BiopharmaLynx software from Waters (Manchester, UK) was very proficient in fishing out the predicted peptides and also indicating modifications such as oxidation in peptides containing methionine. The second approach was to subject the eluate after immunocapture to chromatographic separation on the ProSwift monolithic HPLC column and fractionate the most abundant protein peak. The eluate of this fraction was digested, and the peptides separated and detected using the MS E mass spectrometry acquisition mode. Using the BiopharmaLynx software, both runs could be easily compared to identify common peptides or those unique to one run. The procedure allowed us to gather clues as to whether further low abundant catabolites were present in the sample which could not be detected sufficiently sensitive as intact proteins. In addition, further fractionation of the relevant portions of the LC protein separation could be conducted followed by tryptic digestion of the collected fractions to achieve more sensitive detection of proteins. Using this approach, any catabolite with an intact fusion region of TN-ApoA1 could be identified by extracting EQQALQTVDEPPQSPWDR at m/z 708.6713 as a triply charged ion ([M + 3H]3+) from the respective MSE acquisition file. The tryptic peptide APIVNAK at the N-terminus could be used to monitor the cleavage of AP by analysis of plasma samples taken at sampling times spanning 5 min to 48 h after i.v. administration. The sequence coverages by tryptic peptides for the protein immuno-extracted from the 5 min and 48 h plasma samples were found to be 97.2% and 94.7%, respectively. Extracted ion chromatograms (XIC) of the doubly charged tryptic peptide APIVNAK at m/z 356.7212 were established using sampling times 5 min, 7 h, 24 h, and 48 h. This signature peptide monitored the disappearance of the parent drug TN-ApoA1. One major shortcoming of this approach was that the abundances of the signature peptides APIVNAK and IVNAK were poor even though the tryptic digest was optimized for this purpose. Further experiments with LysN were conducted yielding a sequence coverage for the 5 min and 48 h plasma samples of 98.2% and 97.9%, respectively. Digestion of the immunoextracted protein portion using LysN generated the signature peptide APIVNA, which provided a 30-fold improvement of sensitivity using the singly charged APIVNA at m/z 584.3402 compared to the tryptic peptide APIVNAK.
Figure 5. Analysis of signature peptide APIVNA to monitor cleavage at the N-terminus.
ApoA1 was left after 48 h of drug administration. These results nicely confirmed the findings of the intact protein analysis as shown in Figure 6. Unfortunately, the formation of the
Figure 6. Peak heights of deconvoluted mass peaks from intact protein analysis of rabbit plasma samples taken at different sampling times plotted versus sampling times. The blue curve corresponds to the sum of all analyte MS responses.
signature peptide IVNA of the catabolic metabolite, formed by AP cleavage, could not be monitored due to its low abundance in the LysN digests. For this reason, monitoring of the formation of catabolite [3-285], using its key signature peptide IVNA, could not be performed. Occurrence of Catabolites of TN-ApoA1 in Rabbit Plasma. The major catabolite [3-285] was found to be formed by cleavage of AP from the N-terminus of TN-ApoA1. As demonstrated in Figure 4 and Figure 5, the formation of this catabolite was fast in rabbit, as catabolite [3-285] could be detected in plasma within 5 min of i.v. administration of TNApoA1. Moreover, the residual level of TN-ApoA1 at 24 h after dosing was less than 1% compared to that after 5 min. For an exploratory estimate of the half-lives of these entities, the intensities of the deconvoluted mass peaks of the intact protein E
DOI: 10.1021/acs.analchem.6b03252 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
