Quantitative Biopharmaceutical Applications of Capillary Electrophoresis
A Thesis
Submitted to the Faculty
of
Drexel University
by
Junge Zhang
in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy
November 2009
ii
© Copyright 2009
Junge Zhang. All Rights Reserved.
iii
Dedications
This thesis is dedicated to my wife Rene and my two beautiful daughters, Zannan and Melanie,
for their love, endless support, and understanding; to my loving father, who left this world in
2008 after fighting with stroke and cancer for eight years; and to my mother, who has been
giving me so much that I can never pay back.
iv
Acknowledgments
I would like to thank my research advisor, Professor Joe Foley for his support and guidance
during my study at Drexel University. I have learned so much from Dr. Foley thorough group
meetings, face-to-face discussions, and numerous phone calls and emails. You challenged me
for using scientifically incorrect citations that I thought to be authentic. You encouraged me to
keep working on my projects until I found the solutions. Because of your in-depth knowledge
and passion in separation science, I became a better analytical scientist. I would like to express
my sincerest gratitude to Dr. Foley, and without you this thesis would not be possible.
I would also like to thank my thesis committee: Dr. Peter Wade, Dr. Anthony Addison, Dr.
Reinhard Schweitz-Stenner, Dr. Felice Elefant, and Dr. Jonathan Shackman. Your time and
comments on my thesis are highly appreciated.
I must mention my former supervisor at J&J, Dr. Utpal Chakraborty. It was he who advised me
to study for the PhD after I joined the company. We became close friends after working
together for six years. Although we are in different groups now, your help and continuous
support are still tremendous and meant a lot to me. Thank you so much, Utpal!
I would like to thank my current supervisor at J&J, Dr. Sudhir Burman, who has reviewed two
of my manuscripts and provided valuable comments.
My thanks also go to the Foleite research group. At Drexel, the Foleite research group is like a
family to me. Many thanks to Dr. Melissa Mertzman, Dr. Kimberly Kahle, Dr. Stephanie
Schuster, Dr. Dave Thomas, Addy Kojtari, Alex Adair, Jeanine Erdner-Tindall, Anna
Caltabiano, Barbara Hale, Donna Blackney, Mirlinda Biba, Laquisha Hamilton, Li Li, and
v
Adam Socia. Your ideas and ways of solving a problem have inspired me in many ways.
Sometimes, even a simple question raised by one of you in a group meeting changed my way
of thinking. Your friendship is a blessing to me. Dr. Melissa was so caring and considerate. Dr.
Kim Kahle was so sharp and I wish I could be as smart as you are. Dr. Stephanie Schuster was
always smiling and nice to me even when I asked too many questions. Dr. Dave Thomas was
so patient and encouraging. Addy was always energetic and enthusiastic about science. Alex
was so perseverant. Although many times I had to make effort to go to a group meeting after a
hard working day, I’ll miss that after I graduate because it was such a good learning
opportunity for me.
I would like to thank many faculty and staff members in Chemistry Department at Drexel
University, particularly, to Dr. Anthony Addison, Dr. Alan Bandy, Dr. Peter Wade, Dr. Jun Xi,
Dr. Keven Owens, Dr. Reinhard Schweitz-Stenner, Virginia Nesmith, Edward Thorne, Edith
Smith, Edward Doherty, Susan Rutkowsky, and Thomas Cachaza.
Last, but not least, I would like to thank my wife Hong (Rene) Lu for her love. You have
sacrificed so much for me. I clearly remember those days and nights I had classes or group
meetings that you took care of the two kids. So many weekends I went to work on my project
and left you guys alone. At home whenever I worked on my research project, you never asked
me to do anything. Instead, you brought me a cup of tea to drink or some fruits to eat. I would
also thank my two daughters for their sacrifice too. I owed you guys so much time. I promise I
will take you somewhere or play with you more after this.
vi
Table of Contents
LIST OF TABLES ......................................................................................................................... xii
LIST OF FIGURES ...................................................................................................................... xiii
LIST OF SYMBOLS ................................................................................................................... xvii
LIST OF ABBREVIATIONS ....................................................................................................... xix
ABSTRACT .................................................................................................................................. xxi
Chapter 1:
Introduction to capillary electrophoresis: fundamentals and applications .................. 1
1.1 Development of capillary electrophoresis .................................................................................. 1
1.2 Development of capillary electrophoresis instrumentation ........................................................ 1
1.3 Fundamental theory of capillary electrophoresis separation ...................................................... 3
1.3.1 Efficiency ................................................................................................................................ 3
1.3.2 Electroosmotic flow ................................................................................................................ 6
1.3.3 Ohm’s law ............................................................................................................................. 12
1.3.4 Electrophoretic mobility ....................................................................................................... 15
1.3.5 Separation mode.................................................................................................................... 15
1.4 Application of capillary electrophoresis in biopharmaceutical analysis .................................. 27
1.5 List of references...................................................................................................................... 29
Chapter 2: Characterization of glycoprotein erythropoietin by its isoform distribution using
capillary zone electrophoresis ........................................................................................................ 36
2.1 Introduction .............................................................................................................................. 36
2.2 Materials and methods ............................................................................................................. 39
2.2.1 Chemicals and reagents ......................................................................................................... 39
2.2.2 EPO sample preparation........................................................................................................ 39
2.2.3 Formulated epoetin alfa sample preparation ......................................................................... 40
2.2.4 Capillary zone electrophoresis (CZE) ................................................................................... 41
vii
2.2.5 Modified capillary conditioning............................................................................................ 43
2.3 Results and discussion ............................................................................................................. 44
2.3.1 Restoration of the fused silica capillary surface ................................................................... 44
2.3.2 System suitability .................................................................................................................. 48
2.3.3 Characterization and qualification of epoetin alfa secondary standard................................. 49
2.3.4 Precision ................................................................................................................................ 55
2.3.5 Use of duplicate injections for test sample ........................................................................... 57
2.3.6 Use of 2-hour conditioning period on new capillaries .......................................................... 57
2.3.7 Removal of polysorbate 80 from formulated epoetin alfa .................................................... 60
2.4 Conclusions .............................................................................................................................. 62
2.5 List of references...................................................................................................................... 65
Chapter 3: Separation of residual cell culture media components by micellar electrokinetic
chromatography ............................................................................................................................. 68
3.1Introduction ............................................................................................................................... 68
3.2 Materials and methods ............................................................................................................ 70
3.2.1 Chemicals and reagents ......................................................................................................... 70
3.2.2 Standard solutions ................................................................................................................. 72
3.2.3 Sample preparation ............................................................................................................... 72
3.2.4 Capillary electrophoresis....................................................................................................... 73
3.2.5 Capillary conditioning........................................................................................................... 73
3.3 Results and discussion ............................................................................................................. 74
3.3.1 Capillary zone electrophoresis (CZE) vs. micellar electrokinetic chromatography (MEKC)
....................................................................................................................................................... 74
3.3.2 Effect of surfactant concentration ......................................................................................... 78
3.3.3 Effect of pH........................................................................................................................... 81
3.3.4 Detection wavelength............................................................................................................ 81
3.4 Validation of the analytical procedure ..................................................................................... 83
viii
3.4.1Limit of detection (LOD) and limit of quantitation (LOQ) ................................................... 90
3.4.2 Specificity ............................................................................................................................. 97
3.4.3 Linearity and range ............................................................................................................... 97
3.4.4 Accuracy and precision ......................................................................................................... 99
3.5 Concluding remarks ................................................................................................................. 99
3.6 List of references.................................................................................................................... 103
Chapter 4: Characterization of monoclonal antibody using capillary sodium dodecyl sulfate gel
electrophoresis ............................................................................................................................. 107
4.1Introduction ............................................................................................................................. 107
4.2 Materials and methods ........................................................................................................... 111
4.2.1Reagents and solutions ......................................................................................................... 111
4.2.2 Preparation of solutions ...................................................................................................... 112
4.2.3 Preparation of samples ........................................................................................................ 113
4.2.4 Capillary SDS (cSDS) analysis ........................................................................................... 114
4.2.5 Capillary conditioning......................................................................................................... 114
4.3 Results and discussion ........................................................................................................... 115
4.3.1 Effect of sample buffer pH.................................................................................................. 115
4.3.2 Alkylation condition ........................................................................................................... 119
4.3.3 Incubation temperature and time......................................................................................... 119
4.3.4 Reduction conditions........................................................................................................... 123
4.3.5 Characterization of the aglycosylated heavy chain (AGHC) .............................................. 123
4.3.6 Summary of optimized parameters of the cSDS Method.................................................... 126
4.4 Method validation .................................................................................................................. 129
4.4.1 Specificity ........................................................................................................................... 129
4.4.2 Accuracy ............................................................................................................................. 135
4.4.3 Precision (repeatability and intermediate precision) ........................................................... 135
ix
4.4.4 Limit of quantitation (LOQ) ............................................................................................... 138
4.4.5 Linearity and range ............................................................................................................. 144
4.4.6 Sample stability ................................................................................................................... 146
4.5 Concluding remarks ............................................................................................................... 146
4.6 List of references.................................................................................................................... 150
Chapter 5: Optimization of injection length into a capillary for detection sensitivity
enhancement in micellar electrokinetic chromatography............................................................. 157
5.1 Introduction ............................................................................................................................ 157
5.2 Materials and methods ........................................................................................................... 163
5.2.1 Reagents and solutions ........................................................................................................ 163
5.2.2 Capillary electrophoresis (CE) ............................................................................................ 165
5.2.3 Capillary conditioning......................................................................................................... 166
5.3 Results and discussion ........................................................................................................... 166
5.3.1 Injection length vs. retention factor .................................................................................... 166
5.3.2 Injection length vs. efficiency ............................................................................................. 172
5.4 Concluding remarks ............................................................................................................... 177
5.5 List of References .................................................................................................................. 180
Chapter 6: Conclusions and future directions .............................................................................. 184
6.1 Conclusions ............................................................................................................................ 184
6.2 Future directions .................................................................................................................... 186
6.3 List of references.................................................................................................................... 188
Appendix A: UV spectra of the six cell culture media components ............................................ 189
Appendix B: Detection sensitivity comparison of cell culture media components at fixed
wavelength ................................................................................................................................... 195
x
List of Tables
Table 1.1List of parameters that affect EOF in CE [41-42, 30] ..................................................... 13
Table 2.1 Precision of migration times (MTs) and relative migration times (RMTs) for two sets
of analyses (Run-1 and Run-2) based on triplicate injections of each preparation ........................ 50
Table 2.2 Precision of relative peak area (%) for two sets of analyses (Run-1 and Run-2) based
on triplicate injections of each preparation .................................................................................... 51
Table 2.3 System suitability data of the BRP primary standard .................................................... 53
Table 2.4 Precision of relative peak area for Run-3 and Run-4 based on triplicate injections of
each preparation ............................................................................................................................. 56
Table 2.5 Average relative areas (%) of the isoforms in all secondary standard preparations ...... 58
Table 2.6 Average relative areas (%) of the isoforms in all secondary standard preparations for
the evaluation of duplicate injection .............................................................................................. 59
Table 2.7 Precision of relative migration time with 2-hour conditioning vs. 12-hour
conditioning ................................................................................................................................... 61
Table 2.8 Relative area (%) of each isoform of the epoetin alfa secondary standard (EPO Std)
and 2000 (2K) IU/mL F-EPO and 40,000 (40K) IU/mL F-EPO ................................................... 63
Table 2.9 Relative area (percent) of each isoform of the epoetin alfa secondary standard (EPO
Std) and 2000 (2K) IU1/mL F-EPO2 and 40,000 (40K) IU/mL F-EPO........................................ 64
Table 3.1 Limit of detection and limit of quantitation data for hypoxanthine ............................... 91
Table 3.2 Limit of detection and limit of quantitation data for riboflavin ..................................... 92
Table 3.3 Limit of detection and limit of quantitation data for xanthine ....................................... 93
Table 3.4 Limit of detection and limit of quantitation data for mycophenolic acid....................... 94
Table 3.5 Limit of detection and limit of quantitation data for folic acid ...................................... 95
Table 3.6 Limit of detection and limit of quantitation data for nicotinic acid ............................... 96
Table 3.7 Linearity and range data (10-100 μM) ......................................................................... 100
Table 3.8 Accuracy and precision data ........................................................................................ 101
Table 4.1 Area percentage of IgG and Impurity using sample buffers with different pH and
varied ionic strength ..................................................................................................................... 116
Table 4.2 Area percentage of IgG and Impurity using sample buffers with different pH with the
same conductivity ........................................................................................................................ 120
xi
Table 4.3 Optimization of alkylating agent (IAM) concentration ............................................... 122
Table 4.4 Comparison of mAb purity under different incubation conditions for reduced and
non-reduced samples. NA: Not Available ................................................................................... 124
Table 4.5 Effect of reduction conditions on the analysis of reduced mAb .................................. 125
Table 4.6 Spike recovery of PNGase F treated samples .............................................................. 128
Table 4.7 Summary of optimized method parameters ................................................................. 130
Table 4.8 Matrix effects study of non-reduced mAb ................................................................... 134
Table 4.9 Matrix effects study of reduced mAb .......................................................................... 134
Table 4.10 Accuracy of mAb purity under non-reduced conditions ............................................ 136
Table 4.11 Accuracy data for mAb purity under reduced conditions .......................................... 137
Table 4.12 Repeatability and intermediate precision of non-reduced mAb cSDS assay ............. 139
Table 4.13 Repeatability and intermediate precision for reduced mAb cSDS assay ................... 140
Table 4.14 LOQ data for the non-reduced mAb cSDS assay....................................................... 142
Table 4.15 LOQ data for the reduced mAb cSDS assay .............................................................. 143
Table 4.16 Linear regression equations for mAb cSDS assay ..................................................... 145
Table 4.17 Stability data for three freeze and thaw cycles of mAb RRS ..................................... 147
Table 4.18 Stability data for mAb samples prepared at 2-8° C ................................................... 148
Table 5.1 Injection time and calculated injection length (n=3).................................................... 173
Table 5.2 Migration time and calculated retention factors (n=3)................................................. 174
xii
List of Figures
Figure 1.1 Schematic of a typical CE instrument. ©Agilent Technologies, Inc. 2009. Courtesy
of Agilent Technologies, Inc ............................................................................................................ 2
Figure 1.2 Voltage-driven flow in capillary electrophoresis vs. pressure-driven flow in column
chromatography [24]. Reprinted with permission from Physics Today, vol. 54, ©copyright
2001, American Institute of Physics ................................................................................................ 5
Figure 1.3 Formation of zeta potential. The inside wall of the capillary is covered by silanol
groups (Si-OH) that are deprotonated (Si-O-) at pH ≥ 3.0. Si-O- attracts cations to the inside
wall of the capillary. The distribution of charge at the surface is described by the Stern doublelayer model and results in the zeta potential [33]. Reprinted from J. Chromatogr. 1991, 559,
69-80, © Copyright (1991), with permission from Elsevier ............................................................ 8
Figure 1.4 Effect of buffer pH on silanol groups at the surface of fused silica capillary. At low
pH the silanol groups are not ionized. At high pH the silanol groups are ionized ......................... 10
Figure 1.5 Effect of buffer pH on electroosmotic mobility [37]. Reprinted with permission
from J. High Resolut. Chromatogr. Chromatogr. Commun. 1985, 8, 407-11, © Copyright 1985,
John Wiley & Sons, Inc. ................................................................................................................ 11
Figure 1.6 Ohm’s Law Plot. A. Plot of observed current vs. applied voltage for each of three
buffers: 100 mM phosphate, pH 2.5; 100 mM borate, pH 8.3; 100 mM CAPS, pH 11.0. B.
Direct plot of current vs. applied voltage for 100 mM CAPS, pH 11.0. A straight line drawn
from the front edge of the plateau illustrates the ability of the cooling system to dissipate the
heat generated by the current [30, 40]. Reproduced from Handbook of capillary and microchip
electrophoresis and associated microtechniques, © Copyright (2008), with permission from
CRC Press. ..................................................................................................................................... 14
Figure 1.7 Illustration of CZE separation of cations, anions, and neutral compounds .................. 17
Figure 1.8 Illustration of MEKC separation. SDS micelles travel against the EOF but because
of stronger EOF, they are swept to the cathode. Analytes are separated due to their interaction
time with SDS micelles. ................................................................................................................. 20
Figure 1.9 Elution modes in MEKC, where veo, is the electroosmotic velocity; vsc is the velocity
of separation carrier; vs is the analyte velocity. Adapted from reference [48] ............................... 23
Figure 1.10 Illustration of CGE separation. Different sizes of analytes are separated by
electrophoresis under the condition of minimized EOF. vep: electrophoretic velocity. ................. 26
Figure 2.1 Experiment setup of an electronic infusion pump using syringes connected to nonionic detergent (NID) trap cartridges to remove polysorbate 80. The electronic infusion pump
is programmed for different syringe sizes and flow rates, which enables the process to be
automated ....................................................................................................................................... 42
xiii
Figure 2.2 Demonstration of capillary restoration. The peak area and migration time appear
consistent after 36th injection. A: 1st EPO sample injection; B: 36th EPO sample injection.
Peaks 3-8 are six isoforms of EPO................................................................................................. 47
Figure 2.3 Electropherograms of BRP primary standard vs. epoetin alfa secondary standard. A:
BRP primary standard; B: epoetin alfa secondary standard. Isoforms 1-8 were very well
separated. Isoforms 1 and 2 were not present in the epoetin alfa sample ...................................... 52
Figure 3.1 Structures and pKa’s of the cell culture media components. a: folic acid (2.3 (pKa1),
8.3 (pKa2)); b: nicotinic acid (4.85); c: hypoxanthine (8.7); d: xanthine (7.4 (pKa1), 11.1 (pKa2));
e: riboflavin (10.2); f: mycophenolic acid (4.5) [15, 19, 24-26] .................................................... 71
Figure 3.2 Comparison of the separation of the cell culture media components under (A) CZE
and (B) MEKC conditions. Compound identification (10 µM each): 1. hypoxanthine; 2.
riboflavin; 3. xanthine; 4. mycophenolic acid; 5. folic acid; 6. nicotinic acid; and 7. 2naphthalenemethanol (neutral marker). Analytical conditions: 20 mM NaH2PO4, 20 mM
Na2B4O7, pH 9.0, 40 mM SDS (MEKC only), 25 kV, 20°C, absorbance detection at 210 nm.
Fused silica capillary dimensions: 50 µm i.d. x 40 cm (40 cm effective length) ........................... 75
Figure 3.3 Matrix interference under CZE condition. (A) standards under CZE conditions (B)
sample matrix under CZE conditions. 1. hypoxanthine 2. riboflavin 3. xanthine 4.
mycophenolic acid 5. folic acid 6. nicotinic acid. Other conditions as in Figure 3.2 .................... 77
Figure 3.4 Effect of SDS concentration on the separation of the cell culture media components:
(A) 10 mM SDS, (B) 20 mM SDS, (C) 50 mM SDS, (D) 75 mM SDS, and (E) 100 mM SDS.
Compound identification and other conditions as in Figure 3.2 .................................................... 79
Figure 3.5 Effect of SDS concentration on the retention time of the cell culture media
components. Conditions as in Figure 3.2 ...................................................................................... 80
Figure 3.6 Effect of buffer pH on separation (A) At pH 8.0, folic acid and nicotinic acid (5 and
6) co-eluted; (B) At pH 8.5, peak shape of riboflavin (2) and nicotinic acid (6) were distorted.
Other conditions as in Figure 3.2B ................................................................................................ 82
Figure 3.7 Comparison of detection wavelength (210 nm, 220 nm, 254 nm, 275 nm) on the
calibration sensitivity for hypoxanthine. Other conditions as in Figure 3.2 .................................. 84
Figure 3.8 Comparison of detection wavelength (210 nm, 220 nm, 254 nm, 275 nm) on the
calibration sensitivity for riboflavin. Other conditions as in Figure 3.2 ........................................ 85
Figure 3.9 Comparison of detection wavelength (210 nm, 220 nm, 254 nm, 275 nm) on the
calibration sensitivity for xanthine. Other conditions as in Figure 3.2 .......................................... 86
Figure 3.10 Comparison of detection wavelength (210 nm, 220 nm, 254 nm, 275 nm) on the
calibration sensitivity for mycophenolic acid. Other conditions as in Figure 3.2 .......................... 87
Figure 3.11 Comparison of detection wavelength (210 nm, 220 nm, 254 nm, 275 nm) on the
calibration sensitivity for folic acid. Other conditions as in Figure 3.2 ......................................... 88
Figure 3.12 Comparison of detection wavelength (210 nm, 220 nm, 254 nm, 275 nm) on the
calibration sensitivity for nicotinic acid. Other conditions as in Figure 3.2 .................................. 89
xiv
Figure 3.13 Electropherograms of spiked (A) and non-spiked (B) samples from late stage
protein purification process. Conditions as in Figure 3.2 ............................................................. 98
Figure 4.1 Schematic diagram of Immunoglobulin (IgG) [1]. Fab: antigen-binding fragment
generated by proteolysis with papain; Fc: "crystallizable" fragment of generated by proteolysis
with papain. Reproduced with permission from Lehninger Principles of Biochemistry, 5th
edition, ©Copyright (2008) W. H. Freeman and Company ......................................................... 108
Figure 4.2 Effect of sample buffer pH on cSDS separation of non-reduced mAb. Peak
identification: No. 1: Internal standard with molecular weight of 10,000 Dalton; No. 2: largest
impurity peak; No. 3: mAb (IgG). Peaks earlier than No.1 were from sample buffer blank.
Other impurity peaks were not labeled but integrated for total peak areas.
A: 25 mM citrate-phosphate sample buffer at pH 6.5. Relative area (%) of IgG (peak No.3) was
98.3% and relative area of impurity (peak No. 1) was 1.0%;
B: Original Beckman sample buffer (100 mM Tris-HCl, 1.0% SDS, pH 9.0). Relative area (%)
of IgG (peak No.3) was 96.4% and relative area of impurity (peak No. 2) was 2.3%.
Separation conditions: 30 cm x 50 μm bare silica, electrokinetic injection (-5 kV for 15 s),
separation voltage: -15 kV for 35 minutes ................................................................................... 121
Figure 4.3 PNGase F treated reduced mAb vs. non-treated reduced mAb. 1: Light chain; 2:
Aglycosylated heavy chain; 3: Heavy chain A: Reduced mAb not being treated by PNGase F;
B: Reduced mAb treated by PNGase F. Separation conditions: 30 cm x 50 μm bare silica,
electrokinetic injection (-5 kV for 15 s), applied separation voltage: -15 kV for 35 minutes,
buffer: 25 mM citrate-phosphate, 1.0% SDS, pH 6.5 .................................................................. 127
Figure 4.4 Typical electropherogram of system suitability sample. 1.Internal reference standard
(10 kDa); 2. carbonic anyhydrase (31 kDa); 3. bovine serum albumin (66 kDa); 4. βgalactosidase (116 kDa); 5. mAb RRS (148 kDa) Separation conditions as in Figure 4.3 .......... 131
Figure 4.5 Matrix effect of non-reduced mAb. A: water blank; B: formulation buffer; C: mAb
sample. Peak identification 1: internal standard; 2: impurity; 3: mAb (IgG). Other peaks prior
to 1 are from the sample buffer. Separation conditions as in Figure 4.3 ..................................... 132
Figure 4.6 Matrix effect of reduced mAb. A: water blank; B: formulation buffer; C: mAb
sample. Peak identification 1: internal Standard; 2: light chain; 3: heavy chain. Separation
conditions as in Figure 4.3. .......................................................................................................... 133
Figure 5.1 Basis of stacking in CZE using cations as an example. The region containing sample
ions is a low conductivity solution, while the background region is a high conductivity solution.
Upon application of voltage, the low conductivity region will experience a high electric field
compared to the high conductivity background region. Sample ions then move faster in the low
conductivity region than in the higher conductivity region resulting in the reduction of sample
zone or higher sample concentration. [21] ................................................................................... 158
Figure 5.2 Sweeping in a homogenous electric field in MEKC using a negatively charged
micelle as a pseudostationary phase (PSP) and a negligible EOF environment. (A) A longer
than a typical injection of sample zone prepared in a matrix having a conductivity similar to
micellar background electrolyte (BGE). (B) Upon application of a negative voltage, the BGE
enters the sample zone and sweeps (concentrates) the analyte molecules. (C) The final swept
zone is formed when the BGE completely fills the sample zone. [22] ........................................ 159
xv
Figure 5.3 Structures of the studied compounds: (a) phenyl methyl ketone (acetophenone); (b)
phenyl ethyl ketone (propiophenone); (c) phenyl propyl ketone (butyrophenone)...................... 164
Figure 5.4 Separation of swept zone in a diluted pseudostationary phase (PSP) environment.
Explanation is in the text.............................................................................................................. 169
⎛ l inj
⎜l −l
⎝ eff inj
Figure 5.5 Plot of log (1/k) versus log⎜
⎞
⎟ with eight data points corresponding to
⎟
⎠
injection times of 5-100 seconds .................................................................................................. 175
⎛ l inj
⎜l −l
⎝ eff inj
Figure 5.6 Plot of log(1/k) versus log⎜
⎞
⎟ with five data points corresponding to
⎟
⎠
injection times of 20-50 seconds .................................................................................................. 176
Figure 5.6 Electropherograms of 50 s injection (A) vs. 100 s injection (B). Analyte
identification: (1) acetophenone; (2) propiophenone; (3) butyrophenone. Analytical conditions:
50 mM phosphate, 50 mM sodium dodecyl sulfate, pH 2.0, -25 kV, 210 nm, 20°C, 72 cm x 50
µm capillary . Analyte concentrations: 40 µM (A); 20 µM (B)................................................... 178
Figure 5.7 Plots of plate number vs. injection time. Injection time of 35 seconds gives highest
plate number for butyrophenone and acceptable plate number for acetophenone ....................... 179
Figure A1. UV spectrum of hypoxanthine with λmax = 200 nm. Solvent: 20 mM NaH2PO4, 20
mM Na2B4O7, 40 mM SDS, pH 9.0. Analyte concentration: 0.1 mM. Collected on line using a
photodiode array detector with a 50 μm i.d. fused silica capillary. ............................................. 189
Figure A2. UV spectrum of riboflavin with λmax = 270 nm. Solvent and solute concentration as
in Figure A1. ................................................................................................................................ 190
Figure A3. UV spectrum of xanthine with λmax = 200 nm. Solvent and solute concentration as
in Figure A1. ................................................................................................................................ 191
Figure A4. UV spectrum of mycophenolic acid with λmax = 225 nm. Solvent and solute
concentration as in Figure A1. ..................................................................................................... 192
Figure A5. UV spectrum of folic acid with λmax = 190 nm. Solvent and solute concentration as
in Figure A1. ................................................................................................................................ 193
Figure A6. UV spectrum of nicotinic acid with λmax = 190 nm. Solvent and solute concentration
as in Figure A1. ............................................................................................................................ 194
Figure B1. Comparison of analyte absorbance at wavelength of 210 nm.................................... 195
Figure B2. Comparison of analyte absorbance at wavelength of 222 nm.................................... 196
Figure B3. Comparison of analyte absorbance at wavelength of 254 nm.................................... 197
xvi
Figure B4. Comparison of analyte absorbance at wavelength of 275 nm.................................... 198
xvii
List of symbols
α
Δ
ΔP
Δv
Δμep
ε0
εr
η
σ
σ2
μeo
μep
μep,avg
μnet
ζ
ζa
ϕ
δ
λmax
AC
Cmic
Cs
D
e
E
k
k1
k2
selectivity factor
distance between two zones
pressure across the capillary
difference in velocities
difference in electrophoretic mobilities of the two zones
permittivity of a vacuum
relative permittivity of the buffer
viscosity
standard deviation of residue
variance or standard deviation squared
electroosmotic mobility
electrophoretic mobility
average electrophoretic mobility of the analytes
net mobility
zeta potential
zeta potential of analyte
phase ratio
double layer thickness
absorbance wavelength at maximum intensity
Isoform of EPO anion
sample concentration
micelle concentration in the running buffer
micelle concentration in the separation zone
diffusion coefficient
total excess charge in solution per unit area
electric field
retention factor
retention factor, peak #1
retention factor, peak #2
k
average retention factor
conductivity of running buffer
conductivity of sample buffer
length of the capillary
effective capillary length (inlet to the detector)
total capillary length
injection length
swept length
molar concentration
efficiency or number of theoretical plates
number of replicates
partition coefficient
injection amount
coefficient of determination
capillary inner radius
resolution
migration time of a neutral solute
migration time of the micelle
ka
kb
L
Leff
Ltot
linj
lsweep
M
N
n
P
Qinj
r2
r
Rs
t0
tmc
xviii
tmob
tpsp
tR
tsc
vavg
vep
V
Vsc
Vaq
Vmob
Wavg
residence time in mobile phase
migration time of the pseudostationary phase
migration time of the solute
residence time with the separation carrier
average velocity
electrophoretic velocity
voltage
volume of separation carrier
volume of aqueous phase
volume of mobile phase
average width of the two zones measured via tangents to the baseline
xix
List of abbreviations
ACS
AGHC
BGE
BRP
BSA
CAPS
CE
CGE
CIEF
CITP
CMC
cSDS
CZE
DAB
EKC
EOF
EPO
Fab
FB
Fc
F-EPO
GC
H3PO4
HAc
HC
HCl
HPLC
IAA
IAM
IgG
i.d.
IU
LC
mAb
2-ME
MEKC
MeOH
NaHAc
NaOH
MT
NEM
NID
Ph. Eur.
PNGase F
PS80
PSP
QC
RMT
American Chemical Society
aglycosylated heavy chain
background electrolyte
biological reference preparation
bovine serum albumin
N-cyclohexyl-3-aminopropanesulfonic acid
capillary electrophoresis
capillary gel electrophoresis
capillary isoelectric focusing
capillary isotachophoresis
critical micelle concentration
capillary sodium dodecyl sulfate gel electrophoreis
capillary zone electrophoresis
1, 4-diaminobutane
electrokinetic chromatography
electroosmotic flow
erythropoietin
antigen binding fragment
formulated bulk
crystallizable fragment
formulated erythropoietin
gas chromatography
phosphoric acid
acetic acid
heavy chain
hydrochloric acid
high performance liquid chromatography
iodoacetic acid
iodoacetamide
immunoglobulin G
inner diameter
International Units
light chain
monoclonal antibody
2-mercaptoethanol
micellar electrokinetic chromatography
methanol
sodium acetate
sodium hydroxide
migration time
N-ethylmaleimide
non-ionic detergent
The European Pharmacopoeia
Peptide N-Glycanase
polysorbate 80
pseudostationary phase
quality control
relative migration time
xx
RRS
RSD
SDS
SDS-PAGE
SEC
TLR
Tris-base
Tris-HCl
Tricine
UV
research reference standard
relative standard deviation
sodium dodecyl sulfate
sodium dodecyl sulfate-polyacrylamide gel electrophoresis
size exclusion chromatography
toll-like receptor
tris-(hydroxymethyl)aminomethane
tris-(hydroxymethyl)aminomethane hydrochloride
N-tris-(hydroxymethyl)methylglycine
ultraviolet
xxi
Abstract
Quantitative Biopharmaceutical Applications of Capillary Electrophoresis
Junge Zhang
Dr. Joe P. Foley
Various modes of capillary electrophoresis offer numerous possibilities for biopharmaceutical
analysis. Chapter 2 studied capillary zone electrophoresis (CZE) for characterization of
glycoprotein erythropoietin by its isoform distribution. The CZE method for identification of
recombinant human erythropoietin described in the European Pharmacopoeia has shown poor
reproducibility. In this study, it was found that the root cause of the irreproducibility was due to
inadequate capillary conditioning. The method was optimized to make it more robust and
suitable for quality control laboratory for analysis of epoetin alfa. Further, a procedure for
removal of polysorbate 80 from formulated epoetin alfa was developed, allowing the material
to be analyzed using the modified CZE method.
Chapter 3 describes a micellar electrokinetic chromatography (MEKC) method for the
determination of residual cell culture media components including folic acid, hypoxanthine,
mycophenolic acid, nicotinic acid, riboflavin, and xanthine. Acceptable limits of detection and
quantitation were obtained. The MEKC method was compared to the corresponding CZE
method using the same running buffer containing no sodium dodecyl sulfate (SDS). The effect
of SDS concentration on separation, the pH of the running buffer, and the detection wavelength
were studied and optimal MEKC conditions were established.
