A High Resolution Bench-Top Orbitrap LC-MS Workflow Solution for Comprehensive Intact Monoclonal Antibody Characterization

Transcription

1 A High Resolution Bench-Top Orbitrap LC-MS Workflow Solution for Comprehensive Intact Monoclonal Antibody Characterization Zhiqi Hao, 1 David M. Horn, 1 Shiaw-Lin Wu, 2 Fan Zhang, 2 and Patrick Bennett 1 1 Thermo Fisher Scientific, San Jose, CA; 2 Barnett Institute, Northeastern University, Boston, MA

2 Overview Purpose: To develop a high resolution LC/MS-based workflow solution for robust, accurate and comprehensive intact monoclonal antibody (mab) characterization. Methods: Thermo Scientific Q Exactive hybrid quadrupole-orbitrap mass spectrometers were used for intact mass measurement and top-down sequencing. Full MS spectra of intact and reduced mabs were analyzed using Thermo Scientific Protein Deconvolution software version 2. that utilizes the ReSpect TM algorithm for molecular mass determination. The top-down msx HCD spectra were analyzed using Thermo Scientific ProSightPC software version 2.. Results: A mass error of less than 1 ppm was routinely achieved for intact mab mass measurement. Low mass modifications, such as oxidation, can be confidently identified on substructure level such as intact Fab, or light chain. Using an on-line high resolution topdown MS/MS approach, over 3% of the fragmentation site was covered for intact light chain as well as for Fab heavy chain. Sequence coverage from top-down approach also confirmed disulfide linkage on partially reduced samples. Introduction Monoclonal antibodies (mabs) are increasingly being developed and utilized for diagnosing and therapeutic treatment of diseases including cancer. Due to the heterogeneity of mab products, thorough characterization is necessary for their reproducible and safe production. Among the analytical tools used for the analysis of therapeutic mabs, mass spectrometry has become more and more important in providing valuable information on various protein properties. Such information includes intact mass, amino acid sequence, post-translational modifications including glycosylation form distribution, minor impurities due to sample processing and handling, and high order structure. Characterization at the intact protein level is usually the first step. In this study, a high resolution LC-MS based workflow solution was developed for robust, accurate and comprehensive mab characterization at the intact protein level. The fast chromatography, superior resolution and mass accuracy provided by the Q Exactive LC/MS system, and accurate data analysis of this workflow, provides a high-confident screening tool to accelerate biopharmaceutical product development. input spectrum were u the masses were compare commonly found glycoform ProSightPC software in the A B C Methods Samples: Four intact mabs were used in this study. To reduce intact mab, the sample was incubated for one hour at 6 C or 37 C in 6 M guanidine-hcl containing mm DTT for complete or partial reduction, respectively. Fab was generated using papain in 1mM EDTA, 1 mm Cys, mm sodium phosphate buffer, ph 7.. Before digestion, the enzyme suspension (1 mg/ml) was activated for 1 min at 37 C in the same buffer at an enzyme: buffer ratio of 1:9. The digestion was performed at 37 C overnight using an enzyme: antibody ratio of 1:99 w/w. HPLC: A Thermo Scientific TM ProSwift RP-1R monolithic column (1 x mm) was used for desalting and separation of light and Fab heavy chain. LC solvents were.1% formic acid in H 2 O (Solvent A) and.1% formic acid in acetonitrile (Solvent B). The column was heated to 8 C during analysis. The flow rate was 6 µl/min. After injection of 1 µg mab, a 1 min gradient was used to elute mabs from the column (. min, 2% B; 1. min, 3% B; 3. min, % B; 4. min, 98% B; 7. min, 98% B; 7.1 min, 2% B; 1. min, 2% B). Figure 2: Deconvoluted s errors of average molecu 2

