Scholarly article on topic 'Non-immunospecific association of immunoglobulin G with chromatin during elution from protein A inflates host contamination, aggregate content, and antibody loss'

Non-immunospecific association of immunoglobulin G with chromatin during elution from protein A inflates host contamination, aggregate content, and antibody loss Academic research paper on "Biological sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Journal of Chromatography A
OECD Field of science
Keywords
{"Protein A" / IgG / Chromatin / "Host contamination" / Aggregation / Recovery}

Abstract of research paper on Biological sciences, author of scientific article — Pete Gagnon, Rui Nian, Yuansheng Yang, Qiaohui Yang, Chiew Ling Lim

Abstract Monoclonal IgG at pH 3.5 expressed a tendency to self-associate and associate non-specifically with surfaces, including the surfaces of precipitated chromatin heteroaggregates. The tendency was elevated with protein A-eluted IgG still in elution buffer (100mM acetate, pH 3.5). Association of IgG with chromatin elements under protein A elution conditions amplified host protein contamination of the elution fraction about 15-fold, caused formation of aggregates that persisted after pH neutralization, and imposed an approximate 5% loss on IgG recovery. Neutralization released eluted IgG from its low pH associations with chromatin and caused heteroaggregate remnants to associate into large particles easily removed by microfiltration. Most effective host contaminant clearance was achieved by filtration after neutralization to pH 5.5. All chromatin-mediated liabilities were suspended by extraction of chromatin heteroaggregates in advance of protein A.

Academic research paper on topic "Non-immunospecific association of immunoglobulin G with chromatin during elution from protein A inflates host contamination, aggregate content, and antibody loss"

ELSEVIER

Contents lists available at ScienceDirect

Journal of Chromatography A

journal homepage www.elsevier.com/locate/chroma

Non-immunospecific association of immunoglobulin G with chromatin during elution from protein A inflates host contamination, aggregate content, and antibody loss

Pete Gagnon *, Rui Nian, Yuansheng Yang, Qiaohui Yang, Chiew Ling Lim

Bioprocessing Technology Institute, 20 Biopolis Way, Centros #06-01, Singapore 138668, Singapore

ARTICLE INFO ABSTRACT

Monoclonal IgG at pH 3.5 expressed a tendency to self-associate and associate non-specifically with surfaces, including the surfaces of precipitated chromatin heteroaggregates. The tendency was elevated with protein A-eluted IgG still in elution buffer (100 mM acetate, pH 3.5). Association of IgG with chromatin elements under protein A elution conditions amplified host protein contamination of the elution fraction about 15-fold, caused formation of aggregates that persisted after pH neutralization, and imposed an approximate 5% loss on IgG recovery. Neutralization released eluted IgG from its low pH associations with chromatin and caused heteroaggregate remnants to associate into large particles easily removed by microfiltration. Most effective host contaminant clearance was achieved by filtration after neutralization to pH 5.5. All chromatin-mediated liabilities were suspended by extraction of chromatin heteroaggregates in advance of protein A.

© 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 history:

Received 28 April 2015

Received in revised form 20 June 2015

Accepted 5 July 2015

Available online 8 July 2015

Keywords: Protein A

Chromatin Host contamination Aggregation Recovery

1. Introduction

Protein A denatures IgG upon binding [1,2], and low pH elution compounds that denaturation, creating a highly disordered conformation about half the hydrodynamic size of native IgG that persists until the antibody is exposed to physiological conditions [3]. The denatured conformation exhibits an elevated tendency to associate with relatively inert surfaces, indicated by its failure to elute from a size exclusion chromatography (SEC) column equilibrated to protein A elution conditions. It also manifests higher vulnerability to aggregate formation than native IgG exposed to equivalent conditions.

Host contaminants remain accessible to IgG during elution from protein A even after extensive and aggressive washing [4]. Chro-matin heteroaggregates consisting of non-histone host proteins accreted onto nucleosomes bind protein A more strongly than IgG. Some contaminant subsets leach from these heteroaggregates during IgG elution while others remain bound to the protein A. Eluting IgG has access to both. This creates potential for IgG-contaminant interactions that would not occur under physiological conditions.

The study describing the reduced-size conformation of protein A-eluted IgG was performed exclusively with highly purified IgG.

* Corresponding author. Tel.: +65 6407 0941; fax: +65 6478 9561. E-mail address: pete_gagnon@bti.a-star.edu.sg (P. Gagnon).

The present study explores the additional dimension of associations between protein A-eluted IgG and chromatin elements under elution conditions, and evaluates their practical impact on purification performance of protein A affinity chromatography.

2. Materials and methods

2.1. Reagents and equipment

Buffers, salts, and reagents were obtained from Sigma-Aldrich (St. Louis, MO), except allantoin, which was obtained from Merck Millipore (Darmstadt, Germany). Toyopearl® AF-rProtein A-650F was obtained from Tosoh Bioscience (Tokyo). UNOsphere™ Q was obtained from Bio-Rad Laboratories (Hercules, CA, USA). Capto™ adhere and MabSelectTM SuRe were obtained from GE Healthcare (Uppsala, Sweden). Eshmuno™ HCX was obtained from Merck-Millipore (Bedford, MA, USA). Chromatography media were packed in XK or Tricorn™ series columns (GE Healthcare). Chromatography experiments were conducted on an ÄKTA™ Explorer 100 or Avant 25 (GE Healthcare).

2.2. Experimental methods

A prospective biosimilar IgG1 monoclonal antibody (Herceptin®) was expressed by mammalian cell culture in Chinese hamster ovary (CHO, DG44, Life Technologies, Carlsbad, CA) cells

http://dx.doi.org/10.1016/j.chroma.2015.07.017

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/).

using a tricistronic vector developed by Ho et al. [5]. Antibody was produced in 5 L BIOSTAT® B stirred-tank glass bioreactor (Sartorius Stedim Biotech) fed-batch cultures using protein free medium consisting of an equal ratio of CD CHO (Life Technologies) and HyQ PF (GE Healthcare). Cultures were harvested at 30-50% viability. Pumps were avoided during harvest to minimize potential cell disruption.

A parental line of CHO DG44 cells licensed from Life Technologies was cultured as above and used as a null cell line (not expressing IgG).

Traditional harvest clarification was performed by centrifu-gation at 4000 x g for 20 min at room temperature, followed by filtration through 0.22 |im membrane (Nalgene® Rapid-Flow Filters, Thermo Scientific, Waltham, MA). Clarified harvests were stored at 2-8°C for short-term usage or -20°C for long-term storage.