Chapter 4 describes a capillary SDS gel electrophoresis method for purity and impurity
analysis of monoclonal antibodies. The study investigated the effect of sample buffer pH,
incubation temperature and duration, alkylation conditions with iodoacetamide, and reduction
conditions with 2-mercaptoethanol. It was observed that the sample buffer at slight acidic
xxii
conditions (pH 5.5-6.5) greatly decreased thermally induced fragmentation of non-reduced
CNTO3157. As such, a citrate-phosphate buffer at pH 6.5 was used for sample preparation to
replace the commonly used Beckman sample buffer (pH 9.0).
Chapter 5 details an investigation of the effect of sample injection length on efficiency and
detection sensitivity enhancement in MEKC using the sweeping technique under conditions of
suppressed electroosmotic flow. The relationship between injection length and retention factor
was derived and experimentally confirmed. A method for predicting the optimal injection
length for a given analyte was developed.
1
Chapter 1: Introduction to capillary electrophoresis: fundamentals and
applications
1.1 Development of capillary electrophoresis
Capillary electrophoresis (CE) is a separation technique that employs narrow-bore
capillaries to perform electrophoresis in free solution or nonconductive medium,
such as a gel [1]. CE was first introduced in 1967 by Hjerten [2] and further
developed and improved by Mikkers et al. [3] and Jorgenson and Lukacs [4] by
reducing capillary diameters that achieved high column efficiencies. Other notable
developments after the introduction of CE include transition from the slab-gel to the
capillary format in 1980s [5-8] for protein and DNA separations, the use of micellar
solution as a medium for separation of neutral and charged species [9], and the
breakthrough of on-line concentration phenomenon for trace analysis [10-11].
These developments led to a variety of separation modes including capillary zone
electrophoresis (CZE), capillary gel electrophoresis (CGE), and micellar
electrophoresis (MEKC). Different separation modes greatly enhanced the
versatility of CE technique and expanded the horizon of its applications [12-18].
1.2 Development of capillary electrophoresis instrumentation
A schematic presentation of a CE system is given in Figure 1.1. Analytes are
separated in a small diameter capillary filled with running buffer. Samples are
introduced by either hydrodynamic (pressure or vacuum) or electrokinetic (positive
or negative voltage) techniques. Upon application of an electric field, compounds
are separated due to their differences in mobility. On-capillary detection
2
Figure 1.1 Schematic of a typical CE instrument. ©Agilent Technologies, Inc. 2009,
Courtesy of Agilent Technologies, Inc.
3
is normally used. The detector signal is collected and processed through a
computerized data system.
Fused silica with polyimide covering in various lengths up to 100 cm and 25-100
µm inner diameters are normally used as separation capillaries. Fused silica is
transparent to UV light, flexible when coated with polymer, and inexpensive,
making it an ideal material for CE capillaries [19] . The commonly available
detectors include absorbance, fluorescence, conductivity, or a mass spectrometer.
Most high voltage supplies are restricted to 30 kV due to the limitation of electrical
breakdown of insulators and atmosphere [20].
A high level of automation and safety features are the main considerations for
instrumentation development. In 1988, the first commercial CE instrument was
made available for analysts [21] and followed by numerous models from several
leading scientific instrument companies. The recent improvements in CE
instrumentation bring unprecedented separation power that resolves many analytical
challenges for CE applications.
1.3 Fundamental theory of capillary electrophoresis separation
1.3.1
Efficiency
Most chromatographic separation techniques, such as high performance liquid
chromatography (HPLC) and gas chromatography (GC), are pressure-driven
separations. CE is a voltage-driven analytical separation technique in which charged
or neutral analytes move with different mobility under the influence of an electric
4
field. The major difference between pressure-driven and voltage-driven separations
lies in the bulk flow profile due to the different separation mechanisms. In open
tubular pressure-driven systems, the frictional forces of the mobile phase interacting
at the walls of the tubing result in radially symmetric velocity gradients throughout
the tube. As a result, the fluid velocity is greatest at the middle of the tube and
approaches zero near the walls. A consequence of this pressure-driven flow is a
parabolic radial velocity profile, also known as a laminar profile, which attributes to
band broadening [22].
In voltage driven system, the flow is generated uniformly down the entire length of
the capillary. There is no pressure drop in CE, and the radial flow profile is uniform
across the capillary except very close to the wall, where the flow rate approaches
zero [23]. Figure 1.2 shows a comparison of a pressure-driven flow profile in
column chromatography vs. a voltage driven flow profile in capillary
electrophoresis [24]. One advantage of flat flow over laminar flow is that it
eliminates the contributions from eddy dispersion to the column plate height [25,
19]. The separation efficiency (N) in CE can be determined, assuming that
longitudinal diffusion is the only source of band broadening [26]:
N=
μ netV ⎛ Leff ⎞
⎜
⎟=
2 D ⎜⎝ Ltot ⎟⎠
μ net ELeff
2D
(1.1)
where D is the diffusion coefficient of an ion in electrolyte solution; µnet is net
5
Figure 1.2 Voltage-driven flow in capillary electrophoresis vs. pressure-driven flow
in column chromatography [24]. Reprinted with permission from Physics Today,
vol. 54, ©copyright 2001, American Institute of Physics.
6
(apparent) electrophoretic mobility of an ion; V is the voltage applied to the
capillary; Leff is the effective capillary length (from inlet to the detection window);
Ltot is the total capillary length; E is the electric field . As can be seen from the
equation, the separation efficiency can be enhanced by higher electric field. A
shorter capillary can be used to reduce analysis time without compromising the
column efficiency. However, in practice, there are other sources that cause band
broadening in CE, including Joule heating, capillary wall binding, and finite width
of injection, which result in lower than calculated efficiency [27-29]. These factors
can be controlled by proper column cooling, capillary conditioning, short injection
plug, etc.
1.3.2
Electroosmotic flow
The net mobility of an analyte is given by the equation [30]:
μ net = μ eo + μ ep
(1.2)
where µeo is the electroosmotic mobility and µep is the electrophoretic mobility.
The electroosmotic mobility (m2/[Vs]) is the constant of proportionality between
electroosmotic velocity and electric field (V/m). Generally speaking, electroosmotic
flow (EOF) is the motion of an electrolyte solution driven by a combination of a
charge imbalance at a solid-liquid interface and an electrical field in that liquid [31,
19]. In capillary electrophoresis, EOF is induced by the electric field in the parallel
direction with the capillary wall on the free electric charge in the diffuse part of the
electric double layer at the solid–liquid interface inside the capillary [32]. The
origin of EOF is described below.
7
The inner surface of a fused silica capillary has ionizable silanol groups (Si-OH).
These silanol groups readily dissociate (Si-O-) when in contact with an electrolyte
solution with a pH ≥ 3, producing a capillary wall with an intrinsic negative charge.
At the silica surface, a slight excess of positively charged ions from the buffer
solution are attracted to the negatively charged capillary wall, forming an electrical
double layer and a potential difference, called zeta potential, denoted by ζ, as
depicted in Figure 1.3 [33]. The zeta potential is described by Stern’s model in
many CE text books [1, 34, 30]. Stern’s model for an electrical double layer is
composed of a compact layer of adsorbed ions (Stern layer) and a diffuse layer. The
thickness of the electrical double layer is normally about 10 nm [19]. The zeta
potential decreases exponentially with increasing distance from the capillary wall
surface. Zeta potential is expressed by:
ζ =
4πδe
ε
(1.3)
where ε is the buffer’s dielectric constant; e is total excess charge in solution per
unit area; δ is the double layer thickness or Debye ionic radius. When a voltage is
applied across the capillary, cations in the diffuse layer are forced to migrate
towards the cathode, carrying the bulk solution with them. The result is a net flow in
the direction of the cathode, with an electroosmotic mobility described as [1]:
μ eo =
εζ
4πη
where η is the viscosity of the buffer in the electrical double layer.
(1.4)
8
Figure 1.3 Formation of zeta potential. The inside wall of the capillary is covered
by silanol groups (Si-OH) that are deprotonated (Si-O-) at pH ≥ 3. Si-O- attracts
cations to the inside wall of the capillary. The distribution of charge at the surface is
described by the Stern double-layer model and results in the zeta potential [33].
Reprinted from J. Chromatogr. 1991, 559, 69-80, © Copyright (1991), with
permission from Elsevier.
9
EOF is a very important parameter in CE. On the one hand, it provides separation
of ions with positive and negative charges and on the other hand, it is a major
contributor to migration time variability form run to run if it is not properly
controlled. There are many factors that affect EOF, including buffer pH and
concentration, temperature, viscosity, capillary surface, field strength, and buffer
additives such as surfactants and organic modifiers. Understanding how these
factors affect the EOF is critical for CE method development.
For fused silica, at high pH (> 9) the silanol groups are mostly ionized and produce
a high zeta potential and a dense electrical double layer, so the EOF is high [35]. At
low pH (< 3), the silanol groups are hardly ionized and the zeta potential is low, so
is the EOF [36]. Figures 1.4 illustrates the effect of buffer pH on silanol groups.
Figure 1.5 depicts the effect of buffer pH on electroosmotic mobility.
Buffer ionic strength also affects EOF. As ionic strength increases, the double layer
becomes compressed, which results in a decreased zeta potential and lower EOF.
This can be seen from equation (1.3) as well. Normally a high ionic strength buffer
is preferred to suppress ion exchange effect between the charged analyte ions and
ionized silanol groups on the capillary wall. However, a high buffer ionic strength
will generate high current and results in substantial Joule heating that the capillary
cooling system cannot handle. In addition, Joule heating can cause band broadening
and reduce resolution. Manipulation of EOF by buffer alone (pH and ionic strength)
is difficult because some other factors such as the contributions of
10
Figure 1.4 Effect of buffer pH on silanol groups at the surface of fused silica
capillary. At low pH the silanol groups are not ionized. At high pH the silanol
groups are ionized.
11
Figure 1.5 Effect of buffer pH on electroosmotic mobility [37]. Reprinted with
permission from J. High Resolut. Chromatogr. Chromatogr. Commun. 1985, 8,
407-11, © Copyright 1985, John Wiley & Sons, Inc.
12
specific adsorption, the competition of different ions for binding sites, and effects
like secondary adsorption are not yet fully understood [36]. Other factors that affect
EOF are summarized in Table 1.1.
1.3.3
Ohm’s law
Because buffer concentration has both positive and negative effect on CE
separation, there is a high demand for finding an optimized buffer concentration for
CE method development. Nelson et al. described a simple method, termed “Ohm’s
Law Plot”, that can easily determine the optimal buffer concentration and the
highest voltage that can be used with this particular buffer system [38]. In this
method, the applied voltage is varied over short intervals across a capillary that is
filled with buffer at a set temperature, and the current is recorded at each voltage. A
graph is made of current vs. voltage. The plot should be linear if Joule heat is
effectively dissipated. Whenever the curve starts showing a positive deviation from
linearity, the cooling system has reached its capacity. Operating on the linear
portion of the curve will give high capillary efficiency. Generally for an optimal
separation, the heat should not exceed 1.5 W/m [39]. Figure 1.6 shows a typical
Ohm’s law plot with 100 mM concentrations of each of these three buffers:
phosphate pH 2.5, borate pH 8.3, and N-cyclohexyl-3-aminopropanesulfonic acid
(CAPS) titrated by NaOH to pH 11.0 [40]. At every applied voltage above, the
current is lowest with borate buffer and significantly higher with phosphate and
CAPS. At 25 kV, the calculated power for borate, phosphate, and CAPS is 0.58,
10.07, and 5.88 W/m, respectively. From this plot, it demonstrates that phosphate
13
Table 1.1 List of parameters that affect EOF in CE [41-42, 30]
Parameter
Buffer pH
Result
EOF increases at high
pH and decrease at low
pH
Comments
1.
2.
1.
Buffer ionic strength
Decreases zeta potential
and EOF when increased
Temperature
Changes viscosity 2-3%
per °C
Organic modifier
Changes zeta potential
and viscosity (usually
decreases EOF)
Electric field
Proportional change in
EOF
2.
1.
1.
2.
1.
2.
1.
Surfactant
Absorbs to capillary wall
via hydrophobic and/or
ionic interactions
2.
3.
Most convenient and useful
method to change EOF
May change charge or structure
of analyte
High ionic strength generates
high current and possible Joule
heating
Low ionic strength problematic
for sample adsorption
Often useful since temperature
is controlled automatically by
instrument
Complex changes, effect most
easily determined
experimentally
May change selectivity
Efficiency and resolution may
decrease when lowered
Joule heating may result when
increased
Anionic surfactants can
increase EOF
Cationic surfactant can
decrease or reverse EOF
Can significantly alter
selectivity
14
Phosphate
CAPS
Borate
Figure 1.6 Ohm’s Law Plot. A. Plot of observed current vs. applied voltage for each
of three buffers: 100 mM phosphate, pH 2.5; 100 mM borate, pH 8.3; 100 mM
CAPS, pH 11.0. B. Direct plot of current vs. applied voltage for 100 mM CAPS, pH
11.0. A straight line drawn from the front edge of the plateau illustrates the ability
of the cooling system to dissipate the heat generated by the current [30, 40].
Reproduced from Handbook of capillary and microchip electrophoresis and
associated microtechniques, © Copyright (2008), with permission from CRC Press.
15
and CAPS buffer should not be used at voltage above 10 kV and 15 kV,
respectively. It also shows that for 100 mM borate buffer, a voltage of up to 30 kV
can be used.
1.3.4
Electrophoretic mobility
In equation 1.1, besides electroosmotic mobility, the electrophoretic mobility
contributes to the net mobility of the analytes too. Electrophoretic mobility (µep) is
given by [1]:
⎛ 2 ⎞ ε 0ε r ζ a
⎝3⎠ η
μ ep = ⎜ ⎟
(1.5)
where ε0 is the permittivity of a vacuum; εr is the relative permittivity of the buffer
(εr=ε0εbuffer/4π); ζa is the zeta potential of the analyte; η is the viscosity of the buffer.
From equation (1.2), the apparent mobility of an analyte is not directly related to its
electrophoretic mobility, but to the combination of both its electrophoretic mobility
and the EOF mobility. In CE, if the buffer pH is high (pH ≥ 9.0) and a positive
voltage is applied to the inlet, the EOF will be in a direction of inlet to detector.
Because the electrophoretic mobility of anions will be slower than the EOF, the net
mobility of anions, cations, and neutral analytes will migrate towards the detector
regardless of their charge. The migration order will be cations, neutrals, and anions.
1.3.5
Separation modes
16
Separation mode in HPLC is determined by column stationary phase and mobile
phase. Separation mode in CE is determined only by buffer. In CE, one type of
capillary can be used for different modes of separations.
1.3.5.1 Capillary zone electrophoresis (CZE)
CZE is the most commonly used mode of separation in CE. Under high EOF
conditions, CZE is able to separate both cations and anions in the same run. The
separation in CZE is based on the differences in analytes’ electrophoretic mobilities
that result in different velocities of migration of ionic species [43].
Figure 1.7 illustrates a CZE separation. The EOF in uncoated fused silica capillaries
is usually significantly greater than the electrophoretic mobility of the individual
ions in the injected sample. Upon the application of an electric field, cations are
migrating towards the cathode and their speed is augmented by the EOF. Anions,
although electrophoretically migrating towards the anode, are swept towards the
cathode with the bulk flow of the running buffer. Under these conditions, cations
with the highest charge/frictional drag migrate first, followed by cations with lower
charge/frictional drag. All the neutral compounds migrate unresolved because their
charge is zero. Anions with lower charge/frictional drag ratio migrate earlier than
those with greater charge/frictional drag ratio. The anions with the greatest
electrophoretic mobilities migrate last. One important point to note is that it is
possible to change the charge of many ions by adjusting pH of the running buffer to
alter their ionization and hence electrophoretic mobility. The resolution (Rs) in CZE
is governed by the following relationships [44-45]:
17
EOF
+
Anion -2
Cation -2
Cation -1
0
Cation -3
Neutral
Anion -1
tm (min)
Figure 1.7 Illustration of CZE separations of cations, anions, and neutral
compounds.
18
Rs =
Δ
N ⎛⎜ Δv
=
4 ⎜⎝ v avg
Wavg
⎞
⎛
Δμ ep
⎟= N ⎜
⎟
4 ⎜⎝ μ ep ,avg + μ eo
⎠
⎞
⎟
⎟
⎠
(1.6)
Replacing N with equation (1.1), Rs can be written as:
⎛
⎞
V
⎛ 1 ⎞
⎟
Rs = ⎜
⎟(Δμ ep )⎜
⎜ D( μ + μ ) ⎟
⎝4 2⎠
ep
eo ⎠
⎝
1
2
(1.7)
where Δ is the distance between two zones; Wavg is the average width of the two
zones measured via tangents to the baseline; Δv and Δµep are the differences in the
velocities and electrophoretic mobilities of the two zones, respectively; vavg and
µep, avg are the average velocity and electrophoretic mobility of two zones; V is the
applied voltage; N is the theoretical plate number; and D is the diffusion coefficient
of one analyte.
Equation (1.7) shows that increasing voltage will improve resolution but only by its
square root. However, if Ohm’s law permits, the highest voltage should be used to
obtain fastest separation. EOF plays a major role in resolution. As can be seen from
the equation, the highest resolution will be attained when the EOF is approaching
the average electrophoretic mobility of the analytes but in the opposite direction of
analytes (µeo ≈ -µep,avg). This will result in almost infinite separation time. It is
obvious that higher resolution will also be attained when the difference in analyte
electrophoretic mobility is large. Other main parameters affecting CZE resolution
are capillary dimension and nature, separation electrolyte composition (pH, ionic
strength, salt nature, additives), and capillary temperature. Recently, more
methodologies have been applied to enhance resolution using non-aqueous and
19
hydro-organic electrolytes, isoelectric buffers and additives such as ionic liquids
[46-47].
1.3.5.2 Micellar electrokinetic chromatography (MEKC)
MEKC is derived from electrokinetic chromatography (EKC). EKC is a separation
technique based on a combination of electrophoresis and interactions of the analytes
with additives (e.g. surfactants) that form a dispersed secondary phase moving at a
different velocity [48], also called a pseudostationary phase or separation carrier
[49-50]. MEKC is a special case of EKC, in which the secondary phase is a micellar
dispersed phase in the capillary. MEKC can separate neutral as well as charged
analytes. In MEKC, surfactants are added to the running buffer to form micelles.
The separation of neutral analytes is based on the hydrophobic interaction of solutes
with the micelles. The stronger the interaction, the longer the solutes migrate with
the micelle. The selectivity of MEKC can be controlled by the choice of surfactant
and also by the addition of modifiers to the buffer.
Figure 1.8 shows an illustration of a MEKC separation using sodium dodecyl
sulfate (SDS) as a pseudostationary phase. SDS is a widely used anionic surfactant.
One end of SDS is the hydrophilic sulfate group, and the other end is the
hydrophobic C12 group. When a surfactant is in solution at a concentration higher
than its critical micelle concentration, CMC, it forms micelles which are
aggregation of individual surfactant molecules. Micelles have a three-
20
SDS
EOF
+
tR1
t0
tR2
tR3
-
tmc
tm
Figure 1.8 Illustration of MEKC separation. SDS micelles travel against the EOF
but because of stronger EOF, they are swept to the cathode. Analytes are separated
due to their interaction time with SDS micelles.
21
dimensional structure with the hydrophobic moieties of the surfactant in the interior
and the charged moieties at the exterior . When a hydrophobic compound is added
to an aqueous solution that contains micelles, it partitions into the hydrophobic
portions of the micelles. As a result, the analytes are maintained in the solution.
MEKC can separate both neutral and ionic compounds.
As shown in Figure 1.8, the analyte migration order in MEKC is different than
CZE. In MEKC, hydrophilic water soluble analytes do not partition into the
micelles. They are carried through the capillary at the rate of the EOF and are the
first to migrate at t0. The hydrophobic analytes that are totally solubilized by the
micelles spend all of the time in the micelles, and are carried through the capillary
at the same rate as the micelles and elute last at tmc. Analytes that spend part of their
time in the micelles migrate between t0 and tmc.
Unlike chromatography, the pseudostationary phase in MEKC has an effective
electrophoretic mobility due to its charge property. The employment of a
pseudostationary phase changes the dynamics of the analyte zone. The analyte
velocity in MEKC is expressed by [48]:
vs =
t mob
t rsc
1
k
v mob +
v sc =
v mob +
v sc
t mob + t rsc
t mob + t rsc
k +1
k +1
(1.8)
where tmob is the residence time in the mobile phase; trsc is the residence time with
the separation carrier; vsc is the observed velocity of the separation carrier; vmob is
the velocity of mobile phase (EOF) and k is the retention factor (nsc/nmob), where nsc
22
and nmob are the number of moles of solute in the separation carrier and mobile
phase, respectively.
Like chromatography, the retention factor (k) in MEKC is defined as residence time
in the separation carrier divided by residence time in the surrounding liquid [48].
The separation process is due to the distribution between two distinct phases having
two different observed mobilities [48]:
k = ϕP =
Vsc
P
Vmob
(1.9)
where ϕ is phase ratio; P is partition coefficient; Vsc is volume of separation carrier;
and Vmob is volume of surrounding mobile phase.
The retention factor and resolution in MEKC are more complex than in CZE. There
are three elution modes in MEKC, namely, normal mode, restricted mode, and
reversed mode, as shown in Figure 1.9. In each mode, the retention factor is
calculated differently.
Normal elution mode:
k=
t s − t0
t 0 (1 − t s / t sc )
k=
ts + t0
t 0 (t s t sc − 1)
(1.11)
ts − t0
t 0 (t s t sc + 1)
(1.12)
Reversed elution mode:
k=
Restricted elution mode:
(1.10)
23
veo
vsc
vs
(A) Normal Elution Mode
veo
vsc
vs
(B) Restricted Elution Mode
veo
vsc
vs
(C) Reversed Elution Mode
Figure 1.9 Elution modes in MEKC, where veo, is the electroosmotic velocity; vsc is
the velocity of separation carrier; vs is the analyte velocity. Adapted from reference
[48].
24
Corresponding to each elution mode, the resolution is also expressed differently
[51].
Normal elution mode:
Rs =
N ⎛ α − 1 ⎞⎛ k ⎞⎛ 1 − t 0 / t sc
⎟⎜
⎜
⎟⎜
4 ⎝ α ⎠⎜⎝ k + 1 ⎟⎠⎜⎝ 1 + (t 0 / t sc )k
⎞
⎟⎟
⎠
(1.13)
Reversed elution mode:
Rs =
N ⎛ α − 1 ⎞⎛ k ⎞⎛ 1 + t 0 / t sc ⎞
⎟
⎟⎜
⎜
⎟⎜
4 ⎝ α ⎠⎜⎝ k + 1 ⎟⎠⎜⎝ (t 0 / t sc )k − 1 ⎟⎠
(1.14)
Restricted elution mode:
Rs =
N ⎛ α − 1 ⎞⎛ k ⎞⎛ 1 + t 0 / t sc
⎟⎜
⎜
⎟⎜
4 ⎝ α ⎠⎜⎝ k + 1 ⎟⎠⎜⎝ 1 − (t 0 / t sc )k
⎞
⎟⎟
⎠
(1.15)
where α is selectivity factor (k2/k1); k is the average retention factor.
From above equations, in normal elution mode, the time window for a neutral
compound to elute is between t0 and tsc and the time ratio t0/tsc has a significant
impact on resolution.
Foley discovered that for neutral solutes in the normal elution mode, the highest
resolution can be achieved when kOPT ( Rs ) =
t sc / t 0 [52]; for charged solutes, the
_
highest resolution occurs when k OPT ( Rs ) = t mc t 0 (1 + μ r ) , where µr is the relative
electrophoretic mobility (µr =µep/µeo), i.e., the electrophoretic mobility of an analyte
(positive, zero, or negative) relative to the coefficient of electroosmotic flow [53].
In restricted and reversed modes, there is no time window for a neutral analyte to
25
elute because with the increase of retention factor, resolution reaches infinity. Even
for very low selectivity, high resolution can be obtained. However, the migration
time could be very long.
1.3.5.3
Capillary gel electrophoresis (CGE)
CGE is the capillary format of traditional slab-gel electrophoresis and is used for the
size-based separation of biological macromolecules such as oligonucleotides, DNA
restriction fragments and proteins [54]. The basic difference between traditional
slab gel electrophoresis and capillary gel electrophoresis is the use of narrow bore,
fused-silica columns filled with buffer and a sieving medium [53]. Figure 1.10
shows an illustration of a CGE separation. The capillaries are either wall coated or
chemically bonded to minimize electroosmotic flow so that gel will not be extruded
from the capillary [1]. Capillaries are filled with cross-linked or linear polymerized
gels, usually polyacrylamide [7]. Samples are introduced into the capillaries
preferably by electrokinetic injection, a method that usually results in sharp peaks
due to stacking. Analytes move through the capillary by electrophoresis and are
separated by the sieving mechanism of the gel.
Several mechanisms have been described for the size separation [23]. One of them
is Ogsten model, which treats a molecule as a non-deformable sphere with the
migration velocity determined by an analyte’s mobility modified by the probability
of an encounter with a restricting pore [55, 21]. The separation of analytes with a
radius of gyration less than or equal to the average pore size can be explained by
this model. CGE can be performed using either crosslinked or linear polymers.
26
Capillary Gel Electrophoresis (CGE)
+
vep
-
Figure 1.10 Illustration of CGE separation. Different sizes of analytes are separated
by electrophoresis under the condition of minimized EOF. vep: electrophoretic
velocity.
27
Crosslinked polyamide gel can be prepared in situ [56]. The size and distribution of
the pores for a crosslinked gel are determined by the distribution of crosslinks, the
distance between the crosslinks, and the liquid present in the gel. The pore structure
is responsible for separation of analytes. The pore size is controlled by the total
amount of acrylamide (T%) and the amount of crosslinker (C%). T% and C% are
calculated by equation (1.16) and (1.17), respectively [21].
T% =
C% =
acrylamide ( g ) + bisacrylam ide ( g )
100 mL
(1.16)
bisacrylam ide ( g )
× 100
bisacrylam ide ( g ) + acrylamide ( g )
(1.17)
Non-cross-linked, linear polymer networks are also used in CGE. A linear polymer
network is not attached to the inside wall of the capillary and very flexible. The
pore sizes of these network is defined by dynamic interactions between the polymer
chains and can be varied by changing capillary temperature, separation voltage, salt
concentration, or pH. Linear polymer gels are not heat sensitive and are easily
replaceable by pressure between the runs. Such gels permit both electrokinetic and
hydrodynamic injections. Linear polyacrylamide gels were introduced first into
coated capillary columns and applied to the separation of double-stranded DNA
fragments containing as many as several thousand base pairs [57].
1.4
Application of Capillary Electrophoresis in biopharmaceutical analysis
28
CE has demonstrated to be a complementary alternative to chromatographic
techniques in a broad range of applications such as environmental, clinical, forensic,
biomedical, pharmaceutical, and biopharmaceutical analysis. Various modes of CE
offer numerous possibilities for biopharmaceutical analysis including glycosylated
therapeutic proteins, monoclonal antibodies, and pharmaceutical and
biopharmaceutical impurities [58-59]. CZE has been used for analytes as diverse as
inorganic ions, organic pharmaceutical drugs, and proteins as long as there are
differences in charge-to-frictional drag ratios among the analyte species [60].
MEKC has been used for analysis of water-soluble neutral compounds, weak acids
and bases [61-62]. CGE has been used in DNA fragments, SDS protein and
macromolecules [63-67].
Biopharmaceutical drugs such as erythropoietin and various therapeutic monoclonal
antibodies need to be fully characterized for glycosylation compositions, IgG purity,
and impurities for quality control purposes. Therapeutic biomolecules have a
highly complex composition and structure. Also, the products may vary in structure
due to the complexity of cell culture and purification processes [68].
There is a huge demand for the development of novel, straightforward, efficient and
comprehensive analytical methodology, which is able to describe and secure
product quality for this diverse class of complex therapeutic biomolecules.
This study investigated three CE separation modes and their applications in
biopharmaceutical analysis. Several analytical methodologies were developed and
validated for their potential use in a quality control environment.
29
1.5
List of references
1.
Baker, D. R. Capillary electrophoresis. Wiley: New York, 1995.
2.
Hjerten, S. Free zone electrophoresis. Chromatogr. Rev. 1967, 9, 122-219.
3.
Mikkers, F.; Ringoir, S.;De Smet, R. Analytical isotachophoresis of uremic
blood samples. J. Chromatogr. 1979, 162, 341-50.
4.
Jorgenson, J. W.;Lukacs, K. D. Free-zone electrophoresis in glass
capillaries. Clin. Chem. 1981, 27, 1551-3.
5.
Cohen, A. S.;Karger, B. L. High-performance sodium dodecyl sulfate
polyacrylamide gel capillary electrophoresis of peptides and proteins. J.
Chromatogr. 1987, 397, 409-17.
6.
Hjerten, S. High-performance electrophoresis. Elimination of
electroendosmosis and solute adsorption. J. Chromatogr. 1985, 347, 191-8.
7.
Hjerten, S. High-performance electrophoresis: the electrophoretic
counterpart of high-performance liquid chromatography. J. Chromatogr.
1983, 270, 1-6.
8.
Hjerten, S.; Elenbring, K.; Kilar, F.; Liao, J. L.; Chen, A. J.; Siebert, C.
J.;Zhu, M. D. Carrier-free zone electrophoresis, displacement
electrophoresis and isoelectric focusing in a high-performance
electrophoresis apparatus. J. Chromatogr. 1987, 403, 47-61.
9.
Terabe, S. Electrokinetic chromatography: an interface between
electrophoresis and chromatography. TrAC, Trends Anal. Chem. (Pers. Ed.)
1989, 8, 129-34.
10.
Chien, R. L.;Burgi, D. S. On-column sample concentration using field
amplification in CZE. Anal. Chem. 1992, 64, 489A-96A.
30
11.
Quirino, J. P.;Terabe, S. Exceeding 5000-fold concentration of dilute
analytes in micellar electrokinetic chromatography. Science 1998, 282, 4658.
12.
Ganzler, K.; Greve, K. S.; Cohen, A. S.; Karger, B. L.; Guttman, A.;Cooke,
N. C. High-performance capillary electrophoresis of SDS-protein
complexes using UV-transparent polymer networks. Anal. Chem. 1992, 64,
2665-71.
13.
Lausch, R.; Scheper, T.; Reif, O. W.; Schloesser, J.; Fleischer, J.;Freitag, R.
Rapid capillary gel electrophoresis of proteins. J. Chromatogr., A 1993,
654, 190-5.
14.
Nishi, H.;Terabe, S. Application of micellar electrokinetic chromatography
to pharmaceutical analysis. Electrophoresis 1990, 11, 691-701.
15.
Nishi, H.; Tsumagari, N.; Kakimoto, T.;Terabe, S. Separation of watersoluble vitamins by micellar electrokinetic chromatography. J. Chromatogr.
1989, 465, 331-43.
16.
Quirino, J. P.; Iwai, Y.; Otsuka, K.;Terabe, S. Determination of
environmentally relevant aromatic amines in the ppt levels by cation
selective exhaustive injection-sweeping-micellar electrokinetic
chromatography. Electrophoresis 2000, 21, 2899-903.
17.
Sasaki, T. Analysis of proteins by capillary electrophoresis/sodium
dodecylsulfate-polyacrylamide gel electrophoresis (CE/SDS-PAGE). Jasco
Rep. 1992, 34, 10-14.
18.
Terabe, S.; Yashima, T.; Tanaka, N.;Araki, M. Separation of oxygen
isotopic benzoic acids by capillary zone electrophoresis based on isotope
effects on the dissociation of the carboxyl group. Anal. Chem. 1988, 60,
1673-7.
19.
Poole, C. F. The essence of chromatography. 1st ed.; Elsevier: Amsterdam ;
Boston, 2003.
20.
Altria, K. D. Capillary electrophoresis guidebook : principles, operation,
and applications. Humana Press: Totowa, N.J., 1996.
31
21.
Weinberger, R. Practical capillary electrophoresis. 2nd ed.; Academic
Press: San diego, CA, 2000.
22.
Farkas, T.; Zhong, G.;Guiochon, G. Validity of Darcy's law at low flowrates in liquid chromatography. J. Chromatogr., A 1999, 849, 35-43.
23.
Grossman, P. D. Factors affecting the performance of capillary
electrophoresis separations: Joule heating, electroosmosis, and zone
dispersion. Capillary Electrophor. 1992, 3-43.