3 e a 1 min gradient was used to elute mabs from the column (. min, 2% B; 1. min, 3% B; 3. min, % B; 4. min, 98% B; 7. min, 98% B; 7.1 min, 2% B; 1. min, 2% B). Mass Spectrometry: Q Exactive instruments were used for this study. Intact and reduced mabs were analyzed by ESI-MS for intact molecular mass. Top-down MS/MS was performed using high energy collision dissociation with a unique spectrum multiplexing feature (msx HCD). In this data acquisition mode, fragment ions produced from several individual HCD events, each on a precursor of a different charge state of the reduced mab, were detected together in the Thermo Scientific Orbitrap mass analyzer. The spray tion Bench-Top voltage was 4 kv. Orbitrap Sheath gas flow rate was set at LC-MS 1. Auxiliary gas flow rate Workflow was set at. Sol Capillary temperature was 2 C. S-lens level was set at. In-source CID was set at 4 nal Antibody Characterization ev. For full MS, resolution was 17, for intact mab and intact Fab average mass measurement, or 14, for light chain and Fab heavy chain monoisotopic mass measurement. Resolution was set at 14, for top-down MS/MS. The AGC target was rn 1, Shiaw-Lin Wu set at 2 3E6, for Xiaoyue full scan and 2E for Jiang MS/MS. Maximum 1, Andreas IT was set at 2 ms. FR Huhmer 1 and Pa Data Processing: Full MS spectra were analyzed using Protein Deconvolution software ntific, San Jose, that CA, utilizes USA; the ReSpect 2 algorithm Barnett for molecular Institute, mass determination. Northeastern Mass spectra for Univer deconvolution were produced by averaging spectra across the most abundant portion of ed workflow solution for robust, (mab) characterization. brid quadrupole-orbitrap mass nt and top-down sequencing. Full using Thermo Scientific Protein ReSpect TM algorithm for molecular ra were analyzed using Thermo the elution profile for the mab. A minimum of at least 8 consecutive charge states from the input spectrum were used to produce a deconvoluted peak. To identify glycoforms, the masses were compared to the expected masses with the various combinations of commonly found glycoforms. The top-down msx HCD spectra were analyzed using ProSightPC software in the single protein mode with a fragment ion tolerance of ppm. Results Figure 1: Intact mab analysis using LC MS Figure 2: Deconvoluted s errors of average molecu To measure the mass accu MS in conjunction with Prot several times using two diff ppm mass accuracy are sh various glycoforms are sho Table 1: ppm mass deviat abundant glycoforms RAW file Q Exactiv ely achieved for intact mab mass on, can be confidently identified on sing an on-line high resolution topn site was covered for intact light age from top-down approach also. ing developed and utilized for including cancer. Due to the erization is necessary for their cal tools used for the analysis of re and more important in providing h information includes intact mass, ns including glycosylation form ng and handling, and high order sually the first step. In this study, a eveloped for robust, accurate and in level. The fast chromatography, e Q Exactive LC/MS system, and high-confident screening tool to Time (min) A B C GF+G1F GF+GF GF+G2F Total Ion Chromatogram Average Spectrum of Intact mab Charge Envelope GF+GF The average ppm error for a instruments was 6.9 ppm w shown here). This indicates confirmation of protein prim Table 2. Relative abundan RAW file Q Exactive CV For the top glycoforms, th Figure 3: Identification of o reduce intact mab, the sample was dine-hcl containing mm DTT for erated using papain in 1mM EDTA,.. Before digestion, the enzyme in the same buffer at an enzyme: 7 C overnight using an enzyme: lithic column (1 x mm) was used One microgram of mab was desalted and eluted from a ProSwift RP-1R monolithic column using a 1 min gradient and analyzed using ESI-MS on the Q Exactive MS. As shown in Figure 1, the mab was eluted over one minute as shown in (A). The average spectrum over the elution time shows a nicely distributed complete charge envelope of the mab (B). A zoom-in view of each charge state reveals five major glycosylation forms that are baseline separated (C). Thermo Scientific Poster Note BioPharma_PN63944_E Intact 11/13S Fab After each of the mab datasets were analyzed using the Protein Deconvolution software, Resolution =

