Simultaneous Catabolite Identification and Quantitation of Large

Subscriber access provided by UB + Fachbibliothek Chemie | (FU-Bibliothekssystem)

Article

Simultaneous catabolite identification and quantitation of large molecular therapeutic protein at intact level by immuno-affinity capture liquid chromatography-high resolution mass spectrometry (LC-HRMS) Lijuan Kang, Raul Camacho, Wenyu Li, Katharine D'Aquino, Seohee You, Vanessa Chuo, Naidong Weng, and Wenying Jian Anal. Chem., Just Accepted Manuscript • Publication Date (Web): 29 Apr 2017 Downloaded from http://pubs.acs.org on May 3, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Simultaneous catabolite identification and quantitation of large molecular therapeutic protein at intact level by immuno-affinity capture liquid chromatography-high resolution mass spectrometry (LC-HRMS) Lijuan Kang1, Raul Camacho2, Wenyu Li2, Katharine D’Aquino2, Seohee You2, Vanessa Chuo 2, Naidong Weng1, Wenying Jian1,3 Departments of 1Pharmacokinetics, Dynamics, and Metabolism (PDM) and 2Cardiovascular Metabolism Janssen Research & Development, Johnson & Johnson 1400 McKean Road, Spring House, Pennsylvania, USA, 19477 3

Corresponding author: [email protected]

Key words: Protein, Bioanalysis, Catabolite Identification, Biotransformation, Intact Analysis, LC-HRMS

1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 32

ABSTRACT As therapeutic recombinant fusion proteins become more widely applicable for the treatment of various types of diseases, there is an increased demand for universal methods such as liquid chromatography (LC)-mass spectrometry (MS) for the determination of their pharmacokinetic properties, particularly, their catabolism. The most common approach of analyzing proteins by LC-MS is to digest them into peptides, which can serve as surrogates of the protein.

Alternatively, we have

developed a novel high resolution mass spectrometry (HRMS) based approach for analyzing large molecule proteins at intact level in biological samples without digestion. We established an immunoaffinity capture LC-HRMS method to quantify the intact parent molecule while simultaneously identifying catabolites for recombinant fusion proteins. We describe this method using dulaglutide, a glucagon-like peptide 1 (GLP1)-Fc fusion protein.

Two proteolytic sites within the GLP1 peptide

sequence of dulaglutide were identified using this novel LC-HRMS analysis in vivo, in mice. These proteolytic sites were identified while the intact molecule being quantified simultaneously. Together with the trypsin digestion based LC-MS/MS analysis using surrogate peptides from different domains of the analyte, an insightful understanding of the pharmacokinetics (PK) and in vivo biotransformation of dulaglutide was obtained. Thus, this method enables simultaneous acquisition of both intact drug concentration and important catabolite information for this recombinant fusion protein, providing valuable insight about the integrity of the molecule and its catabolism in vivo. This is critical for designing and screening novel protein therapeutics and for understanding their pharmacokinetics and pharmacodynamics. With continuing advancement of LC-HRMS and software, this method can be very beneficial in drug discovery and development.


Page 3 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


INTRODUCTION The use of protein therapeutics has grown rapidly in recent decades and continues to play a significant role in many therapeutics areas. In addition to monoclonal antibodies (mAbs) and vaccines, which have wide applications and account for around half of total protein therapeutics that are on the market, there are a variety of other effective protein therapeutics such as hormones, enzymes, and fusion proteins. With advances in molecular biology and protein chemistry, novel forms of protein therapeutics are emerging and flourishing. Examples include genetically or chemically engineered protein therapeutics (like antibody drug conjugates [ADC]), PEG(polyethylene glycol)ylated proteins, and recombinant fusion proteins. These new protein therapeutics are designed to extend half-life or targeted distribution, resulting in enhanced pharmacokinetics (PK)/pharmacodynamics (PD) relationships 1-3. However, many fail in early discovery and development due to a number of challenges including reduced efficacy, increased immunogenicity, and inadequate stability.

Unlike mAbs,

recombinant fusion proteins are subject to proteolytic cleavage or other biotransformation processes in vivo which play a critical role in decreasing half-life and/or activity. Since protein therapeutics are assumed to undergo similar catabolism as endogenous proteins, and bioanalysis of large molecular proteins has been challenging, comprehensive studies of biotransformation of protein therapeutics have rarely been conducted 4. With continuing advancement of new protein therapeutics, the need for novel bioanalytical methods for in-depth determination of the biotransformation of protein therapeutics remains. This will allow for a comprehensive elucidation of biotransformation to aid in the design of next-generation protein therapeutics, as well as enhance the understanding of pharmacokinetics PK data and PK-PD relationships in preclinical and clinical studies 2,4,5. Current commonly used bioanalytical platforms for measuring protein therapeutics include ligand binding assay (LBA) and liquid chromatography coupled with mass spectrometry (LC-MS). LBA,



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 32

the gold standard for quantification of protein therapeutics in biologics matrices, faces challenges in the detection and identification of biotransformation events. It is possible to probe the integrity of a protein using differential LBA, in which combination of two or more assays are used to compare the time-concentration profile of different domains

2,6

. However, the outcome is highly dependent on the

availability of specific antibodies, and still does not reveal exactly where the proteolytic cleavages occur. Because of its capability to elucidate molecular structure with high specificity, mass spectrometry (MS) in contrast, is a more desirable platform for detection and identification of catabolites. Insights on in vivo biotransformation of peptides and proteins using LC-MS have been documented in the literature, although mainly focused on peptides or proteins of small sizes. For example, Liao et al. studied biotransformation of the 34 amino acid human parathyroid hormone in rat tissue homogenate using LChigh resolution mass spectroscopy (LC-HRMS) and successfully identified the major proteolytic site between L15 and N16 7. Copley et al. identified different primary proteolytic sites for exenatide, a 39 amino acid peptide with gluco-regulatory activities similar to glucagon-like peptide-1 (GLP1), in kidney membranes of different species 8.

Na et al. identified degradation products of an antimicrobial

decapepetide in saliva and simulated gastric fluids 9. Hager et al. demonstrated the identification of proteolytic sites of an Fc-fibroblast growth factor 21 (FGF21) fusion protein (molecular weight [MW] 92KDa), a potential drug for treatment of type 2 diabetes, in monkey plasma using differential LBA and HRMS 6. Recently, HRMS methods have also been successfully applied to the characterization of in vivo biotransformation of ADC and to the determination of drug-to-antibody ratio (DAR) distribution

10-13

.