24.
Whitesides, G. M.;Stroock, A. D. Flexible methods for microfluidics. Phys.
Today 2001, 54, 42-48.
25.
Harris, D. C. Quantitative chemical analysis. 7th ed.; W.H. Freeman and
Co.: New York, NY, 2007.
26.
Khaledi, M. G. High-performance capillary electrophoresis : theory,
techniques, and applications. Wiley: New York, 1998.
27.
Gas, B.;Kenndler, E. Dispersive phenomena in electromigration separation
methods. Electrophoresis 2000, 21, 3888-97.
28.
Gas, B.; Stedry, M.;Kenndler, E. Contribution of the electroosmotic flow to
peak broadening in capillary zone electrophoresis with uniform zeta
potential. J. Chromatogr., A 1995, 709, 63-8.
29.
Gas, B.; Stedry, M.;Kenndler, E. Peak broadening in capillary zone
electrophoresis. Electrophoresis 1997, 18, 2123-2133.
30.
Landers, J. P. Handbook of capillary and microchip electrophoresis and
associated microtechniques. 3rd ed.; CRC Press: Boca Raton, 2008.
31.
Mammen, M.; Carbeck, J.; Simanek, E. E.;Whitesides, G. M. Treating
Electrostatic Shielding at the Surface of Silica as Discrete Siloxide.Cation
Interactions. J. Am. Chem. Soc. 1997, 119, 3469-3476.
32
32.
Sazelova, P.; Kasicka, V.; Koval, D.; Prusik, Z.; Fanali, S.;Aturki, Z.
Control of EOF in CE by different ways of application of radial electric
field. Electrophoresis 2007, 28, 756-766.
33.
Salomon, K.; Burgi, D. S.;Helmer, J. C. Evaluation of fundamental
properties of a silica capillary used for capillary electrophoresis. J.
Chromatogr. 1991, 559, 69-80.
34.
Heiger, D. N. High performance capillary electrophoresis : an
introduction. 2nd ed.; Hewlett Packard Co.: France, 1992.
35.
Altria, K. D.;Simpson, C. F. High voltage capillary zone electrophoresis:
operating parameters effects on electroendosmotic flows and
electrophoretic mobilities. Chromatographia 1987, 24, 527-32.
36.
Brechtel, R.; Hohmann, W.; Ruediger, H.;Waetzig, H. Control of the
electroosmotic flow by metal-salt-containing buffers. J. Chromatogr., A
1995, 716, 97-105.
37.
Lukacs, K. D.;Jorgenson, J. W. Capillary zone electrophoresis: effect of
physical parameters on separation efficiency and quantitation. HRC CC, J.
High Resolut. Chromatogr. Chromatogr. Commun. 1985, 8, 407-11.
38.
Nelson, R. J.; Paulus, A.; Cohen, A. S.; Guttman, A.;Karger, B. L. Use of
Peltier thermoelectric devices to control column temperature in highperformance capillary electrophoresis. J. Chromatogr. 1989, 480, 111-27.
39.
Sepaniak, M. J.;Cole, R. O. Column efficiency in micellar electrokinetic
capillary chromatography. Anal. Chem. 1987, 59, 472-6.
40.
Landers, J. P.; Oda, R. P.; Madden, B.; Sismelich, T. P.;Spelsberg, T. C.
Reproducibility of sample separation using liquid or forced air convection
thermostatted high performance capillary electrophoresis systems. J. High
Resolut. Chromatogr. 1992, 15, 517-25.
41.
Grossman, P. D. Free-solution capillary electrophoresis. Capillary
Electrophor. 1992, 111-32.
33
42.
Grossman, P. D.;Colburn, J. C. Capillary electrophoresis : theory &
practice. Academic Press: San Diego, 1992.
43.
Jouyban, A.;Kenndler, E. Theoretical and empirical approaches to express
the mobility of small ions in capillary electrophoresis. Electrophoresis
2006, 27, 992-1005.
44.
Friedl, W.;Kenndler, E. Resolution as a function of the pH of the buffer
based on the analyte charge number for multivalent ions in capillary zone
electrophoresis without electroosmotic flow: Theoretical prediction and
experimental evaluation. Anal. Chem. 1993, 65, 2003-9.
45.
Weekley, B. S.;Foley, J. P. Dual-opposite-injection CZE: theoretical
aspects and application to organic and pharmaceutical compounds.
Electrophoresis 2007, 28, 697-711.
46.
Francois, Y.; Varenne, A.;Gareil, P. Capillary electrophoresis and ionic
liquids. From characterization to use as new electrolyte constituents.
Spectra Anal. 2008, 37, 23-29.
47.
Varenne, A.;Descroix, S. Recent strategies to improve resolution in
capillary electrophoresis. A review. Anal. Chim. Acta 2008, 628, 9-23.
48.
Pyell, U. Electrokinetic chromatography: theorey, instrumentation and
application. John Wiley & Sons, Ltd: 2006.
49.
Quirino, J. P.;Terabe, S. Electrokinetic chromatography. J. Chromatogr., A
1999, 856, 465-82.
50.
Terabe, S. Micellar electrokinetic chromatography. Chromatogr. Sci. Ser.
1993, 64, 65-87.
51.
Gareil, P. Migration time and resolution in micellar electrokinetic capillary
chromatography. Chromatographia 1990, 30, 195-200.
52.
Foley, J. P. Optimization of micellar electrokinetic chromatography. Anal.
Chem. 1990, 62, 1302-8.
34
53.
Camilleri, P. Capillary electrophoresis : theory and practice. 2nd ed.; CRC
Press: Boca Raton, Fla., 1998.
54.
Guttman, A. On the separation mechanism of capillary sodium dodecyl
sulfate-gel electrophoresis of proteins. Electrophoresis 1995, 16, 611-16.
55.
Mercier, J. F.; Tessier, F.;Slater, G. W. An exactly solvable Ogston model
of gel electrophoresis: VIII. Nonconducting gel fibers, curved field lines,
and the Nernst-Einstein relation. Electrophoresis 2001, 22, 2631-8.
56.
Baba, Y.; Tsuhako, M.; Sawa, T.; Akashi, M.;Yashima, E. Specific base
recognition of oligodeoxynucleotides by capillary affinity gel
electrophoresis using polyacrylamide-poly(9-vinyladenine) conjugated gel.
Anal. Chem. 1992, 64, 1920-5.
57.
Heiger, D. N.; Cohen, A. S.;Karger, B. L. Separation of DNA restriction
fragments by high performance capillary electrophoresis with low and zero
crosslinked polyacrylamide using continuous and pulsed electric fields. J.
Chromatogr. 1990, 516, 33-48.
58.
Jouyban, A.;Kenndler, E. Impurity analysis of pharmaceuticals using
capillary electromigration methods. Electrophoresis 2008, 29, 3531-51.
59.
Kostal, V.; Katzenmeyer, J.;Arriaga, E. A. Capillary electrophoresis in
bioanalysis. Anal. Chem. 2008, 80, 4533-50.
60.
Racaityte, K.; Kiessig, S.;Kalman, F. Application of capillary zone
electrophoresis and reversed-phase high-performance liquid
chromatography in the biopharmaceutical industry for the quantitative
analysis of the monosaccharides released from a highly glycosylated
therapeutic protein. J. Chromatogr., A 2005, 1079, 354-365.
61.
Terabe, S. In Micellar electrokinetic chromatography, Handbook of
Capillary and Microchip Electrophoresis and Associated Microtechniques
(3rd Edition), 2008; 2008.109-133.
62.
Terabe, S. Micellar electrokinetic chromatography for high-performance
analytical separation. Chem. Rec. 2008, 8, 291-301.
35
63.
Little, M. J.; Paquette, D. M.;Roos, P. K. Electrophoresis of pharmaceutical
proteins: status quo. Electrophoresis 2006, 27, 2477-2485.
64.
Que, A. H.; He, X. Z.; Toal, M. M.;Cook, K. S. Capillary electrophoresis in
the biopharmaceutical industry of monoclonal antibodies. Abstracts of
Papers, 231st ACS National Meeting, Atlanta, GA, United States, March
26-30, 2006 2006, ANYL-037.
65.
Rochu, D.;Masson, P. Capillary electrophoresis for monitoring stability of
pharmaceutical proteins. Ann. Pharm. Fr. 2003, 61, 185-195.
66.
Rodriguez-Diaz, R.; Tuck, S.; Ng, R.; Haycock, F.; Wehr, T.;Zhu, M.
Capillary electrophoresis of biopharmaceutical proteins. Anal. Tech.
Biopharm. Dev. 2005, 161-226.
67.
Salas-Solano, O. Capillary electrophoresis in biotechnology: A practical
tool to analyze and characterize therapeutic proteins. Abstract of Papers,
238th ACS National Meeting, Washington, DC, United States, August 1620, 2009 2009, BIOT-319.
68.
Hunt, G.;Nashabeh, W. Capillary electrophoresis sodium dodecyl sulfate
nongel sieving analysis of a therapeutic recombinant monoclonal antibody:
A biotechnology perspective. Anal. Chem. 1999, 71, 2390-2397.
36
Chapter 2: Characterization of glycoprotein erythropoietin by its isoform
distribution using capillary zone electrophoresis
2.1
Introduction
Erythropoietin (EPO) is a glycoprotein hormone that regulates erythropoiesis, or red
blood cell production [1]. EPO is also called hematopoietin or hemopoietin, which
is produced by the liver and kidney [2]. After being released into the blood stream it
binds to receptors in the bone marrow, where it stimulates the production of red
blood cells (erythrocytes). Since EPO accelerates erythrocyte production, it also
increases oxygen carrying capacity. This fact caused attention of its illegal use as a
performance-enhancing agent in the athletic community [3]. Purification and
sequencing of EPO led to commercial production of recombinant human
erythropoietin, which is used to treat certain forms of anemia (e.g., due to chronic
kidney failure) [4].
Glycosylation is a posttranslational modification of proteins that plays an important
role in biological activities [5]. The composition and structure of the attached
carbohydrates is essential to the molecular recognition processes and biological
activities of glycoproteins. Two types of glycosylations, O-glycosylation and Nglycosylation, have been discussed by Mechref and Novotny [5]. The structures of
N-linked glycans are of special analytical interest due to their size, heterogeneity,
and variability. N-Linked glycans contain a common core with one or more
antennas attached to the external mannose residues [4]. The primary structure of Nglycans is defined by the number of antennas, sialic acids, the presence of fucoses,
and the number of LacNAc units [4]. Therefore, the structural variability of the
37
glycans residues, the number of glycosylation sites, and the different degree of site
occupancy result in a mixture of glycoforms [6].
The EPO molecule has a molecular mass of approximately 30,600 Daltons,
consisting of 165 amino acids [7]. Several strategies have been developed for the
study of glycoform distribution in EPO [6]. These strategies are based on the
analysis of glycopeptides obtained by tryptic digestion, the analysis of released
carbohydrates, and the direct separation of intact glycoproteins [6].
Capillary electrophoresis (CE) has been one of the widely used tools for the direct
analysis of glycoform distribution of glycoprotein [8], and several CE methods have
been developed for the separation of EPO glycoforms [9-13, 8, 14-19]. Watson and
Yao developed the first capillary zone electrophoresis (CZE) method using a fusedsilica capillary with weakly acidic electrolyte solution containing 1,4diaminobutane (DAB) [18]. After the re-examination of this method by Kinoshita et
al., the method was found to be troublesome due to significant loss on resolution
after several injections, and the life of the capillary was limited [8]. Lopez-SotoYarritu et al. proposed a higher DAB concentration, and the migration time was
controlled, but the zone broadening was significantly increased, and the peak areas
varied due to protein adsorption onto the capillary wall [14]. The European
Pharmacopoeia (Ph. Eur.) has published a monograph describing the separation of
glycoforms of EPO by CZE [7], but its method was prone to poor reproducibility
[17]. Sanz-Nebot et al. optimized the Ph. Eur. method by adjusting the pH value of
separation electrolyte and capillary length in order to achieve better resolution for
38
the separation of erythropoiesis-stimulating protein. Although the intraday
reproducibility was improved, the inter-day reproducibility was relatively poor, and
not all the isoforms were completely resolved [16]. Benavente et al. made further
improvements on the CZE method and proposed a multivariate calibration method
using partial least-squares (PLS) in order to characterize binary mixtures of two
types of recombinant human erythropoietin (epoetin alpha and beta) [9]. Although
the reproducibility was improved, the precision of migration times for all the peaks
still failed to meet the system suitability acceptance criterion of < 2% required by
Ph. Eur. Other CE techniques such as isoelectric focusing (CIEF) method have also
been developed for EPO analysis [20], but CIEF is not required by the Ph.Eur. [7].
Therefore, CIEF was not studied and compared in this work. After the European
Department for the Quality of Medicine first published specifications for the CZE
method in the 2002 European Pharmacopoeia, the EPO manufacturers have been
required to adopt it as a quality control (QC) method for product release.
The objective of this study was to modify the current Ph. Eur. method to make it
suitable for routine QC testing of EPO drug substance and product. The root cause
for poor reproducibility of the CZE method was studied and identified. The study
also qualified a secondary standard as a substitute for primary standard, the latter of
which is only available in limited supply from Ph. Eur. In addition, the study
evaluated a strategy for removal of polysorbate 80 from the EPO formulated with
polysorbate 80. Formulations containing polysorbate 80 interfere with the analysis
of isoforms of EPO in CE separations, and the desalting procedure using a
molecular mass cut-off filter [7] is unable to remove polysorbate 80.
39
2.2
2.2.1
Materials and methods
Chemicals and reagents
All chemicals used in the preparation of buffers and solutions were of ACS reagent
or electrophoresis grade. Acetic acid (HAc) (glacial), anhydrous sodium acetate
(NaAc), 1,4- diaminobutane (DAB), sodium chloride (NaCl), sodium citrate
dihydrate, citric acid, Tris-(hydroxymethyl)aminomethane hydrochloride (TrisHCl), Tris-(hydroxymethyl)aminomethane (Tris-base), and N-Tris(hydroxymethyl)methylglycine (tricine) were supplied by Sigma-Aldrich (St. Louis,
MO, USA). Sodium hydroxide solutions (1 M and 0.1 M) and hydrochloric acid
(0.1 M) were provided by Agilent Technologies (Wilmington, DE, USA). Urea was
supplied by BioRad Laboratories (Hercules, CA, USA). Water with a conductivity
value lower than 5 µS/cm was obtained using a Milli-Q water purification system
from Millipore (Billerica, MA, USA). The running buffer was prepared by
dissolving appropriate amounts of tricine, NaCl, NaAc, DAB, and urea in water to
obtain the desired concentration of 10 mM tricine, 10 mM NaCl, 10 mM NaAc, 2.5
mM DAB, and 7 M urea. The pH was adjusted to 5.55 ± 0.02 with 10% acetic acid.
The 20 mM citrate/250 mM NaCl buffer solution was prepared by dissolving
appropriate amounts of sodium citrate and NaCl in water. The 10 mM Tris-HCl was
prepared by dissolving appropriate amounts of Tris-HCl in water and pH adjusted to
7.00 ± 0.05 with 10% Tris-base. The above solutions were filtered through a 0.22
μM nylon filter.
2.2.2
EPO sample preparation
40
Primary standard samples of recombinant human EPO produced in a Chinese
Hamster Ovary (CHO) cell line were provided by the Ph. Eur. as Biological
Reference Preparation (BRP Batch 1). Epoetin alfa samples were obtained from
Johnson & Johnson Ortho Biotech (Raritan, NJ, USA). One vial of the BRP sample,
which contained 250 μg of EPO, was dissolved in 250 μL of water to give a 1
Mg/mL protein solution. The epoetin alfa test sample was also diluted in water to
obtain a 1 mg/mL protein solution. The sample solution was desalted by passing
through a 10,000 Dalton molecular mass cut-off filter (Microcon centrifugal filter)
from Millipore (Billerica, MA, USA). The filter was pre-rinsed with 250 μL of
water for three times in a micro-centrifuge at 13,000 x g for 20 minutes at 4 °C. The
sample desalting was accomplished by washing the filter three times with 250 μL of
water in the same way as pre-rinsing. The retentate was recovered from the upper
reservoir by upside-down centrifugation to a new vial at 1,000 x g for 3 minutes at
4°C. The recovered retentate was reconstituted with appropriate amount of water to
obtain the protein concentration of 1 mg/mL. The protein concentration was
measured by UV absorbance at 280 nm. The epoetin alfa samples and the BRP
sample were stored at –20 °C when not in use.
2.2.3
Formulated epoetin alfa sample preparation
Epoetin alfa was formulated with polysorbate 80 to contain 2000 (2K) and 40,000
(40K) International Units (IU), which are equivalent to 16.8 μg/mL and 336.0
μg/mL epoetin alfa, respectively. The removal of polysorbate 80 from formulated
epoetin alfa was achieved using a non-ionic detergent (NID) trap cartridge from
Michrom BioResources (Auburn, CA, USA). The NID trap cartridge was connected
41
to different sizes of syringes for cartridge conditioning, sample loading, cartridge
washing, and sample eluting. The syringe with cartridge was connected to an
electronic infusion pump from KD Scientific (Holliston, MA, USA) that was
programmed for different syringe sizes and flow rates, which enabled the process to
be automated, yielding high throughput, accuracy, and precision. Figure 2.1 shows
the experiment setup. The flow rate was set at 0.5 mL/min for all syringe sizes. The
NID trap cartridge was first conditioned with 900 μL of 20 mM citrate/250 mM
NaCl solution to remove any residual substance that may co-elute with epoetin alfa,
followed by 900 μL of 10 mM Tris-HCl buffer solution. After the given volume
was delivered, the pump was programmed to stop and the syringe was switched for
next step. The formulated epoetin alfa from a pre-filled syringe or vial was diluted
10-fold with 10 mM Tris-HCl solution. For 2K IU formulated epoetin alfa, 5 mL of
sample was diluted to 50 mL with 10 mM Tris-HCl, and 48 mL of the diluted
sample was loaded to the NID trap cartridge. For 40K IU formulated epoetin alfa,
0.5 mL of sample was diluted to 5 mL with 10 mM Tris-HCl, and 4.5 mL of the
diluted sample was loaded to the NID trap cartridge. After the given volume of the
sample passed through the cartridge, the cartridge was washed with 1 mL of 10 mM
Tris-HCl to remove polysorbate 80. The EPO was eluted with 200 μL of 20 mM
citrate/250 mM NaCl. The sample was then taken through the desalting procedure
as described in Section 2.2. In order to obtain the protein concentration of 1 mg/mL,
the retentate of 2 K IU and 40 K IU samples were reconstituted with 80 μL and 150
μL of water, respectively. The formulated epoetin alfa samples were stored at 4 °C
when not in use.
2.2.4
Capillary zone electrophoresis (CZE)
42
Figure 2.1 Experiment setup of an electronic infusion pump using syringes
connected to non-ionic detergent (NID) trap cartridges to remove polysorbate 80.
The electronic infusion pump is programmed for different syringe sizes and flow
rates, which enables the process to be automated.
43
The CZE experiments were performed on an Agilent 3D-CE instrument equipped
with a photo diode array detector and ChemStation software from Agilent
Technologies (Wilmington, DE, USA). Bare fused-silica capillary columns (104 cm
x 50 μm id) were supplied by Agilent Technologies (Wilmington, DE, USA). The
detection window was placed at 8.5 cm from the outlet of the capillary. The
separation conditions for running buffer, capillary temperature, voltage, and UV
detection were as described in the Ph. Eur. method [7]. The capillary temperature
was maintained at 25 °C for all the experiments. A voltage of 15.4 kV was applied
during electrophoretic separations. Samples were injected hydrodynamically at 50
mbar for 30 seconds followed by a water injection at 5 mbar for 3 seconds.
Detection was performed at 214 nm. Data were collected at a sampling rate of 5 Hz.
The current was also monitored, and was typically 4-5 μA, corresponding to a Joule
heating of 0.07-0.08 W/m, which was only 5% of the upper limit in an Ohm’s law
plot.
2.2.5
Modified capillary conditioning
The capillary was conditioned between every 10 injections (one water and nine
samples) by flushing the capillary with the following sequence of solutions for the
specified periods: water for 30 minutes, 1 M sodium hydroxide for 60 minutes,
water for 60 minutes, 0.1 M hydrochloric acid for 30 minutes, water for 30 minutes,
0.1 M sodium hydroxide for 45 minutes, water for 30 minutes, CZE running buffer
for 15 minutes. If the capillary was new or had been stored in air, a voltage of 20
kV was applied to the capillary filled with running buffer for 120 minutes. Between
44
injections, the capillary was rinsed with water for 10 minutes and with running
buffer for 10 minutes.
2.3
2.3.1.
Results and discussion
Restoration of the fused silica capillary surface
Fused silica possesses different surface silanol groups—isolated, vicinal, and
geminal, which create the occurrence of charge on the capillary surface, depending
on electrolyte pH [21]. At pH above 3, the silanol groups will begin to ionize, which
results in a negative charge on the capillary wall. In solutions containing ions, the
cations will migrate to the negatively charged wall forming the electric double
layer, which is described simply by the Stern–Gouy–Chapman model [22]. When an
electrical potential is applied to the capillary, the cations will migrate towards the
cathode. This is the origin of the electroosmotic flow (EOF). EOF is affected by pH,
the chemical nature of the capillary wall, the composition of the background
electrolyte (BGE), the ionic strength, and the temperature.
The separation of proteins is complicated due to their adsorption onto the negatively
charged surface of fused silica capillaries [23]. The most common approach to
minimize the adsorption is to coat the capillary wall [24]. Coatings can be broadly
categorized as (i) covalently linked polymeric coatings, (ii) physically adsorbed
polymer coatings, or (ii) small molecule additives [25]. Coating agents (additives)
that are positively charged and have a few bonding centers can effectively interact
with the capillary wall, changing the capillary surface charge and, in consequence,
changing the magnitude of the EOF or even its direction.
45
The Ph. Eur. method for the analysis of EPO isoforms uses 2.5 mM DAB as a small
molecule amine additive to dynamically coat the capillary wall. The binding of
amines to the silanol sites on the capillary wall reduces EOF, a necessary condition
for separation of all the isoforms of the glycoprotein. Coating of the capillary
surface with the amine cations in the BGE was via a dynamic equilibrium [26]. At
pH 5.5 used in the Ph. Eur. running buffer, all the isoforms are negatively charged,
as are some of the silanol groups on the capillary surface. The available silanol
groups for DAB to bind depend on the pH and DAB concentration. The interactions
between DAB and the silanol groups are either through hydrogen bonding or ionic
interaction. The pH of the running buffer, the ionic strength of the buffer, the DAB
concentration, as described in the Ph. Eur. method are experimentally proved to be
adequate for the separation of all the isoforms [17].
The remaining question is how to regenerate the capillary conditions after a certain
number of injections so that the run can be continued without changing the
capillary. During the separation, a portion of the isoform anion (A-) is tied up as an
association complex through hydrogen bonding, ionic interaction, or hydrophobic
interactions, which undergoes little if any electrophoretic migration, the result
being a slower net migration velocity and a longer retention time [26]. The
hydrogen bonding and ionic interactions may be described as simple equilibriums:
-Si-O-H…NH2-(CH2)4-NH3+ + A− = -Si-O-H…NH2-(CH2)4-NH3+A−
(hydrogen bonding)
(2.1)
46
–Si-O− +H3N-(CH2)4-NH3+ + A− = –Si-O− +H3N-(CH2)4-NH3+A−
(ionic interaction)
(2.2)
K = [-Si-O-H…NH2-(CH2)4-NH3+A−]/[ -Si-O-H…NH2-(CH2)4-NH3+][A−]
(2.3)
K = [–Si-O− +H3N-(CH2)4-NH3+A−]/[ –Si-O− +H3N-(CH2)4-NH3+][A−]
(2.4)
The K value and hence the degree with which the net migration of the sample anion
is slowed, is seen to be a function of both K and the concentration of pairing cation
[26]. The value of K will be different for each isoform anion, thus providing an
additional parameter for the resolution of anion mixtures by CZE.
Taking this into consideration, the complete removal of the complex from the
silanol groups and the re-establishment of pH are crucial to restore the surface of
the fused silica capillary. The proposed capillary conditioning procedure using the
combination of base, acid, water, and buffer flush with extended period of time was
able to effectively remove all the isoform complexes from the capillary wall and to
provide approximately the same number of available silanol groups for DAB to
bind again. As a result, reproducible results for both migration times and peak areas
performed on the same capillary were obtained. Figure 2.2 shows a comparison of
the 1st and 36th injection on the same capillary after using the proposed conditioning
procedure. The number of injections beyond 36 was not examined only due to the
limited number of samples available within one run. More injections on one
capillary would be possible if needed.
47
56
mAU
15
A
12.5
3 4
10
7
7.5
5
2.5
3
8
0
-2.5
-5
0
20
40
60
80
min
56
mAU
15
12.5
4
B
10
7
7.5
5
2.5
3
0
8
-2.5
-5
0
20
40
60
80
min
Figure 2.2 Demonstration of capillary restoration. The peak area and migration time
appear consistent after 36th injection. A: 1st EPO sample injection. B: 36th EPO
sample injection. Peaks 3-8 are six isoforms of EPO.
48
2.3.2
System suitability
The system suitability criteria stated in the Ph. Eur. Method are as follows:
•
In the electropherogram obtained with the BRP reference solution, a pattern
of well-separated peaks corresponding to the peaks in the Ph. Eur. reference
electropherogram of EPO is seen.
•
The largest peak is at least 50 times greater than the baseline noise. If
necessary, adjust the sample load to give peaks of sufficient height.
•
Identify the peaks corresponding to isoforms 1 to 8.
•
The peak corresponding to isoform 1 is detected.
•
The resolution between isoforms 5 and 6 is not less than 1.
•
Repeat the separation at least 3 times. The baseline is stable, showing little
drift, and the distribution of peaks is qualitatively and quantitatively similar
to the distribution of peaks in the BRP reference electropherogram of EPO.
•
The RSD (%) of the migration time of the peak corresponding to isoform 2
is less than 2%.
Two sets of analyses designated as Run-1 and Run-2 were performed to test system
suitability. In each of the electropherograms, there was a negative peak that
appeared before the isoform peaks. This negative peak was due to the refractive
index difference between the sample zone (water) and running buffer zone. After
injecting a neutral marker, benzyl alcohol, a large peak appeared at the same
migration time as the negative peak, which confirmed that the negative peak was
caused by water. As such, the negative peak migrates only due to electroosmotic
flow and is therefore a suitable reference point for calculating relative migration
time (RMT). RMT is calculated as follows:
49
RMT = Sample Migration Time / Water Migration Time (negative peak)
(2.5)
Each run included triplicate injections of one BRP primary standard and triplicate
injections of two preparations of epoetin alfa from a batch that was designated as a
secondary standard. The RSD (%) values of absolute migration time (MT) and
RMT are shown in Table 2.1. Although the RSD (%) of migration time
corresponding to isoform 2 passed the Ph. Eur. acceptance criterion of less than 2%,
the RMT had a much lower RSD; using RMT is therefore a better approach to
identify the isoforms.
The RSD (%) values for relative peak area (%) are shown in Table 2.2. The
electropherograms of BRP primary standard (Figure 2.3 A) appeared identical to the
reference electropherogram provided in the Ph. Eur. monograph [7]. All the
electropherograms showed well-separated peaks, low baseline noise, good
resolution and detection, stable baseline, and relatively constant migration time. As
summarized in Table 2.3, all the system suitability data for the primary standard
passed the acceptance criteria stated in the Ph. Eur. monograph.
2.3.3
Characterization and qualification of epoetin alfa secondary standard
Run-1 and Run-2 also served to characterize and qualify the epoetin alfa secondary
standard. It was necessary to qualify the secondary standard because of insufficient
supply of the BRP primary standard. The samples from the same lot of epoetin alfa
secondary standard was prepared in duplicate and injected in triplicate in the two
runs. The RMT, pattern of peaks, and relative peak area were used to identify the
50
Table 2.1 Precision of migration times (MTs) and relative migration times (RMTs) for two sets of analyses (Run-1 and Run-2) based on
triplicate injections of each preparation.
Iso-11
Run-1
Run-2
1
Iso-2
Iso-3
Iso-4
Iso-5
Iso-7
Iso-8
MT2
RMT3
MT
RMT
MT
RMT
MT
RMT
MT
RMT
MT
RMT
MT
RMT
MT
RMT
BRP
1.7
0.1
1.8
0.2
1.8
0.2
1.9
0.3
2.0
0.4
2.0
0.4
2.0
0.4
2.1
0.5
EPO-14
NA5
NA
NA
NA
3.0
0.3
3.0
0.3
3.1
0.4
3.1
0.4
3.1
0.4
3.2
0.5
EPO-26
NA
NA
NA
NA
0.3
0.1
0.3
0.1
0.3
0.1
0.3
0.2
0.3
0.2
0.3
0.2
BRP
0.1
0.1
0.1
0.1
0.1
0.0
0.1
0.0
0.2
0.0
0.2
0.0
0.2
0.0
0.2
0.1
EPO-1
NA
NA
NA
NA
0.3
0.1
0.3
0.1
0.3
0.1
0.3
0.1
0.3
0.1
0.3
0.0
EPO-2
NA
NA
NA
NA
0.1
0.0
0.1
0.0
0.1
0.0
0.1
0.0
0.1
0.0
0.1
0.0
ISO: Isoform
MT: Migration Time
3
RMT: Relative Migration Time, RMT=Sample MT/ corresponding negative peak MT
4
EPO-1: First preparation of epoetin alfa secondary standard
5
NA: Not Applicable
6
EPO-2: Second preparation of epoetin alfa secondary standard
2
Iso-6
51
Table 2.2 Precision of relative peak area (%) for two sets of analyses (Run-1 and Run-2) based
on triplicate injections of each preparation.
Iso-1
Iso-2
Iso-3
Iso-4
Iso-5
Iso-6
Iso-7
Iso-8
19.5
4.7
1.3
0.6
0.8
0.8
0.1
18.4
2
ND
1.1
0.3
0.5
0.1
0.3
10.9
EPO-2
ND
ND
2.5
0.1
0.1
0.1
0.4
1.4
BRP
7.2
2.4
0.7
0.2
0.2
0.2
0.3
2.4
EPO-1
ND
ND
1.9
0.2
0.2
0.2
0.1
8.1
EPO-2
ND
ND
3.8
0.5
0.3
0.2
0.5
11.8
BRP1
Run-1
Run-2
1
2
EPO-1
ND
BRP: Biological Reference Preparation (Batch 1)
ND: Not Detected
52
mAU
15
12.5
56
A
10
4
7.5
7
5
2.5
1
2
3
8
0
-2.5
-5
0
*
20
40
60
80
mAU
15
min
56
12.5
B
10
4
7
7.5
5
3
2.5
8
0
-2.5
-5
0
20
40
60
80
min
Figure 2.3 Electropherograms of BRP primary standard vs. epoetin alfa secondary standard.
A: BRP primary standard; B: epoetin alfa secondary standard. Isoforms 1-8 were very well
separated. Isoforms 1 and 2 were not present in the epoetin alfa sample.
53
Table 2.3 System suitability data of the BRP primary standard
Neg.
Replicate
Peak
MT
(min)
Run-1
Run-2
Iso-2
MT
(min)
Iso-2
RMT
S/N of
Largest
R5,6
Peak
Pattern vs. Ph. Eur.
Iso 1-8
Electropherogram
Identified
1
67.6
72.2
1.069
97
1.5
Pass
Pass
2
66.3
70.7
1.066
108
1.4
Pass
Pass
3
68.4
73.2
1.070
106
1.5
Pass
Pass
Mean
72.0
1.068
RSD (%)
1.8
0.16
1
62.7
67.0
1.069
130
1.3
Pass
Pass
2
62.8
67.1
1.067
143
1.4
Pass
Pass
3
63.0
67.2
1.067
156
1.4
Pass
Pass
Mean
67.1
1.068
RSD (%)
0.1
0.1
54
isoforms. Table 2.1 shows the RSD (%) of MT and RMT. Table 2.2 shows the RSD (%) of
relative peak area. The epoetin alfa secondary standard contains a subset of the isoforms that
are present in the BRP primary standard because the BRP primary standard possesses both alfa
and beta forms of EPO [27]. The RMT values confirmed that the isoforms in epoetin alfa
secondary standard correspond to isoforms 3 through 8 in BRP primary standard. Figure 2.3
shows a comparison of the BRP primary standard with epoetin alfa secondary standard. Figure
2.2 and Figure 2.3 were generated on different instruments using different capillaries and
running buffers. The difference in absolute migration time between these two runs was due to
the running buffer pH, which was adjusted according to a different pH meter. An investigation
was conducted and the conclusion was confirmed. Nevertheless, the sample MT lined up well
with the standard MT.