Some of the experiments in this study employed highly purified IgG to eliminate ambiguity among experimental results. Chromatin was extracted in advance as described in [6]. In brief, caprylic acid was added to cell-free culture harvest to a final concentration of 0.4%, and allantoin to a final concentration of 2%. pH was adjusted to 5.3 with 1 M acetic acid, and the mixture stirred for 2 h. UNO-sphere Qpre-equilibrated with 50 mM MES, 150 mM NaCl, pH 5.3 was added at a proportion of 5% (v/v) and mixing continued for at least 4h. Solids were removed by centrifugation and/or microfiltration. Protein A affinity chromatography was performed with 20 mL of media packed in a XK 16/20 column (10 cm bed), run at linear flow rate of 300 cm/h (volumetric flow rate 10 mL/min). The column was equilibrated with 5 column volumes (CV) of 50 mM HEPES, 120 mM NaCl, pH 7.0 (HBS). 500 mL of chromatin-extracted cell culture supernatant was loaded and the column washed with 20 CV HBS, in some experiments, a second wash with 10 CV of 50 mM Tris, 2 M NaCl, pH 8 was included and followed by a third wash with 10 CV HBS. Antibody was eluted with a 10 CV step to 100 mM acetic acid, pH 3.5. Protein was collected from the point where UV absorbance at 280 nm reached 50 mAU to the point where it descended below that value. The column was cleaned with 20 CV of 0.1 M NaOH. Aggregates, antibody fragments, DNA and residual host cell proteins were further removed by titrating the protein A eluate to pH 8.0, adding NaCl to 1 M, and loading 40 mg IgG onto a 4mL Capto adhere (Tricorn 10/50 at a linear flow rate of 150 cm/h, volumetric flow rate 2 mL/min). The column was washed with 10 CV of equilibration buffer and antibody and eluted with a 10 CV step to 50 mM MES, 0.35 M NaCl, pH 6.0. Protein was collected from the point where UV absorbance at 280 nm reached 50 mAU to the point where it dropped below that value. The column was cleaned with 10 CV of 100 mM acetic acid, pH 3, then 20 CV of 1 M NaOH. Antibodies purified by this process typically contained <2ppm host cell protein (HCP), <1 ppb DNA, and <0.1% aggregates. DNA was extracted from CHO cell culture harvest with a NucleoBond® CB kit for genomic DNA purification (Macherey-Nagel, Düren, Germany), according to the manufacturer's recommendations. 10 mL of the extract was then flowed at 150cm/hr through a 4mL HCX column packed in Tricorn 10/5 equilibrated to 50 mM MES, 100 mM NaCl, pH 6.0 to reduce histone and nucleosomal remnants. The column was cleaned with 20 CV of 1 M NaOH.

Histones were extracted from CHO cell culture harvest beginning with 1 h incubation in 200 mM hydrochloric acid, 1.5 M NaCl, 0.1% Nonidet™ NP 40, 0.2% ethacridine, followed by filtration through a 0.22 | m membrane filter to remove solids. The filtrate was purified by void exclusion anion chromatography on UNOsphere Q according to [7], in 50 mM Tris, pH 8.0. Further concentration was performed by Vivaspin 15R centrifugal concentrators (Sartorius Stedim Biotech, Goettingen, Germany) with 2 kDa molecular weight cut-off (MWCO).

DNA was extracted from CHO cell culture harvest with a NucleoBond® CB kit for genomic DNA purification (Macherey-Nagel, Düren, Germany), according to the manufacturer's recommendations. 10 mL of the extract was then flowed at 150 cm/h through a 4mL HCX column packed in Tricorn 10/5 equilibrated to 50 mM MES, 100 mM NaCl, pH 6.0 to reduce histone and nucleosomal remnants. The column was cleaned with 20 CV of 1 M NaOH.

Smaller scale protein A experiments with the same buffers were used to produce some experimental materials. These experiments employed a Tricorn 5/50 column packed with 1 mL of media, run at 150 cm/h. In one series, a column was loaded with 35 mL of null cell culture harvest clarified by the traditional method described above, then washed 55 CV with HBS and eluted as described above. In a variant of that experiment, the column was loaded with 35 mL clarified null harvest and washed 30 CV, then loaded with 10 CV highly purified IgG as described above, then washed 15 CV and eluted as above. In another variant of that experiment, the column was loaded with 10 mL of a mixture containing DNA extracted from traditionally clarified harvest and histones extracted from traditionally clarified harvest, then washed 30 CV. The column was subsequently loaded with 10 CV highly purified IgG then washed 15 CV and eluted as above. In a different scale-down variation, traditionally clarified cell culture harvest was loaded, washed 30 CV with HBS, then washed 10 CV with 50 mM Tris, 2M NaCl, pH 8.0, and washed again with 15 CV of HBS prior to elution.

Experiments to characterize solubility and turbidity of different IgG preparations were conducted with highly purified IgG as prepared above. IgG for some experiments was prepared by titrating it to pH 3.5 with 1 M acetic acid. IgG for other experiments was prepared loading the highly purified IgG onto a clean protein A column, washing and eluting it, but leaving the IgG in the 100 mM acetate pH 3.5 elution buffer. DNA and/or histones extracted as described above were added to IgG in the following proportions/volumes: 165 |g DNA to 8.5 mg IgG in 10 mL; 220 |g histones to 8.5 mg IgG in 10 mL; 165 |g and 220 |g histones to 8.5 mg IgG in 10 mL. In a follow-on series of experiments, samples at pH 3.5 were subsequently titrated to pH 6.5 with 3 M Tris.

2.3. Analytical methods

Non-histone host cell protein (HCP) content was estimated by ELISA with a Generation III CHO HCP kit from Cygnus Technologies Inc. (Southport, NC). The qualification of non-histone HCP is used because this assay is unable to detect histones [4].

Histone concentration was estimated with a Total H3 Histone kit from Active Motif (Tokyo), or with a PathScan® Total Histone H3 Sandwich ELISA Kit (Cell Signaling Technology Inc., Danvers, MA), following sample extraction as described above. Total histone values were estimated as 4.5 times the amount of H3 to adjust for the normal distribution of histones in chromatin of living cells: H1(H2a,H2b,H3,H4)2.