However, examples of comprehensive catabolite identification for large proteins are still very rare. One of the primary reasons is the technical difficulty in analyzing large proteins in biological samples at intact levels using LC-MS. With advancements in both sample preparation techniques and quantitation instrumentation, the capability to analyze large proteins has been greatly enhanced. Recently, we published a method for 4 ACS Paragon Plus Environment

Page 5 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


intact quantitation of large therapeutic proteins by using LC-HRMS. The workflow employed immunoaffinity capture for sample preparation, LC-HRMS for data acquisition, and deconvolution of MS peaks for quantitation 14. This was the first example of direct absolute quantitation of an intact therapeutic antibody in biological samples using deconvoluted MS.

Here we further expand the method to

demonstrate its application in catabolite identification, as well as its capability for simultaneous quantitative and qualitative analysis. The full scan HRMS data acquired in a single run can be processed for both quantitation of intact parent protein and for identification of its catabolites, giving comprehensive insight on the fate of the protein in vivo. This approach not only provides efficient and effective sample analysis, but also maximizes the use of precious biological samples. To the best of our knowledge, this is the first demonstration of a single, integrated workflow to simultaneously conduct catabolite identification and quantitation of a protein therapeutic in biological samples.

EXPERIMENTAL SECTION Reagents and Materials Dulaglutide (Eli Lilly, Indianapolis, IN) was purchased from Myoderm (Norristown, PA). Stable isotope [13C6, 15N4-arginine] and [13C6, 15N2- lysine] labeled internal standard SILu™Mab K4 (IgG4 kappa, Catalog number MSQC7; referred to as SILuMab in this manuscript) were obtained from Sigma (St. Louis, MO). Blank mouse plasma was obtained from BioreclamationIVT (Hicksville, NY). Dynabeads M-280 streptavidin beads were purchased from Life Technologies (Carlsbad, CA). Mouse anti-human IgG Fc antibody was produced internally and biotinylation was done by using EZ-Link Sulfo-NHS-Biotin reagent (Thermo Fisher Scientific, Waltham, MA) according to the protocol provided by the vendor. Protein LoBind plates, LoBind tubes, and plate vortex incubators were purchased from Eppendorf (Hamburg, Germany). Sequencing grade modified trypsin and magnetic nest for 96-well plates were purchased



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 32

from Promega (Madison, WI). Complete protease inhibitor was purchased from Sigma (St. Louis, MO). Dipeptidyl peptidase IV (DPPIV) inhibitor was purchased from EMD Millipore (Billerica, MA). In vivo Mouse Studies All procedures using experimental were approved by the Institutional Animal Care and Use Committee of Janssen Pharmaceutical Companies. Dulaglutide was administered subcutaneously to 27 diet-induced obese (DIO) mice (~45 grams) that had been on a high-fat diet (Research Diet, D12492, for 12 weeks) at a dose of 10 nmol/kg. Blood was collected into K2EDTA coated tubes containing complete protease and DPPIV inhibitors, from 3 mice per time point at 2, 4, 7, 24, 48, 72, 96, 120, and 144 hours. Blood was collected in a similar manner from three mice not dosed with dulaglutide, for the 0 hour time point. Plasma samples were subsequently harvested and stored at -80 °C prior to analysis. Sample Preparation for Intact Assay Dulaglutide (3 mg/mL) was spiked into blank mouse plasma to make a 100 µg/mL primary plasma stock, which was further serially diluted in mouse plasma to prepare calibration standards at 10, 8, 4, 2, 0.5, 0.1, 0.05, and 0.01 µg/mL. The quality control (QC) samples were prepared in the same way in blank mouse plasma at 0.15 (Low A QC), 0.3 (Low B QC), 5 (Mid QC), and 7.5 (High QC) µg/mL. SILuMab was used as the internal standard (IS). SILuMab was dissolved with 0.1% formic acid in water to 0.2 mg/mL, which was diluted in phosphate buffered saline (PBS) to 0.015 mg/mL. LoBind tubes and plates were used for the entire process. For each sample, 100 µL of Dynabeads M-280 streptavidin beads were used. The beads were prepared as a pool in tubes based on the total volume of each analytical run. Each 100 µL aliquot of the beads was washed with 100 µL of PBST (0.02% Tween 20 in PBS v:v) twice, had 7 µg of biotinylated mouse anti-human IgG Fc monoclonal antibody added, and was incubated on a shaker at 2000 rpm for 1 hour at room temperature. At the


Page 7 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


end of the incubation, the beads were washed with 100 µL of PBST twice, re-suspended in 100 µL of PBST and pipetted into a 96-well plate. Next, 600 µL of PBST, 100 µL of a plasma sample, and 20 µL of 0.015 mg/mL SILuMab (IS solution) was added to each well, and was incubated on a plate shaker at 1100 rpm for 2 hours at room temperature. The supernatant was removed and beads were washed two times with 300 µL of PBS, and once with 300 µL of 50 mM ammonium bicarbonate (pH 7.5) using an automated liquid transferring system for 96-well plates (Tomtec, Hamden CT). A magnet nest for 96well plates (Promega) was used to separate the beads. 100 µL of 0.1 M glycine HCl (pH 2.5-3) was added to each well for elution. The eluent was transferred to a new 96-well plate by Tomtec. Samples were neutralized by adding 50 µL of 50 mM ammonium bicarbonate (pH 9.3). 60 µL of sample was injected to LC-MS system. Sample Preparation for Trypsin Assay Dulaglutide (3 mg/mL) was spiked in blank mouse plasma to make a 50 µg/mL primary plasma stock, which was further serially diluted in mouse plasma to prepare calibration standards at 10, 8, 4, 2, 1, 0.2, and 0.05 µg/mL. The QC samples were prepared in the same way in blank mouse plasma at 0.15 (Low QC), 5 (Mid QC), and 7.5 (High QC) µg/mL. SILuMab was used as the IS. SILuMab was dissolved with 0.1% formic acid in water to 0.2 mg/mL, which was further diluted in PBS to 0.01 mg/mL. The procedure for preparation of beads was the same as for the intact assay. Then, 400 µL of PBST, 20 µL of plasma sample, 50 µL of beads and 10 µL of 0.01 mg/mL SILuMab (IS solution) was added to each well, and was incubated on a plate shaker at 1100 rpm for at least 2 hours at room temperature. The supernatant was removed and the beads were washed three times with 300 µL of PBS using Tomtec. A magnet nest for 96-well plate (Promega) was used to separate the beads. The digestion solution was prepared by mixing 50 mM ammonium bicarbonate (pH 8.0), acetonitrile, and 0.5 µg/µL of sequencing grade modified trypsin in 50 mM acetic acid (8:2:0.4, v:v:v). 100 µL of the digestion 7 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 32