Based on the above results, the consistency of the migration time for isoform 2 as a system
suitability criterion is no longer applicable to an epoetin alfa secondary standard. Since isoform
3 in epoetin alfa secondary standard has a similar relative peak area to isoform 2 in the BRP
primary standard, the RSD (%) of RMT for isoform 3 can be used as system suitability
criterion in future analyses. Based on the epoetin alfa secondary standard, a new set of system
suitability criteria has been established:
•
In the electropherogram obtained with the epoetin alfa secondary standard solution, a
pattern of well-separated peaks corresponding to the peaks in the epoetin alfa secondary
standard reference electropherogram is seen. The electropherogram in Figure 2.3 B is
established as the epoetin alfa secondary standard reference electropherogram.
•
The largest peak is at least 50 times greater than the baseline noise. If necessary, adjust the
sample load to give peaks of sufficient height.
55
•
Identify the peaks corresponding to isoforms 3 to 8.
•
The peak corresponding to isoform 3 is detected.
•
The resolution between isoforms 5 and 6 is not less than 1.
•
Repeat the separation at least 3 times. The baseline is stable, showing little drift, and the
distribution of peaks is qualitatively and quantitatively similar to the distribution of peaks
in the epoetin alfa secondary standard reference electropherogram.
•
The RSD (%) of the RMT of the peak corresponding to isoform 3 is less than 2%. RMT is
calculated against the water negative peak.
2.3.4
Precision
In addition to Run-1 and Run-2, two more sets of analyses, Run-3 and Run-4, were carried out
to provide data for evaluation of repeatability and intermediate precision. Runs 3 and 4
included three independent preparations of epoetin alfa secondary standard. The first
preparation was used to test system suitability, and all three preparations were used for
precision testing. All runs met system suitability acceptance criteria. Table 2.4 shows the RSD
(%) for relative peak area of the isoforms analyzed in Run-3 and Run-4. Taking all epoetin alfa
secondary standard results from runs 1-4 (Tables 2.2 and 2.4) and using intra-run data as a
measure of repeatability, RSD (%) of percent area ranged from 0.0% to 0.9% for isoforms 4-7,
and 0.5% to 14.0% for isoforms 3 and 8. A higher RSD (%) for isoforms 3 and 8 is expected
due to their low relative peak areas (e.g., 1-2%). Overall, these RSD (%) results for
repeatability are considered acceptable.
56
Table 2.4 Precision of relative peak area for Run-3 and Run-4 based on triplicate injections of
each preparation
Run-3
Run-4
Iso-3
Iso-4
Iso-5
Iso-6
Iso-7
Iso-8
EPO Std
3.6
0.6
0.1
0.2
0.2
9.8
EPO-1
1.1
0.6
0.4
0.4
0.9
7.3
EPO-2
5.1
0.2
0.7
0.5
0.8
14.0
EPO Std
4.4
0.1
0.2
0.2
0.8
1.8
EPO-1
5.7
1.8
0.9
0.5
2.5
4.7
EPO-2
0.0
0.3
0.0
0.0
0.3
0.0
57
Intermediate precision was assessed with inter-run data by comparing percent area of all
different preparations of epoetin alfa secondary standard from Run-1 through Run-4 (Table
2.5). The RSD (%) values of isoforms 4-7 were 0.2-0.7%. The RSD (%) values of isoforms 3
and 8 were 3.3-6.8%. Based on these results, intermediate precision is considered acceptable.
2.3.5
Use of duplicate injections for test sample
To evaluate the reliability of using duplicate injections in the analysis of the test sample rather
than triplicate injections, the third injection result for each preparation of epoetin alfa
secondary standard was deleted and the remainder was used for precision evaluation. The
results after re-calculation of average percent area and RSD (%) are presented in Table 2.6.
Using intra-run data as a measure of repeatability, RSD (%) ranged from 0.0% to 0.8% for
isoforms 4-7, and 0.4% to 8.3% for isoforms 3 and 8. Using inter-run data as a measure of
intermediate precision, RSD (%) ranged from 0.2% to 0.8% for isoforms 4-7, and 4.0% to
6.9% for isoforms 3 and 8. These repeatability and intermediate precision results are similar to
those with triplicate injections. Use of duplicate injections for test sample, which improves the
efficiency of the test procedure, is considered acceptable. Primary and secondary standard will
continue to be run with triplicate injections, as specified in the Ph. Eur. monograph.
2.3.6
Use of 2-hour conditioning period on new capillaries
Run 1 and Run 2 were run on new capillaries with a 2-hour conditioning period. The Ph. Eur.
monograph method requires applying voltage for 12 hours for a new capillary. To compare
these two timings, Run 5 was performed on a new capillary with 12 hour conditioning. One
sample of epoetin alfa secondary standard was prepared and injected 6 times. The RSD (%) of
RMT for isoform 3 through 8 was calculated. As shown in Table 2.7, all RSD (%)
58
Table 2.5 Average relative areas (%) of the isoforms in all secondary standard preparations.
Iso-3
Iso-4
Iso-5
Iso-6
Iso-7
Iso-8
2.1
18.8
30.4
30.2
17.2
1.3
2.1
18.9
30.5
30.2
17.1
1.2
Intra-run
Mean (N=2)
2.1
18.9
30.5
30.2
17.2
1.3
%RSD
0.0
0.4
0.2
0.0
0.4
5.7
2.2
19.0
30.6
30.2
17.0
1.1
2.2
19.0
30.5
30.1
17.0
1.2
Intra-run
Mean (N=2)
2.2
19.0
30.6
30.2
17.0
1.2
%RSD
0.0
0.0
0.2
0.2
0.0
6.1
2.2
18.6
30.3
30.3
17.3
1.3
2.3
18.7
30.3
30.1
17.3
1.4
2.2
18.8
30.3
30.2
17.2
1.3
Intra-run
Mean (N=3)
2.2
18.7
30.3
30.2
17.3
1.3
%RSD
2.6
0.5
0.0
0.3
0.3
4.3
2.2
18.9
30.4
30.1
17.1
1.2
2.0
18.8
30.4
30.2
17.3
1.2
2.1
19.0
30.5
30.1
17.1
1.2
Intra-run
Mean (N=3)
2.1
18.9
30.4
30.2
17.2
1.2
%RSD
3.3
0.7
0.3
0.2
0.6
6.8
Inter-run
Mean (N=10)
2.1
18.9
30.4
30.2
17.2
1.2
%RSD
3.3
0.7
0.3
0.2
0.6
6.8
Run-1
Run-2
Run-3
Run-4
Note: Run-1 and Run-2 include two preparations and Run-3 and Run-4 include three
preparations. Each preparation was injected in triplicate. The average area% was calculated
from three injections for each preparation. The intra-run mean (N=2) was calculated from the
two preparations for Run-1 and Run-2. The intra-run mean (N=3) was calculated from the
three preparations for Run-3 and Run-4.
59
Table 2.6 Average relative areas (%) of the isoforms in all secondary standard preparations for
the evaluation of duplicate injection.
Iso-3
Iso-4
Iso-5
Iso-6
Iso-7
Iso-8
2.1
18.8
30.4
30.2
17.1
1.3
2.1
18.9
30.5
30.2
17.2
1.2
Intra-run
Mean (N=2)
2.1
18.9
30.4
30.2
17.2
1.2
%RSD
0.4
0.1
0.2
0.0
0.1
7.0
2.1
19.0
30.6
30.2
17.0
1.1
2.1
18.9
30.5
30.1
17.1
1.2
Intra-run
Mean (N=2)
2.1
19.0
30.6
30.2
17.0
1.2
%RSD
0.7
0.2
0.1
0.2
0.2
8.3
2.3
18.5
30.3
30.3
17.3
1.3
2.3
18.6
30.3
30.1
17.3
1.4
2.2
18.8
30.2
30.1
17.3
1.3
Intra-run
Mean (N=3)
2.2
18.7
30.3
30.2
17.3
1.3
%RSD
0.6
0.8
0.1
0.4
0.2
2.5
2.1
18.9
30.4
30.1
17.2
1.2
2.0
18.7
30.4
30.3
17.4
1.3
2.1
19.0
30.5
30.1
17.2
1.2
Intra-run
Mean (N=3)
2.1
18.9
30.4
30.2
17.3
1.2
%RSD
3.0
0.7
0.2
0.3
0.8
3.2
Inter-run
Mean (N=10)
2.1
18.8
30.4
30.2
17.2
1.2
%RSD
4.0
0.8
0.4
0.2
0.7
6.9
Run-1
Run-2
Run-3
Run-4
Note: Run-1 and Run-2 include two preparations and Run-3 and Run-4 include three
preparations. Each preparation was injected in triplicate. The average area% was calculated
from the first two injections for each preparation. The intra-run mean (N=2) was calculated
from the two preparations for Run-1 and Run-2. The intra-run mean (N=3) was calculated
from the three preparations for Run-3 and Run-4.
60
values were low and RMT was very consistent over all the runs. The data indicate that
applying voltage for 12 hours has no advantage over applying voltage for 2 hours. For the
purpose of efficiency, a 2-hour conditioning period will be used for future analyses.
2.3.7
Removal of polysorbate 80 from formulated epoetin alfa
Polysorbate 80 (PS 80), used in the formulation of drug product from epoetin alfa, interferes
with the analysis of isoforms of EPO in CZE separation. PS 80 is an amphiphilic molecule
containing hydrophilic and hydrophobic moieties; its removal has been a significant challenge
for the formulation scientist due to its amphiphilic character [28]. Although we were
unsuccessful in removing PS 80 using molecular mass cut-off filter as described in the Ph. Eur.
Monograph [7], we were successful when we employed a nonionic detergent (NID) trap. The
NID trap cartridge contains a mixed bed of weak anion and weak cation exchange packing.
The NID trap is designed to bind proteins, while allowing non-ionic detergents such as PS 80
to be eluted out. The 2000 (2K) IU/mL and the 40,000 (40K) IU/mL formulated EPO samples
(F-EPO) were prepared by spiking the formulation buffer, which contains PS 80, with an
epoetin alfa secondary standard. As such, the F-EPO sample has a known amount of epoetin
alfa. PS 80 was removed and desalted according to the proposed procedure as described in
Section 2.3. The epoetin alfa secondary standard, the 2K IU/mL F-EPO and the 40K IU/mL FEPO samples were analyzed by CZE in one sequence within one day. No extra or missing
peaks were found in 2K IU/mL F-EPO and 40K IU/mL F-EPO electropherograms when
compared to that of the epoetin alfa secondary standard. The relative percent area of each
isoform was recorded in Table 2.8. The percent areas of each isoform for 2K IU/mL F-EPO
and 40K IU/mL F-EPO differed from the epoetin alfa secondary standard by 0.7% or less, and
were considered to be equivalent to that of the epoetin alfa standard. Recovery of the protein
61
Table 2.7 Precision of relative migration time with 2-hour conditioning vs. 12-hour
conditioning.
New capillary
conditioning time
(hrs)
Iso-3
Iso-4
Iso-5
Iso-6
Iso-7
Iso-8
2 (Run-1)
0.4
0.5
0.5
0.6
0.6
0.7
2 (Run-2)
0.2
0.2
0.1
0.1
0.1
0.1
12 (Run-5)
0.4
0.5
0.5
0.6
0.6
0.6
Note: The precision was calculated from 6 injections for each run.
62
from the PS 80 removal procedure was evaluated with both 2K IU/mL F-EPO and 40K IU/mL
F-EPO samples. The protein concentration was measured by UV absorbance at 280 nm. In
order to obtain appropriate absorbance readings, the samples were diluted to 500 μL for 2K
IU/mL F-EPO and 600 μL for 40K IU/mL F-EPO. The absorbances at 280 nm of the samples
were measured and the concentrations were calculated. The recoveries of both 2K IU/mL and
40K IU/mL F-EPO samples were 95-98%.
The precision was determined on two days. Three independent preparations of 2K IU/mL and
40K IU/mL formulated epoetin alfa samples were prepared and analyzed by CZE on different
days. All samples were evaluated for RSD (%) of percent areas of all the isoforms as shown in
Table 2.9. Precision ranged from 0.0% to 6.5%.
2.4
Conclusions
Significant modification of the capillary conditioning procedure for the Ph. Eur. monograph
method for the analysis of Erythropoietin concentrated solution resulted in greatly improved
repeatability and intermediate precision, sufficient to make the method suitable for use in
quality control laboratories. A high throughput polysorbate 80 removal procedure for epoetin
alfa formulated with polysorbate 80 was developed and validated. This technique may be used
for separation of other types of glycoproteins that have charge differences among the isoforms.
63
Table 2.8 Relative area (%) of each isoform of the epoetin alfa secondary standard (EPO Std)
and 2000 (2K) IU1/mL F-EPO2 and 40,000 (40K) IU/mL F-EPO
1
Iso-3
Iso-4
Iso-5
Iso-6
Iso-7
Iso-8
EPO Std
2.7
18.7
30.2
30.1
17.1
1.2
2K F-EPO
2.2
18.0
30.2
30.5
17.6
1.5
40K FEPO
2.3
18.5
30.4
30.2
17.3
1.3
IU: International Unit. 2K IU/mL=16.8 mg/mL epoetin alfa 40K IU/mL = 336.0 mg/mL
epoetin alfa.
2
F-EPO: epoetin alfa secondary standard prepared in formulation buffer
64
Table 2.9 Relative area (percent) of each isoform of the epoetin alfa secondary standard (EPO
Std) and 2000 (2K) IU1/mL F-EPO2 and 40,000 (40K) IU/mL F-EPO.
Day 1
Day 2
1
Iso-3
Iso-4
Iso-5
Iso-6
Iso-7
Iso-8
EPO Std
2.3
19.0
30.2
29.9
17.2
1.4
2K-1
3.8
23.8
31.1
27.3
13.2
0.9
2K-2
3.7
23.9
31.2
27.3
13.1
0.9
2K-3
3.9
23.8
31.0
27.2
13.1
0.9
RSD (%)
3.7
0.1
0.3
0.1
0.5
2.9
EPO Std
2.1
19.0
30.6
30.2
17.2
1.4
40K-1
3.5
20.9
29.8
28.5
15.9
1.3
40K-2
3.5
20.8
29.9
28.5
16.0
1.3
40K-3
3.5
20.8
29.9
28.5
16.0
1.3
RSD (%)
3.2
0.4
0.3
0.0
0.6
2.3
EPO Std
2.2
18.9
30.3
30.1
17.2
1.3
2K-1
3.7
23.7
31.1
27.1
13.3
1.1
2K-2
3.6
23.8
31.4
27.1
13.1
1.1
2K-3
3.7
23.7
31.1
27.2
13.3
1.0
RSD (%)
2.5
0.4
0.5
0.2
1.1
6.1
EPO Std
2.0
18.9
30.7
30.2
17.0
1.2
40K-1
3.2
21.1
30.0
28.6
15.8
1.3
40K-2
3.3
20.7
30.0
28.8
15.9
1.3
40K-3
3.6
20.9
29.9
28.6
15.8
1.2
RSD (%)
6.5
0.9
0.2
0.3
0.6
5.7
IU: International Unit 2K IU/mL=16.8 mg/mL epoetin alfa 40K IU/mL = 336.0 mg/mL
epoetin alfa.
2
F-EPO: epoetin alfa secondary standard prepared in formulation buffer
Note: The RSD (%) was calculated from triplicate injections for each sample.
65
2.5
List of references
1.
Siren, A. L.; Fratelli, M.; Brines, M.; Goemans, C.; Casagrande, S.; Lewczuk, P.;
Keenan, S.; Gleiter, C.; Pasquali, C.; Capobianco, A.; Mennini, T.; Heumann, R.;
Cerami, A.; Ehrenreich, H.;Ghezzi, P. Erythropoietin prevents neuronal apoptosis after
cerebral ischemia and metabolic stress. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 4044-9.
2.
Fisher, J. W. Erythropoietin: physiology and pharmacology update. Exp. Biol. Med.
2003, 228, 1-14.
3.
Savulescu, J.; Foddy, B.;Clayton, M. Why we should allow performance enhancing
drugs in sport. Br. J. Sports Med. 2004, 38, 666-70.
4.
Krantz, S. B. Erythropoietin. Blood 1991, 77, 419-34.
5.
Mechref, Y.;Novotny, M. V. Structural investigations of glycoconjugates at high
sensitivity. Chem. Rev. 2002, 102, 321-69.
6.
Balaguer, E.;Neusuess, C. Glycoprotein Characterization Combining Intact Protein
and Glycan Analysis by Capillary Electrophoresis-Electrospray Ionization-Mass
Spectrometry. Anal. Chem. 2006, 78, 5384-5393.
7.
Erythropoietin Concentrated Solution. European Pharmacopoeia 2005, 2, 1528-1529.
8.
Kinoshita, M.; Murakami, E.; Oda, Y.; Funakubo, T.; Kawakami, D.; Kakehi, K.;
Kawasaki, N.; Morimoto, K.;Hayakawa, T. Comparative studies on the analysis of
glycosylation heterogeneity of sialic acid-containing glycoproteins using capillary
electrophoresis. J. Chromatogr., A 2000, 866, 261-71.
9.
Benavente, F.; Gimenez, E.; Olivieri, A. C.; Barbosa, J.;Sanz-Nebot, V. Estimation of
the composition of recombinant human erythropoietin mixtures using capillary
electrophoresis and multivariate calibration methods. Electrophoresis 2006, 27, 40084015.
10.
Benavente, F.; Hernandez, E.; Guzman, N. A.; Sanz-Nebot, V.;Barbosa, J.
Determination of human erythropoietin by on-line immunoaffinity capillary
electrophoresis: a preliminary report. Anal. Bioanal. Chem. 2007, 387, 2633-2639.
66
11.
Bietlot, H. P.;Girard, M. Analysis of recombinant human erythropoietin in drug
formulations by high-performance capillary electrophoresis. J. Chromatogr., A 1997,
759, 177-184.
12.
Cifuentes, A.; Moreno-Arribas, M. V.; de Frutos, M.;Diez-Masa, J. C. Capillary
isoelectric focusing of erythropoietin glycoforms and its comparison with flat-bed
isoelectric focusing and capillary zone electrophoresis. J. Chromatogr., A 1999, 830,
453-463.
13.
de Frutos, M.; Cifuentes, A.;Diez-Masa, J. C. Differences in capillary electrophoresis
profiles of urinary and recombinant erythropoietin. Electrophoresis 2003, 24, 678-680.
14.
Lopez-Soto-Yarritu, P.; Diez-Masa, J. C.; De Frutos, M.;Cifuentes, A. Comparison of
different capillary electrophoresis methods for analysis of recombinant erythropoietin
glycoforms. J. Sep. Sci. 2002, 25, 1112-1118.
15.
Nieto, O.; Hernandez, P.;Hernandez, L. Capillary zone electrophoresis of human
recombinant erythropoietin using C8 coated columns without additives in the running
buffer. Anal. Commun. 1996, 33, 425-427.
16.
Sanz-Nebot, V.; Benavente, F.; Gimenez, E.;Barbosa, J. Capillary electrophoresis and
matrix-assisted laser desorption/ionization-time of flight-mass spectrometry for
analysis of the novel erythropoiesis-stimulating protein (NESP). Electrophoresis 2005,
26, 1451-1456.
17.
Sanz-Nebot, V.; Benavente, F.; Vallverdu, A.; Guzman, N. A.;Barbosa, J. Separation
of recombinant human erythropoietin glycoforms by capillary electrophoresis using
volatile electrolytes. assessment of mass spectrometry for the characterization of
erythropoietin glycoforms. Anal. Chem. 2003, 75, 5220-5229.
18.
Watson, E.;Yao, F. Capillary electrophoretic separation of human recombinant
erythropoietin (r-HuEPO) glycoforms. Anal. Biochem. 1993, 210, 389-93.
19.
Yu, B.; Cong, H.; Liu, H.; Li, Y.;Liu, F. Separation and detection of erythropoietin by
CE and CE-MS. TrAC, Trends Anal. Chem. 2005, 24, 350-357.
67
20.
Lopez-Soto-Yarritu, P.; Diez-Masa, J. C.; Cifuentes, A.;de Frutos, M. Improved
capillary isoelectric focusing method for recombinant erythropoietin analysis. J.
Chromatogr., A 2002, 968, 221-8.
21.
Poole, C. F. The essence of chromatography. 1st ed.; Elsevier: Amsterdam ; Boston,
2003.
22.
Weinberger, R. Practical capillary electrophoresis. 2nd ed.; Academic Press: San
Diego, Calif., 2000.
23.
Bushey, M. M.;Jorgenson, J. W. Capillary electrophoresis of proteins in buffers
containing high concentrations of zwitterionic salts. J. Chromatogr. 1989, 480, 30110.
24.
Doherty, E. A.; Meagher, R. J.; Albarghouthi, M. N.;Barron, A. E. Microchannel wall
coatings for protein separations by capillary and chip electrophoresis. Electrophoresis
2003, 24, 34-54.
25.
Lucy, C. A.; MacDonald, A. M.;Gulcev, M. D. Non-covalent capillary coatings for
protein separations in capillary electrophoresis. J. Chromatogr., A 2008, 1184, 81-105.
26.
Steiner, S. A.; Watson, D. M.;Fritz, J. S. Ion association with alkylammonium cations
for separation of anions by capillary electrophoresis. J. Chromatogr., A 2005, 1085,
170-5.
27.
Bristow, A., Charton, E. Assessment of the suitability of a Capillary Zone
Electrophoresis method for determining Isoform Distribution of Erythropoietin.
Pharmeuropa 1999, 11, 290-300.
28.
Mahler, H.-C.; Printz, M.; Kopf, R.; Schuller, R.;Muller, R. Behaviour of polysorbate
20 during dialysis, concentration and filtration using membrane separation techniques.
J. Pharm. Sci. 2007, 97, 764-774.
68
Chapter 3: Separation of residual cell culture media components by micellar
electrokinetic chromatography
3.1
Introduction
Folic acid, hypoxanthine, mycophenolic acid, nicotinic acid, riboflavin, and
xanthine are widely used as cell culture media components in monoclonal antibody
manufacturing. These components are impurities to the drug product and are
removed during downstream purification processes. Complete removal of these
impurities is important in order to maintain safety and quality of drug product [1].
Several high performance liquid chromatographic (HPLC) methods have been
developed for the determination of some of these compounds in food, nutritional
supplements, human serum, or human plasma [2-10]. Capillary electrophoresis (CE)
has also been used for these analyses [11-15]. Some of the distinct advantages of
CE include faster analysis time, high separation efficiency, less complex sample
preparation, low sample volume required, etc. [16]. The major drawback of CE is
low concentration sensitivity; however, this problem has been largely solved and
equivalent detection can be achieved on CE with the use of on-line concentration
techniques such as stacking or sweeping [17-19].
While some CE methods have been developed for the determination of watersoluble vitamins as main components in food or nutritional supplements, very little
has been focused on the trace level analysis. Priego-Capota and Castro used
capillary zone electrophoresis (CZE) with a native fluorescence spectroscopy and a
charge coupled detector (CCD) to measure vitamin B2 (riboflavin) and B6
(pyridoxine) [20]. Although the detection limit was obtained at part per billion (ppb)
69
levels, extraction was required for eliminating interferences and off-line sample
concentration was also needed when the method was applied to serum samples.
Crevillen et al. developed a method using carbon nano-tube disposable detectors in
microchip capillary electrophoresis for the determination of folic acid, asorbic acid,
and pyridoxine in pharmaceutical formulations [12], and determined that the limit
of detection (LOD) for folic acid was about 10 μM. Trace amounts of xanthine and
hypoxanthine have been determined by Causse et al. recently using CE [13]. The
LODs, evaluated in samples containing serum matrix, were 0.4 μM and 0.6 μM for
hypoxanthine and xanthine, respectively.
Ohyama et al. developed a simple and rapid CZE method for the analysis of
mycophenolic acid and its acyl and phenol glucuronide metabolites in human serum
and achieved a detection limit of 0.1 μg/mL [11].
CZE is the simplest and most commonly used mode of CE [16]. CZE is widely used
for the separation of inorganic and organic ions, ionizable compounds, zwitterions,
and biopolymers. Micellar electrokinetic chromatography (MEKC) is another mode
of CE in which the separation of neutral analytes is based upon the difference in the
solute distribution between the aqueous phase and the micellar phase
(pseudostationary phase) [21]. For charged analytes, the separation is based upon
both differential partitioning and differences in electrophoretic mobility. Like CZE,
the separation takes place in an electric field applied across a capillary. The
difference between CZE and MEKC is the running buffer. In MEKC, the running
70
buffer contains a surfactant that forms micelles when its concentration exceeds the
critical micelle concentration, while in CZE the running buffer contains no
surfactant. The micellar pseudostationary phase moves with a migration velocity
and/or direction that is different to the mobile phase. Separation occurs within a
migration window (elution range) defined by the difference in the migration time of
the running buffer (teo), which is same as the migration time of the electroosmotic
flow (EOF), and the migration time of the micelles (tmc). Analytes are separated
within the migration window.
MEKC has become a powerful technique for the separation of mixtures of charged
and neutral analytes that are difficult to separate by CZE [22-23]. No research has
been done on the simultaneous separation of folic acid, hypoxanthine,
mycophenolic acid, nicotinic acid, riboflavin, and xanthine in biopharmaceutical
matrices. The objective of this study was to establish a simple MEKC method that is
able to simultaneously analyze all the mentioned analytes in cell culture media and
in protein-containing matrices from monoclonal antibody manufacturing processes.
3.2
Materials and methods
3.2.1
Chemicals and reagents
All chemicals used in the preparation of buffers and solutions were of ACS reagent
or electrophoresis grade. Folic acid, hypoxanthine, mycophenolic acid, nicotinic
acid, riboflavin, xanthine, and 2-naphthalenemethanol were supplied by SigmaAldrich (St. Louis, MO, USA). Their structures and pKa’s are shown in Figure 3.1
[15, 19, 24-26]. Methanol, sodium tetraborate (Na2B4O7.10H2O), sodium
71
NH2
HN
N
N
N
N
OH
HN
OH
O
O
NH
O
O
OH
b
a
O
O
H
N
HN
HN
O
N
N
H
N
N
N
H
d
c
OH
HO
OH
HO
OH
OH
CH3
O
O
H3C
N
H3C
N
CH3
O
O
N
NH
CH3
OH
O
O
e
f
Figure 3.1 Structures and pKa’s of the cell culture media components. a: folic acid (2.3
(pKa1), 8.3 (pKa2)); b: nicotinic acid (4.85); c: hypoxanthine (8.7); d: xanthine (7.4
(pKa1), 11.1 (pKa2)); e: riboflavin (10.2); f: mycophenolic acid (4.5). [15, 19 24-26]
72
dihydrogen phosphate (NaH2PO4), and sodium dodecyl sulfate (SDS) were
purchased from J. T. Baker (Phillipsburg, NJ, USA). Sodium hydroxide (NaOH)
solutions (1 M and 0.1 M) were provided by Agilent Technologies (Wilmington,
DE, USA). Water with a conductivity value lower than 5 µS/m was obtained using a
Milli-Q water purification system from Millipore (Billerica, MA, USA). The
running buffer was prepared by dissolving appropriate amounts of Na2B4O7.10H2O,
NaH2PO4, and SDS in water to obtain the desired concentration of 20 mM
NaH2PO4, 20 mM Na2B4O7, and 40 mM SDS. The pH was adjusted to 9.0 with 10%
phosphoric acid. The above solutions were filtered through a 0.22 μm nylon filter.
3.2.2
Standard solutions
A 1 mM stock solution was prepared by dissolving the appropriate amounts of each
analyte in 0.01 M NaOH. Five working standard solutions of 0.01 mM, 0.02 mM,
0.04 mM, 0.08 mM, and 0.1 mM were obtained by dilution of the stock solution (1
mM) in Milli-Q water.
3.2.3
Sample preparation
Protein-containing samples were obtained from different protein purification steps.
In-process samples were diluted with Milli-Q water when analyte concentration was
higher than the highest working standard. For the samples containing high protein
concentrations, 10,000 (10K) Dalton molecular mass cutoff filters (Pall Life
Sciences, East Hills, New York, USA) were used to remove protein prior to CE
injection. An aliquot of 500 μL of the sample was transferred into a 10K molecular
73
mass cutoff filter and centrifuged at 15,000 × g for 30 minutes. When not in use,
samples were stored in the dark at 2-8 °C.
3.2.4
Capillary electrophoresis (CE)
The CE experiments were performed on an Agilent 3D-CE instrument equipped
with a photo diode array detector and ChemStation software from Agilent
Technologies (Wilmington, DE, USA). Bare fused-silica capillary columns (40 cm
x 50 μm id) were supplied by Agilent Technologies (Wilmington, DE, USA). The
detection window was placed at 8.5 cm from the outlet of the capillary. The
capillary temperature was maintained at 20 °C for all the experiments. A voltage of
25 kV was applied during electrophoretic separations. Samples were injected
hydrodynamically at 50 mbar for 5 seconds. Detection was performed at 210 nm,
222 nm, 254 nm, and 275 nm. Data were collected at a sampling rate of 5 Hz. If not
specified, corrected peak areas (peak area divided by migration time) were used for
all the calculations. The current was also monitored to ensure that the Joule heating
was below the upper limit determined by an Ohm’s law plot.
3.2.5
Capillary conditioning
New capillaries were conditioned by rinsing with 1 M NaOH for 10 minutes,
methanol for 5 minutes, water for 5 minutes, and running buffer for 10 minutes.
After the rinse, a voltage of 20 kV was applied to the capillary filled with running
buffer for 10 minutes. Between injections, the capillary was rinsed with running
buffer for 5 minutes. To ensure reproducibility, at the end of each set of runs (15
injections per run), the capillary was rinsed with 1 M NaOH for 10 minutes,
74
methanol for 5 minutes, water for 5 minutes, and running buffer for 10 minutes.
3.3 Results and Discussion
3.3.1
Capillary zone electrophoresis (CZE) vs. micellar electrokinetic
chromatography (MEKC)
Figure 3.2A shows a separation of the studied compounds, riboflavin,
hypoxanthine, xanthine, mycophenolic acid, folic acid, and nicotinic acid, under
CZE conditions. The neutral compound, 2-naphthalenemethanol, was used as an
EOF marker. All six compounds were very well separated within 6 minutes using a
running buffer of 20 mM NaH2PO4 and 20 mM Na2B4O7 containing no SDS with
pH adjusted to 9.0. The buffer pH of 9.0 is within the effective buffer range of 20
mM NaB4O7, which is 8-10 [26]. The addition of 20 mM NaH2PO4 increased the
ionic strength of the solution, which decreased the thickness of the electrical double
layer in the capillary, and hence decreased the EOF. Using a combination of
NaH2PO4 and Na2B4O7 over Na2B4O7 also increased the number of theoretical
plates [27]. This buffer composition provided the optimal separation condition for
the six analytes.
One of the effective parameters in CZE is the pH value of the buffer. Nishi et al.
studied the pH dependence of the separation of eight water-soluble vitamins in the
range of pH 6-9, and found that the peak shape and resolution were gradually
improved and migration times increased with the increase of pH due to the increase
in ionization of the solutes at the specified pH [24]. The ionization of an analyte at a
75
mAU
10
A
8
1
4
6
3
6
4
7
2
5
2
0
0
2
4
6
8
min
8
min
mAU
7
10
B
8
1
4
6
6
3
2
4
5
2
0
0
2
4
6
Figure 3.2 Comparison of the separation of the cell culture media components under
(A) CZE and (B) MEKC conditions. Compound identification (10 µM each): 1.
hypoxanthine; 2. riboflavin; 3. xanthine; 4. mycophenolic acid; 5. folic acid; 6.
nicotinic acid; and 7. 2-naphthalenemethanol (neutral marker). Analytical
conditions: 20 mM NaH2PO4, 20 mM Na2B4O7, pH 9.0, 40 mM SDS (MEKC only),
25 kV, 20°C, absorbance detection at 210 nm. Fused silica capillary dimensions: 50
µm i.d. x 40 cm (40 cm effective length).
