42
Accelerating Biologics Characterization With Agilent’s LC/MS Solutions Life Sciences Group Thomas M Trainor, PhD LC/MS Product Specialist May 1, 2013 1
Accelerating Biologics Characterization
With Agilent’s LC/MS Solutions
Life Sciences Group
Thomas M Trainor, PhD
LC/MS Product Specialist
May 1, 2013
1
Overview - Agilent BioPharm LC/MS Solutions:
Examples showing both hardware and software capabilities,
including:
Intact (“top-down”) protein characterization with enhanced deconvolution
software (MassHunter)
Peptide digest (“bottom-up”) protein characterization with peptide libraries
(MassHunter Bioconfirm, Spectrum Mill, Skyline)
Peptide quant by QQQ LC/MS/MS (MassHunter, Skyline)
Fast Glycan profiling with on-chip enzymatic digestion, (MassHunter, Mass
Profiler Professional)
2
LC/MS for Biologics – Discovery to Manufacturing:
Protein Expression
Bioreactor
Production
Protein Efficacy
3
Accurate mass 6200 series single TOF LC/MS
Accurate mass 6500 series QTOF LC/MS/MS
Dual ion source automatic delivery of mass calibrant
Nominal mass 6100 series SQ and 6400 series QQQ
MassHunter B.06 software, Windows 7 64-bit
Choice of electrospray ion sources – standard ESI,
Agilent JetStream, Agilent ifunnel, and nano chipMS
LC/MS approaches to mAb Characterization:
4
Analytical Chemistry 2013, 85, 715-736
Agilent’s LC/MS Biopharma Workflows:
Our LC/MS biopharma workflows utilize the MassHunter
Bioconfirm software to provide confirmation of putative
sequences and modifications for:
• Intact proteins
• Protein digests (for peptide mapping)
• Synthetic peptides
• Oligonucleotides
5
5990-7065EN MH Biopharma Workflow Overview
5990-7064EN MH Biopharma Workflow Guide
MassHunter Acquisition
TOF or Q-TOF
MS-only data for:
--Intact Protein
--Synthetic Peptide
--Oligonucleotide
Mirror Plot allows
comparison of
biosimilars and to
monitor batch-to-
batch variation
LC/MS Intact Protein Confirmation Workflow
6
Acquisition Qual/
BioConfirm BioConfirm Comparison
BioConfirm
Report
Define Target
Sequence
Define Potential
Modifications
Match?
Integrate
Chromatogram
Extract Spectra
Deconvolute Spectra:
--Maximum Entropy
--Peak Modeling
NEW!
NEW!
Molecular Weight
Potential Modifications
7
Intact mAb Analysis on the Agilent 6550 QTOF
with iFunnel : 1µg of mAb on column
Integrate Chromatogram
Single Scan Single Scan Zoom Deconvolute Max
Entropy
Mobile Phase: 0.1%FA in H2O, 0.1% FA in ACN
Flow rate of 0.5 mL/min; 7 min gradient
Column: Agilent Poroshell 300SB-C8, 1.0 x 75 mm, 5 um
Column Temp: 80C
Easy Access Software Enables Walk-up for
Biologics LC/MS – from sample to final reporting:
8
Analyzing mAb Using 6550 Q-TOF Raw Data
9
Analyzing mAb Using 6550 Q-TOF Enhanced Deconvolution Algorithm - pMod
10
Deconvolution Maximum Entropy vs. pMod
11
Maximum Entropy pMod
pMod Resolves Overlapped Peaks
12
IgG 1
IgG2
IgG1 and IgG2
mix
These two peaks are 39 Da apart
Not resolved by Maximum Entropy
Resolved by pMod IgG1 and IgG2
mix
pMod
Analyzing mAb Using 6550 Q-TOF
Enhanced Sensitivity From 5ng to 5mg mAb on Column
13
5 ug
1 ug
500 ng
100 ng
10 ng
5 ng
6550 QTOF - Linear Response for 4 of the Intact
Isoforms
14
What is Peptide Mapping?
• Peptide mapping is an identity test that confirms the primary structure of recombinant proteins and demonstrates process consistency
• Proteins are enzymatically digested into peptide fragments (most commonly with a proteolytic enzyme called trypsin) which are then separated and identified
• The identified peptides are then “mapped” back to the known protein sequence to ensure the intended product was created.
• Single amino acid changes can be identified, making it possible to monitor genetic stability, identifying errors in cDNA reading or point mutations.
Intact mAb
Digested peptides
15
Compare Protein
Digest module allows
comparison of
biosimilars and to
monitor batch-to-
batch variation
MassHunter Acquisition
TOF MS-only or Q-
TOF MS and
MS/MS data for:
--Protein Digest
(peptide mapping)
LC/MS Protein Digest Confirmation Workflow
16
Acquisition Qual BioConfirm Comparison
BioConfirm
Report
Define Target
Sequence
Define Digest
Enzymes and
Potential
Modifications
Match?