the opportunity to use the area under the curves (AUC) of these plots as exemplified in Figure 7 to compare the exposure of the components to each other. The exposure was defined as the percentage of AUC of an analyte versus that of total drugrelated material (AUC of ELISA 1 PK profile). However, a further condition needed to be fulfilled to establish PK profiles for the intended purpose. The antigen binding capacity of the mAb, used for immunocapture of the analytes, was only linear over a limited range. Since the plasma samples used for metabolic investigations had been analyzed with ELISA 1, the respective concentrations were known. Thus, each plasma sample was diluted with rabbit control plasma to obtain a concentration of the analytes which fell into the linear range of the antigen binding capacity. Observing all these measures was key to obtain accurate results without the use of reference compounds and/or internal standards. Since the composition of the analytes for each ELISA concentration was known from intact protein analysis by LCMS, the concentration for each analyte could be derived from the ELISA concentrations. Applying this procedure, the respective PK profiles of the parent drug TN-ApoA1 and its two catabolites were established (Figure 7). The catabolite [3-285] with an average mass of 32604 Da was the major component accounting for 66% of total drugrelated material in terms of AUC0−72h from ELISA 1 data. The corresponding exposure data for TN-ApoA1 and the catabolite [29-270] were 25% and 9.3%, respectively. These data were in good agreement for the catabolites with that generated with MS data (see Figure 6, Table 1). However, the exposure of TN-
were plotted for each component versus the sampling time as depicted in Figure 6. TN-ApoA1, catabolite [3-285] at 32604 Da, and catabolite [29-270] at 27863 Da demonstrated terminal half-lives of 2.96 h, 18.8 h, and 47.7 h, respectively. The blue curve in Figure 6 corresponds to the sum of the MS responses of all analytes. The calculated half-life for the sum of all analytes was found to be 21.1 h. This finding was in good keeping with the results from ELISA 1, where a half-life of 17.0 h was found (see Figure 7, blue curve). Thus, ELISA 1, used to quantify TN-ApoA1 in
Figure 7. PK profile (blue curve) using the concentrations determined by ELISA (mAb 5.138.259) composed of the parent drug and its catabolites. Other curves derived from the ELISA 1 PK profile applying the composition of peak height ratios of analytes from Figure 6.
Table 1. PK Parameters of TN-ApoA1 and Its Catabolites in Rabbit
the same rabbit plasma samples, determined the parent drug and its catabolites as a combined concentration per sampling point. Exposure of TN-ApoA1 and Its Catabolites in Rabbit Plasma. Of course, the ambitious goal was to obtain exploratory exposure data of the parent drug and its catabolites in the rabbit. In fact, this can only be performed by mass spectrometry with the use of reference compounds or a specific ELISA for each component. Nevertheless, if the ionization efficiency of each analyte is not too different from each other, then other approaches using mass spectrometry may come into play. Since the number of positive charges on the TN-ApoA1 protein will be the same as for its major catabolite [3-285], owing to the same number of strongly basic amino acids such as K and R, it could be inferred that the ionization efficiencies for these two compounds were identical. Concerning the ionization efficiency of catabolite [29-270], it might become more difficult. However, this protein has six K less out of 26 K, and one R less out of 17 than that of TN-ApoA1. For this reason, it might be inferred that its ionization efficiency was in the same range as that for TN-ApoA1. Another favorable issue was the coelution of TN-ApoA1 and its catabolites on the ProSwift monolithic HPLC column to achieve consistent ionization conditions and matrix effect for all these analytes in the electrospray interface. The basis for these measures was that the response factors of these analytes were expected to be in a range of ±30% of TN-ApoA1. For catabolite [3-285], this prerequisite was fulfilled, owing to the same composition of the basic amino acids K and R as in TN-ApoA1. For catabolite [29270], the response factor was inferred from the MS analysis of truncation products with a similar sequence. This opened up
MS data
ELISA Sum of analytes (MS) TN-ApoA1 Catabolite [3-285] Catabolite [29-270]
ELISA 1 derived data
t1/2
% AUC(0−72 h)
t1/2
(h)
(%)
(h)
(%)
17.0
100
2.63 15.4 30.6
24.7 66.0 9.28
21.1
100
2.96 18.7 47.7
14.4 74.7 10.8