76
given pH depends on its pKa value. Riboflavin has the highest pKa value (pKa
=10.2) among the six compounds, resulting in only about 10% ionization at pH 9.0.
Because its ionization was so small, the elution of riboflavin was mainly due to
electroosmotic flow. However, given its relatively good separation from the neutral
marker, the ionized form is also contributing to the mobility and the separation. On
the other hand, nicotinic acid has the lowest pKa value (pKa = 4.85), and is
completely (> 99.99%) ionized at pH 9.0, which is why it eluted last. The migration
order of 2-naphthalenemethanol (EOF marker) and the analytes indicates that all six
compounds were negatively charged at the specified pH. (Although not necessarily
obvious from their structures in Figure 3.1, hypoxanthine and xanthine are weakly
acidic with pKa’s of 8.7 and 7.4, respectively [19].)
Figure 3.2B shows the separation of the studied compounds under MEKC
conditions. The same running buffer used for CZE was used for MEKC except for
the addition of 40 mM SDS. Here, the neutral compound, 2-naphthalenemethanol,
was used as a marker for the net micelle migration velocity since it was expected to
interact very strongly with SDS micelles. The migration order of riboflavin and
hypoxanthine was reversed when the separation mode switched from CZE to
MEKC. In MEKC, riboflavin interacts with SDS more effectively than the other
analytes due to its high hydrophobicity. The marker, 2-naphthalenemethanol, eluted
first in CZE and last in MEKC, the latter providing strong evidence that the MEKC
separation was operating in the normal elution mode, i.e., the net velocity of micelle
(νmc) and the electroosmotic velocity (νeo) have the same directions and |νeo| > |νmc|.
77
mAU
14
12
10
4
1
8
A
6
3
6
4
2
5
2
0
0
1
2
3
4
5
6
7
min
2
3
4
5
6
7
min
mAU
14
12
10
B
8
6
4
2
0
0
1
Figure 3.3 Matrix interference under CZE condition. (A) standards under CZE
conditions (B) sample matrix under CZE conditions. 1. hypoxanthine 2. riboflavin
3. xanthine 4. mycophenolic acid 5. folic acid 6. nicotinic acid. Other conditions as
in Figure 3.2.
78
Both CZE and MEKC modes provided good baseline-resolved separation for the
standards. However, a matrix interference with riboflavin occurred when the CZE
method was applied to the sample matrices that contain large amounts of anionic,
cationic and neutral compounds. As shown in Figure 3.3, under CZE conditions,
riboflavin migrated at the same time as the large matrix peak, and even with the
varied separation conditions (voltage and temperature), the interference still exists.
Due to this observed limitation of CZE in this particular application, MEKC was
chosen as a methodology for further investigation.
3.3.2
Effect of surfactant concentration
One characteristic of a surfactant is the formation of micelles when its solution
concentration exceeds a threshold value, the critical micelle concentration (CMC)
[28]. SDS is an anionic surfactant and its CMC has been reported to be in the range
of 3-8 mM depending on the buffer composition, pH, and temperature [28]. The
effect of SDS concentration on migration time was investigated using five pH 9.0
buffer solutions with SDS concentrations from 10 mM to 100 mM.
Figure 3.4 shows the electropherograms obtained with the five SDS buffer
solutions. A plot of the migration time versus SDS concentration was also made as
shown in Figure 3.5. A linear relationship between migration time and SDS
concentration was observed for all the compounds. An increase in SDS
concentration results in an increase in micelle concentration, and according to the
phase ratio (φ) model of Terabe (k = φK, where K = partition coefficient) [29-30],
the migration factor (k) will increase, which in turn results in an increase in
79
A
1
2
B
C
1
2
4
3
6
5
1
4
3
6
5
4+5
1
2
E
6
5
2
D
4
3
1
3
3
2
6
4
6
5
Figure 3.4 Effect of SDS concentration on the separation of the cell culture media
components: (A) 10 mM SDS, (B) 20 mM SDS, (C) 50 mM SDS, (D) 75 mM SDS,
and (E) 100 mM SDS. Compound identification and other conditions as in Figure
3.2.
80
8
Hypoxanthine
7
Migration Time (min)
Riboflavin
6
Xanthine
5
Mycophenolic
Acid
4
Folic Acid
3
Nicotinic Acid
2
0
20
40
60
80
100
120
SDS Concentration (mM)
Figure 3.5 Effect of SDS concentration on the migration time of the cell culture
media components. Conditions as in Figure 3.2.
81
migration time if EOF remains approximately constant. The increase in migration
time with increasing surfactant concentration was also in good agreement with the
findings of Poole et al [31]. Among the six analytes, the SDS concentration has the
greatest effect on the migration of riboflavin. The plot also shows that the two
curves of riboflavin and hypoxanthine intersect when the SDS concentration is
about 30 mM, indicating that they would co-elute at this SDS concentration and the
migration order would reverse beyond this concentration. The folic acid and
mycophenolic acid curves also intersect similarly, but at 75 mM SDS. Clearly such
intersection points should be avoided in order to resolve all of the analytes. In
addition, the effect of (i) SDS concentration on peak shape and analysis time was
considered, as well as the need to simultaneously minimize (ii) the possibility of
sample overload (with respect to the micelles) and (iii) the level of Joule heating.
Based on the data in Figures 3 and 4 and the criteria summarized here (resolution,
peak shape, analysis time, sample overload, and Joule heating), the optimal SDS
concentration was found to be about 40 mM (different resolution, but same elution
order as in Figure 3.3C).
3.3.3
Effect of pH
Like CZE, the pH of the separation buffer for MEKC also has a major influence on
selectivity. Three buffer solutions with pH values of 8.0, 8.5, and 9.0 were
evaluated for selectivity. The SDS concentration for the three buffer solutions was
40 mM. At pH 8.0, folic acid and nicotinic acid co-eluted (Figure 3.6A). At pH 8.5,
although baseline resolution was achieved for all the analytes, the peak shapes of
3.3.4
Detection wavelength
82
mAU
14
12
10
A
4
8
6
2
1
5+6
4
3
2
0
0
1
2
3
4
5
6
7
min
7
min
mAU
14
12
10
B
8
4
6
1
3
4
5
2
6
2
0
0
1
2
3
4
5
6
Figure 3.6 Effect of buffer pH on separation (A) At pH 8.0, folic acid and nicotinic
acid (5 and 6) co-eluted; (B) At pH 8.5, peak shape of riboflavin (2) and nicotinic
acid (6) were distorted. Other conditions as in Figure 3.2B
83
riboflavin and nicotinic acid were distorted (Figure 3.6B). The best separation was
obtained with pH at 9.0, where neither of the previous disadvantages was observed.
The UV spectrum of each analyte was taken in the range of 190 nm to 300 nm
(Appendix A). Good absorbances were obtained at wavelengths of 210 nm, 222 nm,
254 nm, and 275 nm for all analytes. The absorbance at shorter wavelength is
attributed to C=N bond present in xanthine, hypoxanthine, and the three watersoluble vitamins. The absorbance at longer wavelength is attributed to conjugated
bonds in all the six compounds. The signals at the four wavelengths were collected
and the response curves were made for each analyte as shown in Figures 3.7-3.12.
The slope at 210 nm appeared to be the highest for hypoxanthine, xanthine, and
folic acid and second highest for mycophenolic acid and nicotinic acid. Although
the slope at 210 nm was the lowest for riboflavin, the response at 210 nm was only
reduced in half compared to that at 222 nm. The 210 nm wavelength was chosen for
all the experiments based on the overall response data.
3.4 Validation of the analytical procedure
The application and validation of this technique were demonstrated through
determination of the six components in protein-containing biopharmaceutical
manufacturing process samples. The principle of the validation of an analytical
procedure is widespread today in most domains of chemical analysis [32].
Nevertheless, for a given application, the validation procedures remain incompletely
determined in several cases despite the various regulations relating to the good
practices. An effort to harmonize the validation strategies [32-34] is ongoing, and
84
14
12
210 nm
Peak Area
10
8
222 nm
6
254 nm
4
275 nm
2
0
0
20
40
60
80
100
120
Hypoxanthine Concentration (mM)
Figure 3.7. Comparison of detection wavelength on the calibration sensitivity for
hypoxanthine. Other conditions as in Figure 3.2.
85
40
35
210 nm
30
Peak Area
25
222 nm
20
254 nm
15
275 nm
10
5
0
0
20
40
60
80
100
Riboflavin Concentration (mM)
Figure 3.8. Comparison of detection wavelength on the calibration sensitivity for
riboflavin. Other conditions as in Figure 3.2.
120
86
16
14
210 nm
12
222 nm
Peak Area
10
8
254 nm
6
275 nm
4
2
0
0
20
40
60
80
100
Xanthine Concentration (mM)
Figure 3.9. Comparison of detection wavelength on the calibration sensitivity for
xanthine. Other conditions as in Figure 3.2.
120
87
35
30
210 nm
Peak Area
25
222 nm
20
15
254 nm
10
275 nm
5
0
0
20
40
60
80
100
Mycophenolic acid concentration (mM)
Figure 3.10 Comparison of detection wavelength on the calibration sensitivity for
mycophenolic acid. Other conditions as in Figure 3.2.
120
88
12
10
210 nm
Peak Area
8
222 nm
6
254 nm
4
275 nm
2
0
0
20
40
60
80
100
120
Folic acid concentration (mM)
Figure 3.11 Comparison of detection wavelength on the calibration sensitivity for
folic acid. Other conditions as in Figure 3.2.
89
35
30
25
Peak Area
210 nm
20
222 nm
15
254 nm
10
275 nm
5
0
0
20
40
60
80
100
Nicotinic acid concentration (mM)
Figure 3.12 Comparison of detection wavelength on the calibration sensitivity for
nicotinic acid. Other conditions as in Figure 3.2.
120
90
many of these strategies were adopted in the current International Conference on
Harmonisation (ICH) guidelines [35]. The validation procedures used in this study
were taken mostly from the current ICH guidelines since they have been accepted
by the industry.
3.4.1
Limit of detection (LOD) and limit of quantitation (LOQ)
The LOD and LOQ were determined according to the ordinary least-squares
regression method [35]. Standards at the low end of the standard curve were
prepared and analyzed. Standard curves were generated by plotting the corrected
peak areas of the analytes against the corresponding analyte concentrations (2-20
mM) to determine the slope (S) and the standard deviation of the residuals (σ).
These values were used to calculate the LOD and the LOQ using the formula:
LOD = 3.3 ∗ σ Slope
LOQ = 10 ∗ σ Slope
The coefficients of determination for the linear regressions for each analyte at the
five levels were > 0.99. The precision at each level was demonstrated by relative
standard deviation (RSD in %). The second level that shows relatively high RSD
values for all analytes might be attributed to a poor injection since it appeared to
have much lower corrected areas for each analyte. The LOD and LOQ results are
presented in Tables 3.1-3.6. Another approach for LOD and LOQ determination is
to take the analyte concentrations corresponding to signal to noise ratio of 3.3 for
LOD and 10 for LOQ [36]. This approach was used to verify the LOD values
obtained using ordinary least-squares linear regression. An in-process sample was
spiked with standards at the LOD levels. Six determinations were made. For each
91
Table 3.1 Limit of Detection and Limit of Quantitation data for hypoxanthine.
Hypoxanthine Conc.
(μM)
1.98
1.98
1.98
2.47
2.47
2.47
4.95
4.95
4.95
9.89
9.89
9.89
19.79
19.79
19.79
σ
Slope
Intercept
r2
LOD (3.3*σ)
LOQ (10*σ)
Corrected Peak Area
1.30
1.12
1.44
1.71
1.41
1.61
2.83
2.65
2.57
5.38
5.57
5.05
10.99
11.01
10.83
RSD (%)
12.3
9.7
5.0
4.9
0.9
0.19
0.54
0.13
0.9987
1.15 (μM)
0.16 (μg)
3.47 (μM)
0.47 (μg)
92
Table 3.2 Limit of Detection and Limit of Quantitation data for riboflavin.
Riboflavin Conc. (μM)
Corrected Peak Area
3.96
3.96
3.96
4.95
4.95
4.95
9.89
9.89
9.89
19.78
19.78
19.78
39.57
39.57
39.57
σ
Slope
Intercept
r2
1.18
1.15
1.11
2.05
2.03
1.89
3.54
3.31
3.00
6.30
6.58
5.81
11.53
12.65
12.78
LOD (3.3*σ)
LOQ (10*σ)
RSD (%)
3.0
4.3
8.2
6.2
5.6
0.36
0.31
0.21
0.9963
3.91 (μM)
1.47 (μg)
11.84 (μM)
4.46 (μg)
93
Table 3.3 Limit of Detection and Limit of Quantitation data for xanthine.
Xanthine Conc. (μM)
Corrected Peak Area
1.98
1.98
1.98
2.47
2.47
2.47
4.95
4.95
4.95
9.89
9.89
9.89
19.79
19.79
19.79
σ
Slope
Intercept
r2
0.87
0.98
0.96
1.13
1.41
1.47
2.47
2.21
2.24
3.92
4.69
4.55
8.68
9.49
9.27
LOD (3.3*σ)
LOQ (10*σ)
RSD (%)
6.4
13.6
6.1
9.3
4.6
0.27
0.23
0.07
0.9964
3.86 (μM)
0.59 (μg)
11.69 (μM)
1.78 (μg)
94
Table 3.4 Limit of Detection and Limit of Quantitation data for mycophenolic acid.
Mycophenolic acid
Conc. (μM)
1.99
1.99
1.99
2.47
2.47
2.47
4.95
4.95
4.95
9.89
9.89
9.89
19.79
19.79
19.79
σ
Slope
Intercept
r2
LOD (3.3*σ)
LOQ (10*σ)
Corrected Peak Area
1.25
1.25
1.22
1.38
1.51
1.27
2.54
2.41
2.48
4.21
5.25
4.76
9.77
10.52
10.24
RSD (%)
1.3
8.8
2.6
10.9
3.8
0.31
0.50
0.07
0.9960
2.03 (μM)
0.65 (μg)
6.14 (μM)
1.97 (μg)
95
Table 3.5 Limit of Detection and Limit of Quantitation data for folic acid.
Folic acid Conc. (μM)
Corrected Peak Area
2.02
2.02
2.02
2.51
2.51
2.51
5.03
5.03
5.03
10.05
10.05
10.05
20.11
20.11
20.11
σ
Slope
Intercept
r2
0.86
0.82
0.78
0.64
0.85
0.91
1.25
1.35
1.35
2.25
2.93
2.73
5.00
5.28
5.38
LOD (3.3*σ)
LOQ (10*σ)
RSD (%)
5.0
17.5
4.2
13.3
3.7
0.22
0.25
0.18
0.9943
2.50 (μM)
1.10 (μg)
8.72 (μM)
3.85 (μg)
96
Table 3.6 Limit of Detection and Limit of Quantitation data for nicotinic acid.
Nicotinic acid Conc.
(μM)
2.01
2.01
2.01
2.49
2.49
2.49
4.98
4.98
4.98
9.95
9.95
9.95
19.90
19.90
19.90
σ
Slope
Intercept
r2
LOD (3.3*σ)
LOQ (10*σ)
Corrected Peak Area
1.28
1.16
1.30
1.44
1.54
1.45
2.83
2.59
2.47
4.37
5.57
4.97
10.60
11.45
11.38
RSD (%)
6.1
3.8
6.9
12.1
4.2
0.45
0.55
-0.04
0.9945
2.67 (μM)
0.33 (μg)
8.10 (μM)
1.00 (μg)
97
determination, the signal to noise ratios obtained were in the range of 1.9 to 3.3 for
all analytes (data not shown). Given the variation of LOD determination using
different approaches, the results obtained from ordinary least-square regression
method used in this study were consistent with those from signal to noise ratio
method [37]. The LOQ values were verified in the linearity section as the lowest
point on the linear curve.
3.4.2
Specificity
Specificity is the ability to assess unequivocally the analyte in the presence of
components which may be expected to be present [35]. To access specificity, a
sample from the late stage of the process purification steps was prepared by spiking
a standard solution into the sample containing no analytes. A non-spiked sample
was also prepared as a background. No interferences were found from the sample
matrices. Representative electropherograms are shown in Figure 3.13.
3.4.3
Linearity and range
The linearity of an analytical procedure is its ability to obtain test results, within a
given range, that are directly proportional to the concentration (amount) of analyte
in the test sample [35]. The range of an analytical procedure is the interval between
the upper and the lower concentrations (amounts) of analyte in the test article for
which it has been demonstrated that the analytical procedure has a suitable level of
precision, accuracy, and linearity [35]. The linearity of each analyte was evaluated
at five levels ranging from 10 μM to 100 μM, where the lower end was near the
LOQ. Calibration standards were prepared by series of dilutions of a stock standard
98
mAU
7
6
A
5
4
6
3
1
2
2
1
3
4
5
0
0
2
4
6
min
2
4
6
min
mAU
7
B
6
5
4
3
2
1
0
0
Figure 3.13 Electropherograms of spiked (A) and non-spiked (B) samples from late
stage protein purification process. Other conditions as in Figure 3.2.
99
in water. Standard curves were generated by plotting analyte peak area (corrected)
versus analyte concentration. Samples were prepared by spiking appropriate
amounts of analytes into samples that contained none of the analytes. Each sample
solution was prepared in triplicate and each preparation was analyzed once. Sample
concentrations were calculated against corresponding standard curves. The linear
regression analysis was performed by plotting the measured analyte concentrations
versus the theoretical analyte concentrations. The coefficients of determination from
the linear regression analysis were > 0.999 for all analytes. Table 3.7 shows the
linearity and range data.
3.4.4
Accuracy and precision
As this method will be used for impurity clearance, the accuracy and precision near
the LOQ level were evaluated. A sample was spiked with a standard to obtain the
level of 10 μM for each analyte. Nine determinations were made. Non-spiked
samples served as blanks. Recoveries of all samples analyzed were calculated to
demonstrate accuracy. The RSD of nine measured analyte concentrations were also
calculated to demonstrated precision. Accuracy and precision data were presented in
Table 3.8.
3.5
Concluding remarks
A simple MEKC procedure was developed for the simultaneous determination of
folic acid, hypoxanthine, mycophenolic acid, nicotinic acid, riboflavin, and xanthine
in protein-containing matrices from monoclonal antibody manufacturing processes.
The MEKC running buffer was 40 mM SDS, 20 mM sodium phosphate, and 20
100
Table 3.7 Linearity and range data (10-100 μM)
*
Analyte
r2
RSD (%)*
Recovery (%)
Hypoxanthine
0.9995
0.6-3.5
96-103
Riboflavin
0.9997
0.5-2.6
97-102
Xanthine
0.9996
0.5-2.8
96-108
Mycophenolic acid
0.9993
1.2-3.0
95-107
Folic acid
0.9995
0.7-3.7
96-104
Nicotinic acid
0.9992
0.6-3.6
95-109
The RSD was calculated from three triplicates at each level.
101
Table 3.8 Accuracy and precision data
Analyte
Hypoxanthine
Riboflavin
Xanthine
Mycophenolic acid
Folic acid
Nicotinic acid
RSD (%) (n = 9)
2.4
1.7
4.7
6.7
4.8
3.1
Recovery (%)
88-93
94-98
98-112
91-106
89-99
99-106
102
mM sodium borate (pH 9.0). Other MEKC conditions were established, including
an applied voltage of 25 kV, an injection length of 50 mbar for 5 seconds, and a
detection wavelength of 210 nm. The samples required minimal preparation prior to
analysis. The MEKC method was able to determine the six analytes concentrations
with good linearity, accuracy and precision.
103
3.6
List of references
1.
Carleton, F. J.; Agalloco, J. P.; NetLibrary Inc. Validation of
pharmaceutical processes sterile products.
http://ezproxy.tamu.edu:2048/login?url=http://www.netLibrary.com/urlapi.
asp?action=summary&v=1&bookid=11520
2.
Sorensen, L. M.; Nielsen, K. F.; Jacobsen, T.; Koch, A. G.; Nielsen, P. V.;
Frisvad, J. C., Determination of mycophenolic acid in meat products using
mixed mode reversed phase-anion exchange clean-up and liquid
chromatography-high-resolution mass spectrometry. J. Chromatogr., A
2008, 1205, 103-108.
3.
Mino, Y.; Naito, T.; Matsushita, T.; Kagawa, Y.; Kawakami, J.,
Simultaneous determination of mycophenolic acid and its glucuronides in
human plasma using isocratic ion pair high-performance liquid
chromatography. J. Pharm. Biomed. Anal. 2008, 46, 603-608.
4.
Gutzeit, D.; Moench, S.; Jerz, G.; Winterhalter, P.; Rychlik, M., Folate
content in sea buckthorn berries and related products (Hippophae
rhamnoides L. ssp. rhamnoides): LC-MS/MS determination of folate
vitamer stability influenced by processing and storage assessed by stable
isotope dilution assay. Anal. Bioanal. Chem. 2008, 391, 211-219.
5.
Yau, W.-P.; Vathsala, A.; Lou, H.-X.; Zhou, S. F.; Chan, E., Simple
reversed-phase liquid chromatographic assay for simultaneous
quantification of free mycophenolic acid and its glucuronide metabolite in
human plasma. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2007,
846, 313-318.
6.
Elbarbry, F. A.; Shoker, A., Simple high performance liquid
chromatographic assay for mycophenolic acid in renal transplant patients. J.
Pharm. Biomed. Anal. 2007, 43, 788-792.
7.
Cooper, N.; Khosravan, R.; Erdmann, C.; Fiene, J.; Lee, J. W.,
Quantification of uric acid, xanthine and hypoxanthine in human serum by
HPLC for pharmacodynamic studies. J. Chromatogr. B: Analyt. Technol.
Biomed. Life Sci. 2006, 837, 1-10.
8.
Marklund, N.; Ostman, B.; Nalmo, L.; Persson, L.; Hillered, L.,
Hypoxanthine, uric acid and allantoin as indicators of in vivo free radical
reactions. Description of a HPLC method and human brain microdialysis
data. Acta Neurochir. 2000, 142, 1135-41; discussion 1141-2.
104
9.
Czauderna, M.; Kowalczyk, J., Simultaneous measurement of allantoin,
uric acid, xanthine and hypoxanthine in blood by high-performance liquid
chromatography. J. Chromatogr. B: Biomed. Sci. Appl. 1997, 704, 89-98.
10.
Albala-Hurtado, S.; Veciana-Nogues, M. T.; Izquierdo-Pulido, M.; MarineFont, A., Determination of water-soluble vitamins in infant milk by highperformance liquid chromatography. J. Chromatogr., A 1997, 778, 247-53.
11.
Ohyama, K.; Kinoshita, N.; Kishikawa, N.; Kuroda, N., A simple and rapid
CZE method for the analysis of mycophenolic acid and its phenol
glucuronide metabolite in human serum. Electrophoresis 2008, 29, 36583664.
12.
Crevillen, A. G.; Pumera, M.; Gonzalez, M. C.; Escarpa, A., Carbon
nanotube disposable detectors in microchip capillary electrophoresis
vitamin for water-soluble vitamin determination: analytical possibilities in
pharmaceutical quality control. Electrophoresis 2008, 29, 2997-3004.
13.
Causse, E.; Pradelles, A.; Dirat, B.; Negre-Salvayre, A.; Salvayre, R.;
Couderc, F., Simultaneous determination of allantoin, hypoxanthine,
xanthine, and uric acid in serum/plasma by CE. Electrophoresis 2007, 28,
381-7.
14.
Cataldi, T. R.; Nardiello, D.; Scrano, L.; Scopa, A., Assay of riboflavin in
sample wines by capillary zone electrophoresis and laser-induced
fluorescence detection. J. Agric. Food Chem. 2002, 50, 6643-7.
15.
Akhtar, M. J.; Khan, M. A.; Ahmad, I., Photodegradation of folic acid in
aqueous solution. J. Pharm. Biomed. Anal. 1999, 19, 269-275.
16.
Landers, J. P., Handbook of capillary electrophoresis. 2nd ed.; CRC Press:
Boca Raton, 1997.
17.
Quirino, J. P.; Iwai, Y.; Otsuka, K.; Terabe, S., Determination of
environmentally relevant aromatic amines in the ppt levels by cation
selective exhaustive injection-sweeping-micellar electrokinetic
chromatography. Electrophoresis 2000, 21, 2899-903.
18.
Silva, M., MEKC: an update focusing on practical aspects. Electrophoresis
2007, 28, 174-92.
19.
Dawson, R. M. C.; Elliott, D. C.; Elliott, W. H.; Jones, K. M.; Editors, Data
for Biochemical Research. 1959.
20.
Priego-Capote, F.; Luque de Castro, M. D., Determination of B2 and B6
vitamers in serum by capillary electrophoresis-molecular fluorescencecharge coupled detector. Electrophoresis 2005, 26, 2376-2383.
105
21.
Quirino, J. P.; Terabe, S., Electrokinetic chromatography. J. Chromatogr.,
A 1999, 856, 465-82.
22.
Quirino, J. P.; Terabe, S., Large volume sample stacking of positively
chargeable analytes in capillary zone electrophoresis without polarity
switching: use of low reversed electroosmotic flow induced by a cationic
surfactant at acidic pH. Electrophoresis 2000, 21, 355-9.
23.
Quirino, J. P.; Terabe, S.; Bocek, P., Sweeping of neutral analytes in
electrokinetic chromatography with high-salt-containing matrixes. Anal.
Chem. 2000, 72, 1934-43.
24.
Nishi, H.; Tsumagari, N.; Kakimoto, T.; Terabe, S., Separation of watersoluble vitamins by micellar electrokinetic chromatography. J.
Chromatogr., A 1989, 465, 331-43.
25.
Budavari, S.; Merck & Co., The Merck index : an encyclopedia of
chemicals, drugs, and biologicals. 12th ed.; Merck: Whitehouse Station,
N.J., 1996.
26.
Reijenga, J. C.; Verheggen, T. P. E. M.; Martens, H. H. P. A.; Everaerts, F.
M., Buffer capacity, ionic strength and heat dissipation in capillary
electrophoresis. J. Chromatogr., A 1996, 744, 147-153.
27.
Brown, P. R.; Grushka, E.; NetLibrary Inc., Advances in chromatography.
Vol. 40. In Advances in chromatography v. 40, 1st ed.; Marcel Dekker:
New York, 2000; xxiv, 651 p.
28.
Pyell, U., Electrokinetic chromatography: theorey, instrumentation and
application. John Wiley & Sons, Ltd: 2006.
29.
Terabe, S.; Otsuka, K.; Ichikawa, K.; Tsuchiya, A.; Ando, T., Electrokinetic
separations with micellar solutions and open-tubular capillaries. Anal.
Chem. 1984, 56, 111-13.
30.
Terabe, S.; Otsuka, K.; Ando, T., Electrokinetic chromatography with
micellar solution and open-tubular capillary. Anal. Chem. 1985, 57, 834-41.
31.
Poole, S. K.; Poole, C. F., Characterization of surfactant selectivity in
micellar electrokinetic chromatography. Analyst 1997, 122, 267-274.
32.
Hubert, P.; Nguyen-Huu, J. J.; Boulanger, B.; Chapuzet, E.; Chiap, P.;
Cohen, N.; Compagnon, P. A.; Dewe, W.; Feinberg, M.; Lallier, M.;
Laurentie, M.; Mercier, N.; Muzard, G.; Nivet, C.; Valat, L., Harmonization
of strategies for the validation of quantitative analytical procedures. A
SFSTP proposal--Part I. J. Pharm. Biomed. Anal. 2004, 36, 579-86.
106
33.
Hubert, P.; Nguyen-Huu, J. J.; Boulanger, B.; Chapuzet, E.; Chiap, P.;
Cohen, N.; Compagnon, P. A.; Dewe, W.; Feinberg, M.; Lallier, M.;
Laurentie, M.; Mercier, N.; Muzard, G.; Nivet, C.; Valat, L.; Rozet, E.,
Harmonization of strategies for the validation of quantitative analytical
procedures. A SFSTP proposal--part II. J. Pharm. Biomed. Anal. 2007, 45,
70-81.
34.
Hubert, P.; Nguyen-Huu, J. J.; Boulanger, B.; Chapuzet, E.; Cohen, N.;
Compagnon, P. A.; Dewe, W.; Feinberg, M.; Laurentie, M.; Mercier, N.;
Muzard, G.; Valat, L.; Rozet, E., Harmonization of strategies for the
validation of quantitative analytical procedures: a SFSTP proposal part IV.
Examples of application. J. Pharm. Biomed. Anal. 2008, 48, 760-71.
35.
Validation of analytical procedures: Text and Methodology Q2(R1). URL:
http://www.ich.org. In International Conference on Harmonisation: 2005.
36.
Vial, J.; Jardy, A., Experimental Comparison of the Different Approaches
To Estimate LOD and LOQ of an HPLC Method. Anal. Chem. 1999, 71,
2672-2677.
37.
Armbruster, D. A.; Pry, T., Limit of blank, limit of detection and limit of
quantitation. Clin. Biochem. Rev. 2008, 29 Suppl 1, S49-52.
107
Chapter 4: Characterization of monoclonal antibody using capillary sodium
dodecyl sulfate gel electrophoresis
4.1
Introduction
Immunoglobulin G (IgG) is a major class of antibody molecules and one of the
most abundant proteins in the blood serum [1]. IgG consists of two identical heavy
chains and light chains with approximate molecular weight of 150,000 Daltons [1]
(Figure 4.1). The two heavy chains are linked to each other and to a light chain each
by disulfide bonds. The resulting tetramer has two identical halves which together
form the Y-like shape. Each end of the fork contains an identical antigen binding
site. IgG is used for various purposes in medical science because its interaction with
a particular antigen is highly specific. CNTO3157 (mAb) is an engineered human
IgG (IgG4) monoclonal antibody. The amino acid sequence of mAb is still under
investigation. It is known that the constant domain of heavy chains that links to the
light chains through disulfide bond (hinge region) is at position 131 in the amino
acid sequence. Like any IgG, glycosylation was also observed on the heavy chains
of mAb. mAb is a toll-like receptor three (TLR3) antagonist. There are
approximately ten to fifteen types of TLRs in most mammalian species [2]. Thirteen
TLRs (named simply TLR1 to TLR13) have been identified in humans and mice
together, and equivalent forms of many of these have been found in other
mammalian species [3, 2]. TLR3 is broadly expressed throughout the pulmonary
airway and, when unregulated, triggers and perpetuates inflammation [3]. The
primary indication of interest for mAb1 is asthma, and secondary indications under
consideration include sarcoidosis, inflammatory bowel disease, rheumatoid arthritis,
and chronic obstructive pulmonary disease.
108
Figure 4.1 Schematic diagram of Immunoglobulin (IgG) [1]. Fab: antigen-binding
fragment generated by proteolysis with papain; Fc: "crystallizable" fragment of
generated by proteolysis with papain. Reproduced with permission from Lehninger
Principles of Biochemistry, 5th edition, ©Copyright (2008) W. H. Freeman and
Company.
109
Like any recombinant monoclonal antibody, mAb must be well characterized and
subjected to rigorous analytical testing for release prior to commercialization as
required by the regulatory agencies [4]. Size exclusion high performance liquid
chromatography (SEC) is a common method for protein characterization based on
its molecular weight [5]. However, due to the complexity of protein structure,
orthogonal methods are necessary for thorough protein characterization.
In 1960s, the discovery that the molecular weight of proteins can be determined by
sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE)
revolutionized protein characterization and boosted the use of electrophoresis [6-7].
Over the decades, SDS-PAGE has become one of the most extensively used
analytical techniques for characterization of complex protein mixtures [8-10]. SDSPAGE has substantial resolving power with good sensitivity especially with silver
staining technology. Although highly sensitive, SDS-PAGE procedure suffers from
some limitations including manual operation, long run time, and inaccurate
quantification [11-12]. Capillary gel electrophoresis (CGE) is the automated and
instrumental version of slab-gel electrophoresis. The advantages of CGE over slab
gel electrophoresis are automated sample injection, on-line detection, softwarebased data processing, high efficiency, and high throughput. Different modes of CE,
such as capillary zone electrophoresis, micellar electrokinetic chromatography, and
capillary gel electrophoresis (CGE), etc., are available for various applications.