Molecular Feature
Extractor (MFE):
--Filter features
found by quality
using Q- score
NEW!
Sequence Coverage %
Potential Modifications
and their Locations
HPLC-Chip-Based Peptide Mapping Base Peak Chromatogram (BPC) of Trypsin Digested mAb on C18 Chip nano-ESI
17
Peptide Mass Accuracy, 6520 QTOF
light chain A(112-130) 2101.1245 2101.1208 1.76
light chain A(113-130) 1945.0199 1945.0197 0.09
light chain A(131-146) 1796.8922 1796.888 2.38
light chain A(150-173) 2676.2669 2676.2627 1.55
light chain A(150-187) 4160.01 4160.0033 1.61
light chain A(154-173) 2134.9657 2134.9615 1.99
light chain A(193-211) 2140.0778 2140.0735 1.99
light chain A(195-211) 1874.9254 1874.9197 3.06
Sequence
Name Label Mass
Target
Sequence
Mass
Match
Difference
(ppm) light chain A(1-19) 1996.088 1996.0841 1.98
light chain A(26-47) 2438.1986 2438.1979 0.31
light chain A(48-56) 992.5656 992.5655 0.15
light chain A(64-93) 3388.5238 3388.5194 1.31
light chain A(94-107) 1538.7191 1538.7154 2.43
Sequence Name Label Mass Target Sequence
Mass
Match Difference
(ppm)
heavy chain A(1-19) 1880.0511 1880.048 1.65 heavy chain A(20-38) 2126.9587 2126.9554 1.56 heavy chain A(39-67) 2927.4492 2927.4414 2.68 heavy chain A(39-65) 2714.3216 2714.3188 1.05 heavy chain A(44-65) 2233.0584 2233.0539 2 heavy chain A(44-67) 2446.1802 2446.1765 1.51 heavy chain A(66-72) 835.4663 835.4664 -0.11 heavy chain A(68-72) 622.3431 622.3439 -1.28 heavy chain A(73-87) 1768.8812 1768.8778 1.93 heavy chain A(77-87) 1324.6816 1324.6809 0.51 heavy chain A(88-98) 1317.5937 1317.5911 2 heavy chain A(99-105) 714.4018 714.4024 -0.94 heavy chain A(106-129) 2531.2534 2531.2479 2.2 heavy chain A(107-129) 2375.1509 2375.1468 1.74 heavy chain A(130-141) 1185.6426 1185.6394 2.72 heavy chain A(142-155) 1320.6705 1320.6708 -0.21 heavy chain A(156-218) 6712.309 6712.3072 0.27 heavy chain A(156-221) 7054.4995 7054.4975 0.28 heavy chain A(156-222) 7182.5962 7182.5925 0.52 heavy chain A(227-256) 3333.643 3333.6349 2.45 heavy chain A(231-256) 2843.4544 2843.4503 1.45 heavy chain A(257-263) 834.4274 834.4269 0.55 heavy chain A(264-282) 2138.0249 2138.0202 2.22 heavy chain A(283-296) 1676.7985 1676.7947 2.25 heavy chain A(297-309) 3115.3418 3115.3315 2.13 heavy chain A(310-325) 1807.0038 1806.9992 2.51 heavy chain A(310-328) 2227.2034 2227.2001 1.5 heavy chain A(335-342) 837.496 837.496 0 heavy chain A(347-368) 2509.3347 2509.3289 2.32 heavy chain A(349-368) 2310.1993 2310.1968 1.08 heavy chain A(353-363) 1285.6677 1285.6667 0.79 heavy chain A(353-368) 1871.9648 1871.9629 1.03 heavy chain A(369-378) 1160.6228 1160.6223 0.4 heavy chain A(379-400) 2543.1289 2543.1241 1.9 heavy chain A(401-417) 1872.9184 1872.9146 2.06 heavy chain A(423-447) 3043.3964 3043.393 1.12 heavy chain A(425-447) 2800.2679 2800.2598 2.89 heavy chain A(448-454) 659.3488 659.349 -0.35
Average Error = 1.3 ppm 100% Total mAb sequence coverage
18
Using BioConfirm Qualitative Comparison to
Compare Sample and Reference One-to-One
Table
MS spectra
MS/MS spectra Chromatograms
Common
features Reference Sample
R
S
R
S
S R
mirror
19
BioConfirm Matches MS/MS Spectra
20
Product ion assignment
Product ion label
21
Mass Spectra
Analysis/Export
Spectral Libraries
Two Approaches for Peptide Libraries:
MSMS Spectra
Spectrum Mill
MacCoss Lab
University of Washington
MSMS Spectra
BioConfirm/
Qualitative Analysis
Personal Compound
Database/Library (PCDL)
22
Peptide Libraries: Personal Compound Database/Library
MSMS Spectra BioConfirm/
Qualitative Analysis
Personal Compound
Database/Library (PCDL)
Peptide
Sequences
MSMS Spectrum
Precursor
m/z Collision
Energy
23
Peptide Libraries: Skyline