DNA content was measured using a QX100™ Droplet Digital™ PCR System (Bio-Rad Laboratories) designed for absolute quanti-tation of DNA copy number. Samples were prepared according to manufacturer's recommendations. In brief, they were digested by proteinase K (adding 10% v/v of 2 mg/mL proteinase K in 5% SDS to sample) for 16 h at 50 °C, followed by DNA extraction using either a DNA extractor kit (Wako, P/N 295-50201) or QIAamp viral RNA mini kit (Qiagen, P/N 52906). TaqMan PCR reaction mixture was assembled from a 2 x ddPCR Mastermix, 10 x primer and probes (resDNASEQ™ Quantitative CHO DNA Kit, Applied Biosystems, Foster City, CA) and DNA sample in a final volume of 20 | L. Each reaction mixture was loaded into a sample well of an eight-channel disposable droplet generator cartridge, then 70 | L of droplet generation oil. Generated droplets were transferred to a 96-well PCR plate, heat-sealed, then placed on a thermal cycler and amplified

to end-point by denaturation at 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s then 60 °C for 1 min. Analysis was performed with QuantaSoft analysis software (Bio-Rad Laboratories). Correlation between DNA copy number and DNA concentration was based on CHO host cell DNA standards from Applied Biosystems (resDNASEQ™ Quantitative CHO DNA Kit).

Aggregate content was measured by analytical size exclusion chromatography (SEC) with a G3000SWxl column (Tosoh Bioscience) on a Dionex UltimateTM 300 HPLC system (Thermo Scientific) operated at a flow rate of 0.6 mL/min, using a buffer formulation of 50 mM MES, 20 mM EDTA, 200 mM arginine, pH 6.0. Sample injection volume was 100 |L. Details for other SEC experiments are described in the following section.

Recovery of non-aggregated IgG was also monitored by SEC, comparing experimental results with a calibration curve prepared from known quantities of injected purified IgG.

Non-Reduced SDS-PAGE was performed on 4-15% Criterion™ TGX Stain-FreeTM Gel (Bio-Rad). Protein bands were visualized with silver to detect low-level proteins.

Sub-| m solute size distributions in free solution were characterized by dynamic light scattering (DLS) using a Zetasizer ZS (Malvern Instruments, Worcestershire, UK). The sample (200 |L) was mixed gently for 10 s on a vortex before being placed into a quartz cuvette (ZEN2112, Malvern Instruments) using a gel loading tip to avoid bubbles. Viscosity of the carrier solution was determined using a SV-10 viscometer (A&D Company, Tokyo). The backscattered light at 173° was measured and 3 measurements were averaged. Attenuation index was maintained at a value of 7-8. Analysis of the data was performed using version 7.02 of the Dispersion Technology Software provided by the manufacturer.

Analysis of |im range particle size distributions was performed by laser diffraction with a Microtrac Bluewave particle analyser (Microtrac, Montgomeryville, PA, USA). Refractive index of the precipitate and carrier solution was measured with a refractometer (Refracto 30PX, Mettler Toledo Internation, Inc, Columbus, OH, USA). The particle size of the sample was obtained by priming the laser diffraction unit with 50 mM HEPES pH 7.0 buffer solution, while flow was set at 70%. Samples were gently mixed using a vortex before slowly loaded into the unit until optimal obscuration of 0.1-0.3 was obtained. The protein refractive index was 1.59, absorbing and irregular.

Turbidity expressed in nephelometric turbidity units (NTU) was measured with an Orion AQ4500 Handheld Turbidity Meter (Thermo Scientific).

Other experimental details are described or reiterated for clarity in the following section.

been shown to create minor disorder in antibody structure independently from other sources of stress [3,8].

3.1. Associative tendencies oflgG under acidic conditions

Solubility of IgG was evaluated as a function of pH with the premise that its behavior in free solution might provide insights about its behavior inside a protein A column during elution. For purified native size IgG in buffer, there was a shallow but apparent trend toward reducing solubility against decreasing pH, with 98% remaining soluble at pH 3.5 (Fig. 1). With nothing but IgG and buffer in the samples, the 2% loss was interpreted to indicate a tendency toward self-association, leading to formation of insoluble particles that were removed by centrifugation. Other data showed a broader tendency for IgG to form stable non-specific associations with surfaces at low pH.

When native size IgG was applied to a SEC column equilibrated with 100 mM acetate, pH 3.5, antibody was absent from its normal elution time of about 14.7 min under physiological conditions (Fig. 2, panel 1). Most of it eluted at about 20 min, indicating its transport through the column was retarded (Fig. 2, panel 2). This was an important result because it showed that despite solubility being reduced only 2%, the entire IgG population was affected by the low pH. Circular dichroism spectroscopy and high resolution 2D-NMR have both documented minor structural perturbation of the C^2 domain at pH 3.5, but no major changes in domain architecture of the antibody as whole (3.7). The implication is that IgG's tendencies toward self-association and non-specific association with surfaces are mediated by these small structural changes. The further implication is that they might also enhance association of the perturbed IgG with other surfaces.

Solubility of IgG in harvest at pH 3.5 was much lower than purified IgG in buffer, only 82%. This was accompanied by heavy precipitation of other species. Fig. 1 indicates that the primary precipitated species were chromatin heteroaggregates. Consistent with their reduced solubility, chromatin heteroaggregates in cell culture harvests have been shown to range in size from 50 nm to 400 nm [4,6]. The DNA and histone proteins that make up their nucleosomal array cores were precipitated entirely at pH 3.5. Non-covalently associated non-histone HCP contribute 80% or more of the total heteroaggregate mass [4,6]. This suggests that the 40% loss of HCP at pH 3.5 resulted from its pre-existing association with chromatin, and not because of the native solubility properties of the individual HCP species.

Fig. 3, showing non-reduced SDS-PAGE of precipitate and supernatant at each pH point, provided orthogonal confirmation of IgGs

3. Results and discussion

pH 3.5 was used to model protein A elution conditions throughout this study because the second constant region of the heavy chain (C72) is known to become progressively denatured by exposure to pH values below 3.5, leading to complete loss of C72 structure at pH 3.1 [8]. Use of pH values below 3.5 would have made it impossible to discriminate and objectively characterize the influence of other variables. The antibody used to conduct these studies was previously documented to elute fully from protein A in 100 mM acetate, pH 3.5 [3,4].