CGE has become an important separation technique in analytical biochemistry and
molecular biology [13]. In 1983 Hjerten successfully transitioned slab gel SDSPAGE to CGE format (cSDS) using a 150 µm inner diameter capillary filled with
110
crosslinked polyacrylamide [14]. Since the introduction of cSDS technique,
numerous applications and developments have been reported [15-42]. The
separation mechanism for cSDS is similar to SDS-PAGE except for the sieving gel
matrix due to the use of capillary. For both cSDS and SDS-PAGE, proteins are
denatured in SDS to produce SDS-protein complexes before analysis. The
formation of SDS-protein complexes is achieved through hydrophobic bindings at a
ratio of 1.4 g of SDS per gram of protein [43]. This binding ratio is relatively
independent of the protein sequence when its molecular weight is greater than
15,000 Daltons. As a result, the native charge of protein is mostly masked by the
negative charge of SDS, so they can be separated based on their molecular weight
differences [44]. The mobility of SDS-protein complexes is inversely proportional
to the logarithm of their effective molecular weight [7]. When a voltage (reverse
polarity) is applied to the capillary, all SDS-protein complexes will migrate towards
the anode due to their negative charges contributed by SDS.
The replacement of slab gel with cSDS involves a change of sieving matrix. In
cSDS, noncrosslinked linear polymer networks are commonly used due to their low
viscosity or high flexibility [45]. Linear polymer network system can be replaced
after each analysis simply by rinsing the polymer network from the capillary via
pressure or vacuum. Replacing “gels” not only avoids contamination and extends
life time of the capillary but also improves precision and robustness of the assay.
Many different polymer networks have been used in cSDS applications [46-48]. Liu
et al. at Beckman Coulter developed a replaceable polymer matrix for an IgG purity
assay based on the previous work done by Demorest and Karger [47]. Other
111
polymer matrix based separation gels are also commercially available from some
manufacturers. These separation gels are gaining popularity in biopharmaceutical
laboratories due to their ease of use, but only limited applications of cSDS on
monoclonal antibody have been reported [22, 29, 36]. There have been some ongoing efforts in the biopharmaceutical industry to adopt the cSDS technique in
quality control (QC) laboratories as a purity assay for therapeutic protein and
antibody product release [43]. However, many challenges such as assay robustness
and reproducibility, were encountered during the method validation and
implementation using commercially available assay kits [49]. Although other cSDS
methods using laser-induced fluorescent detection for protein characterization have
been reported [50-53], they require chemical derivatization, which added
complexity to the method, are therefore not widely accepted in QC environment in
biopharmaceutical industry. It is of great interest to further investigate this method
and make it more robust, reproducible, and QC friendly.
In this study, the cSDS assay was optimized for the analysis of mAb drug substance
under reduced and non-reduced conditions. Some of the sample preparation
parameters including sample buffer pH, incubation temperature and time, alkylation
conditions and reduction conditions were investigated. The optimal sample
preparation conditions were established and the method was qualified for potential
QC release.
4.2
Materials and methods
4.2.1
Reagents and solutions
112
All chemicals and reagents used in this study were of analytical grade. 2mercaptoethanol (2-Me), carbonic anhydrase, citric acid monohydrate,
iodoacetamide (IAM), iodoacetic acid (IAA), N-Ethylmaleimide (NEM), bovine
serum albumin (BSA), β-galactosidase and 10% SDS stock solution were
purchased from Sigma-Aldrich (St. Louis, MO, USA). Phosphoric acid (H3PO4),
methanol (MeOH), sodium phosphate dibasic dihydrate, and sodium chloride
(NaCl) were provided by J. T. Baker (Phillipsburg, NJ, USA). Sodium hydroxide
(NaOH) solution (0.1 M), hydrochloric acid solution (HCl) (0.1 M), sieving gel, 10
kD internal standard (5 mg/ml protein in 0.5% SDS, 0.2% sodium azide), and
sample buffer (0.1 M tris-HCl, 1.0% w/v SDS, pH 9.0) were provided by Beckman
Coulter (Fullerton, CA, USA). The SDS gel buffer creates a physical gel of an
entangled polymer network for separation of the SDS-protein complexes. The gel
buffer comprises a proprietary polymer buffer formulation (at pH 8.0) with 0.2%
SDS. Water with a conductivity value lower than 5 μS/cm was obtained using a
Milli-Q water purification system from Millipore (Billerica, MA, USA). mAb
formulated bulk (FB) (50 mg/mL) and mAb research reference standard (RRS) (10
mg/mL) were manufactured in house (Centocor R&D, Malvern, PA).
4.2.2
Preparation of solutions
A 0.25 M iodoacetamide (IAM) solution was used as the alkylation reagent. The
solution was prepared by dissolving 46 mg of IAM in 1 mL of water. A 0.25 M
IAA and a 0.25 M NEM solutions were also prepared for comparability studies.
Stock solutions of carbonic anhydrase (5 mg/mL), BSA (5 mg/mL), and βgalactosidase (5 mg/mL) were prepared by dissolving 5 mg of respective material in
113
each of 1 mL water. The 10 kD internal standard (1 mg/mL) was prepared by
diluting the stock standard (5 mg/mL) 5-fold in water. A 2-Me diluted solution was
prepared by 5-fold dilution of 2-Me in water. A 0.1 M citric acid was prepared by
dissolving 2.1 g of citric acid monohydrate in 100 mL of water. A 0.2 M sodium
phosphate was prepared by dissolving 3.56 g of sodium phosphate dibasic dihydrate
in 100 mL of water. Some sample buffers at various pH and concentrations were
prepared by adding appropriate 0.2 M sodium phosphate and 0.1 M citric acid and
10 mL of 10% SDS stock solution and diluting in water to each 100 mL of the total
volume.
4.2.3
Preparation of samples
mAb formulated bulk (50 mg/mL) was diluted to 10 mg/mL with deionized water
before subsequent preparation. Samples were diluted to a final concentration of 2.5
mg/mL per the following procedure: 25 µL of the sample solution (10 mg/mL) was
combined with 10 µL of IAM (non-reduced) or 2-ME (reduced), 10 µL of 10,000
Dalton internal reference standard, 55 µL of the sample buffer to a total volume of
100 µL. A system suitability sample was prepared by mixing 5 µL of each BSA,
carbonic anhydrase, β-galactosidase, 2.5 µL of mAb research reference standard, 10
µL of IAM, 10 µL of 10,000 Dalton internal reference standard, and 57.5 µL of
sample buffer. Each of the samples and system suitability solution was mixed, spun
down and incubated at 65°C in a water bath for 5 minutes. Each mixture was cooled
to room temperature, centrifuged at 5000 rpm for 5 minutes, and 90 µL was
transferred from each mixture to a sample vial for injection.
114
4.2.4
Capillary SDS (cSDS) gel electrophoresis
The cSDS experiments were performed on a Beckman ProteomeLab PA800 CE
system equipped with a photo diode array detector and 32 Karat data acquisition
software from Beckman Coulter. Bare fused-silica capillaries (30.2 cm x 50 μm id)
were also supplied by Beckman Coulter. The detection window was placed at 10.2
cm from the outlet of the capillary. The capillary temperature was maintained at 25
°C for all the experiments. A voltage of 15 kV (reverse polarity) was applied
during electrophoretic separations. Samples were injected electrokinetically at -5
kV for 15 seconds. Detection was performed at 220 nm. Data were collected at a
sampling rate of 5 Hz. The current was also monitored to ensure that the Joule
heating was below the upper limit in an Ohm’s law plot.
4.2.5
Capillary conditioning
New capillaries were conditioned by rinsing with 0.1 M NaOH for 10 minutes, 0.1
M HCl for 5 minutes, water for 5 minutes, and running buffer for 10 minutes using
70 psi pressure. After the rinse, a negative voltage of 15 kV was applied to the
capillary filled with running buffer for 5 minutes. To ensure reproducibility,
between injections, the capillary was rinsed with 0.1 M NaOH for 2 minutes, 0.1 M
HCl for 1 minute, water for 1 minute, and running buffer for 10 minutes using 70
psi pressure.
115
4.3 Results and discussion
4.3.1
Effect of sample buffer pH
Sample buffer pH has a great effect on fluorescent labeling for the cSDS analysis of
monoclonal antibody because the derivatization rate is highly dependent on the
acid-base properties of the target sites [51, 53]. It was necessary to study the effect
of sample buffer pH on reduced and non-reduced mAb without fluorescent labeling.
In addition, from the size exclusion HPLC analysis, the area percentage (A%) for
intact IgG analyzed under native conditions (no sample treatment) was 98.5%, and
the largest impurity was 0.8%, while the A% of the intact IgG for non-reduced mAb
obtained from the cSDS analysis using the Beckman sample buffer (pH 9.0) was
only 96.4% and the largest impurity observed in the cSDS analysis was 2.3% (Table
4.1). The disagreement on the results from these two orthogonal methods indicates
that the impurities may be related to cSDS artifacts due to the sample preparation
using high pH sample buffer.
In order to analyze intact IgG molecule (i.e., unfragmented and unaggregated), it is
necessary to block the free thiol (SH) groups of cysteine to prevent the molecules
from forming inter-molecular disulfide bonds and to reduce the presence of free
light chain and heavy chain fragments produced through the reduction of the
existing disulfide bonds by nearby thiol groups of cysteine residues [54]. Disulfide
bonds are usually formed from the oxidation of thiol groups:
2 R-SH → R-S-S-R + 2H+ + 2e-
(1)
Reversible thiol/disulfide exchange reactions occur by the nucleophilic attack of a
thiol on one of the two sulfurs of a disulfide [55]:
R-S-S-R + R-'SH ⇔ R-'S-S-R + R-SH
(2)
116
Table 4.1 Area percentage of IgG and Impurity using sample buffers with different
pH and varied ionic strength
IgG Peak
Area
IgG
Area%
Impurity
Peak Area
Impurity
Area%
16474*
98.47*
140*
0.83*
100 mM Tris-HCl, 1.0%SDS, pH 9.0
(Original Beckman buffer)
51972
96.35
1227
2.27
100 mM Tris-HCl, 1.0%SDS, pH 7.5
(Beckman Buffer with pH adjusted)
17283
98.34
224
1.27
50 mM citrate-phosphate, 1.0%SDS,
pH 6.75
26126
98.22
319
1.20
75 mM citrate-phosphate, 1.0%SDS,
pH 6.75
12702
98.35
152
1.18
Sample Buffer
Size exclusion HPLC*
*Courtesy of Centocor R&D
117
Thiol groups may be blocked by alkylation using alkylating reagents such as
iodoacetic acid (IAA, pKa 3.12), IAM, or N-ethylmaleimide (NEM). IAM is a
widely used reagent for alkylation of cysteine [55]. The alkylation reactions with
IAA and IAM can be written as:
RCH2-SH + I-CH2-COOH (IAA) → RCH2-S-CH2-COOH + H+ + I-
(3)
RCH2-SH + I-CH2-CONH2 (IAM) → RCH2-S-CH2-CONH2 + H+ + I-
(4)
As can be seen from the reaction schemes above, alkylation and disulfide bond
formation are favored under basic conditions. The protonated thiol (-SH) is
unreactive, i.e., thiols cannot attack disulfide bonds, only thiolates. Hence, thioldisulfide exchange is inhibited at low pH (< 7.0) where the protonated thiol form is
favored relative to the deprotonated thiolate form (The pKa of a typical thiol group
is roughly 8.3.).
For cSDS analysis, the alkylation reactions (3) and (4) are preferable, and disulfide
bond or thiol-disulfide exchange reactions (1) and (2) are undesirable. For example,
a basic sample buffer condition (pH > 7.0) will favor both alkylation and disulfide
bond formation and thiol-disulfide exchange. On the other hand, an acidic sample
buffer (pH < 7.0) will prevent disulfide bond formation and thiol-disulfide
exchange but may hinder alkylation depending on the reaction kinetics. It was
reported that at pH 8.0, the second order rate constant for reaction of a typical
alkylation of thiol with IAM was 4.6 M-1s-1[56], while for thiol-disulfide exchange
reaction the second order rate was 8600 M-1s-1 [57]. Because of the slower
alkylation reaction rate, using a high concentration of alkylating agent does not
ensure adequate suppression of the thiol/disulfide exchange [56]. The large
118
difference on reaction kinetics between alkylation and thiol/disulfide rearrangement
indicates that an acidic sample buffer may be used without affecting the alkylation
reaction.
To determine the effect of sample buffer pH, mAb formulated bulk samples were
prepared and analyzed using the following buffers: (1) 100 mM Tris-HCl, pH 9.0
(Beckman buffer); (2) 100 mM Tris-HCl, pH 7.5 (Beckman buffer with pH
adjusted); (3) 50 mM citrate-phosphate, pH 6.75; (4) 75 mM citrate-phosphate, pH
6.75. The samples were alkylated using 10 µL IAM and incubated at 65 °C for 2
minutes. The results are shown in Table 4.1. By decreasing the sample buffer pH
from 9.0 to 7.5, the area percentage (A%) of IgG was increased from 96.4% to
98.3%. It was observed that the peak areas varied when using different buffer
solutions. This was due to the difference in conductivity of each sample buffer.
For electrokinetic injection, the quantity injected, Qinj, is given by [58]:
⎛ V ⎞⎛ k
Qinj = μ app ⎜ ⎟⎜⎜ b
⎝ L ⎠⎝ k a
⎞ 2
⎟⎟tπr C
⎠
where μapp is the apparent mobility of analyte, V is the voltage, L is the total
capillary length, r is the capillary radius, C is the sample concentration, kb is the
conductivity of the running buffer, and ka is the conductivity of the sample solution.
The ions injected into a capillary are inversely proportional to the conductivity of
the sample solution when the running buffer remains the same.
To confirm the pH effect on purity, more citrate-phosphate buffers were prepared,
and the conductivity was maintained at the same level by adjustment with 1 M
sodium chloride. When the pH was decreased from 8.0 to 6.5, the A% increased
119
about 0.4%. Further lowering the pH from 6.5 to 5.5, the A% of IgG did not change
significantly (Table 4.2). Based on the experimental data, the sample buffer with pH
6.5 was chosen for the future experiments. Under pH 6.5, over 98% of the thiol
(pKa ~ 8.3) groups remain protonated, which will prevent disulfide bond formation
effectively. The effect of sample buffer pH on cSDS separation of non-reduced
mAb is shown in Figure 4.2.
4.3.2
Alkylation condition
Common alkylating agents are IAM, IAA, or NEM. The choice of reagent is
governed by applications. Although NEM has been reported to have greater
reactivity than that of IAM and IAA [56], equivalent results were obtained between
NEM and IAM in this study. However, IAA produced more fragments than IAM.
The alkylation conditions with IAM were then further examined by incubating mAb
at 65°C for 2 minutes using various volumes of 0.25 M IAM. It was found that the
MAB1 purity increased as the volume of 0.25 M IAM increased from 0 µL to 10
µL, which confirmed that blocking thiol groups by alkylation was necessary.
Further increases in the volume of 0.25 M IAM did not change the mAb purity.
Reproducible results for measurements of purity were obtained from duplicate
sample preparations. Based on these results, 10 µL of 0.25 M IAM was a choice of
alkylation condition for this application (see Table 4.3).
4.3.3
Incubation temperature and time
Sample incubation temperature and time were evaluated under reduced and nonreduced conditions. Under reduced conditions, samples were incubated at 65 °C,
120
Table 4.2 Area percentage of IgG and Impurity using sample buffers with different
pH with the same conductivity
Sample Buffer
25 mM citrate-phosphate,
1.0%SDS, pH 5.5
25 mM citrate-phosphate,
1.0%SDS, pH 6.0
25 mM citrate-phosphate,
1.0%SDS, pH 6.5
25 mM citrate-phosphate,
1.0%SDS, pH 7.0
25 mM citrate-phosphate,
1.0%SDS, pH 8.0
IgG Peak
Area
IgG Area%
Impurity
Peak Area
Impurity
Area%
40638
98.39
408
1.00
49388
98.28
521
1.04
49483
98.30
512
1.02
52678
97.91
688
1.28
52923
97.91
687
1.27
121
3
0.10
Absorbance at 210 nm
0.08
0.06
1
2
A
0.04
0.02
B
0.00
-0.02
0
5
10
15
20
25
30
35
Minutes
Figure 4.2 Effect of sample buffer pH on cSDS separation of non-reduced mAb.
Peak identification: No. 1: internal standard with molecular weight of 10,000
Dalton; No. 2: largest impurity peak; No. 3: mAb (IgG). Peaks earlier than No.1
were from sample buffer blank. Other impurity peaks were not labeled but
integrated for total peak areas.
A: 25 mM citrate-phosphate sample buffer at pH 6.5. Relative area (%) of IgG
(peak No.3) was 98.3% and relative area of impurity (peak No. 2) was 1.0%;
B: Original Beckman sample buffer (100 mM Tris-HCl, 1.0% SDS, pH 9.0).
Relative area (%) of IgG (peak No.3) was 96.4% and relative area of impurity (peak
No. 2) was 2.3%.
Separation conditions: 30 cm x 50 µm bare silica, electrokinetic injection (-5 kV for
15 s), separation voltage: -15 kV for 35 minutes.
122
Table 4.3 Optimization of alkylating agent (IAM) concentration
0
Purity (%)
(n=2)
98.04
Average Peak Area of mAb
(n = 2)
38529
5
99.06
38731
10
99.23
39231
15
99.28
40001
20
99.28
38691
0.25 M IAM (μL)
123
70°C, and 75°C for 2 minutes and 5 minutes. Under non-reduced conditions,
samples were incubated at 60 °C, 65 °C, 70 °C, and 75 °C for 2 minutes and 5
minutes. Each sample was analyzed in duplicate and the average results were shown
in Table 4.4. The results showed that incubation at 65 °C for 5 minutes resulted in
high purity with little fragmentation for reduced samples. For non-reduced samples,
incubation at 65 °C for 2 minutes and 5 minutes yielded comparable results based
on the t-test analysis (tcalculated = 0.9, while ttabulated, 95% confidence = 2.6)[59]. Therefore,
65 °C for 5 minutes was chosen as an incubation condition for both reduced and
non-reduced samples.
4.3.4
Reduction conditions
Reduction conditions were studied with 2-mecaptoethanol (2-ME) as reducing
agent. mAb FB was reduced using 10 µL of 2-ME at various concentrations diluted
in water. The samples were incubated at 65 °C for 5 minutes. As shown in Table
4.5, adequate reduction was obtained using 10 µL of 5-fold diluted 2-ME, which is
equivalent to 2 µL of pure 2-ME.
4.3.5
Characterization of the aglycosylated heavy chain (AGHC)
Aglycosylation (none glycosylated) in heavy chain has been previously observed
during the analysis of monoclonal antibodies under reduced conditions [60].
Although mAb is a glycoprotein and the aglycosylated heavy chain (AGHC) is
considered to be part of the product, the amounts of AGHC need to be controlled to
ensure product consistency. The AGHC peak appeared at approximately 56,500
Daltons and migrated earlier than the heavy chain. During the method development
124
Table 4.4 Comparison of mAb purity under different incubation conditions for
reduced and non-reduced samples. NA: Not Available
Temperature
(°C)
Time
(min)
A% of
IgG
HC1
Peak
Area
LC2
Peak
Area
Purity
(Reduced)
(%)
Ratio of
Peak Area
of HC/LC
2
5
2
5
2
5
2
5
98.76
98.59
98.77
98.54
98.41
98.20
98.50
98.19
NA
NA
32581
30087
24589
30739
31342
32245
NA
NA
14071
13105
12758
13534
13926
14042
NA
NA
97.64
98.54
83.35
96.03
95.19
98.38
NA
NA
2.32
2.30
1.93
2.27
2.25
2.30
60
65
70
75
1
2
HC: Heavy Chain
LC: Light Chain
125
Table 4.5 Effect of reduction conditions on the analysis of reduced mAb
1
2-ME dilution
Purity (%) (n = 2)
1:40
1:30
1:25
1:20
1:10
1:5
93.56
96.58
98.06
98.75
99.24
99.48
HC: Heavy Chain;
LC: Light Chain
Ratio of Peak Area of
HC/LC1 (n = 2)
2.15
2.19
2.23
2.20
2.19
2.18
126
of cSDS assay for mAb, a peak prior to the heavy chain was observed in the
electropherograms of reduced mAb research reference standard samples. This peak
was characterized to be an AGHC peak using the experiments described below.
Reduced mAb heavy chain contains a majority of glycoproteins. By using
deglycosylation agent, some of the oligosaccharides can be removed from the heavy
chain. As a result, the amount of deglycosylated heavy chain (AGHC) will increase.
In this experiment, Glyko enzyme Peptide N-Glycanase (PNGase F) was used as a
deglycosylation agent to remove oligosaccharides from glycoproteins. This PNGase
F is a specific enzyme that cleaves N-linked glycans from glycoprotein. A PNGase
F treated sample was prepared by adding 5 µL of 5 U/mL PNGase F to 100 µL of
reduced sample followed by incubation at 37 ºC overnight. Upon completion of the
incubation, a spiked sample was prepared by adding 10 µL, 15 µL, and 20 µL of the
PNGase F treated sample to each non-treated sample. The PNGase treated sample
was analyzed along with non-treated and spiked samples. The recovery data and
migration time data confirmed that the peak that migrated prior to the heavy chain
was from AGHC (see Figure 4.3 and Table 4.6).
4.3.6
Summary of optimized parameters of the cSDS method
127
0.16
3
0.14
Abso rba nce a t 2 2 0 nm
0.12
0.10
2
0.08
1
A
0.06
B
0.04
0.02
0.00
-0.02
12
13
14
15
16
17
18
19
20
21
22
23
Minutes
Figure 4.3 PNGase F treated reduced mAb vs. non-treated reduced mAb 1: light
chain; 2: Aglycosylated heavy chain; 3: heavy chain
A: reduced mAb not being treated by PNGase F; B: reduced mAb treated by
PNGase F.
Separation conditions: 30 cm x 50 µm bare silica, electrokinetic injection (-5 kV for
15 s), applied separation voltage: -15 kV for 35 minutes, buffer: 25 mM citratephosphate, 1.0% SDS, pH 6.5.
24
128
Table 4.6 Spike recovery of PNGase F treated samples
NonTreated_1
NonTreated_2
PNGase
Treated_1
PNGase
Treated_2
Spike 10_1*
Spike 10_2
Spike 15_1*
Spike 15_2
Spike 20_1*
Spike 20_2
1
HC1
Area
HC A%2
AGHC3
Area
AGHC%
Theoretical
(A%)
Recovery
(%)
32922
91.69
2985
8.31
NA
NA
32424
91.72
2926
8.28
NA
NA
13398
42.09
18433
57.91
NA
NA
11758
41.38
16656
58.62
NA
NA
30434
29379
24578
28039
27887
27900
86.84
86.66
84.47
84.39
81.95
81.97
4614
4523
4517
5187
6144
6137
13.16
13.34
15.53
15.61
18.05
18.03
14.12
14.12
17.03
17.03
19.95
19.95
93.2
94.5
91.1
91.6
90.5
90.4
HC: Heavy Chain
A%: Area%
3
AGHC: Aglycosylated Heavy Chain
*Spike 10, Spike 15, and Spike 20 are spike levels of 10%, 15%, and 20% of
AGHC, respectively. Each spike level was prepared in duplicated.
2
129
The optimized conditions for mAb purity analysis using the cSDS method are
summarized in Table 4.7. Figure 4.4 is a typical cSDS electropherogram for system
suitability.
4.4
Method validation
The optimized method was validated for specificity, accuracy, precision, limit of
quantitation (LOQ), linearity and range, and sample stability. System suitability was
run prior to the samples in each sequence.
4.4.1
Specificity
Specificity of the test method for mAb was demonstrated through matrix
interference of mAb formulation buffer. Reduced and non reduced mAb FB, mAb
formulation buffer, and water samples were prepared and analyzed. No detectable
peaks were found in the formulation buffer and water, indicating that this test
method has no interference with the sample matrix. Figures 4.5-4.6 are the
electropherograms of mAb FB, mAb formulation buffer, and water under nonreduced and reduced conditions, respectively.
The matrix effect on peak area and relative migration time (RMT) was also studied.
The RMT was calculated against the internal standard peak. A mAb FB sample was
diluted in water and formulation buffer and analyzed under reduced and nonreduced conditions. As shown in Tables 4.8 and 4.9, the matrix had no effect on
peak area, RMT, and purity for both reduced and non-reduced conditions.
130
Table 4.7 Summary of optimized method parameters
Non-reduced
Reduced
mAb sample (10 mg/mL)
25 μL
25 μL
Internal standard
(1 mg/mL)
10 μL
10 μL
0.25 M IAM
10 μL
2-ME (1:5 diluted)
10 μL
SDS sample buffer:
25 mM citrate-phosphate, 1.0%
SDS, pH 6.5
55 μL
55 μL
65°C for 5 min
65°C for 5 min
- 5 kV for 15 s
- 5 kV for 15 s
Incubation temperature and time
Injection
131
0.03
A bs o r ba nc e a t 2 2 0 nm
4
2
0.02
5
3
1
0.01
0.00
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
25.0
27.5
30.0
32.5
Minutes
Figure 4.4 Typical electropherogram of system suitability sample. 1.internal
reference standard (10 kDa); 2. carbonic anyhydrase (31 kDa); 3. bovine serum
albumin (66 kDa); 4. β-galactosidase (116 kDa); 5. mAb RRS (148 kDa)
Separation conditions as in Figure 4.3.
35.0
132
0.10
3
A b so rb a n ce a t 2 2 0 n m
0.08
1
0.06
A
0.04
B
0.02
C
0.00
0.0
2.5
5.0
2
7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5 35.0
Minutes
Figure 4.5 Matrix effect of non-reduced mAb. A: water blank; B: formulation
buffer; C: mAb sample. Peak identification 1: internal standard; 2: impurity; 3:
mAb (IgG). Other peaks prior to 1 are from the sample buffer. Separation
conditions as in Figure 4.3.
133
3
A b so r b a n c e a t 2 2 0 n m
0.15
0.10
2
1
A
0.05
B
C
0.00
0.0
2.5
5.0
7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5 35.0
Minutes
Figure 4.6 Matrix effect of reduced mAb. A: water blank; B: formulation buffer; C:
mAb sample. Peak identification 1: internal standard; 2: light chain; 3: heavy chain.
Separation conditions as in Figure 4.3.
134
Table 4.8 Matrix effect study of non-reduced mAb
IgG
Diluent
Purified water
mAb formulation
buffer
Peak
area
RMT1
Diff% in RMT
IgG%
Diff% in IgG%
2.099
Control
98.36
Control
45313
2.095
0.1
98.20
0.2
45552
Table 4.9 Matrix effect study of reduced mAb
LC2
Diluent
Purified
water
mAb
formulation
buffer
1
RMT
Diff%
in RMT
1.223
Control
1.223
0.0
HC3
Peak
area
1438
6
1394
1
RMT
Diff% in
RMT
1.532
Control
1.533
0.1
Purity
Peak
area
3365
2
3266
5
HC/LC
Ratio
Purity
(%)
Diff% in
Purity
99.03
Control
2.3
99.19
0.2
2.3
RMT: Relative migration time (analyte migration time / internal standard migration
time)
2
LC: Light chain
3
HC: Heavy chain
135
4.4.2
Accuracy
The accuracy of an analytical procedure expresses the closeness of agreement
between the “true” value, which is accepted either as a conventional true value or an
accepted reference value and the value found [61]. Accuracy of mAb purity was
evaluated by measuring peak areas of IgG at its concentrations of 0.25-3.0 mg/mL,
which correspond to 10-120% of nominal concentration (2.5 mg/mL). Appropriate
dilutions were made in water from a mAb (50 mg/mL) sample. Reduced and nonreduced samples each were prepared and analyzed in triplicate. Table 4.10 and
Table 4.11 showed peak areas of IgG and heavy chain and light chain under nonreduced and reduced conditions, respectively.
Accuracy was determined as follows. For non-reduced sample, the IgG purity (%)
from each sample concentration was compared to that from the nominal sample
concentration of 2.5 mg/mL. For reduced sample, the purity (%) of HC and LC
from each sample concentration was compared to that from the nominal sample
concentration of 2.5 mg/mL. Accuracy was expressed as error% and calculated
according to the formula:
Error (%) =100 x (Measured purity – theoretical purity) / theoretical purity at 2.5
mg/mL
Accuracy results for mAb samples are shown in Tables 4.10 and 4.11. It was
observed that accuracy of purity was excellent at all concentrations tested.
4.4.3
Precision (repeatability and intermediate precision)
136
Table 4.10 Accuracy of mAb purity under non-reduced conditions
% Normal
Concentration
20
40
80
100
120
Theoretical
Concentration
(mg/mL)
0.50
1.0
2.0
2.5
3.0
Injection #
IgG Peak
Area
1
2
3
Average
RSD (%)
1
2
3
Average
RSD (%)
1
2
3
Average
RSD (%)
1
2
3
7339
7279
6782
7133
4.3
14280
14983
14362
14542
2.6
30389
30350
29455
30065
1.8
36955
35540
35616
Average
36037
RSD (%)
1
2
3
Average
RSD (%)
2.2
41122
36907
41652
39894
6.5
IgG (%)
Relative
Error (%)
97.57
97.93
98.28
-0.23
0.14
0.50
0.4
97.98
97.99
97.71
0.19
0.20
-0.09
0.2
97.79
97.92
98.02
0.00
0.13
0.23
0.1
97.74
97.93
97.71
97.79
(Theoretical)
0.1
97.98
97.82
98.12
NA
NA
NA
NA
0.19
0.03
0.33
137
Table 4.11 Accuracy data for mAb purity under reduced conditions
% Normal
Injection
10
20
50
80
100
120
Theoretical
Concentration Injection #
(mg/mL)
0.25
0.50
1.25
2.00
2.50
3.00
*missing injection
NA: Not Applicable
LC Peak
Area
HC Peak
Area
Purity
(%)
HC:LC
Ratio
Relative
Error (%)
1
1529
3481
100.0
2.28
0.51
2
3
Average
RSD (%)
1
2
3
Average
RSD (%)
1
2
3
Average
RSD (%)
1
2
3
Average
RSD (%)
1
2
3
1498
1233
1420
11.5
2348
2339
2327
2338
0.5
3408
2778
3222
12.0
5339
5385
5305
5343
0.8
100.0
100.0
2.28
2.25
0.51
0.51
99.76
99.78
100.0
2.27
2.30
2.28
0.26
0.28
0.51
7184
6967
7076
NA
10914
10856
16974
16459
16717
NA
25270
25094
*
99.54
100.0
99.80
2.36
2.36
2.36
0.04
0.51
99.50
99.51
*
2.32
2.31
0.00
0.01
10885
NA
13323
13315
13195
25182
NA
30374
30361
30098
2.28
2.28
2.28
NA
NA
NA
Average
13278
30278
99.54
99.43
99.52
99.50
(Theoretic
al)
RSD (%)
1
2
3
Average
RSD (%)
0.5
15868
15555
15300
15574
1.8
0.5
35899
35154
34707
35253
1.7
99.54
99.44
99.62
2.26
2.26
2.27
0.04
-0.06
0.12
138
Repeatability expresses the precision of the procedure under the same operating
conditions over a short interval of time. Repeatability was demonstrated by
preparing six samples in duplicate from the same lot of mAb FB under both nonreduced and reduced conditions. Relative standard deviation (RSD, expressed as
percentage) was calculated on the average values (n=6). For non-reduced samples,
repeatability was calculated on molecular mass, peak area, and IgG purity. For
reduced samples, the RSD was calculated on molecular mass (HC and LC), peak
area (HC and LC), purity, and the HC/LC ratio. The repeatability data under nonreduced conditions are presented in Table 4.12. The RSD of molecular masses, peak
areas, and IgG purities were 1.2 %, 4.6%, and 0.1%, respectively (Day 1). The
repeatability data for reduced conditions are presented in Table 4.13. The RSD of
the molecular masses for the LC and HC were 0.3% and 0.4%, respectively. The
RSD of peak areas for the LC and HC were 1.4% and 1.0%, respectively. The RSD
of purity (%) was 0.3%, and the RSD of HC/LC ratio was 0.5% (Day 1).
Intermediate precision was determined by performing the repeatability study on a
different day. Results from both days were used to determine the intermediate
precision. For non-reduced sample, the RSD (n=12) of molecular masses, peak
areas, and IgG purities were 1.4%, 9.7%, and 0.1%, respectively. For reduced
sample, the RSD (n=12) of molecular mass, peak areas, purities, and HC/LC ratios
were 0.5% (LC) and 1.0 (HC), 3.0% (LC) and 2.7% (HC), 0.2%, and 0.5%,
respectively. Intermediate precision data are shown in Tables 4.12 and 4.13.