MSMS Spectra Spectrum Mill MacCoss Lab
University of Washington
Peptide
Sequences
MSMS Spectrum
Skyline – Mass Hunter QqQ Quant and Optimizer:
Software tools for Peptide Quant
24
Agilent MassHunter with Skyline peptide software,
Peptide Quant with QQQ LC/MS/MS:
25
5990-9886EN Workflow Overview
5990-9887EN Workflow Guide
Reproducibility for 110 Injections (10 fmol SIS
Peptides and 2.5 µg Plasma Digest On-column) Protein
Response
%RSD
Ret. Time
%RSD
Adiponectin:
IFYNQQNHYDGSTGK 9.8 0.13
Antithrombin-III :
DDLYVSDAFHK 4.7 0.16
Apolipoprotein A-II precursor:
SPELQAEAK 6.7 0.12
Apolipoprotein C-III:
GWVTDGFSSLK 2.3 0.08
Ceruloplasmin :
EYTDASFTNR 9.6 0.14
Heparin cofactor II:
TLEAQLTPR 6.1 0.15
Histidine-rich glycoprotein:
DGYLFQLLR 3.4 0.02
Kininogen-1:
TVGSDTFYSFK 3.3 0.13
L-selectin:
AEIEYLEK 9.5 0.15
Plasminogen:
LFLEPTR 2.2 0.13
Vitamin D-binding protein:
THLPEVFLSK 3.0 0.12
von Willebrand Factor:
ILAGPAGDSNVVK 9.5 0.15
The samples were provided by Derek Smith and Christoph H. Borchers from the UVic-Genome BC Proteomics Centre
2.2% RSD
n=110
4.7% RSD, n=4
7.9% RSD, n=4
12.3% RSD, n=4
Plasminogen LFLEPTR
26
2,000 Dynamic MRM transitions in peptide digest of
depleted human sera – no time segment boundaries
Pag
e
27
Protein Validation Workflow
Page 28
Target Proteins
Peptide and MRM
Generation
Data Acquisition
Data review and MRM selection
Quantitation Statistical Validation
MassHunter Quant
Skyline
MassProfiler
Professional
28
HPLC-Chip/MS Solution for Biopharmaceutical Characterization
Glycan Analysis
Using mAb-Glyco chip
29
Glycoprotein
PNGase F
Glycosylamine
N-Glycan
-NH3
• MALDI-TOF/MS• LC/MS• HPAEC-PAD
Anomeric equilibrium
2-AB
labelling agent
Reductant
labelled N-Glycan
• HPLC w/ fluorescence detection • CE w/ fluorescence detection
Glycan Analysis Traditional workflows for N-glycan analysis
• Enzymatic release of N-glycans can often take up to 24 hours • Fluorescence techniques require sufficient time for efficient labeling
30
HPLC-Chip/MS Solution for Biopharmaceutical Characterization mAb-Glyco chip workflow
4 minutes
1 minute
TOF MS Detector
mAb sample
Expe
rime
ntal Tim
e
LC- or CE- Fluorescence
10-30 minutes
3-8 hours
2-4 days
MALDI MS
//
//
mAb-Glyco Chip MS
2 minutes
PNGase F, enzymatic N-glycan release
N-Glycan concentration
N-Glycan separation
31
mAb-Glyco Chip-LC/MS Solution What components are required?
mAb-Glyco Chip kit: mAb-Glyco Chip,
reagents, calibration standard, mAb
standard
1260 Cap/Nano LC stack
Chip Cube interface
Agilent Q-TOF (TOF) mass spectrometer Agilent MassHunter software
(with glycan library)
32
310 nL
160 nL
43 mm
(A) 310 nL
(B) 160 nL
(D) rotor-in rotor design
(C) 43mm
mAb-Glyco Chip-LC/MS Solution Chip design
(A) Enzyme reactor packed with immobilized PNGase F beads for on-chip deglycosylation of mAbs
(B) Porous graphitized carbon enrichment column for trapping cleaved N-glycans
(C) Porous graphitized carbon analytical column for separation of cleaved N-glycans
(D) Unique rotor-in-rotor design of the chip cube
(A)
(A) mAb-Glyco Chip nano electrospray tip
(B) Spray Into Mass Spectrometer (C) Mass Spectrometer
(C) (B)
33
Publ 5990-6924EN
mAb-Glyco Chip-LC/MS Solution How the chromatography works
Total analysis time of 12 min with great chromatographic resolution
A ... Sample loading B ... ER fill C ... Deglycolsyation D ... Glycan transfer E ... Glycan analysis
Sample prep = 6 min
Sample analysis = 6 min
FASTEST ANALYSIS TIME AVAILABLE
34
mAb-Glyco Chip-LC/MS Solution Glycan Quantitation and Report Generation
Glycan database is an accurate mass and structure database with 144 entires of glycans that are frequently found on mAb‘s.
