Workflow for Antibody Aggregates and Fragments Analysis

Protocol for SEC-UV analysis of a monoclonal antibody aggregates and fragments

Jessie Zhixin Miao¹, Linh Nguyen,², Stephan Altmaier³, Katie Falls², Uma Sreenivasan¹

¹MilliporeSigma, Round Rock, TX, USA, ²Sepax, Newark, DE, USA, ³Merck KGaA, Darmstadt, Germany

Preparation

Antibody Stressing

Chromatography

Measurement and Analysis

Table of Contents

• Abstract
• Introduction
• Experimental Procedure
• Results
• Conclusion
• Materials
• References

ABSTRACT

A complete SEC-UV workflow has been developed to enable the analysis of antibody aggregates and fragments. This workflow offers the following:

• System suitability test utilizing a protein standard mix
• Procedures for pH and temperature stressing of antibodies
• SEC-UV method for antibody monomer, aggregates, and fragments separation/analysis

INTRODUCTION

Monoclonal antibodies (mAbs, or immunoglobulins - IgGs) are large glycoproteins with a molecular weight of approximately 150 kDa (150,000 g/mol). These biomolecules are composed of two light chains (LC, molecular weight ca. 25 kDa each) and two heavy chains (HC, molecular weight ca. 50 kDa each) linked through covalent inter- and intra-chain disulfide bonds. mAbs are utilized for the treatment of various types of cancer and other diseases such as multiple sclerosis, Alzheimer’s disease, and migraine.

Careful and thorough characterization of therapeutic mAbs is essential for ensuring drug safety and efficacy. mAbs are typically manufactured in mammalian host cell lines in bioreactors, generating many heterogeneous drug molecules. Establishing several critical quality attributes (CQAs) for each mAb and demonstrating that production batches are within acceptable limits are requirements for both innovator and biosimilar therapeutics.^1,2 In many cases, the characterization of an antibody-based drug is performed using a specific chromatographic technique (e.g., size exclusion, reversed phase or hydrophilic interaction liquid chromatography, respectively – SEC, RP or HILIC)^3,4 coupled with UV or mass spectrometry (MS) detection. This combination allows for different types of analyses to be carried out, e.g., accurate mass measurement of aggregates,⁵ the intact mAb and subunits, fragment analysis, peptide mapping, and the determination of post-translational modifications such as glycosylation, oxidation, and deamidation. SEC is a frequently utilized method for the determination of mAb fragments and aggregates.

Monoclonal antibody formulations can undergo various degradation processes during production, formulation, transport and storage, leading to the formation of protein aggregates and fragments - impurities that are CQAs and that must stay within specific limits. Typically, the total amount of impurities (high and low molecular weight - HMW/LMW) is required to be below 5% and no individual peak area should exceed 1-2%. Usually, the concentration of aggregates in a mAb formulation is higher than the fragment concentration and ranges from 1-5%. Specific limits are validated based on holistic characterization data or on stability trends.

During aggregation, two or more mAb monomers form a stable complex, a process that is very difficult to reverse. This aggregation can be induced by electrostatic interactions, for example, leading to self-association of mAbs in concentrated formulations. Another pathway for aggregate formation is misfolding or (partial) unfolding of antibodies creating hydrophobic regions which subsequently bind to each other. Chemical reactions such as disulfide bonding of thiol groups can also form mAb aggregates. Depending on the process, reversible and irreversible covalent and non-covalent species of different sizes can be observed.

Factors influencing aggregation are, e.g., temperature, pH value, shear stress during shaking or stirring, or exposure to light or hydrophobic surfaces.⁶ Fragmentation of mAbs can be described as the chemically or enzymatically induced disruption of a covalent protein bond.⁷ In the same manner as aggregation, fragmentation is a CQA that needs to be monitored. An inactivation of a mAb by irreversible fragmentation reactions is most frequently caused by temperature, pH or mAb concentration or the presence of enzymes, metals or radicals.

The presence of aggregates or fragments in antibody formulations is considered a paramount issue, since they can cause effects such as decreased or increased activity, increased cytotoxicity and immunogenicity and decreased solubility of the drug.⁸ More specifically, side effects such as headache, shivers, anaphylactic reactions or renal failure have been observed⁹ as well as the development of an immune response by patients. In such a situation, the effectiveness of a drug is strongly reduced.¹⁰ Hence, it is essential to characterize mAb aggregate and fragment formation during drug formulation and to establish a suitable quality control process for these therapeutics.