4.4.4
Limit of quantitation (LOQ)
139
Table 4.12 Repeatability and intermediate precision data for non-reduced mAb
cSDS assay
Replicate
MW1 (kDa)
IgG Peak Area
IgG (%)
1
154.8
46964
97.87
2
152.9
48090
97.96
3
153.2
46370
98.03
4
150.8
48583
97.77
5
151.5
44479
97.94
6
148.7
50875
97.96
152.0
47560
97.92
1.4
4.6
0.1
1
154.6
52095
97.95
2
152.6
52437
97.88
3
152.0
48281
97.77
4
155.1
36113
97.56
5
150.5
46124
97.78
6
148.6
53699
97.80
152.2
48125
97.79
1.6
13.6
0.1
153.9
47843
97.86
1.4
9.7
0.1
Day 1
Average (n=6)
RSD (n=6) (%)
Day 2
Average (n=6)
RSD (n=6) (%)
Intermediate
Precision
(Two Days)
1
Average (n=12)
RSD (n=12) (%)
MW: Molecular weight
140
Table 4.13 Repeatability and intermediate precision for reduced mAb cSDS assay
LC1
Purity
(%)
HC/LC
36489
98.96
2.31
61.9
36670
98.94
2.30
15961
61.7
36896
99.02
2.31
26.3
16066
61.5
37045
99.11
2.31
5
26.3
15436
61.4
36025
99.82
2.33
6
26.2
15890
61.1
36547
99.11
2.30
26.3
15847
61.6
36612
99.16
2.31
0.3
1.4
0.4
1.0
0.3
0.5
1
26.3
17449
61.1
40032
99.21
2.29
2
26.2
16357
61.0
37605
99.13
2.30
3
26.2
16431
60.7
37557
99.12
2.29
4
26.1
16390
60.5
37559
99.27
2.29
5
26.0
16057
60.2
36966
99.16
2.30
6
26.0
16194
60.1
37331
99.13
2.31
26.1
16480
60.6
37842
99.17
2.30
0.4
3.0
0.7
2.9
0.1
0.3
26.2
16163
61.1
37227
99.17
2.30
0.5
3.0
1.0
2.7
0.2
0.5
Replicate
Day 1
MW (kDa)
LC Peak
Area
MW
(kDa)
HC Peak
Area
1
26.3
15807
61.7
2
26.4
15919
3
26.4
4
Average
(n=6)
RSD (n=6)
(%)
Day 2
Intermediate
Precision
(Two Days)
1
Average
(n=6)
RSD (n=6)
(%)
Average
(n=12)
%RSD
(n=12)
LC: Light chain
MW: Molecular weight
3
HC: Heavy Chain
2
HC3
2
141
The LOQ of the assay was obtained by measuring the diluted mAb as a surrogate
for impurity. A mAb (50 mg/mL) sample was diluted 5-fold to obtain the stock
concentration of 10 mg/mL. Further dilutions were made to obtain the
concentrations from 0.01 mg/mL to 0.5 mg/mL, corresponding to 0.1% to 5% w/w
of the nominal sample concentration of 10 mg/mL. Reduced and non-reduced
samples each were prepared in triplicate and analyzed according to the optimized
conditions. Table 4.14 lists the IgG peak areas at various concentrations for nonreduced samples, and its linear regression data. The relationship between peak area
and theoretical protein concentration was found to be linear, with coefficient of
determination (r2) of 0.9969 for non-reduced mAb (n=18). Table 4.15 lists the
heavy chain and light chain peak areas at various concentrations for reduced
samples, and their linear regression data. The relationship between peak area and
protein concentration was found to be linear (n=18), with coefficient of
determination (r2) of 0.9959 for reduced light chain and 0.9998 for reduced heavy
chain. The observed concentrations of non-reduced IgG (Table 4.14), reduced HC
and LC (Table 4.15) were calculated from their regression plots. Accuracy was
expressed as relative error (%) and calculated according to the formula:
Relative error (%) = 100 x (Measured purity – theoretical purity) / theoretical purity
at 2.5 mg/mL
The HC and LC theoretical concentrations were based on the ratio of HC/LC = 2.3.
The LOQ was determined to be 0.2% of the nominal concentration for both nonreduced and reduced conditions. The LOQ value was based on the acceptable
precision and accuracy data as presented in Tables 4.14 and 4.15.
142
Table 4.14 LOQ data for non-reduced mAb cSDS assay
% of Nominal IgG Theoretical
Conc.
Conc. (mg/mL)
0.10
0.00252
0.20
0.0504
0.40
0.0108
1.00
0.0252
2.00
0.0504
5.00
0.1250
Sample
1
2
3
Average
RSD (%)
1
2
3
Average
RSD (%)
1
2
3
Average
RSD (%)
1
2
3
Average
Difference%
1
2
3
Average
RSD (%)
1
2
3
Average
RSD (%)
IgG Peak
Area
Measured IgG
Conc. (mg/mL)
22
35
28
28
23.0
58
68
68
65
8.9
122
156
163
147
14.9
408
partial
injection
397
403
1.5
797
841
840
826
3.0
0.0023
0.0031
0.0027
0.0027
14.3
0.0045
0.0051
0.0051
0.0049
7.1
0.0083
0.0103
0.0107
0.0098
13.4
0.0253
1985
2050
2207
2081
0.1192
0.1231
0.1325
0.1249
5.5
5.4
0.0247
0.0250
1.2
0.0485
0.0511
0.0510
0.0502
3.0
Relative
Error (%)
8.35
-2.55
-2.24
-0.03
0.43
-0.05
143
Table 4.15 LOQ data for reduced mAb cSDS assay
% of
Nominal
Conc.
0.1
0.2
HC
Theo.
Conc.
(mg/mL)
0.0017
0.0035
0.4
0.0070
1.0
0.0174
2.0
0.0348
5.0
0.0871
Inj. #
1
2
3
Average
RSD (%)
1
2
3
Average
Differenc
e%
1
2
3
Average
RSD (%)
1
2
3
Average
RSD (%)
1
2
3
Average
RSD (%)
1
2
3
Average
RSD (%)
HC
Measured
Conc.
(mg/mL)
NA
NA
NA
NA
NA
0.0039
HC Peak
Area
ND
ND
ND
NA
NA
58
HC
Relative
Error (%)
NA
LC
Theo.
Conc.
(mg/mL)
0.0008
LC
Measured
Conc.
(mg/mL)
LC Peak
Area
LC
Relative
Error (%)
ND
ND
ND
29
*
0.0014
*
62
60
0.0041
0.0040
31
30
0.0015
0.0014
6.7
4.7
6.7
6.6
126
127
128
127
0.8
342
346
338
342
1.2
703
701
701
702
0.2
1830
1811
1842
1828
0.9
0.0071
0.0071
0.0072
0.0071
0.9
0.0173
63
68
72
68
6.7
169
173
150
164
7.5
319
328
323
323
1.4
832
810
810
817
1.6
0.0029
0.0032
0.0034
0.0032
6.6
0.0079
0.0080
0.0070
0.0076
7.5
0.0148
0.0152
0.0150
0.0150
1.4
0.0386
0.0376
0.0376
0.0379
1.6
14.21
0.0015
2.45
0.0030
0.0171
0.0172
0.8
0.0343
-1.36
0.0076
0.0342
0.0343
0.2
0.0875
0.0866
0.0880
0.0874
0.8
-1.68
0.0152
0.28
0.0379
-6.96
4.16
0.63
-0.92
0.11
144
4.4.5
Linearity and range
Linearity was examined for the purity of mAb and its impurities. Linearity for mAb
purity was evaluated using the data from the accuracy study. The linear regression
analysis was performed using the peak areas of non-reduced IgG, reduced HC, and
reduced LC vs. corresponding protein concentrations. The assay was found to be
linear for determining mAb purity in the range of 10% to 120% of the nominal
sample concentration of 2.5 mg/mL. Table 4.16 summarizes the linear regression
equations for mAb purity. Linearity for impurities was evaluated using the data
from the LOQ determination. mAb was employed as a surrogate impurity to assess
the linearity. The linear regression analysis was performed using the peak areas of
non-reduced IgG, reduced HC, and reduced LC vs. corresponding protein
concentrations. The assay was found to be linear for determining impurities in the
range of 0.1% to 5% of the nominal sample concentration of 2.5 mg/mL. Table 4.16
summarizes the linear regression equations for the impurity determination.
The range for the cSDS assay was verified using the data from accuracy and LOQ
study. In addition to the linearity data, accuracy and precision data were used to
determine the range. For mAb purity determination, within the range of 10-120% of
the nominal sample concentration, the accuracy was 100.1-100.2% for non-reduced
samples and 100.0%-100.5% for reduced samples, and the precision was 0.120.36% for non-reduced samples and 0.5-12.0% for reduced samples. For impurity
determination, within the range of 0.1-5.0% of the nominal sample concentration,
the accuracy was 97-108% for non-reduced samples and 93-114% for reduced
145
Table 4.16 Linear regression equations for mAb cSDS assay
mAb
Purity
mAb
Impurity
Description
Reduced HC
Reduced LC
Non-Reduced IgG
Reduced HC
Reduced LC
Non-Reduced IgG
*y: peak area; x: protein concentration
r2
0.9951
0.9971
0.9939
0.9998
0.9959
0.9969
Regression Equation*
y = 11927x + 308
y = 5259x + 60
y = 12996x + 3139
y = 21199x -24
y = 21565x – 0.4
y = 16791x - 17
146
samples, and the precision was 1.2-14.3% for non-reduced samples and 0.8-7.5%
for reduced samples.
4.4.6
Sample stability
Sample stability was studied under two conditions: after three freeze-thaw cycles
and storage of at 2-8 °C. Three aliquots of a mAb sample were initially thawed at
37 °C and one aliquot was tested. The remaining aliquots were frozen at -70 °C for
24 hours and thawed at 37 °C for two more cycles. For each cycle, the sample was
prepared and tested under reduced and non-reduced conditions. The results of the
initial thawed sample are used as a control. Data from the reduced sample for the
third cycle was not available due to instrument pressure failure. The results
demonstrated that mAb was stable after three freeze-thaw cycles under non-reduced
conditions. Under reduced conditions, the mAb was stable after two freeze-thaw
cycles. Prepared sample stability at 2-8 °C was studied on mAb sample. Large
volumes of mAb sample were prepared under reduced and non-reduced conditions
and stored at 2-8 °C for a period of time. The samples were tested at fresh, and after
48 hours of storage. The results demonstrated that prepared mAb sample may be
analyzed after 48 hours of storage at 2-8 °C (see Table 4.18).
4.5
Concluding remarks
This study optimized the cSDS assay for the analysis of mAb drug substance under
reduced and non-reduced conditions. Some of the sample preparation parameters
including sample buffer pH, incubation temperature and duration, alkylation
conditions with iodoacetamide (IAM), and reduction conditions with 2-ME
147
Table 4.17 Stability data for three freeze and thaw cycles of mAb RRS
Reduced Conditions
F/T
Cycle
1
2
3*
LC
MW
(kDa)
26.4
26.5
N/A
Diff.%
N/A
0.4
N/A
LC
Peak
Area
14386
14856
N/A
Diff.%
N/A
3.3
N/A
HC
MW
(kDa)
61.9
62
N/A
N/A
0.2
N/A
HC
Peak
Area
33652
34802
N/A
Diff.%
Diff.%
Purity
(%)
Diff.%
N/A
3.4
N/A
99.03
98.87
N/A
N/A
0.2
N/A
New
Peak ≥
0.10%
None
None
N/A
Non-reduced Conditions
F/T
Cycle
1
2
3
IgG MW
(kDa)
154.1
151.1
154.6
Diff.%
IgG Peak Area
Diff.%
IgG%
Diff.%
N/A
1.9
2.3
45313
44506
45103
N/A
1.8
1.3
98.36
98.65
98.64
N/A
0.3
0.0
*Data not available due to missing injections
New Peak≥
0.10%
None
None
None
148
Table 4.18 Stability data of prepared mAb sample at 2-8° C
Reduced conditions
Sample
LC1
MW2
(kDa)
Diff.%
in LC
MW
HC3 MW
(kDa)
Diff.% in
HC MW
Purity
(%)
Diff.% in
Purity
New peak ≥ 0.10%
Fresh
26.0
NA
60
NA
99.13
NA
None
After 48
hours
25.9
0.4
59.9
0.2
99.13
0.0
None
Non-reduced Conditions
1
Sample
IgG MW (kDa)
Diff.% in MW
IgG%
Diff.% in IgG%
New
peak ≥
0.20%
Fresh
148.7
NA
97.96
NA
None
After 48 hours
148.3
-0.3
97.99
0.0
None
LC: Light chain
MW: Molecular weight
3
HC: Heavy chain
2
149
were investigated. It was observed that the sample buffer at slight acidic conditions
(pH 5.5-6.5) greatly decreased thermally induced fragmentation of non-reduced
mAb. As such, a citrate-phosphate buffer at pH 6.5 was used for sample preparation
to replace the original Beckman sample buffer (pH 9.0). The optimal sample
preparation conditions were established and the method was qualified for potential
QC release. Although the cSDS technique was developed using mAb as a model
compound, it applies to all types of IgG molecules. However, some of the alkylation
conditions (amounts of IAM) and reduction conditions (incubation temperature and
duration) may require minor optimization due to the differences in formulation
buffer compositions.
150
4.6
List of references:
1.
Lehninger, A. L.; Nelson, D. L.;Cox, M. M. Lehninger principles of
biochemistry. 5th ed.; W.H. Freeman: New York, 2008.
2.
Tabeta, K.; Georgel, P.; Janssen, E.; Du, X.; Hoebe, K.; Crozat, K.; Mudd,
S.; Shamel, L.; Sovath, S.; Goode, J.; Alexopoulou, L.; Flavell, R.
A.;Beutler, B. Toll-like receptors 9 and 3 as essential components of innate
immune defense against mouse cytomegalovirus infection. Proc. Natl.
Acad. Sci. U. S. A. 2004, 101, 3516-21.
3.
Choe, J.; Kelker, M. S.;Wilson, I. A. Crystal structure of human toll-like
receptor 3 (TLR3) ectodomain. Science 2005, 309, 581-5.
4.
United States Pharmacopeial Convention., USP pharmacists' pharmacopeia.
In United States Pharmacopeial Convention: Rockville, MD, 2005.
5.
Kenney, A.;Fowell, S. Practical protein chromatography. Humana Press:
Totowa, N.J., 1992.
6.
Shapiro, A. L.; Vinuela, E.;Maizel, J. V., Jr. Molecular weight estimation of
polypeptide chains by electrophoresis in SDS-polyacrylamide gels.
Biochem. Biophys. Res. Commun. 1967, 28, 815-20.
7.
Weber, K.;Osborn, M. The reliability of molecular weight determinations
by dodecyl sulfate-polyacrylamide gel electrophoresis. J. Biol. Chem. 1969,
244, 4406-12.
8.
Kresge, N.; Simoni, R. D.;Hill, R. L. SDS-PAGE to determine the
molecular weight of proteins: the work of Klaus Weber and Mary Osborn.
J. Biol. Chem. 2006, 281, e19.
9.
Rodríguez-Díaz, R.; Wehr, T.;Tuck, S. Analytical techniques for
biopharmaceutical development. M. Dekker: Boca Raton, 2005.
10.
Walker, J. M. Protein protocols handbook. 3rd ed.; Springer
Science+Business Media, LLC: New York, NY, 2009.
151
11.
Switzer, R. C., 3rd; Merril, C. R.;Shifrin, S. A highly sensitive silver stain
for detecting proteins and peptides in polyacrylamide gels. Anal. Biochem.
1979, 98, 231-7.
12.
Zilberstein, G.; Korol, L.; Antonioli, P.; Righetti, P. G.;Bukshpan, S. SDSPAGE under focusing conditions: an electrokinetic transport phenomenon
based on charge compensation. Anal. Chem. 2007, 79, 821-7.
13.
Hames, B. D. Gel electrophoresis of proteins : a practical approach. 3rd
ed.; Oxford University Press: Oxford ; New York, 1998.
14.
Hjerten, S. High-performance electrophoresis: the electrophoretic
counterpart of high-performance liquid chromatography. J. Chromatogr.
1983, 270, 1-6.
15.
Cohen, A. S.;Karger, B. L. High-performance sodium dodecyl sulfate
polyacrylamide gel capillary electrophoresis of peptides and proteins. J.
Chromatogr. 1987, 397, 409-17.
16.
Dittmann, M. M.;Rozing, G. P. High-sensitivity separations of sodium
dodecyl sulfate-protein complexes with capillary gel electrophoresis. LCGC 1999, 17, 132, 134, 136-137.
17.
Guttman, A. Capillary sodium dodecyl sulfate-gel electrophoresis of
proteins. Electrophoresis 1996, 17, 1333-1341.
18.
Guttman, A. On the separation mechanism of capillary sodium dodecyl
sulfate-gel electrophoresis of proteins. Electrophoresis 1995, 16, 611-16.
19.
Guttman, A.;Nolan, J. Comparison of the separation of proteins by sodium
dodecyl sulfate-slab gel electrophoresis and capillary sodium dodecyl
sulfate-gel electrophoresis. Anal. Biochem. 1994, 221, 285-9.
20.
Guttman, A.; Nolan, J. A.;Cooke, N. Capillary sodium dodecyl sulfate gel
electrophoresis of proteins. J. Chromatogr. 1993, 632, 171-5.
21.
Guttman, A.; Shieh, P.; Hoang, D.; Horvath, J.;Cooke, N. Effect of
operational variables on the separation of proteins by capillary sodium
152
dodecyl sulfate-gel electrophoresis. Electrophoresis (Weinheim, Fed.
Repub. Ger.) 1994, 15, 221-4.
22.
Hunt, G.;Nashabeh, W. Capillary electrophoresis sodium dodecyl sulfate
nongel sieving analysis of a therapeutic recombinant monoclonal antibody:
A biotechnology perspective. Anal. Chem. 1999, 71, 2390-2397.
23.
Kinghorn, N. M.; Norris, C. S.; Paterson, G. R.;Otter, D. E. Comparison of
capillary electrophoresis with traditional methods to analyze bovine whey
proteins. J. Chromatogr., A 1995, 700, 111-23.
24.
Kroon, D. J.; Goltra, S.;Sharma, B. Analysis of monoclonal antibodies by
sodium dodecyl sulfate-capillary gel electrophoresis with special reference
to quantitation of half-antibody. J. Capillary Electrophor. 1995, 2, 34-9.
25.
Kurosu, Y.;Saito, M. Sodium dodecyl sulfate (SDS)-non-gel molecular
sieving capillary electrophoresis for proteins. Handb. Capillary
Electrophor. Appl. 1997, 158-172.
26.
Lausch, R.; Scheper, T.; Reif, O. W.; Schloesser, J.; Fleischer, J.;Freitag, R.
Rapid capillary gel electrophoresis of proteins. J. Chromatogr., A 1993,
654, 190-5.
27.
Lee, H. G. High-performance sodium dodecyl sulfate-capillary gel
electrophoresis of antibodies and antibody fragments. J. Immunol. Methods
2000, 234, 71-81.
28.
Lee, H. G.;Fritsche, E. Determination of biotin on a protein by quantitative
sodium dodecyl sulfate-capillary gel electrophoresis of monomeric avidin.
J. Chromatogr., A 2003, 994, 213-219.
29.
Liu, J.; Abid, S.;Lee, M. S. Analysis of monoclonal antibody chimeric
BR96-doxorubicin immunoconjugate by sodium dodecyl sulfate-capillary
gel electrophoresis with ultraviolet and laser-induced fluorescence
detection. Anal. Biochem. 1995, 229, 221-8.
30.
Luo, S.; Feng, J.;Pang, H.-m. High-throughput protein analysis by
multiplexed sodium dodecyl sulfate capillary gel electrophoresis with UV
absorption detection. J. Chromatogr., A 2004, 1051, 131-134.
153
31.
Manabe, T. Sodium dodecyl sulfate-gel electrophoresis of proteins
employing short capillaries. Electrophoresis 1995, 16, 1468-73.
32.
Manso, M. A.; Cattaneo, T. M.; Barzaghi, S.; Olieman, C.; Lopez-Fandino,
R.; Leaver, J.; Otte, J.; Sanchez, A.; Svenning, C.; Tirelli, A.;Vegarud, G.
Determination of vegetal proteins in milk powder by sodium dodecyl
sulfate-capillary gel electrophoresis: interlaboratory study. J. AOAC Int.
2002, 85, 1090-1095.
33.
Mellado, M. C. M.; Franco, C.; Coelho, A.; Alves, P. M.;Simplicio, A. L.
Sodium dodecyl sulfate-capillary gel electrophoresis analysis of rotaviruslike particles. J. Chromatogr., A 2008, 1192, 166-172.
34.
Na Dong, H.; Park Eun, J.; Jo Yeong, W.;Lee Kang, C. Capillary
electrophoretic separation of high-molecular-weight poly(ethylene glycol)modified proteins. Anal. Biochem. 2008, 373, 207-12.
35.
Nakatani, M.; Shibukawa, A.;Nakagawa, T. Effect of temperature and
viscosity of sieving medium on electrophoretic behavior of sodium dodecyl
sulfate-proteins on capillary electrophoresis in presence of pullulan.
Electrophoresis 1996, 17, 1210-1213.
36.
Ou, J. P.; Chan, S. T. H.;Yeung, W. S. B. Separation of bovine serum
albumin and its monoclonal antibody from their immunocomplexes by
sodium dodecyl sulfate-capillary gel electrophoresis and its application in
capillary electrophoresis-based immunoassay. J. Chromatogr., B Biomed.
Sci. Appl. 1999, 731, 389-394.
37.
Sasaki, T. Analysis of proteins by capillary electrophoresis/sodium
dodecylsulfate-polyacrylamide gel electrophoresis (CE/SDS-PAGE). Jasco
Rep. 1992, 34, 10-14.
38.
Schmerr, M. J.; Jenny, A.;Cutlip, R. C. The use of capillary sodium dodecyl
sulfate gel electrophoresis to detect the prion protein extracted from
scrapie-infected sheep. J. Chromatogr., B Biomed. Sci. Appl. 1997, 697,
223-229.
154
39.
Tsuji, K. Evaluation of sodium dodecyl sulfate non-acrylamide, polymer
gel-filled capillary electrophoresis for molecular size separation of
recombinant bovine somatotropin. J. Chromatogr., A 1993, 652, 139-47.
40.
Tsuji, K. Factors affecting the performance of sodium dodecyl sulfate gelfilled capillary electrophoresis. J. Chromatogr., A 1994, 661, 257-64.
41.
Tsuji, K. High-performance capillary electrophoresis of proteins. Sodium
dodecyl sulfate-polyacrylamide gel-filled capillary column for the
determination of recombinant biotechnology-derived proteins. J.
Chromatogr. 1991, 550, 823-30.
42.
Wu, D.;Regnier, F. E. Sodium dodecyl sulfate-capillary gel electrophoresis
of proteins using non-cross-linked polyacrylamide. J. Chromatogr. 1992,
608, 349-56.
43.
Gad, S. C. Handbook of pharmaceutical biotechnology. Wiley-Interscience:
Hoboken, N.J., 2007.
44.
Strege, M. A.;Lagu, A. L. Capillary electrophoresis of proteins and
peptides. Humana Press: Totowa, N.J., 2004.
45.
Altria, K. D. Capillary electrophoresis guidebook : principles, operation,
and applications. Humana Press: Totowa, N.J., 1996.
46.
Ganzler, K.; Greve, K. S.; Cohen, A. S.; Karger, B. L.; Guttman, A.;Cooke,
N. C. High-performance capillary electrophoresis of SDS-protein
complexes using UV-transparent polymer networks. Anal. Chem. 1992, 64,
2665-71.
47.
Lucy Y. Liu, C. R., Jeff Chapman, Narasaiah Dontha, Sae Choo, M. P.
Reddy Assay of IgG Purity and Heterogeneity using High-resolution
Sodium Dodecyl Sulfate Capillary Gel Electrophoresis.
www.beckmancoulter.com
48.
Ma, S.;Nashabeh, W. Carbohydrate analysis of a chimeric recombinant
monoclonal antibody by capillary electrophoresis with laser-induced
fluorescence detection. Anal. Chem. 1999, 71, 5185-92.
155
49.
Han, M.; Phan, D.; Nightlinger, N.; Taylor, L.; Jankhah, S.; Woodruff, B.;
Yates, Z.; Freeman, S.; Guo, A.; Balland, A.;Pettit, D. Optimization of CESDS method for antibody separation based on multi-users experimental
practices. Chromatographia 2006, 64, 335-342.
50.
Liu, H.; Cho, B. Y.; Krull, I. S.;Cohen, S. A. Homogeneous fluorescent
derivatization of large proteins. J. Chromatogr. A 2001, 927, 77-89.
51.
Michels, D. A.; Brady, L. J.; Guo, A.;Balland, A. Fluorescent
Derivatization Method of Proteins for Characterization by Capillary
Electrophoresis-Sodium Dodecyl Sulfate with Laser-Induced Fluorescence
Detection. Anal. Chem. 2007, 79, 5963-5971.
52.
Quigley, W. W.;Dovichi, N. J. Capillary electrophoresis for the analysis of
biopolymers. Anal. Chem. 2004, 76, 4645-58.
53.
Salas-Solano, O.; Tomlinson, B.; Du, S.; Parker, M.; Strahan, A.;Ma, S.
Optimization and Validation of a Quantitative Capillary Electrophoresis
Sodium Dodecyl Sulfate Method for Quality Control and Stability
Monitoring of Monoclonal Antibodies. Anal. Chem. 2006, 78, 6583-6594.
54.
Li, L.; Sun, M.; Gao, Q. S.;Paul, S. Low level formation of potent catalytic
IgG fragments mediated by disulfide bond instability. Mol. Immunol. 1996,
33, 593-600.
55.
Friedman, M. The chemistry and biochemistry of the sulfhydryl group in
amino acids, peptides, and proteins. [1st ed.; Pergamon Press: Oxford, New
York,, 1973.
56.
Gilbert, H. F. Thiol/disulfide exchange equilibria and disulfide bond
stability. Methods Enzymol. 1995, 251, 8-28.
57.
Houk, J.; Singh, R.;Whitesides, G. M. Measurement of thiol-disulfide
interchange reactions and thiol pKa values. Methods Enzymol 1987, 143,
129-40.
58.
Landers, J. P. Handbook of capillary and microchip electrophoresis and
associated microtechniques. 3rd ed.; CRC Press: Boca Raton, 2008.
156
59.
Harris, D. C. Quantitative chemical analysis. 7th ed.; W.H. Freeman and
Co.: New York, NY, 2007.
60.
Rustandi, R. R.; Washabaugh, M. W.;Wang, Y. Applications of CE SDS
gel in development of biopharmaceutical antibody-based products.
Electrophoresis 2008, 29, 3612-20.
61.
Branch, S. K. Guidelines from the International Conference on
Harmonisation (ICH). J. Pharm. Biomed. Anal. 2005, 38, 798-805.
157
Chapter 5: Optimization of injection length into a capillary for detection
sensitivity enhancement in micellar electrokinetic chromatography
5.1
Introduction
Micellar electrokinetic chromatography (MEKC) has emerged as a versatile and
powerful technique for the separation of mixtures of charged and neutral analytes
that cannot be separated by capillary zone electrophoresis (CZE) [1-4]. In MEKC,
the running buffer contains micelles, i.e., a solution of surfactant at a concentration
higher than the critical micelle concentration [3]. The separation mechanism of
MEKC is based upon differences in analyte partitioning between the aqueous phase
and the micellar phase (pseudostationary phase) and for charged analytes,
differences in electrophoretic mobility [5]. Like CZE, MEKC is one mode of
capillary electrophoresis (CE). Many advantages of CE over other separation
techniques include high efficiency, short analysis time, and low sample
consumption. However, sensitivity in CE is often insufficient for trace levels of
analytes due to limited injection volume and short optical path length of
photometric detectors in CE [6]. Many on-line sensitivity enhancement techniques
that concentrate analytes inside the capillary prior to separation and detection have
been reported [7-19]. Among these techniques, sample stacking and sweeping are
by far the most widely used methods in MEKC [20]. Stacking in MEKC is an online sample concentration technique where analytes cross a boundary separating
regions of the high electric field (sample zone) and the low electric field (BGE
zone). Sweeping is described as a process of picking and accumulating of analytes,
which occurs due to partitioning or interactions of analytes with pseudostationary
phase when voltage is applied [20]. Figures 5.1 and 5.2 show the basis of stacking
158
Figure 5.1 Basis of stacking in CZE using cation as an example. The region
containing sample ions is a low conductivity solution, while the background region
is a high conductivity solution. Upon application of voltage, the low conductivity
region will experience a high electric field compared to the high conductivity
background region. Sample ions then move faster in the low conductivity region
than in the higher conductivity region resulting in the reduction of sample zone or
higher sample concentration. [21]
159
Figure 5.2 Sweeping in a homogenous electric field in MEKC using a negatively
charged micelle as pseudostationary phase (PSP) and a negligible EOF
environment. (A) A longer than a typical injection of sample zone prepared in a
matrix having a conductivity similar to micellar background electrolyte (BGE). (B)
Upon application of a negative voltage, the BGE enters the sample zone and sweeps
(concentrates) the analyte molecules. (C) The final swept zone is formed when the
incoming BGE (from the anode reservoir) overtakes the more distant boundary of
the sample zone. [22]
160
and sweeping in CE and MEKC, respectively [21, 12, 22]. The basic condition for
sweeping in MEKC is a pseudostationary phase in separation buffer (background
electrolyte, BGE) and a sample matrix free of the pseudostationary phase [20].
Since its original introduction by Quirino and Terabe in 1998, sweeping has been
widely used because it provides higher concentration efficiency than stacking [12,
23]. Moreover, substantial improvements have been made on sweeping in recent
years with over 5000-fold sensitivity enhancement [12]. One of the approaches is
using modified micelles [24]. Monton et al. studied sweeping using mixed micelles
of sodium dodecyl sulfate (SDS)-SB-12, which increased partitioning of
hydrophobic analytes into the core due to large radius of mixed micelles [24]. The
mixed micelles also possess lower negative surface charge relative to pure SDS
micelles so anionic analytes can be retained because of decreased electrostatic
repulsion. As a result, better focusing efficiency was achieved.
Another approach was so called “selective exhaustive injection-sweeping”, which
combines both stacking and sweeping techniques [25, 11]. Selective exhaustive
injection-sweeping has achieved parts per trillion level of detection sensitivity,
which is much higher than regular sweeping. However, few applications have been
reported.
Palmer et al. proposed a high salt sample sweeping technique [26]. By adding salt
to sample matrixes (e.g., sodium chloride), the sample matrix conductivity
increased 2-3 fold higher than the separation buffer, resulting in effective
concentration of analytes. After close examination of this technique by Quirino et
161
al., the focusing effect of sweeping using a matrix with equal or higher conductivity
compared to the separation buffer was roughly the same [7].
A methanol assisted sweeping technique was developed by Huang et al. [27]. This
methanol plug serves as a micelle destruction zone. When the sweeping-MEKC
mode is applied, the analytes become concentrated in the SDS-micelles from the
inlet reservoir along the capillary axis. Once these SDS-micelle-analytes enter the
methanol plug, most of the SDS-micelle-analytes are freed since SDS does not form
complete micelles in methanol. Thus, the methanol plug becomes a “barrier” that
interrupts the micelle-analytes when they pass through the methanol plug. As a
result, the injected analytes are further concentrated, leading to a higher separation
efficiency.
Shih and Lin developed a full-capillary sample stacking/sweeping-MEKC for the
separation of naphthalene-2, 3-dicarboxaldehyde-derivatized tryptophan and
isoleucine [28]. An at least 400-fold improvement was achieved comparing to the
regular sweeping technique.
Sweeping requires a larger than normal volume of sample to be introduced into a
capillary. In CE, the volume of the sample injected into a capillary must be much
smaller than the dispersion generated by the diffusion during the analysis and
detection [29]. If band broadening is to be minimized, the injection volume should
be no more than about 0.1% of the column volume [30]. The contribution of the
injection length to the peak dispersion has been described by Sternberg [31]:
162
σ inj =
2
linj
2
12
where σinj2 is the variance due to the injection and linj is the injection length; the
numerical value of 12 assumes that a certain length of the inside of the tube is
completely displaced by a plug of the sample solution because the sample solution
is very slowly siphoned into the injection end of the tube. Grushka and McCormick
studied zone broadening due to sample injection in capillary zone electrophoresis
and discovered the relationship between the injection length, linj, and the relative
decrease in efficiency [32]:
1
linj = (24DNt ) 2
where D is the analyte diffusion coefficient, N is the acceptable relative decrease
(%) in efficiency (usually assumed to be 105), t is the analysis time. From this
equation, sample injection volume can be calculated for a particular analyte. For a
100 cm capillary with 50 μm inner diameter, the normal d injection length with
acceptable column efficiency is only 0.1 cm, or 0.1% of the capillary length.