Experiments with cell culture harvest or chromatin elements are designated as such. Otherwise, experiments were performed with highly purified IgG (<2 ppm HCP, less than 1 ppm DNA, <0.1% aggregate). The term protein A-denatured ¡gG refers to the 5.5 nm conformation eluted from protein A at pH 3.5. The term native size ¡gG refers to the normal 11.5 nm conformation. Native size should not be understood to mean native conformation since pH 3.5 has

soluble

___Q"'

••cr*.'--*

O IgG in buffer

• IgG in harvest □ HCP in harvest O histone in harvest

♦ DNA in harvest

3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0

Fig. 1. Solubility of purified IgG in buffer and IgG in cell culture harvest as a function of pH.

purified IgG HBS SEC pH 6.5

254 nm 280 nm

purified IgG, pH 3.5 SEC pH 3.5 re-eq pH 6.5

pA-eluted IgG, pH 3.5 SEC pH 3.5 re-eq pH 6.5

10 20 30 10 20 30 40 50 60 10 20 30 40 50 60

elution time, min

Fig. 2. Non-specific association of IgG with SEC media. Absorbance at 254 nm increases at ~50 min because the incoming buffer absorbs at that wavelength.

insoluble soluble

std 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 std std 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 std

Fig. 3. Non-reduced SDS PAGE of harvest precipitates and supernatants at different pH values.

progressive loss of solubility with decreasing pH. This was documented by its increasing presence in the precipitate fraction, but note also the dominating presence of histones across the precipitate fractions and their relative absence across the supernatant fractions. Fig. 3 further highlights the diversity of non-histone host protein species associated with chromatin heteroaggregates.

The shape of the IgG-in-harvest precipitation curve in Fig. 1 meanwhile supports the hypothesis that its loss of solubility at pH 3.5 was mediated largely by its non-specific association with the surfaces of precipitated chromatin heteroaggregates. Note the shape of the DNA and histone curves, and that they are almost completely precipitated even at pH 4.5. If the dominant IgG-associative force in the system was a direct chemical interaction with chromatin, then it should have influenced the shape of the IgG precipitation curve. An example of this is provided by the HCP curve. Its response is less extreme than the chromatin components because a large proportion remains soluble, but it otherwise tracks with chromatin, descending most steeply in the range of pH 6.5 to 4.5, and much less from 4.5 to 3.5.

The in-harvest-IgG showed a different response. Instead, it represented a larger amplitude variant of the in-buffer-IgG curve, with the same relationship to pH as the in-buffer-IgG. This suggests its response was driven principally by its own native solubility properties, and the larger amplitude compared to in-buffer-IgG reflects the availability of a non-specific adsorptive surface, provided in this case by precipitated chromatin heteroaggregates.

3.2. Elevated associative tendencies of protein A-denatured ¡gG

3.2.1. Associations at pH 3.5

Fig. 2, panel 3 illustrates the application of protein A-denatured IgG to a SEC column equilibrated to 100 mM acetate, pH 3.5. It failed to elute, as reported previously [3]. However it desorbed when the

column was re-equilibrated with a solubilizing buffer as above. This documented the elevated tendency of protein A-eluted IgG to participate in non-specific associations, compared to native size IgG under the same buffer conditions (Fig. 2, panel 2). It further suggested that protein A-denatured IgG might be more prone to non-specific association with precipitates than native size IgG.

Fig. 4 illustrates % soluble IgG and turbidity in 0.45 | m membrane filtrates after purified native size IgG was titrated to pH 3.5 and compared with protein A-denatured IgG still at pH 3.5. Note that protein A-denatured IgG was obtained from a protein A column loaded with purified IgG. Both antibody preparations remained almost entirely soluble as pure IgG but all mixtures with chromatin elements produced precipitates, and the response was dramatically greater for protein A-denatured IgG still at pH 3.5.

Addition of histones to the pH 3.5 antibodies produced a modest response, with negligible effect on native size IgG titrated to pH 3.5, but reducing solubility to 87% for protein A-denatured IgG still at pH 3.5 (Fig. 4, panel 1). Addition of DNA, alone or in combination with histones, reduced the solubility of both IgG conformations to -70%. Addition of DNA to protein A-denatured IgG still at pH 3.5 reduced solubility to 61%. Addition of DNA plus histones reduced solubility to 43%. Turbidity showed the same trends (Fig. 4, panel 2).

3.2.2. Associations persisting after pH neutralization

Fig. 5 illustrates % solubility and turbidity following sequential exposure of IgG to pH 3.5 then pH 6.5. Purified IgG of both conformations was fully recovered after neutralization to pH 6.5 (panel 1). Recovery was greater than 95% for both conformations mixed with histones. Recovery was reduced -5% for native size IgG in the presence of DNA titrated to pH 3.5 then pH 6.5, but solubility of protein A-denatured IgG combined with DNA before neutralization was less than 50%. Recovery of native size IgG with DNA and his-tones was nearly 95% after neutralization, but protein A-denatured IgG combined with DNA and histones was less than 20% soluble.

Turbidity reflected the same trends (Fig. 5, panel 2). Turbidity was less than 5 NTU for samples containing IgG only or IgG plus histones, and similarly low for native size IgG mixed with DNA and native size IgG mixed with DNA and histones. In dramatic contrast, protein A-denatured IgG combined with DNA produced a turbidity value of 160 NTU. Protein A-denatured IgG combined with DNA and histones produced turbidity of 212 NTU.

The disproportionate response of protein A-eluted IgG combined with DNA was also revealed by dynamic light scattering (Fig. 6). DNA combined with native IgG at pH 3.5 then titrated to pH 6.5, showed the presence of non-aggregated IgG, 90 nm aggregates, and a smaller proportion of particles approaching 1 | m. Protein

Soluble

Soluble

] nativeIgG, titrated to pH 3.5 I protein A-eluted IgG, still at pH 3.5

0.7 0.7 1.9 1.7

Turbidity NTU

+DNA +his & DNA

+histone +DNA +his & DNA

Fig. 4. Solubility and turbidity at pH 3.5 of IgG and mixtures with DNA and histones.

] native IgG pH 3.5, then 6.5 I protein A-eluted IgG pH 3.5, then 6.5

2.0 2.2 3.9 3.3 3.9

Turbidity NTU

IgG +histone +DNA +his & DNA IgG +histone +DNA +his & DNA

Fig. 5. Solubility and turbidity IgG and mixtures with DNA and histones at pH 3.5 then neutralized.

intensity (%)

size distribution by intensity

native IgG + DNA, pH 3.5, then 6.5 - protein A-eluted IgG + DNA, pH 3.5, then 6.5.....

A ^ ; ■

J \ / V

diameter, nm

Fig. 6. Size distribution by DLS of IgG-DNA mixtures at pH 3.5 then neutralized.

+histone

A-eluted IgG combined with DNA then neutralized was completely dominated by 1-2 |im particles.

Figs. 2-6 emphasize several important points, first that protein A-denatured IgG has a much higher tendency than native size IgG to form non-specific associations with surfaces, including the surfaces of precipitated chromatin heteroaggregates. Second, that protein A-eluted IgG has a particular tendency to associate with DNA at pH 3.5, and those associations produce effects that persist after neutralization. Third, although histone-IgG interactions are weaker than DNA-IgG interactions, histones in combination with DNA impose larger effects than DNA alone.