The focus of this work is to analyze/quantitate a variety of mAbs and their aggregates and fragments. The forced pH and temperature stress study was conducted on a monoclonal antibody, and the resulting aggregates and fragments, as well as monomers, were separated/quantified utilizing SEC-UV analysis.

EXPERIMENTAL PROCEDURE

Reagent Preparation

• 150 mM Phosphate Buffer, pH 7.0(±0.02)
  To prepare 1 L of buffer, 8.751 g sodium phosphate monobasic and 23.191 g sodium phosphate dibasic were added to 950 mL HPLC grade water. Sodium hydroxide solution (2 mol/L) was used to adjust the pH value. HPLC grade water was added to a 1 L graduated cylinder to make the final volume 1 L. The buffer was filtered through a 0.2 µm filter before use.
• 150 mM Phosphate Buffer, pH 7.0 + 300 mM Sodium Chloride
  To prepare 500 mL of salt buffer, 8.766 g sodium chloride was dissolved in 450 mL of 150 mM phosphate buffer.150 mM phosphate buffer was added to a 500 mL graduated cylinder to make the final volume 500 mL. The buffer was filtered through a 0.2 µm filter before use.

System Suitability Sample Preparation

• Protein Standard Mix (15-600 kDa) was used as a system suitability sample.
• Sample preparation/dilution: 1 mL of 150 mM phosphate buffer was added to the protein standard vial. The vial was gently swirled to dissolve the lyophilized protein mix powder. 10 µL of this reconstituted protein standard solution was injected onto the LC system.

Unstressed mAb Sample Preparation

Five mAb samples were prepared using the following protocols:

• Adalimumab (Humira) and Infliximab
  Both adalimumab and infliximab were received as 0.25 mL sample solutions formulated in 12.5 mM histidine buffer with an initial concentration of 10 mg/mL. 2.25 mL of 150 mM phosphate buffer was added to the original vials to make the final mAb concentration 1 mg/mL. Samples were mixed thoroughly before injection onto the LC system.

• SILu™Lite SigmaMAbTM
  SILu™Lite SigmaMAbTM Adalimumab Monoclonal Antibody (MSQC16), SILu™Lite SigmaMAbTM Rituximab Monoclonal Antibody (MSQC17), and SILu™Lite SigmaMAbTM Cetuximab Monoclonal Antibody (MSQC18) were obtained as solid powders. 0.5 mL of 150 mM phosphate buffer was added into the sample vials and the vials were gently swirled to dissolve the solid powder.

Stressed mAb Sample Preparation

Adalimumab (Humira) solution was stressed following the procedures below:
1 M hydrochloric acid solution was added dropwise by a pipette to change the pH value of the adalimumab sample solution from 6.0 to 1.0. Then, 1 M sodium hydroxide was added to raise the pH value to 10.0. Finally, 1 M hydrochloric acid solution was utilized to adjust the pH to 6.0. There was approximately a 1 min wait/equilibration time between the pH shifts while the sample solution was constantly stirred at 500 rpm. Finally, the resulting sample solution was incubated in a water bath at 60 °C for 60 min.

SEC-UV SYSTEM SETUP AND DATA ANALYSIS

Instrument and System Setup

The essential settings of the SEC-UV chromatography system applied in the analysis of antibody aggregates and fragments are listed in Table 1 below.

Table 1. SEC-UV system settings.

Instrument
HPLC system	Agilent 1260 Infinity II
Software	Agilent ChemStation
Column	Sepax Zenix®/Zenix®-C SEC Column, 30 cm × 7.8 mm I.D. 3 µm
Column temp	23 °C
Autosampler temp	5 °C ± 3 °C
Mobile phase	150 mM Sodium phosphate, pH 7.0
Gradient	Isocratic
Flow	1.0 mL/min
Injection volume	10 μL
Run time	15 min
Detector	UV: 280 nm, 214 nm
Data Collection Rate	10.0 Hz
Response Time	0.25 s (20 Hz)

Table 2. A typical sample injection sequence.

Sample	Number of injections
Blank	31
Protein standard mix (system suit)	22
Blank	1
Reference material (unstressed sample)	2
Blank	1
Sample 1 injection 1	1
Sample 1 injection 2	1
Sample 2 injection 1	1
Sample 2 injection 2	13
Blank	1
Guanidine hydrochloride	14

¹ Starting protocol: It is recommended to slowly ramp up the flow rate after installation. Then, the column is not exposed to a high pressure suddenly. For the column used in this study, the startup flow rate is 0.1 mL/min and ramp up at increments of 0.1 mL/min. Once the baseline is stable at each step, the next 0.1 mL/min increment was added until 1 mL/min was achieved. Once the baseline was stable at 1 mL/min, the sample injections were started.