In sweeping, the length of the sample zone injected (linj) and the length of swept
zone (lsweep) have the following relationship [12]:
l sweep =
linj
1+ k
(5.1)
where k is the analyte retention factor. Based on this equation, the concentration
factor can be achieved by approximately 1+k. In order to obtain the most
concentration effect, a large injection volume should be made under acceptable
163
separation efficiency. An increase of injection volume will result in a decrease of
separation efficiency due to the increase of lsweep.
In this study, the effect of injection length on sweeping in MEKC, particularly its
effects on separation efficiency was studied. The relationship between injection
length and retention factor was derived and experimentally confirmed. An optimal
injection length for a given analyte can be obtained based on a linear curve.
5.2
Materials and methods
5.2.1
Reagents and solutions
Reagents of the highest grade were used in the preparation of buffers and solutions.
Acetophenone, propiophenone, butyrophenone, and dodecanophenone were from
Sigma-Aldrich (St. Louis, MO, USA). Their structures are shown in Figure 5.3.
Phosphoric acid (H3PO4), methanol, sodium dodecyl sulfate (SDS), and sodium
chloride were provided by J. T. Baker (Phillipsburg, NJ, USA). Sodium hydroxide
(NaOH) solutions (1 M and 0.1 M) and hydrochloric acid (0.1 M) were provided by
Agilent Technologies (Wilmington, DE, USA). Water with a conductivity value
lower than 5 μS/cm was obtained using a Milli-Q water purification system from
Millipore (Billerica, MA, USA).
The running buffer was prepared by dissolving appropriate amounts of SDS in
water and diluting appropriate volumes of H3PO4 to obtain the desired concentration
of 50 mM SDS and 50 mM phosphate. The pH was measured to be about 2.0.
Sample buffer was prepared by diluting the same volume of H3PO4 in water and the
164
O
CH3
(a)
O
CH3
(b)
O
CH3
(c)
Figure 5.3 Structures of the studied compounds: (a) phenyl methyl ketone
(acetophenone); (b) phenyl ethyl ketone (propiophenone); (c) phenyl propyl ketone
(butyrophenone)
165
conductivity was adjusted with NaCl to be the same as the running buffer. The
above solutions were filtered through a 0.22 μM nylon filter. Individual stock
solution of acetophenone, propiophenone, butyrophenone, and dodecanophenone
was prepared in methanol. Concentrations of the alkyl phenyl ketone stock solutions
were 40 mM. Each stock solution was further diluted 100-fold in sample buffer
except for dodecanophenone, which was diluted in running buffer. Mixed stock
solution was prepared by combining appropriate portion of each stock solution and
diluted in sample buffer. Working solutions were prepared by diluting the mixed
stock solution to obtain the concentrations of 20 μM, 40 μM, 100 μM, 200 μM, and
400 μM.
5.2.2
Capillary electrophoresis (CE)
The CE experiments were performed on an Agilent 3D-CE instrument equipped
with a photo-diode array absorbance detector and ChemStation software from
Agilent Technologies (Wilmington, DE, USA). Bare fused-silica capillary columns
(72 cm x 50 μm i.d.) were supplied by Agilent Technologies (Wilmington, DE,
USA). The detection window was placed at 8.5 cm from the outlet of the capillary.
The capillary temperature was maintained at 25 °C for all the experiments. A
negative voltage of 25 kV was applied during electrophoretic separations. Samples
were injected hydrodynamically at 50 mbar for 5, 10, 25, 50, and 100 seconds for
40 μM, 20 μM, 8 μM, 4μM, and 2 μM working solutions, respectively. Detection
was performed at 210 nm absorbance. Data were collected at a sampling rate of 5
Hz. The current was also monitored to ensure that Joule heating was below the
upper limit in an Ohm’s law plot.
166
5.2.3
Capillary conditioning
New capillaries were conditioned by rinsing with 1 M NaOH for 10 minutes,
methanol for 5 minutes, water for 5 minutes, and running buffer for 10 minutes.
After the rinse, a voltage of 20 kV was applied to the capillary filled with running
buffer for 10 minutes. Between injections, the capillary was rinsed with running
buffer for 5 minutes. To ensure reproducibility, at the end of each run (20
injections), the capillary was rinsed with 1 M NaOH for 10 minutes, methanol for 5
minutes, water for 5 minutes, and running buffer for 10 minutes.
5.3 Results and discussion
5.3.1
Injection length vs. retention factor
Jandera et al. studied the characteristics of retention in MEKC and observed a linear
relationship between the logarithm of the retention factors, k, of analytes and the
logarithm of the concentration of micelles in the running buffer, Cmic, which can be
described as [33]:
log k = a + m log C mic
(5.2)
where a is the value of log k in the pure aqueous phase and m is the contribution to
log k that results when the concentration of micelles is increased.
In MEKC, the retention factor is defined as the ratio of total moles of analyte in the
pseudostationary phase versus those in the aqueous phase [34]. Chiari et al. first
introduced a unique MEKC mode for the separations of neutral isotopic molecules
[35], which was later defined as reversed electroosmotic flow MEKC (RF-MEKC)
by Janini et al. in their study of separation of several classes of hydrophobic
compounds [36]. In RF-MEKC, electroosmotic flow is minimized either by using a
167
very acidic running buffer or coated capillary. A voltage in negative polarity is
applied to the inlet (cathode) and anionic micelles such as SDS move to the outlet
(anode) at a higher velocity than non-ionic solutes, and solute migration order is
reversed and solute migration time is inversely proportional to micelle
concentration, in contrast to normal MEKC [36]. The separation mechanism of RFMEKC is the same as normal MEKC, which for neutral compounds is based solely
on the differential interactions of the analytes with the micelles. The equation for
the retention factor in RF-MEKC can be derived from basic chromatographic theory
provided that electroosmotic flow is negligible:
k=
t mic − t R
t mic
(5.3)
where tR is the retention time of the analyte and tmic is the retention time of the
micelle marker.
In RF-MEKC, k is replaced by 1/k because the role of micelles changed from
pseudostationary phase to mobile phase. Equation (5.2) is written as:
log(1 k ) = a + m log C mic
(5.4)
As shown in Figure 5.2, after the sample is injected into the capillary, upon the
application of a negative voltage, the SDS micelles in the running buffer reservoir
start moving into the capillary due to the electric field. The micelles behind the
sample zone are sweeping the analytes towards the outlet direction. Meanwhile, the
micelles in front of the sample zone also start moving towards the detector, thus
maintaining a zone that is free of micelles. After the analytes are completely swept,
the micelle-free zone, or pseudostationary phase (PSP) vacant zone, has equal
168
length to the original sample zone (linj). Because the conductivity of sample zone is
the same as the background electrolyte zone, the sweeping has taken place in a
homogenous electric field.
The time for the micelles to travel from the trailing edge of the sample zone to the
detector is calculated by:
t trail =
l eff − l inj
μE
(5.5)
The time for the micelles to travel from the leading edge of the PSP vacant zone is
calculated by:
t lead =
l eff − 2linj
μE
(5.6)
From equations (5.5) and (5.6), it is apparent that ttrail > tlead, so the PSP vacant zone
will be maintained until it reaches the detector. In the mean time, the micelles will
be moving by diffusion from both edges into the PSP vacant zone (Figure 5.4).
During the entire separation, two processes are happening: (1) analytes are being
swept by the micelles coming from the inlet reservoir; (2) analytes are separated in
the environment where the micelle concentration is diluted. These two processes
must be considered separately. In the first process, the interaction between the
analyte and the micelle must be maximized, so as to increase the concentration
efficiency, whereas in the second process the interaction between the two must be
optimized to give good resolution or suitable retention factors.
169
Figure 5.4 Separation of swept zone in a diluted pseudo stationary phase
(PSP) environment. Explanation is in the text.
170
Ideally, the separation begins after the analytes are fully swept. But actually a small degree
of separation is achieved during sweeping because the average position of a highly swept
analyte (large analyte-micelle interactions) will be slightly ahead of a lesser swept analyte
(smaller analyte-micelle interactions). Due to micelle diffusion, the separation occurs in a
diluted micelle environment (diffusion zone). The micelle concentration in the separation
zone (CS) is related to the sample injection length, and it may be calculated by:
⎛ l inj
C S = dC mic ⎜
⎜l −l
⎝ eff inj
⎞
⎟
⎟
⎠
(5.7)
where d is a dilution factor, which can be measured experimentally.
Replacing Cmic with CS, equation (5.4) can be written as:
⎛ linj
log(1 k ) = a + m log C mic _ PSP = a + m log dC mic ⎜
⎜l −l
⎝ eff inj
⎞
⎟
⎟
⎠
(5.8)
Where leff is the effective capillary length (inlet to detector)
Equation (5.8) can be further written as:
⎛ linj
log(1 k ) = [a + m log dC mic ] + m log⎜
⎜l −l
⎝ eff inj
⎞
⎟
⎟
⎠
(5.9)
For a given capillary, leff and Cmic are constant, log (1/k) has a linear relationship
⎛ linj
⎜l −l
⎝ eff inj
with log⎜
⎞
⎟ , and equation (5.9) can be written as:
⎟
⎠
⎛ linj
log(1 k ) = A + m log⎜
⎜l −l
⎝ eff inj
⎞
⎟
⎟
⎠
where A is the intercept and m is the slope from the linear curve.
(5.10)
171
To verify this relationship, a series of alkyl phenyl ketone solutions were injected
with different injection times. Seven solutions at concentrations of 40 μM, 20 μM, 8
μM, 6.7 μM, 5 μM, 4 μM, and 2 μM were injected in triplicate under the same
pressure of 50 milibars for 5 s, 10 s, 25 s, 30 s, 40 s, 50 s, and 100 s, respectively.
As such, the total amounts of analytes introduced into the capillary for each
injection remained constant. Other experimental conditions were used as described
in Quirino’s work [22]. The k values were calculated using decanophenone as a
micelle marker. At pH 2.0, the electroosmotic flow is negligible. The length of
sample plug was calculated from the equation [37]:
linj =
ΔP ∗ t ∗ r 2
8ηl total
(5.11)
where ΔP is the pressure across the capillary, r is the capillary inner radius, t is the
time duration when the pressure is applied, η is the viscosity of the sample solution,
and ltotal is the total capillary length.
The experimental data are summarized in Tables 5.1 and 5.2. The retention factors
(1/k) for acetophenone, propiophenone, butyrophenone under non-sweeping
conditions were also calculated for comparison purpose. The retention factors
under non-swept conditions were higher than those under sweeping conditions,
suggesting the dilution of micelles in sweeping. As the sample zone was produced
by the hydrodynamic injection, the length of sample zone was dependent on the
injection duration. The longer the injection time, the longer the sample plug zone
will be. Plots were made using eight data points to show the linear range of log
172
⎛ linj
⎜l −l
⎝ eff inj
(1/k) vs. log⎜
⎞
⎟ . As shown in Figure 5.5, when all eight data points were
⎟
⎠
used to make the plot, corresponding to 5-100 seconds injection time, the
⎛ l inj
⎜l −l
⎝ eff inj
correlation coefficient between log (1/k) and log⎜
⎞
⎟ was <0.99, meaning a
⎟
⎠
weaker correlation. When the injection time was at 20-50 seconds, the correlation
coefficient for all analytes were >0.99, meaning a linear relationship between log
⎛ linj
⎜l −l
⎝ eff inj
(1/k) and log⎜
⎞
⎟ (Figure 5.6).
⎟
⎠
Figure 5.6 indicates that the optimal injection time within a linear range is 50 s.
5.3.2 Injection length vs. efficiency
Zone broadening in capillary electrophoresis caused results in low efficiency [38].
Generally speaking, the injection of a larger sample volume results in a decrease in
the theoretical plate number (efficiency) if no concentration effect takes place. The
decrease of efficiency from injection length is due to diffusion [32]. However, the
diffusion does not have a dramatic effect before the analytes are being swept by the
micelles because they are at static state. As a result, the peak broadening was
suppressed [39]. To evaluate the impact of injection length on efficiency,
theoretical plate numbers were plotted against injection time. With increase of
injection time/length, the plate number changed differently for different analyte.
173
Table 5.1 Injection time and calculated injection length (n=3)
Injection time (s)
Injection length (linj) (cm)
5
10
20
25
30
40
50
100
0.241
0.482
0.964
1.205
1.446
1.928
2.41
4.82
⎛ linj
log⎜
⎜l −l
⎝ eff inj
-2.47
-2.17
-1.87
-1.77
-1.69
-1.56
-1.46
-1.14
⎞
⎟
⎟
⎠
174
Table 5.2 Migration time and calculated retention factors (n=3)
Injection
Migration Time (min)
Retention Factor (1/k4)
Log(1/k)
Time (s)
AP1
5*
1
14.8
PP2
18.4
BP3
22.1
AP
PP
BP
2.18
3.05
5.22
AP
PP
BP
0.338
0.484
0.718
5
14.4
17.8
21.2
2.11
2.86
4.51
0.325
0.457
0.654
10
14.2
17.5
20.8
2.09
2.78
4.19
0.319
0.444
0.622
20
14.1
17.3
20.5
2.07
2.73
4.04
0.315
0.436
0.606
25
14.1
17.3
20.5
2.06
2.72
4.00
0.314
0.434
0.602
30
14.0
17.2
20.4
2.06
2.70
3.93
0.313
0.431
0.594
40
13.9
17.0
20.2
2.04
2.66
3.85
0.310
0.425
0.585
50
13.9
17.0
20.1
2.03
2.64
3.78
0.307
0.421
0.577
100
13.4
16.5
19.5
1.97
2.53
3.51
0.294
0.403
0.545
AP: acetophenone
2
PP: propiophenone
3
BP: butyrophenone
4
Retention factor (k) is calculated using dodecanophenone as a micelle marker (tm =
27.4 min).
*
Under non-sweeping condition
175
⎛ l inj
⎜l −l
⎝ eff inj
Figure 5.5. Plot of log(1/k) versus log⎜
⎞
⎟ with eight data points
⎟
⎠
corresponding to injection times of 5-100 seconds.
176
⎛ l inj
⎜l −l
⎝ eff inj
Figure 5.6. Plot of log(1/k) versus log⎜
⎞
⎟ with five data points
⎟
⎠
corresponding to injection times of 20-50 seconds.
177
The slower decrease of efficiency was attributed to sweeping process. The loss of
the efficiency was due to zone broadening in the MEKC separation process. During
the sweeping process, the interaction between analyte and micelle differs among the
three analytes. The analyte-micelle interaction is related to the analyte
hydrophobicity. Butyrophenone has the most interaction with micelle due to its
relatively higher hydrophobicity, which can be swept more efficiently and the
longest injection length can be obtained without losing capillary efficiency. On the
other hand, acetophenone is less hydrophobic and being swept less effectively and
only a shorter injection length can be applied. These predictions were
experimentally proved in our study, as shown in Figures 5.6. With injection time of
50 s, acceptable plate numbers are still maintained. Figure 5.7 shows a comparison
of a 50 s vs. a 100 s injection. The peak shapes for the 100 s injection are
deteriorated significantly comparing to those for the 50 s injection.
5.4 Concluding remarks
The injection length for RF-MEKC was studied and an optimal injection length for
sweeping was derived theoretically and experimentally confirmed. The
experimental data were in good agreement with the theoretical predictions.
178
mAU
10
8
6
A
3
2
1
4
2
0
0
5
10
15
20
25
min
25
min
mAU
10
8
6
B
4
2
1
3
2
0
0
5
10
15
20
Figure 5.6 Electropherograms of 50 s injection (A) vs. 100 s injection (B). Analyte
identification: (1) acetophenone; (2) propiophenone; (3) butyrophenone. Analytical
conditions: 50 mM phosphate, 50 mM sodium dodecyl sulfate, pH 2.0, -25 kV, 210
nm, 20°C, 72 cm x 50 µm capillary . Analyte concentrations: 40 µM (A); 20 µM
(B).
179
Figure 5.7 Plots of plate number vs. injection time. Injection time of 35 seconds
gives highest plate number for butyrophenone and acceptable plate number for
acetophenone.
180
5.5
List of references:
1.
Serrano, J. M.;Silva, M. Use of SDS micelles for improving sensitivity,
resolution, and speed in the analysis of beta-lactam antibiotics in
environmental waters by SPE and CE. Electrophoresis 2007, 28, 3242-9.
2.
Terabe, S.; Monton, M. R. N.; Le Saux, T.;Imami, K. Applications of
capillary electrophoresis to high-sensitivity analyses of biomolecules. Pure
Appl. Chem. 2006, 78, 1057-1067.
3.
Terabe, S.; Otsuka, K.;Ando, T. Electrokinetic chromatography with
micellar solution and open-tubular capillary. Anal. Chem. 1985, 57, 834-41.
4.
Terabe, S.; Otsuka, K.; Ichikawa, K.; Tsuchiya, A.;Ando, T. Electrokinetic
separations with micellar solutions and open-tubular capillaries. Anal.
Chem. 1984, 56, 111-13.
5.
Terabe, S. In Micellar electrokinetic chromatography, Handbook of
Capillary and Microchip Electrophoresis and Associated Microtechniques
(3rd Edition), 2008; 2008.109-133.
6.
Quirino, J. P.;Terabe, S. Stacking of neutral analytes in micellar
electrokinetic chromatography. J. Capillary Electrophor. 1997, 4, 233-45.
7.
Monton, M. R. N.; Quirino, J. P.; Otsuka, K.;Terabe, S. Separation and online preconcentration by sweeping of charged analytes in electrokinetic
chromatography with nonionic micelles. J. Chromatogr., A 2001, 939, 99108.
8.
Quirino, J. P.; Otsuka, K.;Terabe, S. Field enhanced sample injection in
electrokinetic chromatography. Chromatography 1998, 19, 164-165.
9.
Quirino, J. P.; Otsuka, K.;Terabe, S. Online concentration of neutral
analytes for micellar electrokinetic chromatography. VI. Stacking using
reverse migrating micelles and a water plug. J. Chromatogr., B: Biomed.
Sci. Appl. 1998, 714, 29-38.
181
10.
Quirino, J. P.; Otsuka, K.;Terabe, S. Sample stacking in micellar
electrokinetic chromatography: application to analytes of environmental
interest. Chromatography 1997, 18, 324-325.
11.
Quirino, J. P.;Terabe, S. Approaching a million-fold sensitivity increase in
capillary electrophoresis with direct ultraviolet detection: cation-selective
exhaustive injection and sweeping. Anal. Chem. 2000, 72, 1023-30.
12.
Quirino, J. P.;Terabe, S. Exceeding 5000-fold concentration of dilute
analytes in micellar electrokinetic chromatography. Science 1998, 282, 4658.
13.
Quirino, J. P.;Terabe, S. Online concentration of neutral analytes for
micellar electrokinetic chromatography. 3. Stacking with reverse migrating
micelles. Anal. Chem. 1998, 70, 149-157.
14.
Quirino, J. P.;Terabe, S. Online concentration of neutral analytes for
micellar electrokinetic chromatography. I. Normal stacking mode. J.
Chromatogr., A 1997, 781, 119-128.
15.
Quirino, J. P.;Terabe, S. Online concentration of neutral analytes for
micellar electrokinetic chromatography. II. Reversed electrode polarity
stacking mode. J. Chromatogr., A 1997, 791, 255-267.
16.
Quirino, J. P.;Terabe, S. Online concentration of neutral analytes for
micellar electrokinetic chromatography. IV. Field-enhanced sample
injection. J. Chromatogr., A 1998, 798, 251-257.
17.
Quirino, J. P.;Terabe, S. Sample concentration by sweeping in MEKC.
Chromatography 1999, 20, 196-197.
18.
Quirino, J. P.;Terabe, S. Sweeping with an enhanced electric field of neutral
analyte zones in electrokinetic chromatography. J. High Resolut.
Chromatogr. 1999, 22, 367-372.
19.
Takeda, S.; Wakida, S.-i.; Yamane, M.; Higashi, K.;Terabe, S. Effect of the
polar groups of anionic surfactant on migration behavior in micellar
electrokinetic chromatography. J. Chromatogr., A 1997, 781, 11-16.
182
20.
Quirino, J. P.; Kim, J. B.;Terabe, S. Sweeping: concentration mechanism
and applications to high-sensitivity analysis in capillary electrophoresis. J.
Chromatogr., A 2002, 965, 357-73.
21.
Chien, R. L.;Burgi, D. S. On-column sample concentration using field
amplification in CZE. Anal. Chem. 1992, 64, 489A-96A.
22.
Quirino, J. P.;Terabe, S. Sweeping of Analyte Zones in Electrokinetic
Chromatography. Anal. Chem. 1999, 71, 1638-1644.
23.
Silva, M. MEKC: an update focusing on practical aspects. Electrophoresis
2007, 28, 174-92.
24.
Monton, M. R.; Otsuka, K.;Terabe, S. On-line sample preconcentration in
micellar electrokinetic chromatography by sweeping with anioniczwitterionic mixed micelles. J. Chromatogr., A 2003, 985, 435-45.
25.
Quirino, J. P.; Iwai, Y.; Otsuka, K.;Terabe, S. Determination of
environmentally relevant aromatic amines in the ppt levels by cation
selective exhaustive injection-sweeping-micellar electrokinetic
chromatography. Electrophoresis 2000, 21, 2899-903.
26.
Palmer, J.; Munro, N. J.;Landers, J. P. A Universal Concept for Stacking
Neutral Analytes in Micellar Capillary Electrophoresis. Anal. Chem. 1999,
71, 1679-1687.
27.
Huang, H. M.;Lin, C. H. Methanol plug assisted sweeping-micellar
electrokinetic chromatography for the determination of dopamine in urine
by violet light emitting diode-induced fluorescence detection. J.
Chromatogr., B: Analyt. Technol. Biomed. Life Sci. 2005, 816, 113-9.
28.
Shih, C.-M.;Lin, C.-H. Full-capillary sample stacking/sweeping-MEKC for
the separation of naphthalene-2,3-dicarboxaldehyde-derivatized tryptophan
and isoleucine. Electrophoresis 2005, 26, 3495-3499.
29.
Gas, B.; Stedry, M.;Kenndler, E. Peak broadening in capillary zone
electrophoresis. Electrophoresis 1997, 18, 2123-2133.
183
30.
Poole, C. F. The essence of chromatography. 1st ed.; Elsevier: Amsterdam ;
Boston, 2003.
31.
Sternberg, J. C. Extra column contributions to chromatographic band
broadening. Adv. Chromatogr. 1966, 2, 205-70.
32.
Grushka, E.;McCormick, R. M. Zone broadening due to sample injection in
capillary zone electrophoresis. J. Chromatogr. 1989, 471, 421-8.
33.
Jandera, P.; Fischer, J.; Jebava, J.;Effenberger, H. Characterisation of
retention in micellar high-performance liquid chromatography, in micellar
electrokinetic chromatography and in micellar electrokinetic
chromatography with reduced flow. J. Chromatogr., A 2001, 914, 233-44.
34.
Pyell, U. Electrokinetic chromatography: theorey, instrumentation and
application. John Wiley & Sons, Ltd: 2006.
35.
Chiari, M.; Nesi, M.; Ottolina, G.;Righetti, P. G. Separation of charged and
neutral isotopic molecules by micellar electrokinetic chromatography in
coated capillaries. J. Chromatogr., A 1994, 680, 571-7.
36.
Janini, G. M.; Muschik, G. M.;Issaq, H. J. Micellar electrokinetic
chromatography in zero-electroosmotic flow environment. J. Chromatogr.,
B: Biomed, Appl. 1996, 683, 29-35.
37.
Baker, D. R. Capillary electrophoresis. Wiley: New York, 1995.
38.
Delinger, S. L.;Davis, J. M. Influence of analyte plug width on plate
number in capillary electrophoresis. Anal. Chem. 1992, 64, 1947-59.
39.
Chang, Y.-S.; Shih, C.-M.; Li, Y.-C.;Lin, C.-H. Large-volume sample
sweeping with a high theoretical plate number using a coupled-capillary in
capillary electrophoresis. Anal. Sci. 2006, 22, 557-561.
184
Chapter 6: Conclusions and future directions
6.1 Conclusions
Capillary electrophoresis (CE) as a complementary alternative to chromatographic
techniques offers great opportunities for biopharmaceutical analysis. Although the
CE technique has been used in some quality control laboratories, its main
application is still in research and development arena [1]. This study investigated
three CE separation modes and their applications in biopharmaceutical analysis.
Several analytical methodologies were developed and validated for their potential
use in a quality control environment.
The application of capillary zone electrophoresis (CZE) was studied through the
characterization of glycoprotein erythropoietin by its isoform distribution. The issue
of poor reproducibility of the CZE method described in the European
Pharmacopoeia was resolved by restoration of the fused silica capillary surface. The
restoration procedure developed in this study may be used for most protein
separations by CZE. Another significant aspect of this study was the development
of methodology for the removal of polysorbate 80 (a common formulation agent)
from a protein containing solution using a non-ionic trap. The technique has a wide
application for protein samples that contain polysorbate 80.
The application of micellar electrokinetic chromatography (MEKC) was studied
through the determination of residual cell culture media components including folic
acid, hypoxanthine, mycophenolic acid, nicotinic acid, riboflavin, and xanthine.
185
Acceptable limits of detection and quantitation were obtained. This study provided
an insight for biopharmaceutical impurity analysis. When CZE is insufficient for the
separation of charged analytes in complex matrices such as cell culture media,
MEKC provides a new approach for the separation of charged and neutral
compounds. This study gave an example of how to use MEKC to achieve ideal
separations.
The application of capillary gel electrophoresis was demonstrated through the
separation of a monoclonal antibody (IgG) using a running buffer comprised of
sodium dodecyl sulfate (SDS) and entangled linear polyacrylamide polymers. It was
observed that a slightly acidic sample buffer (pH 5.5-6.5) greatly decreased
thermally induced fragmentation of non-reduced IgG. Although the technique was
developed using a particular IgG (mAb) as a model compound, it can be used for
any type of IgG molecule. However, some of the alkylation conditions (amounts of
IAM) and reduction conditions (incubation temperature and duration) may require
minor optimization due to differences in formulation buffer composition.
The application of CE for the analysis of trace biopharmaceutical impurities has
been limited due to its low detection sensitivity [2]. For this reason, a study was
performed on sample injection length for sensitivity enhancement in MEKC using
the sweeping technique under conditions of suppressed electroosmotic flow. The
relationship between injection length and retention factor was derived and
experimentally confirmed. The injection length equation derived in this study can
be used for predicting the optimal injection length for a given analyte.
186
6.2 Future directions
In each of the studied applications, there are areas for potential improvements. For
the CZE separation of glycoprotein erythropoietin, the separation time could be
reduced by shortening the capillary length. According to equation (1.7), voltage has
to be increased to maintain the desired resolution. A combination of shortest
capillary length and highest voltage may be found through an experimental desire
study.
For the MEKC separation of residual cell culture media components, the detection
sensitivity may be further improved by using a bubble cell capillary and/or a longer
injection length. When increasing injection length, column efficiency needs to be
maintained at an acceptable level. During this study, a diode array detector was
employed and a spectrum of each analyte was taken (Appendix A). The optimum
detection wavelength for each analyte may be used to get the best sensitivity
(Appendix B). Although using multiple wavelength detection adds complexity to
the method, if sensitivity is an issue for a particular application, this approach may
be an option to get the needed sensitivity.
There are also some potential improvements for the capillary SDS gel
electrophoresis application. An alternative gel buffer (running buffer) could be
developed in-house for more optimal separations. The current gel buffer contains a
proprietary polymer formulation, which may be at risk if the supplier discontinues
this product.
187
The injection length study was conducted using only three model compounds. More
compounds and samples may be studied to select an optimal injection length and to
verify the correlation between injection length and retention factor.
In this study, only three CE separation modes were studied. Since CE also
comprises a series of other separation modes, such as capillary isoelectric focusing
(CIEF), capillary isotachophoresis (CITP), and capillary electroosmotic
chromatography (CEC), the applications of these techniques in biopharmaceutical
analysis are also worthy of study. This is especially true for CIEF, since some CIEF
applications have been reported for protein analysis.
This study on the quantitative biopharmaceutical applications of CE has
demonstrated that this powerful technique has potential widespread applications in
quality control laboratories for routine analysis and product release.
188
6.3
List of references
1.
Kostal, V.; Katzenmeyer, J.;Arriaga, E. A. Capillary electrophoresis in
bioanalysis. Anal. Chem. 2008, 80, 4533-50.
2.
Terabe, S.; Monton, M. R. N.; Le Saux, T.;Imami, K. Applications of
capillary electrophoresis to high-sensitivity analyses of biomolecules. Pure Appl.
Chem. 2006, 78, 1057-1067.
189
Appendix A: UV spectra of the six cell culture media components
mAU
16
14
12
10
8
6
4
2
200
220
240
260
280
nm
Figure A1. UV spectrum of hypoxanthine with λmax= 200 nm. Solvent: 20 mM NaH2PO4, 20
mM Na2B4O7, 40 mM SDS, pH 9.0. Analyte concentration: 0.1 mM. Collected on line using a
photodiode array detector with a 50 μm i.d. fused silica capillary.
190
mAU
14
12
10
8
6
4
2
200
220
240
260
280
Figure A2. UV spectrum of riboflavin with λmax = 270 nm. Solvent and solute
concentration as in Figure A1.
nm
191
mAU
20
17.5
15
12.5
10
7.5
5
2.5
200
220
240
260
280
Figure A3. UV spectrum of xanthine with λmax = 200 nm. Solvent and solute
concentration as in Figure A1.
nm
192
mAU
25
20
15
10
5
200
220
240
260
280
nm
Figure A4. UV spectrum of mycophenolic acid with λmax = 225 nm. Solvent and solute
concentration as in Figure A1.
193
mAU
20
17.5
15
12.5
10
7.5
5
2.5
200
220
240
260
280
Figure A5. UV spectrum of folic acid with λmax = 190 nm. Solvent and solute
concentration as in Figure A1.
nm
194
mAU
22
20
18
16
14
12
10
8
200
220
240
260
280
nm
Figure A6. UV spectrum of nicotinic acid with λmax = 190 nm. Solvent and solute
concentration as in Figure A1.
195
Appendix B: Detection sensitivity comparison of cell culture media components at fixed
wavength
Figure B1. Comparison of analyte absorbance at wavelength of 210 nm.
196
Figure B2. Comparison of analyte absorbance at wavelength of 222 nm.
197
Figure B3. Comparison of analyte absorbance at wavelength of 254 nm.
198
Figure B4. Comparison of analyte absorbance at wavelength of 275 nm.
199
VITAE
Junge Zhang was born on December 1, 1966 in Shandong, China. Junge received
his Bachelor’s of Science degree in materials sciences from Wuhan University of
Technology in 1989. Right after his graduation, he worked in Qingdao municipal
government for four years. Junge started his graduate study in Chemistry in 1994 at
the University of Louisiana at Monroe, Louisiana and received a Master of Science
degree in 1997. Thereafter, Junge pursued a career in the pharmaceutical industry as
an analytical chemist, first at Eisai USA, then at Pharmaceutical Intermediate
Division at PPG Industries, and recently at Johnson and Johnson R&D CMC
organization. At J&J, Junge was promoted from Senior Associate Scientist to
Research Scientist, and Senior Scientist. He received twice the Standard of
Leadership Award for his achievement in analytical development. In the fall of
2003, Junge continued his graduate study for Ph.D. as a part-time student at Drexel
University under the direction of Dr. Joe Foley. During his Ph.D. studies at Drexel,
he has presented portions of his research at the Eastern Analytical Symposium and
has published the two articles below with another in preparation for submission to
the Journal of Pharmaceutical and Biomedical Analysis:
Zhang J., Chakraborty U., Villalobos A., Brown J., and Foley J. P., Journal of
Pharmaceutical and Biomedical Analysis, 2009; 50:538-543
Zhang J., Chakraborty U., Foley J.P., Electrophoresis, 2009; 30:3971-3977