solution was added to each well. The plate was incubated overnight on a plate shaker at 1100 rpm at 37 ° C. The next day, the reaction was stopped by adding 10 µL of 10% formic acid in water. The plate was centrifuged at 3500 rpm for 1 minute and then placed on a magnet to separate the beads. The supernatant was transferred to a new plate by Tomtec. 20 µL was used for LC-MS/MS analysis. LC-MS Condition For the intact LC-MS analysis, the HPLC system consisted of a Shimadzu LC20AD Prominence pumps and a SIL-HTc autosampler (Columbia, MD). The HPLC system employed an Aeris widepore C4 column (2.1x50 mm, 3.6 µm, 00B-4486-AN, Phenomenex, Torrance, CA). Mobile phase A was 0.1% formic acid in water and mobile phase B was 0.1% formic acid in 90% acetonitrile. Needle wash was 0.1% trifluoroacetic acid in 50% acetonitrile in water. The gradient elution ramped linearly from 10% to 90% B over 4.5 minutes, held at 90% for 0.3 minute, and then returned to 10% B in 0.1 minute, and finally maintained at 10% B until the end of the run at 6 minutes. The HPLC flow rate was 0.4 mL/min. The time-of-flight (TOF) MS was carried out with a Triple-TOF API5600 mass spectrometer (Applied Biosystem, Foster City, CA) in positive electrospray mode. The ion source parameters in DuoSpray Ion Source were: curtain gas 30 psi, GAS1 50 psi, GAS2 50 psi, ionspray voltage 5000 V, and source temperature 500 °C. The declustering potential was 150 V. The collision energy was 20 V. The TOF masses range (m/z) was between 1000 to 3500. The accumulation time was 2 seconds, and time bins to sum was 40. The mass spectrometer was operated in the Intact Protein Mode. Mass calibration was conducted using APCI positive calibration standard (Applied Biosystem, Foster City, CA) delivered at a speed of 500 µL/min for 2 minutes by the Calibration Delivery System (CDS) every 5-10 injections. For LC-MS/MS analysis of the trypsin assay, the HPLC system consisted of a Shimadzu LC20AD Prominence pumps and a SIL-20AC autosampler (Columbia, MD). The HPLC system employed a Zorbax SB-C8 column (2.1x50 mm, 3.5 µm, 871700-906, Agilent, Santa Clara, CA). Mobile phase A was 0.1% 8 ACS Paragon Plus Environment

Page 9 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


formic acid and 10 mM ammonium formate in water and mobile phase B was 0.1% formic acid and 10 mM ammonium formate in 90% acetonitrile. Needle wash was 0.1% trifluoroacetic acid in 50% acetonitrile in water. The gradient program was as follows: started at 30% and held for 0.5 minute, then increased linearly to 60% B over 1.5 minutes, increased linearly to 90% over 0.5 minute, held at 90% for 0.5 minute, and returned to 30% B in 0.1 minute and finally maintained at 30% B until the end of the run at 4 minutes. The HPLC flow rate was 0.4 mL/min. The multiple reaction monitoring (MRM) analysis was carried out on a triple quadrupole API5000 mass spectrometer (Applied Biosystem, Foster City, CA) in positive electrospray mode. The ion source parameters in Turbo Ionspray mode were: curtain gas 30 psi, GAS1 50 psi, GAS2 50 psi, Ionspray voltage 5000 V, and source temperature 500 °C. The entrance potential (EP) was 10 V, decluster potential (DP) was 120 V, collision energy (CE) was 25 V and exit potential (CXP) was 15 V for all the ions.

The MRM transitions for surrogate peptide

HGEGTFTSDVSSYLEEQAAK (HGE) were of 719.3→ 932.3 (HGE3+ to b9+) and 719.3 →788.4 (HGE3+ to y7+), and the summation of the two transitions were used for quantitation. The MRM transitions for VVSVLTVLHQDWLNGK (VVS) and its internal standard [13C6, 15N2- lysine] VVS were 603.3 → 805.4 (VVS 3+ to y142+) and 606.0 → 809.4 ([13C6, 15N2- lysine] VVS3+ to y142+). Data Processing For catabolite identification, the raw spectra of the chromatographic peak with a window from 2.79 to 4.05 minutes was subjected to deconvolution analysis by BioPharmaViewTM(BPV) Software with a resolution of 2500 and MS range of 1000-3500. For the intact protein quantitation, a novel custom modified version of the Sciex MultiQuantTM software was used, as described in detail 14. The commercial version of this LC-MS processing software performed automated peak-finding and integration of extracted ion chromatograms, followed by calculation of analyte concentrations using calibration standards. The modification to this version of 9 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 32

software involved the addition of an extra deconvolution step to convert the multiply charged m/z data to a neutral spectrum as well as automated data processing for quantitation based on the neutral spectrum. For this work the m/z spectrum in a window of 1 minute around the LC peak (3.14 minutes, window of 2.64-3.64 minutes) was extracted and deconvoluted. A resolution of 2500 was used and the input mass range was restricted to 1000 to 3500 Da. The noise level of 10% and splitting factor of 2 were applied for the peak integration. The calibration curve was established using a linear regression with 1/x2 weighting.

For both dulaglutide and the internal standard, there were multiple peaks

corresponding to different glycosylation forms. Peak heights of the most intensive molecule ion (most abundant glycosylation form) of the intact dulaglutide (62561±1 Da) and that of the internal standard (148048±1 Da) were used for quantitation. Height instead of peak area was used for quantitation because we have found that peak height more reliably quantifies than peak area when deconvoluted peaks are used for quantitation14. For LC-MS/MS quantitation based on trypsin digestion, the peaks of the surrogate peptides HGE and VVS., and the [13C6,

15

N2- lysine] labeled VVS (internal standard) were integrated by the Sciex

AnalystTM software. Peak area ratio of analyte/internal standard was used for quantitation. A quadratic regression with weighing of 1/x2 was used. Functional Bioassay The levels of bioactive dulaglutide in plasma samples were quantitated in a cell based bioassay using clonal HEK293 cells stably expressing the human GLP1 receptor. Cells were treated with either compound standards or diluted plasma and the accumulation of intracellular cyclic adenosine monophosphate (cAMP) was measured with a LANCE FRET based competitive cAMP immunoassay (Perkin Elmer, Akron, OH). The level of cAMP induction measured in plasma samples was used to determine the concentration of functional dulaglutide in the sample by interpolation from a standard 10 ACS Paragon Plus Environment