Reporting templates automatically processes the glycan hits by clustering and merging isomermic structures and thus significantly helps to deliver fast answers
Glycan Database
Reporting Templates
35
mAb-Glyco Chip-LC/MS Solution Comparison of mAb-Glyco Chip with MALDI-MS and FLD 2-AB glycan, top 3 glycans
4.7%
71.1%
10.4%
4.8%
76.7%
9.0%
4.0%
72.3%
8.7%
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
M5 bG0 bG1
mAb-Glyco-Chip/MS
MALDI-MS (+)
FLD 2-AB glycan
36
mAb-Glyco Chip - LC/MS Solution Analyzing the data by principal component analysis (Agilent Mass Profiler Pro)
Biosimilar m-2
Biosimilar m-2
1
2
3 Biosimilar m-1
• Samples from the same manufacturer clearly
cluster together.
• Three clusters obtained for three biosimilars.
Poster: ASMS 2010: Shiaw-Lin Wu1, Yi Wang1, Ning Tang2, Dayin Lin2, William S. Hancock1, and Barry Karger1 1 Barnett Institute and Department of Chemistry and Chemical Biology, Boston, MA 02115. 2 Agilent Technologies, Waldbronn, Germany 76337
COMBINING mAb-Glyco Chip DATA WITH MPP SOFTWARE – MAKING
DATA VISUAL AND USABLE
37
Summary - Agilent BioPharm LC/MS Solutions:
Examples showing both hardware and software capabilities,
including:
Intact (“top-down”) protein characterization with enhanced deconvolution
software (MassHunter)
Peptide digest (“bottom-up”) protein characterization with peptide libraries
(MassHunter Bioconfirm, Spectrum Mill, Skyline)
Peptide quant by QQQ LC/MS/MS (MassHunter, Skyline)
Fast Glycan profiling with on-chip enzymatic digestion, (MassHunter, Mass
Profiler Professional)
38
mAb Structure
Light Chain
Fc
Fab
Antigen
binding
site
Hinge
Glycosylation
site Truncation
(lysine)
Disulfide
shuffling
Pyroglutamate
Deamidation/oxidation
Heavy Chain
It is critical throughout process development to confirm that the antibody drug candidate has the correct intact molecular weight, variants and other potential post translational modifications. Potential degradation products are also identified and will be monitored throughout development and manufacturing. Prior to scale-up, precious amounts of samples are available for analysis, making nano-LC/MS a viable solution.
40
5 x10
0
1
2
3
4
5
23746.50
23727.82 23762.96
Counts vs. Deconvoluted Mass (amu)
23540 23560 23580 23600 23620 23640 23660 23680 23700 23720 23740 23760 23780 23800 23820 23840 23860 23880 23900 23920 23940 23960 23980
Intact mAb Analysis by HPLC-Chip/MS Light/Heavy Chain Analysis
Δ1446.6
Δ162
.7
4 x10
0
1
2
3
4
5 50675.58
49228.96
50838.33
49435.14 51023.43 50121.16 49920.85 49009.52
Counts vs. Deconvoluted Mass (amu) 48600 48800 49000 49200 49400 49600 49800 50000 50200 50400 50600 50800 51000 51200
41
Glycan Analysis mAb structure and heterogeneity
• mAbs are large glycoproteins (~150 kDa) bearing frequently complex N-linked glycans within their structure (Fc-region, the bottom of the “Y”)
• Glycan presence, absence and profile impacts therapeutic efficacy, pharmacokinetics, immunogenicity, folding and stability
• Glycosylation is influenced by many factors; e.g. , organism and cell line in which the mAb is produced or other production conditions (pH, temperature, media)
• Hence, characterization of mAbs and their glycan profiles is of vital importance throughout the various phases of therapeutic development
Glycosylation
Fucose – 146 Da
Mannose – 162 Da
N-Acetylglucosamine – 203 Da
Galactose – 162 Da
1299.3 G0
1445.4 G0F
1607.5 G1F
1769.6 G2F
Average mass Structure Glycan
42