% AUC(0−72 h)
ApoA1 showed a much higher deviation between the ELISA 1 derived data and the MS data. Nevertheless, this deviation was considered acceptable for exploratory purposes in this investigation. In addition, the half-lives for TN-ApoA1, catabolite [3-285] at 32604 Da, and catabolite [29-270] at 27863 Da were 2.63 h, 15.4 h, and 30.6 h, respectively, as determined by conventional noncompartmental analysis. Terminal half-lives were determined from the roughly log− linear terminal phases (0.08−24 h for parent compound, 24− 72 h for catabolites). The pertinent PK parameters for rabbit are summarized in Table 1. Utilizing this approach, we obtained reliable exploratory exposure data of all relevant catabolites in rabbit plasma, circumventing the time-consuming and laborious work to synthesize reference compounds and develop additional quantitative ELISA or LC-MS assays. The PK curve from TN-ApoA1 (see Figure 7) using this procedure complied well with that established by using concentration data provided by ELISA 2 (see Figure 8). This finding clearly indicated that the mAb 11/81 of ELISA 2 was directed to the N-terminus of TNF
DOI: 10.1021/acs.analchem.6b03252 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
rabbits (Figure 8). Only the assay using mAb 5.138.259 correlated nicely with pharmacological activity and was suitable for bioanalytical use. The biotransformation results generated using exactly this mAb showed that TN-ApoA1 is significantly and rapidly catabolized and that the sum of the catabolites nicely correlated with the respective ELISA result, which again suggests that these catabolites are pharmacologically active. PK results obtained with mAb 81/11 (ELISA 2) correlated with the PK of the parent drug only, which indicates that this mAb binds to the N-terminus, which is rapidly cleaved and, thus, is not suitable for the development of an “active TN-ApoA1” assay. PK/PD Considerations. The LC-MS data showed rapid catabolism of the parent TN-ApoA1, leading to formation of two catabolites. The ELISA 1 obviously detected the sum of TN-ApoA1 and its catabolites, in contrast to ELISA 2, which demonstrated rapid loss of TN-ApoA1 in plasma due to loss of the N-terminal AP (see Figure 8). PK/PD modeling using concentration−time data from this ELISA 1 together with cholesterol efflux as PD readout demonstrated a consistent PK/PD relationship over the entire dosing interval (data on file) despite the marked change in the TN-ApoA1/catabolite ratio during this time. This finding indicated a similar pharmacologic activity of parent substance and its catabolites. Therefore, the use of ELISA 1 using mAb 5.138.259 was considered to be appropriate for PK/PD assessment in the project.
Figure 8. PK profiles of two different ELISAs versus the pharmacological effect of TN-ApoA1 expressed as cholesterol efflux measured by a suitable bioassay.32
ApoA1. Apparently, ELISA 2 was not capable of detecting the catabolite [3-285], underscoring its specificity. Formation of Catabolites of TN-ApoA1. The formation of the major catabolite [3-285] of TN-ApoA1, generated by cleavage of AP at the N-terminus of the sequence, was fast and complete in vivo within 24 h after intravenous administration of TN-ApoA1 to rabbits but also in Cynomolgus monkeys and rats (data on file). AP was also cleaved to some extent in vitro in plasma from rabbit, rat, Cynomolgus monkey, and man after incubation at room temperature for 24 h. All these findings suggested the involvement of a ubiquitous peptidase or protease in this cleavage process. In fact, there are only two peptidases which fulfill these requirements. These are the neutral endopeptidase (NEP) 24.11 and dipeptidyl peptidase 4 (DPP4); both of them are, e.g., mediators of the degradation of incretin hormones, such as the glucagon-like peptide 1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP).28−30 DPP4 acts as an exopeptidase whereas NEP 24.11 cuts predominantly within the sequence of a peptide or protein. DPP4, a serine peptidase, is capable of cleaving dipeptides from the N-terminal side of a peptide or protein, provided A or P are in the amino acid two position, such as XAX or XPX.31 DPP4 rapidly cleaves GIP (motif YAE) and GLP-1 (motif HAE) at the N-terminus, leading