3.2.3. Preferential association of chromatin elements with protein A-denatured ¡gG

The strongly defined chemical character of DNA provides insight into the probable cause of its strong interactions with IgG. DNA includes a pair of negatively charged phosphoryl oxygen atoms

at every base pair node, making it essentially a high charge-density liquid-phase cation exchanger. IgG binds strongly to cation exchangers at low pH. This explains their initial attraction, but it seems doubtful that the mature association is limited to this mechanism. DNA-protein interactions are extensively documented to include metal affinity, hydrogen bonding, and van der Waals interactions [8-10].

Histones provide a revealing counterpoint. IgG1 monoclonal antibodies tend to be weakly alkaline. Histones are strongly alkaline [11,12]. Their common electropositivity should manifest as mutual repellency even under physiological conditions. Proteins become more electropositive with decreasing pH. This should increase intensity of repellency between histones and IgG at pH 3.5 and discourage their association. Mutual repellency between IgG and histones might therefore have been expected to moderate the effects of DNA-histone mixtures, but the opposite effect was observed (Figs. 4 and 5).

elution time, min

Fig. 7. Formation of persistent aggregates by association of IgG with DNA at pH 3.5.

This provides another useful insight, leading back to the insolubility of chromatin heteroaggregates in cell culture harvest pH 3.5 (Fig. 1). It seems reasonable to assume that an artificial DNA-histone mixture would also be poorly soluble at low pH, and that poor solubility would produce a larger precipitate mass, as corroborated by the elevated turbidity shown in Fig. 5, panel 2. A larger precipitate mass would logically correspond with a larger cumulative surface. This suggests the greater loss of IgG from DNA-histone mixtures compared to IgG with DNA alone, probably reflects a synergistic contribution by the inherent tendency of IgG to form non-specific associations with surfaces at low pH.

3.3. Formation of persistent aggregates by IgG-chromatin associations at acidic pH

Low pH elution of protein A columns has long been considered a contributor to formation of IgG aggregates [13-15]. The assumption has generally been that low pH itself is independently sufficient to promote aggregate formation. For elution at pH 3.5 and above, recent results from two different research groups suggest this is an oversimplification.

In one study, purified native size IgG titrated to pH 3.5, held for 1 h, then titrated to physiological conditions showed no indication of aggregate formation [3]. Even protein A-denatured IgG was free of aggregates upon titration to physiological conditions. This provides an important baseline reference because it emphasizes that pH 3.5 alone is not sufficient to induce aggregation of protein A-eluted IgG in the absence of other sources of stress, at least not for this particular antibody.

In two other recent studies, Guo and Carta [16,17], revealed that aggregates were formed as a result of an unstable intermediate having been formed while IgG was resident on the surface of a cation exchanger. These findings point to the idea that formation of some aggregate populations may be template-driven, where the interaction of IgG with a particular template creates conformations that do not occur with IgG in free solution under the same buffer conditions, and from which the antibody is unable to spontaneously regain its native conformation upon neutralization.

Fig. 7, panel 3 illustrates the SEC elution profile of native size IgG combined with extracted CHO DNA at neutral pH for reference. UV absorbance was monitored at 254 and 280 nm to provide indications of the relative proportions of DNA and protein along the chromatogram. 254 nm was selected over DNA's absorption maximum at 260 nm because 254 corresponds with an absorption minimum of proteins that improves the ability of wavelength ratios to discriminate DNA from protein [18]. Relative UV absorbance at 254 and 280 nm for proteins and DNA is roughly reversed. This

relationship has been exploited for decades and shown to provide a fair representation of DNA-to-protein ratios in mixed solutions, including in conjunction with SEC [19].

Fig. 7, panel 4 shows creation of a persistent aggregate population when DNA was mixed with native size IgG, titrated to pH 3.5, held for 1 h, then titrated to pH 6.5 and filtered before SEC. Note the UV profile of the created aggregate (red triangle) was dominated by absorbance at 280 nm This indicated its primary constituent was IgG. The same region of the reference profile at pH 7.0 (panel 3) showed dominance by absorbance at 254 nm (DNA). These results can be explained only by the interaction of IgG with DNA at pH 3.5 creating a class of IgG aggregates that were absent from the original sample.

Addition of DNA to protein A-denatured IgG, followed by the same chemical treatment created a larger proportion of aggregates and of fundamentally different composition (Fig. 7, panel 5). Dominant UV absorbance at 254 nm for all species larger than IgG suggested that DNA itself was a significant constituent of these aggregates. The extent to which DNA might also have provided an aggregation template is not clear.

As predicted by solubility and turbidity experiments (Figs. 4 and 5), the combination of histones with DNA resulted in a higher degree of aggregate formation than DNA alone (Fig. 8). Note also the much lower amount of protein in Fig. 8, panel 5, in response to heavy loss of precipitates at the filtration step. According to SEC, treatment of histone-IgG combinations at pH 3.5 did not produce aggregates that persisted at pH 6.5 (data not shown).

The overall results of Figs. 7 and 8 should not be understood to predict the same magnitude of aggregate formation during routine purification of IgG with protein A. DNA and DNA/histone to IgG ratios in the present experiments were chosen to mimic ratios in cell culture harvest, but data from a previous study showed that most of the chromatin in cell harvest flows through protein A columns during sample loading [4]. About 80% of DNA was unbound by protein A and flowed through the column, the elution contained -0.01%, and the NaOH peak contained -1%. For histones, -55% were unbound, -0.07% occupied the elution fraction, and the NaOH peak contained about -4%. These distributions counsel that the current results are cause for increased awareness and further study, but not for alarm.

3.4. Loss ofIgG by association with protein A-retained chromatin heteroaggregates

As noted above, the largest proportion of chromatin heteroaggregates accessible to eluting IgG during protein A chromatography

native IgG

pH 6.5

254 nm

280 nm

DNA + histone pH 6.5

pA-eluted IgG + DNA-his pH 3.5 then pH 6.5

8 10 12 14 16 18 8 10 12 14 16 18 8 10 12 14 16 18 8 10 12 14 16 18 8 10 12 14 16 18

elution time, min

Fig. 8. Formation of persistent aggregates by association of IgG with DNA and histones at pH 3.5.

UV abs 280 nm

0 10 20 30 40 50 60 70 80 90 100 110 120

elution volume, mL

Fig. 9. Comparison of protein A chromatograms from experiments with null harvests. The blue highlighted peak is IgG. The red-yellow highlighted peak is chromatin heteroaggregates.

pH pH pH

Fig. 10. Host contaminant content of protein A eluates after titration to indicated pH values then microfiltration to 0.45 |xm.

resides in elements that remain bound to protein A during elution [4]. These heteroaggregates are understood to reside mostly on the chromatography particle exteriors. On a column loaded to capacity, they would exist down the full length of the column, where they would be readily accessible to eluting IgG for up to several minutes. Since such interactions would take place in the inter-particle void space, restrictions from diffusive mass transport through the pore structure of the particles would be negligible.