4 Figure 3: Identification o o reduce intact mab, the sample was nidine-hcl containing mm DTT for enerated using papain in 1mM EDTA, 7.. Before digestion, the enzyme C in the same buffer at an enzyme: 37 C overnight using an enzyme: nolithic column (1 x mm) was used chain. LC solvents were.1% formic tonitrile (Solvent B). The column was µl/min. After injection of 1 µg mab, olumn (. min, 2% B; 1. min, 3%.1 min, 2% B; 1. min, 2% B). sed for this study. Intact and reduced ular mass. Top-down MS/MS was with a unique spectrum multiplexing ragment ions produced from several ent charge state of the reduced mab, rbitrap mass analyzer. The spray. Auxiliary gas flow rate was set at. et at. In-source CID was set at 4 mab and intact Fab average mass b heavy chain monoisotopic mass -down MS/MS. The AGC target was IT was set at 2 ms. oluted peak. To identify glycoforms, s sing with Protein the various Deconvolution combinations software HCD ass determination. spectra were Mass analyzed spectra using for a across fragment the ion most tolerance abundant of portion ppm. of 8 consecutive charge states from the One microgram of mab was desalted and eluted from a ProSwift RP-1R monolithic column using a 1 min gradient and analyzed using ESI-MS on the Q Exactive MS. As shown in Figure 1, the mab was eluted over one minute as shown in (A). The average spectrum over the elution time shows a nicely distributed complete charge envelope of the mab (B). A zoom-in view of each charge state reveals five major glycosylation forms that are baseline separated (C). After each of the mab datasets were analyzed using the Protein Deconvolution software, the masses were compared to the masses expected for the known amino acid sequence with the various combinations of glycoforms commonly found on mabs. One such result is shown below in Figure 2. Figure 2: Deconvoluted spectrum for a mab with known composition and mass errors of average molecular mass. ppm G+GF 2xMan G+G -7 ppm GF+G1F -.7 ppm -8. ppm GF+GF GF+G2F (or 2G1F) -.9 ppm G2F+G2F +SA To measure the mass accuracy and reproducibility of mab samples on the Q Exactive MS in conjunction with Protein Deconvolution software, the mab sample was analyzed several times using two different instruments over three different days. The results for ppm mass accuracy are shown in Table 1 and the results for relative abundance of the various glycoforms are shown in Table 2. Table 1: ppm mass deviations from expected target masses for the most abundant glycoforms Further characterizatio Figure was generated 4: Top-down using se reduced to generate lig Fab, light chain and F1 (Figure 3, middle). The at resolution 17,2.68 fo z= and Fab heavy chain z=24 (F Sub-structure Resolution = Average Mass Res Fab 1 Fab heavy 1 Fab light 1 RAW file Q Exactive G+GF GF+GF GF+G1F GF+G2F 8 9 Total Ion Chromatogram Hi Average Spectrum of Intact mab Charge Envelope The average ppm error for all 34 measurements of four different mabs on multiple instruments was 6.9 ppm with a standard deviation of 6.4 ppm (not all the data are shown here). This indicates that the Q Exactive MS is a very powerful platform for confirmation of protein primary structure. Table 2. Relative abundance for the most abundant glycoforms RAW file Q Exactive G+GF GF+GF GF+G1F GF+G2F Lig 3% back P Score G2F F+G2F CV 3.4% 1.6% N.A. 3.9% 4.4% For the top glycoforms, the relative intensity reproducibility is within a few percent. Figure 3: Identification of oxidation on intact Fab, light and Fab heavy chain Besides molecular mass, a a top-down LC-MS/MS a multiplexing mode where different number of charge ions were then detected a than 3% of fragments fro heavy chain (Figure 4 botto

5 PPN_noCys_Red_14K # RT: AV: NL: 8.83E4 T: FTMS + p ESI Full ms [.-27.] z= z= z= z= z= z= z= z= z= z= z= z= z= z= z= z instruments was 6.9 ppm with a standard deviation of 6.4 ppm (not all the data are shown here). This indicates that the Q Exactive MS is a very powerful platform for confirmation of protein primary structure. Table 2. Relative abundance for the most abundant glycoforms RAW file Q Exactive G+GF GF+GF GF+G1F GF+G2F 61 z Ligh 3% backb P Score = F G2F m a ProSwift RP-1R monolithic SI-MS on the Q Exactive MS. As ute as shown in (A). The average ted complete charge envelope of als five major glycosylation forms e Protein Deconvolution software, r the known amino acid sequence found on mabs. One such result nown composition and mass -7 ppm GF+G1F m -8. ppm F GF+G2F (or 2G1F) -.9 ppm G2F+G2F +SA CV 3.4% 1.6% N.A. 3.9% 4.4% For the top glycoforms, the relative intensity reproducibility is within a few percent. Figure 3: Identification of oxidation on intact Fab, light and Fab heavy chain Resolution =17. K Average Mass Reduction Oxidized Species of Fab Sub-structure Resolution Delta Mass from non-oxidized (Da) Fab light Fab heavy Fc LC MS Protein level mass error (ppm) Fab 17.K Fab heavy 14K Fab light 14K Fab Heavy Further characterization at substructure level is shown in Figure 3 to Figure. Fab was generated using papain which cleaves this molecule at hinge. Fab was then reduced to generate light chain and Fab heavy chain (Figure 3, top). LC/MS of intact Fab, light chain and Fab heavy chain identified oxidation species as shown above (Figure 3, middle). The mass errors of the identification at protein level were 6.4 ppm at resolution 17, for Fab and less than 1 ppm at resolution 14, for light chain and Fab heavy chain (Figure 3 bottom) Resolution =14 K Monoisotopic Mass Besides molecular mass, am a top-down LC-MS/MS ap multiplexing mode where m different number of charges ions were then detected all than 3% of fragments from heavy chain (Figure 4 bottom Top-down sequencing was less in molecular mass tha ProSight PC software matc molecule (Figure ). Figure : Top-down sequ Spectrum of a partially red (+2 charge +2 ReSpect is a trademark the property of Thermo This information is no manners that might inf Thermo Scientific Poster Note BioPharma_PN63944_E 11/13S