Page 11 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


curve of known dulaglutide concentrations. Standards and quality controls were prepared in plasma diluted to 50% in assay diluent (calcium and magnesium free Hanks buffered saline, 5mM 4-(2hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 0.1% bovin serum albumin (BSA), 5mM ethylenediaminetetraacetic acid (EDTA), 0.5mM 3-isobutyl-1-methylxanthine (IBMX), protease inhibitors (Roche Diagnostics, Indianapolis)). Unknown plasma samples were prepared at various dilutions in control plasma/assay diluent to hold the concentration of plasma at 50%. All diluted samples were assayed in quadruplicate. Data was analyzed using GraphPad Prism (GraphPad Software San Diego). RESULTS AND DISCUSSION Workflow In general, there are two approaches for LC-MS analysis of therapeutic proteins. One is the bottom-up approach which involves digesting the protein into peptides and analyzing a surrogate peptide by LC-MS/MS. This is a commonly used approach for analysis of protein therapeutics in biological samples by LC-MS because it provides high specificity, high sensitivity, and multiplex quantitation. With suitable surrogate peptides, it can provide concentrations of different domains of the protein. It is similar to differential LBA but with higher specificity and is less dependent on critical reagents, which are essential for LBA. Despite its capability for probing the integrity of different domains of protein therapeutics with proper surrogate peptides, it is a targeted analysis and therefore cannot reveal unknown proteolytic sites, which renders it unsuitable for catabolite identification. Peptide mapping, another type of bottom-up approach, involves untargeted, comprehensive analysis of peptides generated using enzymatic digestion to confirm primary sequence of a protein and certain chemical modifications such as oxidation on amino acid residues. However, digesting the protein into peptides significantly increases the complex of the samples while eliminating the context of the sequences. Peptides derived from proteolytic catabolism may not be differentiable from peptides 11 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 32

generated by enzymatic digestion. Overall, peptide mapping currently is not a practical or efficient way of conducting catabolite identification. The other approach for protein bioanalysis is top-down. It analyzes the protein therapeutics at the intact level. In contrast to the bottom-up approach, top-down analysis provides the unique advantage of being able to reveal the integrity of the whole protein and to elucidate sequence. As such, it is the method of choice for structural elucidation. However, until recently, top-down analysis has been limited to qualitative structural characterization of pure protein materials at high concentrations in protein production or manufacture processes. In comparison, analyzing large protein therapeutics at intact levels in biological samples from in vivo studies were very challenging due to technical issues associated with their high molecular weights and complex biological matrix. As molecular weight increases, the ionization efficiency diminishes leading to lower signal intensity which is further diluted by multiple isotope and charging peaks, resulting in significantly poorer sensitivity and resolution. In addition, components in biological samples, if not effectively cleaned-up, causes interference in MS detection.

Recently, with advances in sample preparation, LC-HRMS

instrumentation, and data processing software, we have explored the possibility to conduct intact bioanalysis of large proteins and established a novel workflow to quantify intact therapeutic antibody in biological samples by LC-HRMS 14. In that workflow, the analyte was extracted using immuno-affinity capture and analyzed by LC-HRMS. The full scan data was deconvoluted to reveal a zero charged molecule peak of the analyte, which was used for quantitation. It was the first example to demonstrate quantitation of intact large protein therapeutics (a mAb) in plasma by using deconvoluted HRMS data. In general, the full scan data acquired by HRMS for intact analysis can be processed in two ways: extracted ion chromatogram (XIC) and deconvolution 15. When individual isotope peaks in each charging state are resolved, the isotope peaks can be processed to generate XIC for quantitation.

Varying

charging states, isotope peaks, width of extraction windows can be evaluated to obtain optimal quantitation performance. However, XIC approach is only suitable for small proteins as it requires 12 ACS Paragon Plus Environment

Page 13 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


resolution between isotope peaks, for which the MS should have at least four-times the molecular mass of the target analyte. To quantify an intact antibody by this approach, the optimal resolving power is 600 k, which is not feasible with current state-of-art HRMS 16. In addition, XIC does not reveal unknown components in the data and therefore not applicable for catabolite identification. In comparison, the alternative approach of deconvolution is based on multiple charged peaks and therefore does not require resolution of isotope peaks. For large proteins which are heavily charged and signal being diluted among different charging states, deconvolution can effectively utilize all the signals from the charged ions and therefore improve detection sensitivity and also decrease variability associated variable signal distribution among different charging states. More importantly, deconvolution can effectively reveal existence of proteolytic catabolites. Therefore, in this paper, we describe a further extension of our previously published deconvolution workflow for catabolite identification. When a proper capturing antibody is used in the immuno-affinity procedure, catabolites can be extracted together with the unchanged parent molecule and detected by the LC-HRMS full scan analysis. They can be identified by examining the deconvoluted MS spectra. All the information, including quantitative concentration data of the parent molecule and the qualitative structure data of catabolites, can be acquired simultaneously in a single sample preparation and analytical injection. To our best knowledge, this is the first time that an integrated workflow has been established for simultaneous catabolite identification and quantitation of intact therapeutic proteins in biological samples. In this work, the application of this integrated workflow is demonstrated by using dulaglutide as a model molecule. The sequence and structure of dulaglutide is shown in Figure 1. It is a recombinant protein consisted of dimer of GLP1 variant (GLP1 7-37) fused to Fc portion of a human IgG4 17. A DIO mouse PK study of dulaglutide was conducted and the samples were analyzed using the integrated top-



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 32

down workflow to provide both a time-concentration PK profile of intact molecule and catabolite information. A bottom-up approach using LC-MS/MS to analyze the surrogate peptides generated by trypsin digestion was also performed, allowing for a more sensitive and specific quantitation of different subspecies of the molecule. The data sets provide a comprehensive picture of the in vivo fate of the molecule. The entire workflow is shown in Figure 2. It began with immuno-affinity capture using biotinylated anti-human IgG Fc antibody on streptavidin magnetic beads. For the top-down intact assay, the captured intact protein was eluted from beads and injected into a LC-HRMS system. The full scan HRMS data was processed for quantitation of intact molecule and elucidation of catabolites. For the bottom-up trypsin assay, the captured analyte was digested by trypsin and injected into a LC-MS/MS system. The N-terminal tryptic peptide of GLP1, HGEGTFTSDVSSYLEEQAAK (GLP1 7-27 or Dula 1-20, referred as HGE), is used as the surrogate peptide to represent the fusion protein containing at least one copy of intact GLP1 sequence.

A peptide located on the Fc portion of the molecule,

VVSVLTVLHQDWLNGK (referred as VVS), is used as the surrogate peptide to represent total Fc level, regardless of whether or not the fusion protein still contains intact GLP1 peptides. The location of the surrogate peptides in dulaglutide is depicted in Figure 1. Quantitation Both intact and trypsin assays were conducted to quantify dulaglutide in the DIO mouse PK study. For the intact assay, the quantitation was performed using a deconvoluted dulaglutide peak (MW 62561 Da) by MultiQuantTM software. The software deconvoluted the peak and processed the data for quantitation automatically. Peak height, instead of peak area was used for quantitation because we have found that peak height quantifies more reliably than peak area when deconvoluted peaks are used for quantitation

14

. In theory, in order to reliably quantify analyte using peak height, the peak width

should remain constant, which is expected for deconvoluted peaks because the peak width is


Page 15 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


determined by the deconvolution algorithm.