Strong association of protein A-eluted IgG with DNA and DNA-histone mixtures in free solution at pH 3.5 (Fig. 4), predicted that IgG must form stable associations with protein A-bound chro-matin elements during elution. When protein A was loaded with harvest clarified by centrifugation and microfiltration, washed, eluted, and cleaned with 100 mM NaOH, the cleaning peak contained 4-6% of the total non-aggregated IgG applied to the column.

This was determined by neutralizing the NaOH peak, filtering out solids, applying the filtrate to SEC, and using a pre-established calibration curve to determine the mass of IgG in the non-aggregated peak.

The NaOH peak also contained 4-6% of the applied non-aggregated IgG when protein A was loaded with null harvest and washed, then loaded with purified IgG, washed, eluted and cleaned. The NaOH peak also contained 4-6% of the applied non-aggregated IgG when protein A was loaded with a mixture of DNA and histones and washed, then loaded with highly purified IgG, washed, eluted and cleaned.

These results collectively support the hypothesis that IgG binding to chromatin elements that remain bound to protein A during elution, accounts for product losses of about 5% during protein A affinity chromatography.

200 NaCl 0 mM

Fig. 11. HCP levels of protein A eluates after adjustment to indicated pH values and NaCl concentration then microfiltration to 0.45 |im.

3.5. Amplification of host contamination by IgG-chromatin interactions during elution

Shukla and Hinckley published a landmark study in 2008 that was first to recognize IgG-contaminant associations detracted from purification performance of protein A affinity chromatography [20]. They suggested that stable associations formed between IgG and HCP in cell culture harvest, that the HCP remained associated with IgG during the washing phase of protein A affinity chromatog-raphy, and the associated HCP was carried with the IgG during elution. Associations persisting after neutralization were suggested to contribute to aggregate content. Their hypotheses fit their experimental design and data, but violated one of the most fundamental assumptions concerning monoclonal antibodies.

The entire fields of immunotherapy, immunodiagnosis, and the validity of an immense body of immunological research rely on the premise that monoclonal antibodies are exquisitely specific for their target antigens, especially under physiological conditions. Promiscuous non-specific association of IgG with host-derived contaminants in cell culture would predict the same for other physiological environments, imposing a high probability of interference with their intended applications. Such a phenomenon should surely have been recognized by now.

The elevated associative tendencies of protein A-eluted IgG at pH 3.5 provide the basis for an explanation that makes it unnecessary to challenge the presumption of monoclonal specificity under physiological conditions. A 1 mL protein A column loaded with 35 mL null harvest then washed, eluted 4.9 |ig HCP. This demonstrated that the elution conditions destabilized bound heteroaggregates, causing some of their constituents to be leached from elements that remained bound. Another column loaded with null harvest and washed, then loaded with highly purified IgG and washed, eluted 74.0 |g HCP, an increase of 15-fold. Chromatograms are compared in Fig. 9.

These results corroborate Shukla and Hinckley's findings [20] to the extent of confirming that the presence of IgG elevates the quantity of HCP in protein A eluates. They further confirm

their hypothesis that the phenomenon involves strong associations between IgG and contaminants. Where the studies differ is that there is no possibility in the present experimental design for the associations to have formed in the cell culture harvest under physiological conditions.

In light of the knowledge that bound chromatin heteroaggregates are accessible to IgG during elution [4], and protein A-eluted IgG forms strong associations with chromatin under elution conditions (Fig. 4), the present data point instead to the idea that IgG-chromatin interactions compound the heteroaggregate destabilizing effects of the elution buffer, and promote a higher degree of contaminant leaching than can be accounted for by the elution buffer alone.

This raises the question of how IgG enhances leaching of contaminant subsets from chromatin heteroaggregates. The specific chemical mechanisms are probably various, but electrostatic interactions with DNA suggest themselves as a primary contributor. At pH 3.5, it seems reasonable that a gross excess of eluting IgG interacting with the DNA component of chromatin heteroaggregates would compete with pre-existing DNA associations within those heteroaggregates, and weaken them sufficiently to promote further dissociation.

These findings collectively suggest that antibody-contaminant associations in cell harvest under physiological conditions probably do not occur to a significant extent with the majority of IgG monoclonal antibodies. However they do not suspend the possibility that such associations occurred with the antibodies employed by Shukla and Hinckley [20], or the possibility they could occur with others. Contaminant-IgG association under physiological conditions is routine with anti-chromatin antibodies because chro-matin structure is so highly conserved across phyla. Monoclonal antibodies specific for chromatin from one species bind host chro-matin expelled from dead host cells during cell culture production [21-23]. Lacking specialized purification procedures to dissociate the wrong-species antigen from the antibodies, they cause immunopotency for their authentic-species antigen to vary from lot to lot by up to a factor of 12 [23].

Reports of non-immunospecific antibody-chromatin association during cell culture are presently limited to IgM monoclonal antibodies where the phenomenon was attributed to extreme charge characteristics of the antibodies favoring interactions with both DNA and histones [18,19]. This warns that IgG monoclonal antibodies with extreme charge characteristics may also have an elevated tendency to participate in such interactions.

3.6. Modulation of host contamination as a function of neutralization conditions

Leaching of host contaminant subsets due to destabilization of chromatin heteroaggregates during elution at low pH suggests neutralization of protein A eluates might favor re-association of their components. Fig. 1 also suggests that restoration of physiological conditions should favor dissociation of IgG from its low pH

turbidity NTU

°-f0 0 9 4

chromatin extracted harvest post-PA, after filtration

particle size, ^m

Fig. 12. Turbidity and particle size distribution of protein A eluates titrated to indicated pH values. Particle size by dynamic laser diffraction.

100 25

associations with chromatin elements. These expectations are consistent with previous findings that neutralization causes formation of turbidity, that turbidity can be removed by microfiltration, and that nearly all the IgG is recovered in the filtrate, mostly aggregate-free, with substantially less host contamination than the pre-filtered eluate [4,24]. In this context, Fig. 1 also suggests it should be possible to control HCP content of the filtrate as a function of the pH to which the eluate is titrated prior to filtration.

