Scholarly article on topic 'Double-peak elution profile of a monoclonal antibody in cation exchange chromatography is caused by histidine-protonation-based charge variants'

Double-peak elution profile of a monoclonal antibody in cation exchange chromatography is caused by histidine-protonation-based charge variants Academic research paper on "Chemical sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Journal of Chromatography A
OECD Field of science
Keywords
{"Monoclonal antibody (mAb)" / "Cation exchange chromatography (CEX)" / Double-peak / "Histidine protonation" / DEPC / Hydroxylamine / "Peptide mapping" / "F(ab′)2 "}

Abstract of research paper on Chemical sciences, author of scientific article — Haibin Luo, Mingyan Cao, Kelcy Newell, Christopher Afdahl, Jihong Wang, et al.

Abstract We have systemically investigated unusual elution behaviors of an IgG4 (mAb A) in cation exchange chromatography (CEX). This mAb A exhibited two elution peaks under certain conditions when being purified by several strong CEX columns. When either of the two peaks was isolated and re-injected on the same column, the similar pattern was observed again during elution. The protein distribution between the two peaks could be altered by NaCl concentration in the feed, or NaCl concentration in wash buffer, or elution pH, suggesting two pH-associated strong-and-weak binding configurations. The protein distributions under different pH values showed good correlation with protonated/un-protonated fractions of a histidine residue. These results suggest that the double-peak elution profile associates with histidine-protonation-based charge variants. By conducting pepsin digestion, amino-acid specific chemical modifications, peptide mapping, and measuring the effects of elution residence time, a histidine in the variable fragment (Fab) was identified to be the root cause. Besides double-peak pattern, mAb A can also exhibit peak-shouldering or single elution peak on different CEX resins, reflecting different resins’ resolving capability on protonated/un-protonated forms. This work characterizes a novel cause for unusual elution behaviors in CEX and also provides alternative avenues of purification development for mAbs with similar behaviors.

Academic research paper on topic "Double-peak elution profile of a monoclonal antibody in cation exchange chromatography is caused by histidine-protonation-based charge variants"

ELSEVIER

Contents lists available at ScienceDirect

Journal of Chromatography A

journal homepage www.elsevier.com/locate/chroma

Double-peak elution profile of a monoclonal antibody in cation exchange chromatography is caused by histidine-protonation-based charge variants

Haibin Luoa, Mingyan Caob, Kelcy Newell3, Christopher Afdahla, Jihong Wangb, William K. Wanga, Yuling Lia *

a Purification Process Sciences, Medimmune LLC, One Medimmune Way, Gaithersburg, MD 20878, USA b Analytical Biochemistry Development, Medimmune LLC, One Medimmune Way, Gaithersburg, MD 20878, USA

ABSTRACT

We have systemically investigated unusual elution behaviors of an IgG4 (mAb A) in cation exchange chromatography (CEX). This mAb A exhibited two elution peaks under certain conditions when being purified by several strong CEX columns. When either of the two peaks was isolated and re-injected on the same column, the similar pattern was observed again during elution. The protein distribution between the two peaks could be altered by NaCl concentration in the feed, or NaCl concentration in wash buffer, or elution pH, suggesting two pH-associated strong-and-weak binding configurations. The protein distributions under different pH values showed good correlation with protonated/un-protonated fractions of a histidine residue. These results suggest that the double-peak elution profile associates with histidine-protonation-based charge variants. By conducting pepsin digestion, amino-acid specific chemical modifications, peptide mapping, and measuring the effects of elution residence time, a histidine in the variable fragment (Fab) was identified to be the root cause. Besides double-peak pattern, mAb A can also exhibit peak-shouldering or single elution peak on different CEX resins, reflecting different resins' resolving capability on protonated/un-protonated forms. This work characterizes a novel cause for unusual elution behaviors in CEX and also provides alternative avenues of purification development for mAbs with similar behaviors.

© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

CrossMark

ARTICLE INFO

Article history:

Received 18 August 2015

Received in revised form 27 October 2015

Accepted 1 November 2015

Available online 5 November 2015

Keywords:

Monoclonal antibody (mAb)

Cation exchange chromatography (CEX)

Double-peak

Histidine protonation

Hydroxylamine Peptide mapping F(ab')2

1. Introduction

Cation exchange chromatography (CEX) is well established as an important tool for purification of monoclonal antibodies (mAbs) [1], however, its utility is sometimes influenced by protein chemical features that create unexpected results [2]. For example, resin-protein interactions resulted in conformational changes on an IgGl that led to two elution peaks in Fractogel SO3-(M) CEX chromatography with unfolded proteins contributing to the late eluting peak [3-6]. Our previous work reported salt-induced reversible self-association of an lgG2 resulted in two elution peaks in multiple cation exchangers [7]. Zhang et al. reported a mAb that demonstrated strong retention on CEX due to a highly concentrated charge patch in the variable domain [8]. In addition to mAbs, human serum

* Corresponding author. Tel.: +1 301 398 4845. E-mail address: Liyu@Medimmune.com (Y. Li).

albumin was also reported to exhibit peak-splitting on CEX due to two binding configurations [9,10].

Among the 20 amino acids, histidine is the third least frequently occurring amino acid in proteins, but because of the near-neutral pKa of its imidazole group, histidine is involved in many biological functions [11,12]. Histidine can be either protonated (charged) or un-protonated (neutral) depending on the solution pH. Histidine has been found to be in the active sites of many enzymes [13], pH sensors for viruses' fusions into host cells [14] and pH switch controlling "open" and "close" status of ion channels [15]. The averaged pKa for a histidine residue is pH 6.0, but the actual pKa of a histi-dine residue depends on its surrounding environment. Histidine residues in a hydrophobic environment usually have a more acidic pKa (down shifted) and likely a slow protonation rate as well [16].