Upon inspection of the deconvoluted peaks of the

samples, the peak widths are overall constant except for those near LLOQ which are in general wider due to contribution from neighboring interference peak. For those samples, we have found that peak area is more severely impacted by neighboring peaks and it is often difficult to determine the start and end of a peak. In comparison, peak height is less affected by interference and easier to determine, and therefore provided more reliable quantitation. The deconvoluted spectrum for Blank, STD1 (0.05 µg/mL) and Mid (5 µg/mL) QCs are shown in Figure 3. The dulaglutide peak was shadowed and showed mass accuracy within ±1 Da. The assay had a standard curve range of 0.05 – 10 µg/mL and a linear regression was used with R2 = 0.995. Carryover was observed in the intact LC-MS analysis. The blank sample following STD8 showed analyte peak height at 3.5% of STD8. At least 3 blank samples were injected after STD8 or High QC sample to minimize the impact of carryover. Study samples were injected in the order of sequential pharmacokinetic time points and impact of carryover is minimal. Standard calibrators and QCs passed the 20% accuracy criteria with passing rate of 8/8 and 6/8 for calibrators and QCs, respectively. The intact assay achieved LLOQ of 0.05 µg/mL based on acceptable bioanalytical assay performance of % Bias being within ±20%. This LLOQ was sufficient to measure all the time points except for 144 hours. This LLOQ was identical to that of the trypsin assay, but required five times more plasma volume and three times more injection volume. As the assay required 100 µL of plasma, pooled samples from 3 animals were used. This sample requirement for the intact assay is a limitation for using this approach to analyze biological samples. Improving the sensitivity will be critical for the wider application of intact assay in quantitation. For the trypsin assay, the quantitation was performed using the surrogate peptides HGE and VVS, which represent molecule containing intact GLP1 molecule and total Fc level, respectively.

AnalystTM software was used in data acquisition and

regression. The chromatograms for Blank, STD1 (0.05 µg/mL) and Mid (5 µg/mL) QCs are shown in Figure 4. The retention times for HGE and VVS were 2.33 and 2.66 minutes, respectively. The assay had



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 32

a standard curve range of 0.05 – 10 µg/mL and quadratic fittings were used with R2 = 0.997 and 0.998 for HGE and VVS, respectively. Standard calibrators and QCs passed the 15% accuracy criteria with passing rate of 7/7 and 6/6 for calibrators and QCs respectively. The sample requirement for the trypsin assay was 20 µL of mouse plasma. A stable isotope labeled human monoclonal antibody (SILu™Mab from Sigma) was used as internal standard for both intact and trypsin assay. The main purpose of including this internal standard is to track with the analyte in the immuno-affinity capture, which is based on recognition of Fc region of the analyte by the capturing antibody immobilized on magnetic beads. By using an internal standard also containing human Fc region and being captured based on the same mechanism, recovery loss or binding saturation during this step can be effectively compensated for. For trypsin assay, the internal standard provided additional benefit of compensating for variability in the digestion process. Even though the size of the internal standard is much larger than dulaglutide, it is proven to give satisfying performance. With both top-down and bottom-up approaches, multiplex information reflecting the behavior of different domains of the molecule in vivo was obtained (Figure 5). Total Fc levels measured by VVS surrogate peptide were essentially maintained at the same level after the Cmax was reached. This observation is consistent with the long half-lives of mAbs and Fc in vivo from neonatal Fc receptor (FcRn) recycling 1. The HGE surrogate peptide is located at the N-terminal of GLP1 peptide and represents an intact N-terminus, which is critical for bioactivity. It is well known that loss of even a single amino acid from the N-terminus renders the peptide inactive

18

. Therefore, HGE peptide is expected to reflect

active dulaglutide. The intact assay represents dulaglutide with unaltered molecule weight, and thus, the bioactive form. Samples were also measured by a human GLP1 receptor bioassay shown in Figure 5.


Page 17 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


No significant differences were observed between the bioassay data and HGE or intact data, confirming that the latter two analytes reflect the levels of bioactive species in the samples. In the time-course profile, the most obvious observation is the faster decrease of concentrations of HGE and intact than that of total Fc. This suggests that there are proteolytic activities happening in vivo to the molecule, particularly between the N-terminus of the GLP1 peptide and Fc region. To identify the exact cleavage site(s), the data acquired in the intact assay were analyzed to reveal the existence of catabolites, which in turn, explains the profile of different species in the time-course data. Catabolite Identification The power of this intact workflow is that it enables simultaneous acquisition of quantitation and catabolite data from a single information-rich full scan HRMS data file. The data was processed to generate a deconvoluted mass spectrum, which contains not only the molecule ion of the unchanged parent molecule, but also those of catabolites corresponding to proteolytic products. As shown in Figure 6, the molecule peak of dulaglutide at 62561 Da diminished over time, while two clusters of peaks at lower mass ranges around 57 kDa and 59 kDa started to appear and increase over time. After calculation of mass loss of different potential proteolytic products based on the amino acid sequence of dulaglutide, two major proteolytic sites at the C-terminus side of W25 and L26, respectively, were identified, which contribute to the clusters of catabolite peaks. For better clarification, we used the two adjacent amino acids to describe the proteolytic sites. Proteolysis of the amide bond between W25 and L26 is designated as W25L26, whereas proteolysis of the amide bond between L26 and V27 is designated as L26V27. As shown in figure 6, from 2 to 120 hours, the dulaglutide peak (MW62561 Da) gradually diminished, and peaks corresponding to proteolysis at W25L26 (MW 59777 Da) and L26V27 (MW 59661 Da) on one copy of the two GLP1 peptides (single chain proteolysis) increased and then decreased. Peaks corresponding to proteolysis on both copies of the GLP1 peptide (double chain 17 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 32

proteolysis), W25L26+W25L26 (MW 56991 Da), W25L26+L26V27 (MW 56878 Da) and L26V27+L26V27 (MW 56765 Da) gradually increased and became the major peaks at 120 hours. To better illustrate the identification of the catabolites, the deconvoluted intact spectra of three representative time points of 24, 72 and 120 hours were overlayed (Figure 7). Figure 7A with a broader molecular weight range showed the relative intensity of dulaglutide and clusters of different catabolite peaks at different time points. Dulaglutide was the major peak at 24 hours (blue line) while the products of double chain proteolysis became the major peaks at 120 hours (red lines), revealing in vivo biotransformation of the intact dulaglutide molecule to double chain proteolysis catabolites. Figure 7B shows a focused intact spectrum for the double chain proteolytic peaks. The double chain proteolytic peaks at W25L26+W25L26, W25L26+L26V27, and L26V27+L26V27 are shown (arrows) in figure 7B. Each of the labeled peaks represents the major glycosylation form, the G0F/G0F form

19

.