Fig. 10 confirms this prediction. Protein A was loaded with harvest clarified by centrifugation and microfiltration, washed with equilibration buffer, then eluted with 100 mM acetate, pH 3.5. The prefiltered eluate contained about 5100 ppm non-histone HCP, 410 ppm histone HCP, and 1 ppm DNA. Aliquots were titrated to pH values ranging from 4.0 to 9.0. Filtrate levels for all contaminant classes dropped steeply with gradual neutralization from pH 4.0 to a minimum at pH 5.5 where non-histone HCP was reduced 14-fold to 345 ppm, histone HCP was reduced 10-fold to 40 ppm, and DNA was reduced 1000-fold to <1 ppb. Contamination levels increased gradually thereafter up to pH 8.0, then increased sharply at pH 8.5 and 9.0.

There were indications that the behaviors of DNA and non-histone host proteins were linked. The DNA curve paralleled the HCP curve with both reaching their minima at pH 5.5, but both also showing a shoulder at neutrality, followed by a saddle before ascending steeply above pH 8.0. Histone content seemed to be relatively independent. This tended to suggest it remained more strongly associated with the particle fraction.

Fig. 11 shows the overwhelming influence of NaCl concentration. Prefiltration treatment of the eluate with 200 mM NaCl completely suppressed pH effects, fully destabilized chromatin heteroaggregates, and released essentially all of the HCP into the filtrate. The pH 5.5 optimum for HCP reduction became apparent again at 100 mM NaCl, but HCP reduction was still heavily compromised. To the extent modulation of neutralization pH might be exploited as a tool to maximize overall purification performance, the elution buffer must be free of excess salts.

On a gross level, pre-filtration turbidity was inversely proportional to post-filtration host contaminant levels, but detailed comparison of Figs. 10 and 12 showed turbidity to be unuselful for identifying the conditions supporting lowest filtrate contamination. Pre-filtration particle size showed a sharp transition at pH 5.5 consistent with the idea that pH was a partial determinant of heteroaggregate stability, but otherwise showed no correlation to filtrate contamination. This suggested heteroaggregate stability and filtrate contamination were influenced more directly by the effects of pH on interactions among select contaminant subsets, as shown by the distributions of filtrate non-histone HCP and DNA in Fig. 10.

3.7. Practical management of chromatin-mediated performance liabilities

Suspension of a phenomenon by removal of a suspected causal agent is generally understood to provide experimental proof of causation in itself. Changing the non-specific associative properties of protein A-eluted IgG is not realistically within reach, but making chromatin absent from the intra-column environment during elution is simple and should produce the same practical outcome. Advance extraction of chromatin by the caprylate-solid phase method described in this study essentially eliminated the product losses observed when protein A was loaded with harvest clarified by centrifugation and microfiltration. IgG recovery increased ~6% from an average 93% into the range of 99.2%-99.6%.

Advance chromatin extraction also reduced HCP consistently to <10 ppm, histones beneath detectability, DNA to <1 ppb, and aggregates generally to about half the levels obtained when protein A

was loaded with harvest clarified by centrifugation and microfiltration; in the present experiments ~0.8% versus ~1.6%. Similar results were reported from an earlier study where protein A followed chromatin extraction by an ethacridine-solid phase method [4]. That study also showed advance chromatin extraction increased dynamic binding capacity about 20% through suspension of pore occlusion by 50-400 nm chromatin heteroaggregate binding to the particle exteriors during loading.

Analysis of filtrates following eluate titration to pH 5.5 showed advance chromatin extraction also produced about a 3-fold reduction of turbidity from -4.5 NTU to -1.5 NTU (Fig. 12). This seems a modest improvement but it is consistent with aggregate reduction and worthwile to keep in mind it also correlates with a 99% reduction of HCP compared to protein A loaded with feedstreams clarified by centrifugation and microfiltration [4].

These improvements suggest comparison with the decades-old and still-continuing practice of applying chemically aggressive washes prior to elution [4,15,20,24,25]. Application of a 2M NaCl wash at pH 8.0 produced a 40% reduction of non-histone HCP in the eluted IgG fraction (from 4337 ppm to 1743 ppm), a 50% reduction of histones (from 10 ppm to 5 ppm), and a 14-fold reduction of DNA (from 1.4 ppm to 100 ppb). IgG recovery increased only 0.3%, from 93.4% to 93.7%, because most ofthe balance remained trapped by its persistent association with still-protein-A-bound chromatin heteroaggregate elements. The alkaline NaCl wash also failed to compensate for the 20% loss of dynamic binding capacity [4]. Prevention is clearly the better medicine, in the form of extracting chromatin heteroaggregates before column loading.

It seems prudent in either case to neutralize the eluted IgG to pH 5.5 before filtration as a matter of routine. The technique is so simple and so powerful that there seems little reason to do otherwise. Since its efficacy seems to reside in the associative properties of chromatin, and since the composition of chromatin is highly conserved, it should be expected to provide similar benefits for all IgG monoclonal antibodies eluted from protein A.

4. Conclusions

IgG monoclonal antibodies exhibit an inherent tendency toward self-association and non-specific association with surfaces in 100 mM acetate, pH 3.5. In free-solution experiments, about 2% of IgG becomes insoluble. In the presence of a nominally inert surface such as the internal surface of an SEC column, transport of the entire IgG population is retarded. In cell culture harvest, about 18% of the IgG becomes insoluble, apparently by non-specific association with the surfaces of precipitated chromatin heteroaggregates.

IgG eluted from protein A adopts a half-sized conformation with reduced solubility and an elevated tendency to form nonspecific associations [3]. Exposure of this conformation to DNA or DNA-histone mixtures at pH 3.5 causes gross precipitation and formation of aggregates that persist after neutralization. Electrostatic interactions between alkaline IgG and acidic DNA synergistically enhance the native tendency of IgG to associate non-specifically with precipitate surfaces.

A subpopulation of chromatin heteroaggregates remain bound to protein A after the post-load wash [4]. Elution conditions destabilize those heteroaggregates and cause some elements to leach into the elution fraction. Strong associations between protein A-denatured IgG and chromatin heteroaggregates compound destabilization of the latter, causing a larger contaminant subset to leach into the eluted IgG.

Strong associations between protein A-eluted IgG and chro-matin heteroaggregate elements that remain bound to protein A under elution conditions cause that IgG to be absent from the elution fraction. On protein A columns loaded with harvest clarified by

centrifugation and microfiltration, this loss amounts to about 5% of the non-aggregated IgG applied to the column.

pH-dependent re-association of chromatin heteroaggregates in conjunction with neutralizing the protein A-eluted IgG fraction creates large particles that can be removed by microfiltration. Titration to pH 5.5 followed by microfiltration supports more than 100fold reduction non-histone HCP, 10-fold reduction of histones, and 1000-fold reduction of DNA.