Diethyl pyrocarbonate (DEPC) specifically modifies solvent accessible (protein surface), un-protonated histidine residues. The DEPC modification can be reversibly removed by hydroxylamine. This highly specific and reversible chemical modification has become a valuable tool to study the role of histidine residue. In

http://dx.doi.org/10.1016/j.chroma.2015.11.008

0021-9673/© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

addition, DPEC modified histidine residues can be used to map out their precise location using mass spectrometry [12,17].

In the present work, we have systematically investigated the double-peak behaviors of a mAb A and identified the root cause to be a histidine protonation based charge variants with the histidine locating in its variable heavy chain domain.

2. Materials and methods

2.1. Chemicals, recombinant protein and cation exchanger resins

All the chemicals used in this study were obtained from J.T. Baker (Phillipsburg, NJ, USA). Buffers and cleaning solutions were prepared in house. The mAb A was expressed in Chinese hamster ovary (CHO) cells and produced by Medlmmune LLC (Gaithers-burg, MD, USA). It is a fully human monoclonal antibody (IgG4) composed of two identical heavy chains and two identical light chains, with an overall molecular weight of 148 kDa. The theoretical isoelectric point is 7.4. Unless noted otherwise, the protein feed material was purified using a Protein A affinity column and has purity of 98.5% monomer. POROS HS50 resins were from Applied BioSystems (Grand Island, NY, USA); Fractogel EMD SO3-(M) resins were from EMD Biosciences (Gibbstown, NJ, USA); Eshmuno CPX resins were from EMD Millipore (Billerica, MA, USA); CIMmultus SO3- column was from BIA Separations (Wilmington, DE, USA); Nuvia S resins were from Bio-Rad (Hercules, CA, USA), SP Sepharose Fast Flow, Source 30S, and MabSelect Sure (MSS) protein A resins were from GE Healthcare (Piscataway, NJ, USA); Toyopearl SP 650M resins were from Tosoh Bioscience (King of Prussia, PA, USA). The endoproteinase Lys-C was purchased from Promega (Madison, WI, USA).

2.2. Chromatography instrumentation and operations

Laboratory scale chromatographic experiments were carried out on a GE Healthcare ÄKTA Explorer 100 controlled by Unicorn software version 6.4 (GE Healthcare, Piscataway, NY, USA). Besides CIMmultus SO3- (commerically available as prepacked column), the CEX resins was packed into 0.66 cm inner diameter (ID) Omnifit® columns (Diba Industries, Danbury, CT, USA) to a bed height of 19 ±3 cm. Unless mentioned otherwise, the CEX columns were operated under the following conditions: 3 CV (column volume) the Equilibration buffer (50 mM sodium acetate, pH 5.0) for pre-loading equilibration; loading under pH 5.0 using 5 mg/mL resin as load challenge; 3 CV Wash buffer after loading; followed by 20 CV 0-500 mM NaCl linear gradient elution (LGE) at pH 5.0 (buffered by 50 mM NaAc); 3 CV of 50 mM Tris-HCl, pH 7.0, 1 M NaCl for strip; 3 CV 0.1 M NaOH for sanitation; 3 CV 20% ethanol for column storage; all steps used 5 min as residence time. The absorbance of the protein was monitored at A280 by the built-in ÄKTA probe. The elution was fractionated in one CV fraction based on A28o collection criteria of >50 mAu. In-line pH and conductivity were also monitored for all test runs. Unless mentioned otherwise, chromatographic experiments were carried out at room temperature. Protein concentration was measured using a NanoDrop 2000 (Thermo Fisher Scientific, Wilmington, DE. USA). In the experiment of re-purifying the first and second peak proteins from the Poros HS50 chromatography, the collected pools were buffer exchanged into 50 mM sodium acetate pH 5.0 prior to loading. In the experiments of obtaining elution profile for different pH values, pH levels from 4.0 to 5.5 were provided by 50 mM NaAc and pH levels from 5.5 to 6.5 were provided by 50 mM sodium phosphate, respectively. In the experiments testing temperature effect on elution profile, the POROS HS50 resins were packed in a jacketed column with the temperature controlled by PolyScience circulating water bath

(Burlington, VT, USA). In the POROS HS50 purification of the DEPC modified and hydroxylamine reversed samples, all samples were quenched by adding 6 mM imidazole before loading and 4 mg load mass were used.

2.3. Analytical size-exclusion chromatography (HP-SEC)

Purity analysis was performed using a TSK-GEL G3000SWXL column (7.8 mm x 30 cm) from Tosoh Bioscience (King of Prussia, PA, USA) with an Agilent HPLC system HP1200 from Agilent Technologies (Santa Clara, CA, USA). The column was equilibrated at a flow rate of 0.5 mL/min with 78 mM sodium phosphate, 0.5 M sodium chloride, pH 7.4. All collected fractions were injected and eluted isocratically with the same buffer. The eluted protein was monitored by UV absorbance at 280 nm. Samples were 0.22 |im filtered and injected at a mass load of 0.25 mg. All injected volumes were less than 0.7% of the column volume. The monomer, dimer, and aggregate were estimated by integrating the chromatograms.