They are

accompanied by peaks of minor glycosylation forms (not labeled). For example, one of the minor glycosylation forms of L26V27+L26V27 is at 56928 Da. It differs from the abundant G0F/G0F form by 162 Da, most likely being G0F/G1F. The height of the major glycosylation form of each catabolite was plotted for a time profile to reveal kinetics of each catabolism event (Figure 8). Looking at this plot, it was interesting to find that W25L26 as well as W25L26+W25L26 reached the largest peak height at 24 hours, L26V27 and W25L26+L26V27 at 72 hours, and L26V27+ L26V27 kept increasing till 144 hour. This observation suggests that at early time points W25L26 is the major proteolytic site, and with increasing time L26V27 becomes the major one. A transitioning between these two proteolytic events was clearly captured by the W25L26+L26V27 which peaked at 72 hours. Figure 7C shows a focused intact spectrum for the single chain proteolytic peaks. The appearance of single chain proteolysis at W25L26 and L26V27 at earlier time points and their decrease over time implies that they are the initiating events for the double chain proteolysis and that these two sites are the proteolytic labile sites for dulaglutide in vivo in mice. The existence of single chain proteolysis catabolites explains the slightly lower concentration 18 ACS Paragon Plus Environment

Page 19 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


measured by the intact assay than that of HGE quantified by the trypsin assay (Figure 5). The single chain proteolysis catabolites are not quantifiable in the intact assay because of their MW shift from the parent protein, while the remaining unchanged copy of GLP1 still elicits the HGE signal in the trypsin assay. Figure 7D shows the major proteolytic sites (red arrows) of GLP1 (7-37). The major proteolytic sites, observed mass, observed mass loss and theoretical loss are summarized in Table 1. Only the MW of the most abundant glycosylation form of each species is listed. In addition to the major catabolites, a few minor peaks were also observed but their identification could not be confirmed. In Figure 7B, the peak at 56665 Da with the mass loss of 5897 Da matches the double chain proteolysis of either L26V27+V27K28 or A24W25+G31G32. The peak at 56538 Da with the mass loss of 6024 Da matches the double chain proteolysis of L26V27+K28G29 or W25L26+G30G31. It is impossible to assign a catabolite peak solely based on loss of mass. When the double chain proteolysis peaks are present, their precursors (single chain proteolysis products at corresponding sites) are expected to also be observed.

However, there are no observed peaks

corresponding to proteolysis on single chain at A24W25, V27K28, K28G29, G30G31, and G31G32. In Figure 7C, the peak at 60294 Da with the mass loss of 2268 Da matches proteolysis at E21F22. However, its double chain proteolysis peak is not observed either. The absence of the corresponding single or double chain catabolites of the observed minor peaks may be due to true absence, or due to their instability and lack of accumulation. In these cases, the observed minor peaks remain unidentified. It should be noted that the LC run time for the intact analysis was relatively short (6 min) to provide needed throughput for bioanalytical support of discovery studies. Under this short run time, the analytes, their isoforms, as well as catabolites are all co-eluted.

Initially during method

development, longer run on longer columns for more extensive chromatographic separation was evaluated. No significant improvement in peak shape or separation was achieved even when the run



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 32

was extended up to 60 min. Therefore, instead of using LC to resolve the components in the samples, we relied on the separation power of MS to differentiate them. This reserves the needed throughput without losing the quantitative and qualitative information of the major components in the samples. Under co-eluting condition, care should be given to prevent large proteins from generating fragments in the ion source (e.g. by lowering declustering potential), which may be mistaken for catabolites. In addition, the spectra should be closely inspected for artifacts caused by deconvolution process. For example, peak at molecule weight of 2x, half, 1/3rd, 1/4th, etc, of the main peak may be artifacts due to deconvolution rather than real. In this case, it is helpful to inspect the raw spectra to confirm the identity of the peaks. Previously, Murphy et al. also conducted catabolite identification for GLP1 fused to a mAb using immunoassay and LC-MS 20. A catabolite resulting from the loss of the first two N-terminal amino acids (HA) of the GLP1 peptide mediated by dipeptidyl peptidase IV (DPPIV) was identified in mouse. Because dulaglutide contains the A2G mutation to prevent DPPIV cleavage, our study revealed different proteolytic sites, potentially due to a combination of factors including the difference in GLP1 sequences and the difference in the fusion partner. The knowledge gained by elucidating the biotransformation and resulting catabolites is expected to guide drug candidate optimization to enable engineering of more stable proteins. However, caution must be taken when utilizing the information from observed catabolites. Proteolysis in vivo is a complicated process affected by the ever-changing physical and chemical environments of the circulating protein. Identified catabolites may sometimes be stable intermediates of many sequential proteolysis events 2.

Identification of the initial proteolytic event that leads to relatively stable

intermediates is key for improving overall in vivo stability. Unfortunately, in most cases these initial catabolites/intermediates are short-lived and do not accumulate. There may also be many other


Page 21 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


catabolites which are not observable due to their short half-life in vivo. It is important to bear in mind that modifying proteolytic sites to those of identified stable intermediates may just shift the catabolite profile, which may or may not improve the overall in vivo stability. Nevertheless, there are examples where engineering based on the identified catabolite site has improved resistance to proteolysis 21. Overall, we have elucidated two major proteolytic sites on the GLP1 peptides in dulaglutide in mice in vivo by using LC-HRMS analysis. The sample analysis was conducted simultaneously with quantitation of the intact molecules.

Together with the bottom-up based LC-MS/MS analysis of

surrogate peptides from different domains of the analyte, an insightful understanding of the in vivo PK and catabolism of dulaglutide has been gained. In our laboratory, the workflow has been used for higher MW molecules and has been demonstrated to be effective for proteins larger than antibodies. It must be noted that an appropriate capturing antibody (e.g. an anti-Fc in the current paper) is essential in order to achieve effective catabolite identification. The antibody should recognize a stable portion of the molecule to capture all catabolite-containing components from the samples. If an antibody for GLP1 peptide had been used, depending on the epitopes, certain catabolites may not have been captured. SUMMARY AND FUTURE PERSPECTIVE We have established an immuno-affinity capture LC-HRMS workflow for simultaneous quantitation and catabolite identification of large protein therapeutics in plasma samples. In this integrated workflow, the full scan HRMS data acquired in a single run can be processed for both quantitation of intact parent protein and for identification of its catabolites. Together with data acquired by bottom-up LC-MS/MS analysis, a comprehensive insight of the fate of the protein in vivo can be gained. This approach not only provides efficient and effective sample analysis, but also maximizes the use of precious biological samples.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 32

In our opinion, there are two main areas where this current workflow could be improved for future applications: sensitivity and data processing. With regards to sensitivity, in our lab, the lowest quantitation limit for proteins the size of a mAb (~150 kDa) in biological samples such as plasma is still 20-100 times lower than that of a trypsin assay, not to mention LBA, which usually gives sensitivity orders of magnitude higher. Limited sensitivity may prevent effective detection and quantitation of low level catabolites as well as intact quantitation in cases when the analyte concentration is low. Substantial improvement in sensitivity is critical for wider and more effective application of the current workflow. With regards to data processing, unlike small molecules for which there are numerous datamining tools to facilitate assignment of detected metabolite peaks, there is no software available for automatic catabolite assignment. Deconvoluted spectra have to be manually inspected for catabolite peaks, for which the mass loss from parent protein is calculated and matched with theoretical value based on the protein sequence. Software for automated catabolite assignment is highly desirable and will greatly improve the efficiency of protein catabolite identification. Although still in its infancy, this workflow has already been implemented in our discovery work and provides important information which has greatly facilitated understanding in vivo stability of protein molecules and shed light on design and screening of drug candidates. With the continued advancement of LC-HRMS instrumentation and software, the workflow can have broader application to support drug discovery and development.