to short half-lives of these incretins in vivo. However, DPP4 does not cleave if the motif is as follows: XAP or XPP. For this reason, Bradykinin (RPPGPSPPR) undergoes no cleavage by DPP4. In the case of TN-ApoA1, the N-terminus motif is API, supporting the cleavage by DPP4 using the aforementioned rules. This information on substrate specificity suggests that DPP4 was the peptidase responsible for the fast AP cleavage. Bioanalytical Considerations for ELISA. The goal of the bioanalytical strategy for therapeutic proteins is the development of bioanalytical methods which enabled the generation of appropriate data to establish a correlation between pharmacokinetics and pharmacodynamics (PK/PD). For this reason, the bioanalytical method should be specific for the “active” drug. Thus, a method is needed which monitors all active modifications of the heterogeneous therapeutic protein. A prerequisite for this is a careful selection of appropriate active drug specific reagents.9 In the described case, two drug specific reagents were available for ELISA development. However, the assays provided discrepant concentration data for the same plasma samples following i.v. administration of the drug to
■
CONCLUSIONS This work demonstrated the benefits and the synergy of the combined application of selective enrichment of active catabolites by an immune affinity extraction and LC-MS with its mass selectivity for the identification of catabolites of the therapeutic fusion protein TN-ApoA1 and the functional selectivity of the ELISA to determine active drug exposure data in the rabbit to establish PK/PD of a drug. The methodology applied here served both for identifying as well as for quantifying catabolites. In a first step, the identity of the major catabolite [3-285], formed by cleavage of AP at the N-terminus of the sequence, and also a number of less abundant catabolites were elucidated by intact protein analysis. Subsequent confirmation was done by peptide mapping after digest with trypsin and LysN using column-switching capillary HPLC combined with highresolution MS. Immunocapture, using a monoclonal capture antibody, combined with high-resolution accurate mass measurements, was key for the identification of these catabolites of TN-ApoA1. In a second step, these catabolites were quantified by the combined application of ELISA and LC-MS to assess their exposure in rabbit plasma. After hints that catabolism had become available (see the introduction), not only the structures of the catabolites were to be elucidated, but also their exposures should be assessed. The PK profiles of the parent drug and its catabolites could be derived from the ELISA 1 PK curve applying the mass peak composition of the analytes from deconvoluted mass spectra. On this basis, the respective analyte concentrations in the plasma samples could be obtained from the total concentrations determined by ELISA 1. This procedure provided results which were sufficient for exploratory purposes, obviating the application of additional bioanalytical assays to determine the exposure of catabolites in animals. The described concept of using the specificity determining ELISA reagent in an immunoaffinity extraction procedure prior G
DOI: 10.1021/acs.analchem.6b03252 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
(9) Papadimitriou, A.; Bansal, S.; Heinrich, J.; Staack, R. F. Bioanalysis 2014, 6, 1735−2737. (10) White, J. T.; Bonilla, L. E. Bioanalysis 2012, 4, 2401−2411. (11) Zhang, Y. J.; Olah, T. V.; Zeng, J. Bioanalysis 2014, 6, 1827− 1841. (12) Ezan, E.; Becher, F.; Fenaille, F. Expert Opin. Drug Metab. Toxicol. 2014, 10, 1079−1091. (13) Hager, T.; Spahr, C.; Xu, J.; Salimi-Moosavi, H.; Hall, M. Anal. Chem. 2013, 85, 2731−2738. (14) Murphy, R. E.; Kinhikar, A. G.; Shields, M. J.; Del Rosario, J.; Preston, R.; Levin, N.; Ward, G. H. J. J. Pharm. Biomed. Anal. 2010, 53, 221−227. (15) Hager, T.; Spahr, C.; Xu, J.; Salimi-Moosavi, H.; Hall, M. Anal. Chem. 2013, 85, 2731−2738. (16) Van den Broek, I.; Niessen, W. M. A.; van Dongen, W. D. J. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2013, 929, 161−179. (17) Sandra, K.; Vandenheede, I.; Sandra, P. J. Chromatogr. A 2014, 1335, 81−103. (18) Dekker, L. J.; Bosman, J.; Burgers, P. C.; van Rijswijk, A.; Freije, R.; Luider, T.; Bischoff, R. J. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2007, 847, 65−69. (19) Chandra, R.; Macfarlane, R. D. Anal. Chem. 2006, 78, 680−685. (20) Ackermann, B. L.; Berna, M. Expert Rev. Proteomics 2007, 4, 175−186. (21) Hall, M. P.; Gregg, C.; Walker, K.; Spahr, C.; Ortiz, R.; Patel, V.; Yu, S.; Zhang, L.; Lu, H.; DeSilva, B.; Lee, J. W. AAPS J. 2010, 12, 576−585. (22) Girault, S.; Chassaing, G.; Blais, J. C.; Brunot, A.; Bolbach, G. Anal. Chem. 1996, 68, 2122−2166. (23) Yang, W.; Kernstock, R.; Simmons, N.; Alak, A. Bioanalysis 2015, 7, 307−318. (24) Detobel, F.; Broeckhoven, K.; Wellens, J.; Wouters, B.; Swart, R.; Ursem, M.; Desmet, G.; Eeltink, S. J. Chromatogr. A 2010, 1217, 3085−3090. (25) Zhang, Z.; Pan, H.; Chen, X. Mass Spectrom. Rev. 2009, 28, 147−176. (26) Chen, G.; Warrack, B. M.; Goodenough, A. K.; Wei, H.; WangIverson, D. B.; Tymiak, A. A. Drug Discovery Today 2011, 16, 58−64. (27) Eeltink, S.; Dolman, S.; Vivo-Truyols, G.; Schoenmakers, P.; Swart, R.; Ursem, M.; Desmet, G. Anal. Chem. 2010, 82, 7015−7020. (28) Stephenson, S. L.; Kenny, A. J. Biochem. J. 1987, 241, 237−247. (29) Mentlein, R. Regul. Pept. 1999, 85, 9−24. (30) Deacon, C. F. Horm. Metab. Res. 2004, 36, 761−765. (31) Tagore, D. M.; Nolte, W. M.; Neveu, J. M.; Rangel, R.; GuzmanRojas, L.; Pasqualini, R.; Arap, W.; Lane, W. S.; Saghatelian, A. Nat. Chem. Biol. 2009, 5, 23−25. (32) Ohnsorg, P. M.; Mary, J. L.; Rohrer, L.; Pech, M.; Fingerle, J.; von Eckardstein, A. Biochim. Biophys. Acta, Mol. Cell Biol. Lipids 2011, 1811, 1115−1123.
to dedicated MS analysis enables specific identification of the analytes (parent and potential catabolites) which are detected by the ELISA method and provides a better understanding of the ELISA result. The combination of the quantitative ELISA data with the MS data even allows the assessment of drug/ catabolite-specific pharmacokinetic behavior. The application of the concept significantly improved the understanding of the finally applied ELISA for bioanalytical project support and led to the conclusion that the PD effect is elicited by active catabolites in addition to the parent compound. This approach has the potential to be more widely applied to support the development of therapeutic proteins.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.6b03252. More detailed pieces of information on the immunocapture assay procedure, the digest procedures with trypsin and LysN, and chromatography of the intact protein as well as the description of ELISA (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*Tel: +41 61 687 24 07. E-mail:
[email protected]. Author Contributions
The study was planned by M.Z., R.F.S., W.F.R., and A.P. Experiments were conducted by M.Z., C.H., and G.J. Data was analyzed and interpreted by M.Z., C.H., G.J., R.F.S., W.F.R., S.S., and A.P. The manuscript was written by M.Z., S.S., A.P., R. F.S., and W.F.R. All authors have given approval to the final version of the manuscript. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors thank Kay Stubenrauch, Ulrich Essig, and Rudolf Vogel for scientific discussions, and generation and production of assay reagents; Hugues Matile for providing reagents for ELISA 2; Helga Remy and Jean-Luc Mary for providing the ELISA and bioassay data shown in Figure 8; and Jon Kyle Bodnar for thorough review and language editing of the manuscript.
■
REFERENCES
(1) Leader, B.; Baca, Q. J.; Golan, D. E. Nat. Rev. Drug Discovery 2008, 7, 21−39. (2) Walsh, G. Nat. Biotechnol. 2010, 28, 917−924. (3) Elvin, J. G.; Couston, R. G.; van der Walle, C. F. Int. J. Pharm. 2013, 440, 83−98. (4) Huang, L.; Lu, J.; Wroblewski, V. J.; Beals, J. M.; Riggin, R. M. Anal. Chem. 2005, 77, 1432−1439. (5) Liu, Q.; De Felippis, M. R.; Huang, L. Anal. Chem. 2013, 85, 9630−9637. (6) Alvarez, M.; Tremintin, G.; Wang, J.; Eng, M.; Kao, Y.-H.; Jeong, J.; Ling, V. T.; Borisov, O. V. Anal. Biochem. 2011, 419, 17−25. (7) Wu, J.; Pungaliya, P.; Kraynov, E.; Bates, B. Biomarkers 2012, 17, 125−133. (8) Dudal, S.; Staack, R. F.; Stoellner, D.; Scheel Fjording, M.; Vieser, E.; Pascual, M.-H.; Brudny-Kloeppel, M.; Golob, M. Bioanalysis 2014, 6, 1339−1348. H
DOI: 10.1021/acs.analchem.6b03252 Anal. Chem. XXXX, XXX, XXX−XXX