6 PPN_noCys_Red_14K # RT: AV: NL: 8.83E4 T: FTMS + p ESI Full ms [.-27.] z= z= z= z= z= z= z= z= z= z= z= z= z= z= z= les on the Q Exactive sample was analyzed t days. The results for tive abundance of the for the most Figure 4: Top-down sequencing of light and Fab heavy chain using LC-MS/MS z= z= z= z= z= Full MS of light chain 2F Six Individual isolation and fragmentation events One Orbitrap Detection event High resolution top-down MS/MS mabs on multiple not all the data are werful platform for orms 2F z= Light chain 3% backbone fragments P Score = 4.9E z= z=6 z= z=3 z=4 z= z=6 z=3 z= z= Fab heavy chain 34% backbone fragments P Score = 2.2E % ithin a few percent. ab heavy chain LC MS tion =14 K Fab Heavy Besides molecular mass, amino acid sequence can be obtained at the intact protein level using a top-down LC-MS/MS approach. High resolution top-down HCD was performed using a multiplexing mode where multiple precursors, which were the same protein molecule carrying different number of charges, were isolated, fragmented separately and the resulting fragment ions were then detected all together in a single Orbitrap detection event (Figure 4 top). More than 3% of fragments from backbone cleavage were detected for both light chain and Fab heavy chain (Figure 4 bottom) with excellent P score from ProSight PC software. Top-down sequencing was also performed on a partially reduced light chain which is 4.2 Da less in molecular mass than the fully reduced species. Analysis of the HCD spectrum in ProSight PC software matched two disulfide linkages, which is typical of this type of IgG molecule (Figure ). Figure : Top-down sequencing maps disulfide linkage on partially reduced light chain Spectrum of a partially reduced light chain (+2 charged) +2 charged Top-down HCD ProSight PC partially reduced light chain Monoisotopic Mass ein level mass error (ppm)

7 PPN_noCys_Red_14K # RT: AV: NL: 8.83E4 T: FTMS + p ESI Full ms [.-27.] z= z= z= z= z= z= z= z= z= z= z= z= z= z= z= hin a few percent. b heavy chain C MS =14 K level mass error (ppm) Fab Heavy Monoisotopic Mass e 3 to Figure. Fab hinge. Fab was then top). LC/MS of intact ies as shown above in level were 6.4 ppm 4, for light chain different number of charges, were isolated, fragmented separately and the resulting fragment ions were then detected all together in a single Orbitrap detection event (Figure 4 top). More than 3% of fragments from backbone cleavage were detected for both light chain and Fab heavy chain (Figure 4 bottom) with excellent P score from ProSight PC software. Top-down sequencing was also performed on a partially reduced light chain which is 4.2 Da less in molecular mass than the fully reduced species. Analysis of the HCD spectrum in ProSight PC software matched two disulfide linkages, which is typical of this type of IgG molecule (Figure ). Figure : Top-down sequencing maps disulfide linkage on partially reduced light chain Spectrum of a partially reduced light chain (+2 charged) +2 charged Top-down HCD ProSight PC partially reduced light chain Conclusion In this study, a workflow was developed that combines high resolution Orbitrap MS, fast chromatography, high throughput msx HCD and accurate data analysis to characterize intact mab. The precise mass measurement and extensive, high confident amino acid sequence obtained from this workflow provides the following information for intact mab and its substructure: Accurate measurement of intact molecular mass Reproducible quantification of glycoform relative abundance Confident amino acid sequence information and structural information ReSpect is a trademark of Positive Probability, Ltd. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries. This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others Thermo Fisher Scientific Inc. All rights reserved. ReSpect is a trademark of Positive Probability, Ltd. ISO is a trademark of the International Standards Organization. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries. Specifications, terms and pricing are subject to change. Not all products are available in all countries. Please consult your local sales representative for details. Thermo Fisher Scientific, San Jose, CA USA is ISO 91:28 Certified. Africa Australia Austria Belgium Canada China (free call domestic) Denmark Europe-Other Finland France Germany India Italy Japan Latin America Middle East Netherlands New Zealand Norway Russia/CIS Singapore Spain Sweden Switzerland UK USA BioPharma_PN63944_E 11/13S