2.4. Diethyl pyrocarbonate (DEPC) modification and reversal by hydroxylamine

DEPC was purchased from Sigma-Aldrich (St. Louis, MI, USA) and stored desiccated at 4 ° C to minimize decomposition by hydrolysis. DEPC stock solution was freshly prepared by diluting 100% DEPC solution with cold absolute ethanol at a ratio of 1:19 (v/v). DEPC modification to mAb A histidine residues was initiated by adding the DEPC stock solution to mAb A (at 1 mg/mL) in 50 mM NaAc, pH 5.0. A low ratio of 2.5 |L DEPC stock versus 1 mL of protein solution was used to make the final DEPC concentration to be around 1 mM. The reaction temperature was controlled by incubation in a 25 °C water bath. The reaction process was monitored by measuring UV242 on 8453 UV dissolution systems from Agilent Technologies (Santa Clara, CA, USA). In the experiments showed in Fig. 4, 4mL samples (1), (2) and (3) were taken at 0,10 and 20 min after DEPC modification started. Each sample was added 40 | L of 300 mM imidazole (pH 5.0) stock to quench the unreacted DEPC. The samples were conditioned and then subjected to purification by the POROS HS50 column. The removal of DEPC modification was performed by adding hydroxylamine at a ratio of 10 |L 2 M hydroxylamine (pH 5.2) per 1 mL protein solution. 4mL samples (4) and (5) were taken at 20 and 40 min after hydroxylamine addition. In experiments showed in Fig. 7A, a ratio of 17 |L DEPC stock per 1 mL protein solution was used get 6 mM DEPC final concentration. The reaction was allowed for 30 min until the UV242 signal reached a plateau. The modified sample was purified on POROS HS50. The major peak was collected and analyzed by mass spectrometry pep-tide mapping.

2.5. Far-UV circular dichroic (CD) and fluorescence spectroscopy

Far-UV CD measurements were performed on a Jasco-815 instrument (Easton, MD, USA). The native and modified mAb A proteins were prepared in 50 mM NaAc pH 5.0 and analyzed in a 0.1 cm pathlength cuvette (1-Q10, Starna, Atascadero, CA). Spectra were collected at standard sensitivity range (100 mdeg) with 0.5 nm data pitch, 10 nm/min scanning rate, and 8 s integration time in the range from 200 to 250 nm. Five replicates were averaged for each sample.

8-Anilino-1-naphthalenesulfonic acid (ANS) was from Sigma-Aldrich (St. Louis, MI, USA). In the ANS binding experiments, mAb A was prepared at 0.4 mg/mL in the corresponding buffers. A 1.2 mM ANS stock solution was added to each sample to a final concentration of 120 |M and ANS binding was allowed for 2 h in dark at ambient temperature. The fluorescence signal of each sample was measured by Spectra Max M2 (Molecular Devices,

Volume (mL)

Fig. 1. mAb A demonstrated double-peak elution pattern in POROS HS50 during linear gradient elution. (A) Elution profile of mAb A over a 20 CV linear elution gradient (0-500 mM NaCl) on POROS HS50 at pH 5.0. Load challenge was 40 mg/mL resin. The black solid line represents the A280 trace. The solid squares represent purity of each fraction. (B) Separate re-purification of the proteins from each peak led to similar double-peak pattern. (C) Elution profiles for different load challenges. Different load challenge was achieved by changing injection volume. The equilibration and elution buffer used 50 mM NaAc, pH 5.0 as the buffer system for above runs.

Sunnyvale, CA, USA). The excitation wavelength was 360 nm, and emission was monitored at 500 nm. For the intrinsic fluorescence measurement, the samples at 0.4 mg/mL concentration were prepared in the corresponding buffers then measured by EnVision Plate Readers (Perkin Elmer, Waltham, MA, USA). The excitation wavelength was 280 nm, and emission was monitored at 338 nm. The buffers used in Fig. 5D were made by titrating 50 mM NaAc pH 4.5 with 50 mM sodium phosphate pH 6.5. All experiments were run in triplicate.

2.6. Preparation ofF(ab')2 domain of mAb A using pepsin digestion

The agarose immobilized pepsin was purchased from Sigma-Aldrich (St. Louis, MO, USA). mAb A was digested by pepsin-agarose (1 mg pepsin-agarose dry powder per 6 mg mAb A) at pH 4.0 in a 37 °C water bath for overnight. The digested product was adjusted to pH 7.0 and then was purified by flowing through a MabSelectSure (MSS) column pre-equilibrated at pH 7.0. The intact mAb A and Fc domains were retained on the MSS column. The flow through was collected and was further purified by Toyopearl SP 650M to show a single F(ab')2 peak on HP-SEC.

2.7. Peptide mapping with Mass Spectrometry

The sample was denatured by 6 M guanidine hydrochloride (Sigma-Aldrich, St. Louis, MI, USA) in 50 mM phosphate buffer, pH 7.0 at 37°C for 30 min. After denaturation, the sample was diluted 2.5-fold with 100 mM phosphate buffer containing 0.06 mM EDTA at pH 7.0. Lys-C was added at a 1:10 Lys-C: protein ratio and the reaction mixture was incubated at 37 °C for 16 h. In order to get

complete digestion, additional Lys-C was added at the same ratio and further incubated for a 4 h at 37 °C. Following Lys-C digestion, sample was reduced with DTT at 37°C for 15 min. The digested peptides were separated by a reverse phase Zorbax C18 column on Waters UPLC (Milford, MA, USA) followed by analysis using a UV detector and an on-line Fusion mass spectrometer from Thermo Fisher scientific (Waltham, USA). The RP-UPLC mobile phase A was 0.02% TFA in water and the mobile phase B was 0.02% TFA in acetonitrile. The histidine modification levels were calculated by integrating the modified and un-modified peak areas of selected ion chromatograms from the ESI-MS (electrospray ionization mass spectrometry).