Excess host contamination, excess aggregate formation, and excess antibody losses from association of eluted IgG with protein A-bound chromatin heteroaggregates are all suspended by extraction of chromatin heteroaggregates prior to loading protein A.

Acknowledgements

The authors gratefully acknowledge Aina Hoy and Denise Leong for their contributions to the analysis, and Han Ping Loh, Su Jun Low, Jake Chng for making the cell cultures available to conduct the study. We equally acknowledge the generous support of Exploit Technologies Pte. Ltd., Validated Biosystems, and the Biomedical Research Council of the Singapore Agency for Science and Technology Research.

References

[1] J. Deisenhofer, Crystallographic refinement and atomic models of a human Fc fragment ant its complex with fragment B of protein A from Staphylococcus aureus at 2.9 and 2.8 A resolution, Biochemistry 20 (1981) 2361-2370.

[2] J. Deisenhofer, T.Jones, R. Huber,J. Sjodahl, J. Sjoquist, Crystallization, crystal structure analysis and atomic model ofthe complex formed by a human Fc fragment and fragment B of protein A from Staphylococcus aureus, Hoppe-Seylars Z. Physiol. Chem. 359 (1978) 975-985.

[3] P. Gagnon, R. Nian, D. Leong, A. Hoi, Transient conformational modification of Immunoglobulin G during protein A affinity chromatography, J. Chromatogr. A 1395(2015)136-142.

[4] P. Gagnon, R. Nian, J. Lee, L. Tan, S.M. Abdul Latiff, C.L. Lim, C. Chua, X. Bi, Y.S. Yang, H.T. Gan, Nonspecific interactions of chromatin with Immunoglobulin G and protein A, and their impact on purification performance, J. Chromatogr. A 1340(2014) 68-78.

[5] S.C. Ho, M. Bardor, H. Feng, Mariati, Y.W. Tong, Z. Song, M.G. Yap, Y. Yang, IRES-mediated Tricistronic vectors for enhancing generation of high monoclonal antibody expressing CHO cell lines, J. Biotechnol. 157(2012) 130-139.

[6] P. Gagnon, R. Nian, J. Cheong, V. Yeo, Y.S. Yang, H.T. Gan, Chromatin-mediated depression of fractionation performance on electronegative multimodal chromatography media, its prevention and ramifications for purification of Immunoglobulin G, J. Chromatogr. A 1374 (2014) 145-155.

[7] R. Nian, C. Chuah,J. Lee, H.T. Gan, S.M.B.Abdul-Latiff,W.Y. Lee,V.Vagenende, P. Gagnon, Void exclusion of antibodies by grafted-ligand porous particle anion exchangers, J. Chromatogr. A 1282 (2013) 127-132.

[8] P. Privalov, A. Dragan, C. Crane-Robinson, Interpreting protein/DNA interactions: distinguishing specific from non-specific and electrostatic from non-electrostatic components, Nucl. Acid Res. 39 (2011) 2483-2491.

[9] K. Rippe, J. Mazurkiewicz, N. Kepper, Interactions of histones with DNA: nucleosome assembly, stability and dynamics, in: R. Dias, B. Lindman (Eds.), DNA Interactions with Polymers and Surfactants, J. Wiley and Sons, Hoboken, 2008, pp. 135-172.

10] R. Latypov, S. Hogan, H. Law, H. Gadgil, D. Liu, Elucidation of acid-induced unfolding and aggregation of human immunoglobulin IgG1 and IgG2 Fc,J. Biol. Chem. 287 (2012) 1381-1396.

11] P. Walker, M. Sikorska, Endonuclease activities, chromatin structure, and DNA degradation in apoptosis, Biochem. Cell Biol. 72 (1994) 615-623.

12] N. Luscombe, R. Laskowski,J. Thornton, Amino acid-base interactions: athree-dimensional analysis of protein-DNA interactions at an atomic level, Nucleic Acids Res. 29 (2001) 2860-2874.

13] S. Ghose, M. Allen, B. Hubbard, C. Brooks, S. Cramer, Antibody variable region interactions with protein A: implications forthe development ofgeneric purification processes, Biotechnol. Bioeng. 92 (2005) 665-673.

14] T. Arakawa, J. Philo, K. Tsumoto, R. Yumioka, D. Ejima, Elution of antibodies from a protein-A column by aqueous arginine solutions, Protein Expr. Purif. 36 (2004) 244-248.

15] P. Gagnon, Purification Tools for Monoclonal Antibodies, Validated Biosystems, Tucson, 1996, pp. 155-198.

16] J. Guo, S. Zhang, G. Carta, Unfolding and aggregation of a glycosylated monoclonal antibody on a cation exchange column. Part I. Chromato-graphic elution and batch adsorption behavior, J. Chromatogr. A 1356 (2014) 117-128.

17] 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) 129-137.

18] P. Gagnon, F. Hensel, S. Lee, S. Zaidi, Chromatographic behavior of IgM:DNA complexes, J. Chromatogr. A 1218 (2011) 2405-2412.

19] H.T. Gan, J. Lee, S.M. Abdul-Latiff, C. Chuah, P. Toh, W.Y. Lee, P. Gagnon, Characterization and removal of aggregates formed by nonspecific interactions of IgM monoclonal antibodies with chromatin catabolites during cell culture production, J. Chromatogr. A 1291 (2013) 33-40.

20] A. Shukla, P. Hinckley, Host cell protein clearance during protein A chromatography: development of an improved column wash step, Biotechnol. Progr. 24 (2008) 1115-1121.

21] A. Guth, X. Zhang, D. Smith, T. Detanico, L. Wysocki, Chromatin specificity of anti-double-stranded DNA antibodies and a role for arg residues in the third complementarity-determining region ofthe heavy chain, J. Immunol. 171

(2003) 6260-6266.

22] K. Luhrs, D. Harris, S. Summers, M. Parseghian, Evicting hitchhiker antigens from purified antibodies, J. Chromatogr. B 877 (2009) 1543-1552.

23] L. Mechetner, R. Sood, V. Nguyen, P. Gagnon, M. Parseghian, The effects of hitchhiker antigens co-eluting with affinity-purified research antibodies, J. Chromatogr. B 879 (2011) 2583-2594.

24] S. Vunnum, G. Vedantham, B. Hubbard, Protein A-based affinity chromatography, in: U. Gottschalk (Ed.), Process Scale Purification of Antibodies, J. Wiley and Sons, Hoboken, 2009, pp. 79-102.

25] S. Chollangi, Development of robust wash conditions for protein A capture chromatography using high throughput automation technologies, oral presentation, in: BioProcess International Conference and Exhibition, Boston, October 21,2014.