² Use three injections in case of new columns.

³ Repeat the blank x 1, sample x 2 injection sequence as more samples are needed. It is highly recommended that an injection of 6 M guanidine hydrochloride is performed for every eight hours of consecutive runs. Run sequence is not recommended to last longer than 24 hours.

⁴ After guanidine hydrochloride injection and before shutdown protocol, column should be in the neutral, low salt storage solution. A slow ramping down process is applied like the starting protocol.

Data Analysis

For quantitative analysis of aggregates (HMWs), monomers and fragments (LMWs), all sample peaks are not present in the blank were integrated. For automatic integration, the integration events below were used to integrate the peaks. For smaller peaks such as the LMW1 region that cannot be detected by integration events, manual integrations were applied.

Table 3. Automatic integration parameters.

Integration Events	Value
Tangent Skim Mode	Standard
Tail Peak Skim Height Ratio	0.00
Front Peak Skim Height Ratio	0.00
Skim Valley Ratio	20.00
Baseline Correction	No Penetration
Peak to Valley Ratio	500.00

Table 4. Manual integration parameters.

Time	Integration Events	Value
Initial	Slope Sensitivity	2
Initial	Peak Width	0.04
Initial	Area Reject	1
Initial	Height Reject	1.7
Initial	Shoulders	OFF

Percentage of area for monomers and its HMW, LMW regions were then computed to characterize the purity of sample and understand the aggregation formation of the stressed sample.

COLUMN MAINTENANCE AND STORAGE

• Daily Preventative Care
  To maximize the column lifetime, 100 µL of 6 M guanidine hydrochloride was injected onto the column at the end of the sample sequence before switching to the storage solution. Guanidine hydrochloride can help to gently remove lingering impurities or sample components out of the column. This procedure will prevent blockage of the column head leading to a large buildup of pressure or potentially damaging the column resin bed structure.

• Storage Solution
  For short-term storage (up to 3 or 4 days), mobile phase is used for column storage after guanidine hydrochloride injection. In case of long-term storage, the column is best flushed with 50 mM phosphate buffer, pH 7.0 with 0.02% sodium azide. Before storing the column, the end fittings should be tightly sealed with the end plugs that came with the column to prevent the packing from drying.

RESULTS

Zenix^® and Zenix^®-C SEC columns are identical products except for the surface coating. Zenix^® has a stand-up monolayer and Zenix^®-C has a laydown monolayer. The laydown monolayer is specifically designed to alleviate secondary interactions for sticky or hydrophobic sample types getting adsorbed in traditional surface coatings. For highly charged, hydrophobic proteins, or derivatized mAbs with polymer branches, the Zenix^®-C column will show a significant difference to the Zenix^® column.

In this work, both Zenix^® and Zenix^®-C columns were used to characterize the mAb aggregates and fragments using SEC-UV analysis. The injection sequence showing in Table 2 was used for the sample analysis. Briefly, there were three steps in the data collection:

1) System suitability test

2) Unstressed sample test to characterize the mAb aggregates, fragments, and its monomer using SEC-UV

3) Stressed sample test to compare the aggregation of mAb under the force of pH and temperature

All of the data was collected at wavelengths of both 214 nm and 280 nm.

SYSTEM SUITABILITY TEST RESULTS

A volume of 10 µL of the protein standard mix 15-600 kDa was injected onto the SEC-UV system with the UV wavelength set at 280 nm. 150 mM of sodium phosphate buffer (pH 7.0) was used as mobile phase. Zenix^® and Zenix^®-C SEC columns were compared at a column temperature of 23 ^oC with the flow rate at 1.0 mL/min. A representative chromatogram is displayed in Figure 1a. The chromatogram shows the five peaks corresponding to the five components in the standard mix based on the molecular size order thyroglobulin, bovine γ-globulin, chicken ovalbumin, ribonuclease A, and p-aminobenzoic acid (pABA) which is consistent to the vendor Certificate of Analysis (COA). In addition, some aggregates and fragment peaks were detected as well using this method. The performance of Zenix^® and Zenix^®-C columns on the separation of the protein standard here is comparable with sufficient separation between the five standards. The detailed chromatographic data with the retention time, plate number, tailing factor and resolution is listed in Table 5.

Table 5. Chromatographic data for the SEC analysis of the protein standard mixtures on Zenix^® and Zenix^®-C columns.