3. Results and discussion

3.1. Highly purified mAb A exhibits double-peak elution behavior in POROS HS50

As shown in Fig. 1A, a highly purified mAb A exhibits double-peak pattern of the anticipated major peak in POROS HS50 (strong CEX) during salt linear gradient elution. Both peaks contained mainly pure mAb A proteins, determining by HP-SEC analysis. When isolated and re-injected on the column under the same conditions, the proteins from each peak reformed similar elution pattern (Fig. 1B). Additional purifications of peak fractions resulted in the same elution profile (data not shown), suggesting interchange of the proteins in the two peaks. No significant difference was detected on the proteins from the two peaks by biophysical analyses such as isoelectric focusing, mass spectrometry, dynamic light scattering and differential calorimetry scanning (data not shown). The elution pattern was observed to remain similar

L 1500

© 1000

-UV280 -----Conductivity Control 11 j

100 mM NaCl 1 . ./O.......

80 mM NaCl J

V...../

150 200 250

Volume (mL)

Control ,..-/

Load in the presence of J 100 mM NaCl I / J _

Load in the 0 presence of 1 100 mM ArgHCl / / ..... .- f

s ■s c

50 100 150

Volume (mL)

Fig. 2. mAb A may have two binding configurations to POROS HS50 resins. (A) Effects of 100 and 80 mM NaCl salt wash on elution profile for mAb A. Salt concentration was gradually increased to and maintained at the desired level for 6 CV. 3 CV equilibration was applied before the linear salt gradient elution. (B) Effects of the presence of 100 mM NaCl and ArgHCl during load on elution profile. Equilibration buffer matched the load was applied before loading. All runs were operated at pH 5.0 and using 5 mg/mL load challenge.

regardless of different loading (1-20 mg/mL) changes (Fig. 1C). Therefore, load challenge does not play a role. We previously found salt-induced reversible self-association resulted in peak-splitting of an IgG2 in several strong CEX resins [7]. However, reversible self-association was not detected for mAb A under the CEX conditions tested using several analytical techniques (data not shown). Guo and Carta recently reported that an IgG1 exhibited two elution peaks following a 1000 min hold on the CEX column after loading [3-5]. They found the late eluting peak contained aggregates and was due to resin mediated protein unfolding. For mAb A, no significant aggregation was found during the POROS HS50 purification suggesting that on-column aggregation is not a likely cause.

3.2. Two different binding configurations with weak and strong interactions are the most likely cause of the first and second elution peak respectively

Vitol et al. reported that human serum albumin (HSA) showed two elution peaks that represent weak and strong binding configurations [9,10]. In order to test whether mAb A double-peak

Fig. 3. The double-peak pattern of mAb A may be associated with a histidine residue. (A) Elution profiles for mAb A in POROS HS50 under pH 4.7, pH 5.0, pH 5.2 and pH 5.4. (B) The peak sizes overlapped well with theoretical neutral and proto-nated fractions at different pH values. The theoretical curves were plotted using Henderson-Hasselbalch equation with an assumptive pKa of pH 5.3.

phenomena follows a similar mechanism, two sets of experiments were carried out.

In the first set of experiments (Fig. 2A), after mAb A was bound on the POROS HS50 column, a wash with 100 mM NaCl (the salt concentration for eluting peak 1 in Fig. 1 A) was applied prior to the linear gradient elution. As expected, this 100 mM NaCl wash eluted some proteins off the column. The 100 mM NaCl wash seemed to preferentially elute the proteins in the first peak while the second peak remained similar to that of the control run. When the salt concentration in the wash was decreased to 80 mM NaCl, no protein was eluted from the column. However, the first peak became very small and the second elution peak became proportionally larger, compared to the control run. These results support the hypothesis that mAb A has two binding configurations on POROS HS50. Proteins bound through the weak binding configuration were eluted by a NaCl concentration of 80 mM, but they could re-bind to the resins through the strong binding configuration.

In the second set of experiments (Fig. 2B), binding of mAb A to the column occurred in the absence or presence of 100 mM NaCl. The presence of 100 mM NaCl during loading resulted in smaller peak 1 and larger peak 2. Compared to the control run (mAb A binding happened in the absence of 100 mM NaCl), the loss on the first peak was equal to the gain on the second peak, suggesting the presence of 100 mM NaCl made mAb A adapt the strong binding configuration. These data further supported our hypothesis that the first and second peak represents the weak and strong binding

Fig. 4. The double-peak pattern is associated with a histidine residue on protein surface. (A) Reaction schematic for DEPC chemical modification on histidine residue and hydroxylamine removal of DEPC modification. DEPC modifies the imidazole side group of neutral surface histidine to produce N-carboethoxyhistidine. N-Carboethoxyhistidine can be reversed to Histidine by hydroxylamine. N-Carboethoxyhistidine has characteristic UV242 absorbance. (B)UV242 absorbance of 1 mg/mLmAb A sample (in 50 mM NaAc, pH 5.0) that was modified by DEPC for 20 min then the modification was reversed by hydroxylamine for 40 min. The reactions were carried out at ambient room temperature. The sample was measured by UV spectrum on 1 min interval basis. (C) Elution profiles in POROS HS50 for DEPC modified samples and hydroxylamine reversed samples. Samples (1), (2), and (3) was mAb A modified by DEPC for 0,10 and 20 min, respectively; while samples (4) and (5) was DEPC modified mAb A reversed by hydroxylamine for 20 and 40 min, respectively. 4 mg sample was loaded in each POROS HS50 run. (D) CD spectra for native and DEPC modified mAb A.

configurations, respectively. mAb A can freely bind on the POROS HS50 reins by both configurations when there was no salt present during loading. When the weak binding configuration was not favorable (such as 100 mM NaCl was present), most mAb A proteins bind to the column through the strong binding configuration.