Page 23 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. (A) Amino acid sequence of single chain dulaglutide. The surrogate peptides for bottom-up trypsin assay are underlined. (B) Schematic representation of dulaglutide structure. A

10 20 30 40 50 60 HGEGTFTSDV SSYLEEQAAK EFIAWLVKGG GGGGGSGGGG SGGGGSAESK YGPPCPPCPA 70 80 90 100 110 120 PEAAGGPSVF LFPPKPKDTL MISRTPEVTC VVVDVSQEDP EVQFNWYVDG VEVHNAKTKP

B GLP1 variant peptide Linker peptide

130 140 150 160 170 180 REEQFNSTYR VVSVLTVLHQ DWLNGKEYKC KVSNKGLPSS IEKTISKAKG QPREPQVYTL 190 200 210 220 230 240 PPSQEEMTKN QVSLTCLVKG FYPSDIAVEW ESNGQPENNY KTTPPVLDSD GSFFLYSRLT

IgG4-Fc domain

250 260 270 VDKSRWQEGN VFSCSVMHEA LHNHYTQKSL SLSLG



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 32

Figure 2. Diagram of workflows for top-down intact quantitation and catabolite identification (left) and bottom-up trypsin digestion quantitation (right).

Dulaglutide in Mouse Plasma

Affinity Capture by Biotinylated Anti-human IgG Fc Antibody

Elution from the Beads

Digestion by Trypsin Overnight

LC-High Resolution MS Analysis of Whole Protein

LC-MRM Analysis of Surrogate Peptides

Top-Down Approach: Catabolite ID Intact Quantitation

Bottom-Up Approach: Quantitation Using Surrogate Peptides Trypsin Assay: VVSVLTVLHQDWLNGK (Total Fc) HGEGTFTSDVSSYLEEQAAK (Dula 1-20)


Page 25 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3. MultiQuant deconvoluted intact mass spectra for LC-HRMS quantitation of dulaglutide in mouse plasma, using the intact assay. (A) Blank, (B) STD1 (0.05 µg/mL), and (C) Mid QC (5 µg/mL).



Figure 4. Chromatogram for LC-MS/MS quantitation of dulaglutide in mouse plasma using trypsin assay. (A) HGE Blank; (B) HGE STD1 (0.05 µg/mL); (C) HGE Mid QC (5 µg/mL); (D) VVS Blank; (E) VVS STD1 (0.05 µg/mL); (F) VVS Mid QC (5 µg/mL).

350 HGE blank

D

150 VVS Blank

Intensity, cps

A

100

300

Intensity, cps

250 200 150 100 50 0

B

HGE STD1

2.33

E

600 400 200 0

C

Intensity, cps

800

600

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 Time, min VVS STD1

2.66

500 400 300 200 100 0

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 Time, min

1.2e5 HGE Mid

2.33

F

1.0e5

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 Time, min

1.4e5 1.2e5

VVS Mid

Intensity, cps

Intensity, cps

1.0e5

8.0e4 6.0e4 4.0e4 2.0e4 0.0

50

0

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 Time, min

1000 Intensity, cps

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 32

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 Time, min

2.66

8.0e4 6.0e4 4.0e4 2.0e4 0.0

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 Time, min


Page 27 of 32

Figure 5. Time-course profile of dulaglutide in DIO mice after single subcutaneous dose (10 nmol/kg), n=3, ± SEM (standard error of the mean).

Time-concentration Profile of Dulaglutide in Mice after Single Subcutaneous 10 nmol/kg dose (n=3, ±SEM) 100

Concentration (nM)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


10

HGE (Dula 1-20) 1

VVS (total Fc) Intact Bioassay

0.1 0

20

40

60

80 Time (hr)

100

120

140

160



Figure 6. BioPharmaView deconvoluted intact mass spectra for mouse in vivo samples using intact assay for catabolite identification. (A) Pooled 2 hours, (B) Pooled 24 hours, (C) Pooled 72 hours, (D) Pooled 120 hours.

Reconstructed Intensity

A

1400

62562.44

Pooled 2h

1200 1000 800 600 400 200 0 5.1e4


B

2500

C

5.3e4

5.4e4

5.5e4

5.6e4

5.7e4

5.8e4

5.9e4 6.0e4 Mass, Da

6.1e4

6.2e4

Pooled 24h

6.3e4

6.4e4

6.5e4

6.6e4

6.7e4

6.5e4

6.6e4

6.7e4

62561.97

1500 56991.63 59776.93

1000 56877.67 500

1400


5.2e4

2000

0

59661.17

5.1e4

5.2e4

5.3e4

5.4e4

5.5e4

5.6e4

5.7e4

5.8e4


6.1e4

6.2e4

6.3e4

6.4e4

56878.48

Pooled 72h

56765.87 1200 1000 800

56991.57

600

59776.97

62562.37

400 59662.19

200 0 5.1e4

D

5.2e4

5.3e4

5.4e4

5.5e4

5.6e4

Pooled 120h Reconstructed Intensity

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 32

5.7e4

5.8e4


6.1e4

6.2e4

6.3e4

6.4e4

6.5e4

6.6e4

6.7e4

56766.01

1500

56878.25

1000

500 56991.21 59777.05

62562.03

0

5.1e4

5.2e4

5.3e4

5.4e4

5.5e4

5.6e4

5.7e4

5.8e4


6.1e4

6.2e4

6.3e4

6.4e4

6.5e4

6.6e4

6.7e4


Page 29 of 32

Figure 7. BioPharmaView deconvoluted intact mass spectra for mouse in vivo samples using intact assay for catabolite identification. The deconvoluted intact spectra of three representative time points of 24 hours (blue), 72 hours (purple) and 120 hours (red) are superimposed. (A) MW ranging from 50000 to 65000 Da; (B) Focused spectra for double chain proteolysis peaks, MW ranging from 56100 to 57900 Da; (C) Focused spectra for single chain proteolysis peaks, MW ranging from 59400 to 60400 Da. (D) GLP1 peptide sequence of dulaglutide with red arrows indicating the proteolytic sites.