Compound	Zenix® SEC-300, 3µm, 7.8 x 300 mm				Zenix®-C SEC-300, 3µm, 7.8 x 300 mm
	RT	Plates	Tailing	Resolution	RT	Plates	Tailing	Resolution
Thyroglobulin	6.18	2530	-1.27	NA	5.90	895	-1.25	NA
γ-Globulin	7.73	3687	1.38	3.09	7.30	3901	1.34	2.22
Albumin	9.19	10525	0.93	3.37	8.81	10669	0.94	3.74
Ribonuclease A	10.00	21627	1.06	2.54	9.59	20557	0.95	2.56
pABA	11.88	44611	1.06	7.59	12.33	43384	1.09	10.82

Figure 1. a) Chromatogram of the protein standard mix 15-600 kDa utilizing Zenix^® (Green trace) and Zenix^®-C (Purple trace) column by SEC-UV (280 nm); b) Chromatogram of protein standard sample on Zenix^® column at 214 nm; c) Chromatogram of protein standard sample on Zenix^® column at 280 nm. Peak assignment: 1 Thyroglobulin, 2 bovine γ-globulin, 3 chicken ovalbumin, 4 Ribonuclease A, 5 p-aminobenzoic acid (pABA).

UV detection wavelengths of 214 nm and 280 nm are commonly used for SEC analysis. In this protein standard mix, all components but pABA have higher molar absorptivity at 214 nm than at 280 nm (Figure 1b). Based on Beer’s Law, a peak response depends proportionally on the compound’s molar absorptivity at a certain wavelength. For pABA, its maximum molar absorptivity was at 280 nm, which corresponds to higher peak height at 280 nm than 214 nm (Figure 1c). For better observation, the protein mixture was viewed using a wavelength set at 280 nm as Figure 1a displays. 280 nm is more commonly used in purification analysis because the loading capacity can be increased. In this study, because of the high purity grade of the sample, the 214 nm detection and 10 µL injection volume scheme was selected.

SAMPLE TEST RESULTS – UNSTRESSED SAMPLE

150 mM phosphate buffer, pH 7.0, was used as mobile phase with at flow rate of 1.0 mL/min in an isocratic program to separate the mAbs and their aggregates. An amount of 10 µg of individual, unstressed sample was injected onto the column, and the separation was monitored at a wavelength of 214 nm. The SEC-UV chromatograms of an overlay of five different, unstressed mAbs (adalimumab, infliximab, MSQC16, MSQC18, and MSQC17) on Zenix^® and Zenix^®-C columns at 214 nm were compared in Figure 2 and Figure 3, respectively. For a better view of minor aggregates and fragments peaks, the zoomed-in chromatograms of both Zenix^® and Zenix^®-C columns are also displayed in Figure 2b and Figure 3b, respectively. The peaks eluted after 10 mins in the chromatogram are the formulation buffer and mobile phase buffer peaks. The detailed chromatographic data is listed in Table 6 and Table 7.

The performance of Zenix^® and Zenix^®-C column is comparable in the mAbs analysis shown in this work.

Table 6. Chromatographic data for the SEC-UV analysis of five unstressed mAbs (adalimumab, infliximab, MSQC16, MSQC18, and MSQC17) on Zenix® SEC-300 column at 214 nm.

	Peak	RT (min)	% Area	Plates	Resolution
Adalimumab	HMW	6.73	1.83	1064	NA
	Main	7.63	96.29	10020	1.58
	LMW 1	8.12	1.55	9922	1.54
	LMW 2	9.14	0.33	12486	3.12
Infliximab	HMW	6.52	0.16	2464	NA
	Main	7.64	99.73	7657	2.55
	LMW 1	9.07	0.12	2879	2.8
MSQC16	HMW	6.70	0.63	2049	NA
	Main	7.67	99.27	9810	2.15
	LMW 1	9.16	0.11	5299	3.66
MSQC18	HMW	6.55	4.68	1036	NA
	Main	7.75	92.49	7875	2.06
	LMW 1	8.17	2.41	11284	1.26
	LMW 2	9.21	0.42	5281	2.56
MSQC17	HMW	6.65	0.54	4135	NA
	Main	6.97	1.18	1166	0.52
	LMW 1	7.88	98.10	7486	1.55
	LMW 2	9.31	0.19	5577	3.29

Figure 2. Overlay of five unstressed samples on Zenix^® SEC-300, 30 cm x 7.8 mm I.D., at 214 nm wavelength, a) Full View; b) Zoom-in view. 1 Adalimumab, 2 Infliximab, 3 MSQC16, 4 MSQC18, 5 MSQC17

Table 7. Chromatographic data for the SEC-UV analysis of five unstressed mAbs (adalimumab, infliximab, MSQC16, MSQC18, and MSQC17) on Zenix^®-C SEC-300 column at 214 nm.