Additionally, the presence of 100 mM arginine hydrochloride (ArgHCl) showed similar effects to 100 mM NaCl. ArgHCl has been reported to efficiently weaken resin-protein binding mediated unfolding and therefore decreased the second elution peak in the case for the aforementioned unstable lgG1 [7,8]. In this case, ArgHCl showed opposite effects for mAb A, i.e. it enlarged the second peak. These results further ruled out the possibility of resin mediated unfolding/aggregation for mAb A's double-peak pattern.

3.3. Protein distribution is modulated by elution pH and when plotted against pHfits to the Henderson-Hasselbalch equation with a pKa of 5.3

To evaluate whether pH value impacts on mAb A's adaption of the two binding configurations in POROS HS50, purification of mAb A was performed under a series of pH conditions (pH 4.3-6.5) using the same linear salt gradient elution. The double-peak phenomenon was observed in all tested runs (Fig. 3A) but was variable in extent. The ratio of the two peaks showed high

dependence on the operating pH condition. Higher pH value resulted in larger early-eluting peak but smaller late-eluting peak while low pH conditions favored the late elution peak. At pH 4.6, the late-eluting peak dominated; at pH 5.0, the late-eluting peak still dominated but the early-eluting peak became larger; at pH 5.2, two peaks were roughly equal in size; at pH 5.4 and pH 5.8, the early-eluting peak became dominant. This trend was more obvious in the plotting of peak percentage against the operating pH values (Fig. 3B). The resulting data can be fitted into Henderson Hasslbalch equation that describes the pronation/de-protonation of a charge group. The fitted pKa is about pH 5.3. The curve for the first peak percentage resembled the deprotonation curve while the second peak resembles the protonation curve.

Among the charged groups in mAb A, histidine residue is a proton accepting amino acid with a pKa of pH 6.0. pH 6.0 is actually an averaged value and pKa value of a histidine residue has been found to range from pH 4.5 to 7.5, depending on its surrounding environment [16]. A histidine residue is positively charged when it is in a protonated state and neutral in un-protonated state. Protein isoforms with 1+ charge difference, such as native mAb and its deamidated forms (on asparagine), are known to cause several elution peaks on CEX [18]. Based on these, we hypothesized that a histidine residue is associated with the double-peak pattern of mAb A.

Residence time (min)

Fig. 5. The double-peak causing histidine in mAb A has slow rate of protonation. (A) mAb A elution profiles in POROS HS50 for different elution flow rates. Only the elution flow rate was varied to get different residence time while other conditions remain the same. (B) The plotting of the first elution peak percentage against residence time for 5 mg/mLand 0.5 mg/mL load challenge. The peak 1 percentage was calculated from elution peak area integration. (C)The plotting of the first elution peak percentage against residence time for different operating temperatures. (D) ANS binding and intrinsic fluorescence data under different pH values. mAb A concentration in these samples are 0.4 mg/mL.

3.4. Solvent accessible histidine with a pKa of 5.3 is the most likely cause of the double-peak pattern of mAb A

We attempted to test the histidine hypothesis through a series chemical modifications that were targeted at solvent accessible histidine residues and measuring the effect on the double-peak phenomenon. Diethyl pyrocarbonate (DEPC) specifically modifies histidine residues on protein surfaces at concentrations less than 10 mM [12]. As shown in Fig. 4A, DEPC modifies neutral histidine residues to give the modified product N-carboethoxyhistidine; this product can be reversed back to histidine by hydroxylamine. N-Carboethoxyhistidine also has characteristic absorbance at UV242 allowing the modification and reversal reactions to be monitored by following UV242 absorbance.

Many published studies established that DEPC only modifies histidine residue in un-protonated (neutral) state [12]. Neutral pH is usually used to maximize modification because most histidine residues are in the un-protonated state at neutral pH. However, we selected an acidic pH for the DEPC modification for mAb A to selectively modify histidines with a pKa value near pH 5.3 and reduce modification on other histidine residues that have a more standard higher pKa.

As shown in Fig. 4B, after 1 mM DEPC was added to mAb A protein solution, UV242 signal of the reaction mixture increased quickly over time, indicating the quick production of the DEPC modified histidine product. After 10 mM hydroxylamine was added to the reaction mixture, the UV242 signal decreased with time, indicating the reversal of N-carboethoxyhistidine back to histidine. The reversal of N-carboethoxyhistidine by hydroxylamine was slower than DEPC modification, consistent with the published reports [17].

Five samples were taken at different time points indicated in Fig. 4B. lmidazole was added to each sample to quench the unre-acted DEPC before these samples were run in the POROS HS50 column (Fig. 4C). Compared to the control sample, the 5-min modified sample exhibited larger early elution peak and smaller late eluting peak; the 10-min modified sample had dominant early eluting peak and the late eluting peak became very small. On the contrary, double-peak reappeared on the hydroxylamine-reversed samples. Additionally, the DEPC modified and native mAb A were also analyzed by far-UV CD (Fig. 4D). Their similar CD spectra indicate that DEPC modification on mAb A resulted in no significant changes on the overall structure, consistent with the published reports [17].