A

Pooled 24h Pooled 72h Pooled 120h 2000

56766.01

1500

56765.87

2500

62561.97

56991.63 56991.57 56877.67

1000 500

59776.93 62366.28

62725.30

59662.19 0

5.1e4

5.2e4

5.3e4

5.4e4

5.5e4

5.6e4


5.9e4

6.0e4

6.1e4

6.2e4

6.3e4

6.4e4


B 1800 1600 1400

Pooled 24h Pooled 72h Pooled 120h

56766.01

L26V27 +L26V27

56765.87 56878.48

L26V27 + W25L26

1200 56991.63

1000

W25L26 +W25L26

56878.25

800

56991.57

600 400

56537.55

200

56536.07

0

5.63e4

56665.19

5.65e4

56877.67

5.67e4

56991.21

5.69e4

5.71e4

5.73e4

5.75e4

5.77e4

Mass, Da

C

800


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


700 600

Pooled 24h Pooled 72h Pooled 120h

59776.93

500 400 59776.97

300 200

L26A27

100 0

D

W25L26

5.95e4

5.96e4

59662.19 59661.17

59813.10 59777.05

5.97e4

5.98e4

60294.54

59940.87 5.99e4 Mass, Da

6.00e4

6.01e4

6.02e4

6.03e4

HGEGTFTSDVSSYLEEQAAKEFIAWLVKGG



Figure 8. Deconvoluted intact MS peak height-time plot of the major catabolites of dulaglutide observed in mice. The major glycosylation form (indicated by arrow in Figure 7) of each catabolite was used for the plotting. Intact MS Peak Height-Time Profile of Catabolites of Dulaglutide in Mice After Single Subcutaneous Dose 3.0E+03

Single Chain W25L26 Single Chain L26V27 Double Chain W25L26 + W25L26 Double Chain W25L26 + L26V27 Double Chain L26V27 + L26V27

2.5E+03 2.0E+03

Peak Height

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 32

1.5E+03 1.0E+03 5.0E+02 0.0E+00 0

20

40

60

80 Time (hr)

100

120

140

160


Page 31 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 1. Major catabolites of dulaglutide identified by LC-HRMS in mice in vivo. Predicted Proteolytic Site Single Chain Proteolysis Double Chain Proteolysis

W25L26 L26V27 W25L26 + W25L26 W25L26 + L26V27 L26V27 + L26V27

Observed Mass (Da) 59777 59662 56992 56878 56766

Observed Loss (Da) 2785 2900 5570 5684 5796

Theoretical Loss (Da) 2785.1 2898.2 5570.2 5683.3 5796.4

*Proteolytic sites are designated by the two adjacent amino acids. For example, proteolysis of the amide bond between W25 and L26 is designated as W25L26.



REFERENCES Kontermann, R.E. Curr Opin Biotechnol 2011, 22, 868. Pearson, J.T.; Rock, D.A. Bioanalysis 2015, 7, 3035. Strohl, W.R. BioDrugs 2015, 29, 215. Hall, M.P. Drug Metab Dispos 2014, 42, 1873. Katsila, T.; Siskos, A.P.; Tamvakopoulos, C. Mass Spectrom Rev 2012, 31, 110. Hager, T.; Spahr, C.; Xu, J.; Salimi-Moosavi, H.; Hall, M. Anal Chem 2013, 85, 2731. Liao, S.; Qie, J.K.; Xue, M.; Zhang, Z.Q.; Liu, K.L.; Ruan, J.X. Amino Acids 2010, 38, 1595. Copley, K.; McCowen, K.; Hiles, R.; Nielsen, L.L.; Young, A.; Parkes, D.G. Curr Drug Metab 2006, 7, 367. Na, D.H.; Faraj, J.; Capan, Y.; Leung, K.P.; DeLuca, P.P. Pharm Res 2007, 24, 1544. Grafmuller, L.; Wei, C.; Ramanathan, R.; Barletta, F.; Steenwyk, R.; Tweed, J. Bioanalysis 2016, 8, 1663. Firth, D.; Bell, L.; Squires, M.; Estdale, S.; McKee, C. Anal Biochem 2015, 485, 34. Xu, K.; Liu, L.; Saad, O.M.; Baudys, J.; Williams, L.; Leipold, D.; Shen, B.; Raab, H.; Junutula, J.R.; Kim, A.; Kaur, S. Anal Biochem 2011, 412, 56. Xu, K.; Liu, L.; Dere, R.; Mai, E.; Erickson, R.; Hendricks, A.; Lin, K.; Junutula, J.R.; Kaur, S. Bioanalysis 2013, 5, 1057. Jian, W.; Kang, L.; Burton, L.; Weng, N. Bioanalysis 2016, 8, 1679. van den Broek, I.; van Dongen, W.D. Bioanalysis 2015, 7, 1943. Ruan, Q.; Ji, Q.C.; Arnold, M.E.; Humphreys, W.G.; Zhu, M. Anal Chem 2011, 83, 8937. USFDA, Summary Review of Trulicity (Dulaglutide) 2014. Green, B.D.; Mooney, M.H.; Gault, V.A.; Irwin, N.; Bailey, C.J.; Harriott, P.; Greer, B.; O'Harte, F.P.; Flatt, P.R. J Endocrinol 2004, 180, 379. Beck, A.; Sanglier-Cianferani, S.; Van Dorsselaer, A. Anal Chem 2012, 84, 4637. Murphy, R.E.; Kinhikar, A.G.; Shields, M.J.; Del Rosario, J.; Preston, R.; Levin, N.; Ward, G.H. J Pharm Biomed Anal 2010, 53, 221. Hecht, R.; Li, Y.S.; Sun, J.; Belouski, E.; Hall, M.; Hager, T.; Yie, J.; Wang, W.; Winters, D.; Smith, S.; Spahr, C.; Tam, L.T.; Shen, Z.; Stanislaus, S.; Chinookoswong, N.; Lau, Y.; Sickmier, A.; Michaels, M.L.; Boone, T.; Veniant, M.M.; Xu, J. PLoS One 2012, 7, e49345.

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21)

TOC

Quantitation and Catabolite Identification HGEGTFTSDVSSYLEEQAAKEFIAWLVKGG 100.0

Concentration (nM)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 32

Proteolysis cleavage sites

GLP1 variant peptide Linker peptide

10.0

1.0

IgG4-Fc domain Total Fc Level Intact Molecule Level

0.1 0

50

100 Time (hr)

150

200


Simultaneous Catabolite Identification and Quantitation of Large

Recommend Documents