These results suggested that the double-peak pattern of mAb A in POROS HS50 is associated with histidine residue on the protein surface. lts un-protonated and protonated states correlate with the early eluting peak (the weak binding configuration) and late eluting peak (the strong binding configuration), respectively.

3.5. The low pKa histidine has a slow protonation rate

Protonation and deprotonation of solvent accessible residues is a typically dynamic in nature, with rapid interchange occurring between the two forms to reach equilibrium. When proteins bound in weak binding configuration (the histidine residue is in un-protonated state) were eluted off the resins and entered the mobile phase, the histidine residue on some proteins should protonate to maintain the equilibrium in the mobile phase. The proteins with the newly protonated histidine should be able to re-bind to the

resins. If the protonation is sufficiently fast, all of the early eluting proteins should protonate and bind back to column before going out of the POROS HS50 column, and then the first peak will not be observed. Why then, did mAb A show the first peak? One explanation is that the protonation of mAb A is slow and the timescale of interchange between the two forms is on the time scale of separation for the HS50 column. If this is the case, elution residence time should demonstrate significant impact on the double-peak pattern.

In order to test this hypothesis, the elution flow rate was varied to achieve different residence time while other conditions remained the same. As shown in Fig. 5A, elution residence time did show a significant impact. The first peak decreased and the second peak increased if a longer residence time was used. When 1.3 min was used as the elution residence time, the first and second peaks are equally large; when prolonged residence time of 30 min was used, the first peak became much smaller and the second peak became dominant. Fig. 5B showed the plotting of the sizes for the early eluting peak again the elution residence times. Obviously, longer residence leads to smaller early eluting peak. One may argue that binding capacity is lower at higher flow rate and the rebinding can be limited by binding capacity instead of protonation rate. This possibility was ruled out because similar results were obtained for 5 and 0.5 mg/mL binding capacities (Fig. 5B). Diffusive mass transport of mAb A inside the POROS HS50 beads seems contribute slightly to the slow conversion because such conversion on monolithic CEX column (CIMmultus SO3-) was 2 times faster (data not shown). More study is underway to understand the differences between the two columns. Although the conversion on the above monolithic CEX column is faster, a complete conversion still took more than 10 min. Therefore, these results suggested the double-peak causing histidine residue in mAb A has slow protonating rate.

We also found that operating temperature has significant impacts on the double-peak pattern (Fig. 5). Protein diffusion is usually slower at lower temperatures. If this is the case, the first peak should be smaller at low temperature when using the same residence time. However, Fig. 5C showed the opposite results, suggesting slower protonation under lower temperature. This is consistent with the published results about temperature dependence of histidine protonation [19]. Although it can be affected by factor like temperature, the conversion rate between the un-protonated and protonated states is typically fast and in less-than-a-second scale [20]. Protonation-coupled protein conformational transitions are ubiquitous in biology, such as protein folding, unfolding, and catalysis [21]. Slow conformational changes are found to be associated with slow histidine proto-nation/deprotonation [21]. We hypothesized that conformational change may couple with the protonation/de-protonation of this histidine.

Conformational changes at pH values around the pKa of pH 5.3 were probed by extrinsic (ANS) and intrinsic fluorescence measurements. Both the ANS and intrinsic fluorescence data suggested small conformational changes when pH value altered from pH 6.5 to pH 4.5. Specifically, mAb A surface hydrophobicity (open squares in Fig. 5D) slightly increased with decreasing pH values. While possible, we cannot be certain whether the observed conforma-tional changes are associated with protonation/deprotonation of the histidine and further study would be needed to demonstrate this conclusively.

3.6. Mapping of the histidine residue in mAbA's primary sequence suggest a rationale for the low pKa and slow protonation rate

In the primary protein sequence of mAb A, the heavy and light chain has 8 and 2 histidine residues, respectively. As shown in Fig. 6A, five histidine residues (#1-5) are in the F(ab')2 domain and five (#6-10) are in the Fc domain. However, only the #1 histidine is

Fig. 6. The histidine residue is from the variable domain of mAb A. (A) Schematic illustration of histidine residues in mAb A. 8 histidine residues in the heavy chain and 2 in the light chain. (B) The purified F(ab')2 of mAb A exhibited two elution peaks on POROS HS50 during 0-500 mM NaCl LGE. The inlet shows the HP-SEC profile of the purified F(ab')2. The preparation of the purified F(ab')2 was described in the Materials and Methods section.

in the variable domain while the other 9 histidine residues (#2-10) are in constant domains. Most mAbs have these 9 histidine residues in their primary sequence but they don't usually exhibit double-peak pattern in cation exchanger chromatography. Therefore, we hypothesized that #1 histidine was responsible for the double-peak pattern.

To test this, the F(ab')2 from mAb A was generated by pepsin digestion, purified, and then tested on POROS HS50 (Fig. 6B). Like mAb A, the purified F(ab')2 also exhibited a similar elution profile, supporting the #1 histidine hypothesis but this did not rule out the involvement of #2-5 histidine residues.

Peptide mapping has been used to successfully identify DEPC modified histidine residues on variety of proteins [17,22]. Therefore, we further used peptide mapping to identify the double-peak causing histidine residue. In order to produce homogenously modified sample for peptide mapping, the DEPC modification reaction was performed at pH 5.0 and prolonged to reach an UV242 plateau (the inset of Fig. 7A). This DEPC modified mAb A exhibited a dominant elution peak on POROS HS50 (Fig. 7A), indicating a complete DEPC modification on the expected histidine residue. The modified sample was analyzed by LC-MS peptide mapping and the native mAb A sample was used as a control. Peak differences in UV profiles of these two samples were found to be mainly related to modification on histidine residues. Several peptides showed significant differences (Fig. 7B): the peptides labeled as P1-P6 are the peptides containing histidine residue (from the native mAb A sample); while the peptides P1 '-P6' contain modified histidine residue (from the DEPC modified mAb A sample). The modified histidine residues were identified by comparing the MS/MS spectra of the corresponding native and modified peptides. Fig. 7C and D demonstrated the

Fig. 7. Identification of the histidine residue by MS peptide mapping. (A) Generation of homogenous sample for peptide mapping by allowing complete reaction and further enrichment by POROS HS50. The inlet is the UV242 absorbance for 1 mg/mL mAb A modified by 6 mM DEPC at room temperature for30min. The UV242 reached a plateau at 25 min. (B) Mirror image of LC chromatograms (UV) from the native and DEPC modified mAb A. Compare to the native sample, the DEPC modified sample exhibited some new peptides (P1'-P6') resulted by DEPC modification on histidine residue(s). (C) MS/MS of the peptide with the #1 histidine residue from native mAb A. (D) MS/MS of the peptide with the modified #1 histidine residue from the DEPC modified mAb A. The fragment ions Y10, Y23, Y24, and Y28 are from the DEPC group on DEPC modified #1 histidine. These ions were not seen in the native mAbA.

Table 1

Modification levels of mAb A histidine residues determined by MS peptide mapping.

Histidine residue Location Mod

#1 F(ab')2 VH.CDR 94.5

#2 F(ab')2_CH1 78.8

#3 F(ab')2_CH1 9.7

#4 F(ab')2.CL 26.8

#5 F(ab')2.CL 54.6

#6 FC_CH2 40.1

#7 FCCH2 14.9

#8 FC.CH3 21.0

#9 FCCH3 23.8

#10 FC_CH3 9.4

Primary sequence context'

e4^ н~Ф~Ф~ ^y" фнф^У" ффнф^у" ^y" фнф^У"

~0~0~H00

А^нф^у"

~0~0~Н00 ФФИФФ

a The modification of all histidine's calculated by the modified and unmodified peak area using selected ion chromatograms from ESI-MS detection peaks eluting from HPLC separation of Lys_C fragments containing histidine.

b The symbol ""0" " represents non-polar residue (I. F. V. L, W. M. A. G. and P) and the symbol "Ф" represents polar and charged residue (D. E. K, R. N. Q, S. T. H. Y. and C).

Elution Volume (CV) Elution Volume (CV)

Fig. 8. Elution profiles of mAb A in different strong CEX resins. The evaluated resins were packed in 0.66 cm column with similar height. CIMmultus SO3- column is a pre-packed column. Load challenge was 5 mg/mL resin. The equilibration and elution buffer used 50 mM NaAc, pH 5.0 as the buffer system. The black arrows indicate the shouldered peaks.

identification of the modified #1 histidine residues. The fragment ions Y10, Y23, Y24, and Y28 in Fig. 7D revealed the modification on the #1 histidine residue. Modification levels were calculated using the corresponding selected ion chromatograms from ESI-MS detection and the results were summarized in Table 1. The #1 histidine residue in one of the CDRs (complementary determining regions) in the heavy chain had the highest modification level of 94.5%. Another histidine residue (#2) from the F(ab')2 domain, also had high modification level of 78.8%. The rest of the histidine residues (#3-10) had much lower modification levels and thus were ruled out.

The combination of the two experimental methods used to map out the precise histidine responsible for double-peak pattern support the #1 histidine hypothesis and ruled out all other possible histidine residues except for the #2 histidine. We argue that the #1 histidine residue is the more likely residue because it is surrounded by 4 hydrophobic residues in primary structure, consistent with its estimated acidic pKa (pH 5.3) and slow protonation rate. Furthermore, the #2 histidine is located in the constant domain and

thus commonly occurring in other mAbs that do not all display the double-peak pattern.

3.7. The amount of dual-binding can be modulated by the CEX resins used to purify mAb A

Several CEX resins were tested to see how they resolved the protonated/un-protonated forms. As showed in Fig. 8, mAb A showed double-peak pattern in POROS HS50, Fractogel SO3-(M), Eshmuno CPX and even on monolithic CIMmultus SO3- columns; peak-shouldering on Nuvia S, Source 30S, and Sepharose SP FF columns; but a singular peak on Toyopearl SP 650M column. We believe that these results reflect the resolutions of different CEX resins on the protonated and un-protonated forms. The results suggest resin screening may offer a relatively quick solution to the slow rate of histidine protonation caused double-peak pattern, but this solution likely result in a process removing many of the efficiencies that go along with better behaved mAbs.

4. Conclusion

In this work, we systematically investigated an unusual CEX elution profile of mAb A. The early and late elution peaks correlate with the un-protonated (neutral) and protonated (charged) states ofthe histidine, respectively. The histidine was identified in the highly variable CDR on the heavy chain. Several operating parameters, such as pH, temperature, residence time, and salt concentration of the feed, play a key role on the extent of the double-peak pattern. This double-peak pattern was seen on multiple CEX resins. Depending on the CEX resins' separation capacities, mAb A could show double-peak or peak-shouldering or a major peak. Therefore, this issue can be avoided by resin screening.

Although the double-peak pattern was observed on an IgG4, the double-peak pattern could occur on any IgG isotype containing a similar histidine residue. It is not uncommon to have a histidine residue in CDRs for a mAb. Therefore, these findings can be leveraged to aid development of other mAbs and predict the likelihood of a similar double-peak pattern for a new molecule. This work contains useful information to guide protein engineering for better design of the molecules to avoid histidine causing double-peak pattern. The understanding is also valuable for assessing the implication of the observed phenomena on product quality and for optimization of the purification process.

Acknowledgements

This study was sponsored by MedImmune LLC, the global bio-logics R&D arm of AstraZeneca. We would like to thank Kevin D. Stewart, Wai Keen Chung, Min Zhu, Matthew Dickson and Justin Weaver from the Purification Process Sciences Department for their contributions to this work. We also thank Alan Hunter and Jifeng Zhang for critical review and suggestions.

References

[1] L. Yu, L. Zhang, Y. Sun, Protein behavior at surfaces: orientation, conformational transitions and transport, J. Chromatogr. A 1382 (2015) 117.

[2] A.A. Shukla, J. Thommes, Recent advances in large-scale production of monoclonal antibodies and related proteins, Trends Biotechnol. 28 (2010) 253.

[3] J. Guo, G. Carta, Unfolding and aggregation of a glycosylated monoclonal antibody on a cation exchange column. Part II. Protein structure effects by hydrogen

deuterium exchange mass spectrometry, J. Chromatogr. A 1356 (2014) 129.

[4] J. Guo, S. Zhang, G. Carta, Unfolding and aggregation of a glycosylated monoclonal antibody on a cation exchange column. Part I. Chromatographic elution and batch adsorption behavior, J. Chromatogr. A 1356 (2014) 117.

[5] R. Gillespie, T. Nguyen, S. Macneil, L. Jones, S. Crampton, S. Vunnum, Cation exchange surface-mediated denaturation of an aglycosylated immunoglobulin (IgG1), J. Chromatogr. A 1251 (2012) 101.

[6] J. Guo, G. Carta, Unfolding and aggregation of monoclonal antibodies on cation exchange columns: effects of resin type, load buffer, and protein stability, J. Chromatogr. A 1388 (2015) 184.

[7] H. Luo, N. Macapagal, K. Newell,A. Man,A. Parupudi,Y.Li, Effectsofsalt-induced reversible self-association on the elution behavior of a monoclonal antibody in cation exchange chromatography, J. Chromatogr. A 1362 (2014) 186.

[8] L. Zhang, W. Lilyestrom, C. Li, T. Scherer, R. van Reis, B. Zhang, Revealing a positive charge patch on a recombinant monoclonal antibody by chemical labeling and mass spectrometry, Anal. Chem. 83 (2011) 8501.

[9] A. Voitl, A. Butte, M. Morbidelli, Behavior of human serum albumin on strong cation exchange resins: II. Model analysis, J. Chromatogr. A 1217 (2010) 5492.

10] A. Voitl, A. Butte, M. Morbidelli, Behavior of human serum albumin on strong cation exchange resins: I. Experimental analysis, J. Chromatogr. A 1217 (2010) 5484.

11] S.M. Liao, Q.S. Du, J.Z. Meng, Z.W. Pang, R.B. Huang, The multiple roles of histi-dine in protein interactions, Chem. Cent. J. 7 (2013) 44.

12] V.L. Mendoza, R.W. Vachet, Probing protein structure by amino acid-specific covalent labeling and mass spectrometry, Mass Spectrom. Rev. 28 (2009) 785.

13] A. Schonichen, B.A. Webb, M.P. Jacobson, D.L. Barber, Considering protonation as a posttranslational modification regulating protein structure and function, Annu. Rev. Biophys. 42 (2013) 289.

14] C.M. Mair, T. Meyer, K. Schneider, Q. Huang, M. Veit, A. Herrmann, A histidine residue ofthe influenza virus hemagglutinin controls the pH dependence ofthe conformational change mediating membrane fusion, J. Virol. 88 (2014) 13189.

15] A.N. Thompson, D.J. Posson, P.V. Parsa, C.M. Nimigean, Molecular mechanism of pH sensing in KcsA potassium channels, Proc. Natl. Acad. Sci. U. S. A. 105 (2008) 6900.

16] G.R. Grimsley, J.M. Scholtz, C.N. Pace, A summary ofthe measured pK values of the ionizable groups in folded proteins, Protein Sci. 18 (2009) 247.

17] Lundblad R.L. Chemical Reagents for Protein Modification, 4th edn, 42014.

18] Y. Tao, G. Carta, G. Ferreira, D. Robbins, Adsorption of deamidated antibody variants on macroporous and dextran-grafted cation exchangers: I. Adsorption equilibrium, J. Chromatogr. A 1218 (2011) 1519.

19] S. Bhattacharya, J.T. Lecomte, Temperature dependence of histidine ionization constants in myoglobin, Biophys. J. 73 (1997) 3241.

20] M.A. Slifkin, S.M.Ali, Measurements of protonation reaction kinetics of histidine using a chemical relaxation technique, J. Mol. Liq. 28 (1984) 165.

21] C. Shi, J.A. Wallace, J.K. Shen, Thermodynamic coupling of protonation and conformational equilibria in proteins: theory and simulation, Biophys. J. 102 (2012) 1590.

22] J.L. Dage, H. Sun, H.B. Halsall, Determination of diethylpyrocarbonate-modified amino acid residues in alpha 1-acid glycoprotein by high-performance liquid chromatography electrospray ionization-mass spectrometry and matrixassisted laser desorption/ionization time-of-flight-mass spectrometry, Anal. Biochem. 257(1998) 176.