	Peak	RT (min)	% Area	Plates	Resolution
Adalimumab	HMW	6.41	1.45	991	NA
	Main	7.20	96.17	8879	1.41
	LMW 1	7.61	2.02	10417	1.36
	LMW 2	8.65	0.36	11006	3.33
Infliximab	HMW	6.15	0.15	2770	NA
	Main	7.19	99.73	7014	2.56
	LMW 1	8.62	0.13	3062	2.96
MSQC16	HMW	6.36	0.61	1839	NA
	Main	7.22	99.27	8359	1.88
	LMW 1	8.66	0.12	4250	3.4
MSQC18	HMW	6.17	5.25	1270	NA
	Main	7.25	91.57	7213	2.09
	LMW 1	7.66	2.77	10272	1.27
	LMW 2	8.67	0.41	7762	2.91
MSQC17	HMW 1	6.21	0.45	5406	NA
	HMW 2	6.51	1.17	1164	0.54
	Main	7.33	98.22	7816	1.51
	LMW	8.73	0.16	3580	3.12

Figure 3. Overlay of all unstressed samples on Zenix^®-C SEC-300, 30 cm x 7.8 mm I.D., at 214 nm wavelength, a) Full View; b) Zoom-in view. 1 Adalimumab, 2 Infliximab, 3 MSQC16, 4 MSQC18, 5 MSQC17

To improve the peak resolution, one strategy is to decrease the sample amount loaded onto the column while analyzing at a wavelength, analytes are more sensitive to. As the on-column loading decreased, the resolution between monomer and fragment (LMWs) peaks regions slightly increased, as shown in Figure 4. adalimumab was selected for this evaluation. The on-column loading amount was tested with 100 µg, 10 µg and 0.2 µg of sample. With decreased injection volume, the resolution and plate number for LMWs was improved as shown in Table 8.

Table 8. Data for all different injection volume of samples.

	Peak	RT (min)	% Area	Plates	Resolution
0.2 ug Adalimumab	HMW	6.76	0.66	1000	NA
	Main	7.66	97.72	10025	1.54
	LMW 1	8.10	1.32	11638	1.45
	LMW 2	9.15	0.31	16935	3.60
10 ug Adalimumab	HMW	6.73	1.83	1064	NA
	Main	7.63	96.29	10020	1.58
	LMW 1	8.12	1.55	9922	1.54
	LMW 2	9.14	0.33	12486	3.12
100 ug Adalimumab	HMW	6.78	3.59	1141	NA
	Main	7.68	93.09	2876	1.31
	LMW 1	8.14	2.72	12596	1.05
	LMW 2	9.19	0.61	841	3.04

Figure 4. Overlay of different injection amounts of Adalimumab onto Zenix^® column at 214 nm. a). 2 µg injection, b). 10 µg injection, c). 100 µg injection. a and b were both injected at 10 µL while varying the concentration of Adalimumab; c was injected 100 µL

If the mAb is highly glycosylated or charged, there may be electrostatic interactions during the run. To alleviate this phenomenon, one can add additional sodium chloride to the elution buffer to help eliminate this interaction. With the additional of 300 mM NaCl (Figure 5) in the elution buffer, the resolution between the two HMWs was improved from 0.54 to 0.74.

Table 9. Data for different mobile phases.

Mobile phase	Peak	RT (min)	% Area	Plates	Resolution
150mM Phosphate Buffer, pH 7.0	HMW 1	6.21	0.45	5406	NA
	HMW 2	6.51	1.17	1164	0.54
	Main	7.33	98.22	7816	1.51
	LMW	8.73	0.16	3580	3.12
150mM Phosphate Buffer, pH 7.0 + 300mM NaCl	HMW 1	6.27	0.64	2787	NA
	HMW 2	6.71	1.00	1346	0.74
	Main	7.55	98.22	7125	1.53
	LMW	8.97	0.15	3046	2.82

Figure 5. Overlay of MSQC17 on Zenix^®-C column while varying additive salt concentrations

SAMPLE TEST RESULTS – STRESSED SAMPLE

Adalimumab sample was stressed by pH change from 6.0 to 1.0 to 10.0 and back to 6.0 and then incubated at 60 °C for 60 min. A volume of 100 µL of both unstressed and stressed adalimumab was injected onto an HPLC system to characterize the difference in HMWs percentage. Running condition was at 1.0 mL/min and 150 mM phosphate buffer, pH 7.0, as the mobile phase.

Figure 6 shows the chromatogram of the stressed (red trace) and unstressed adalimumab (blue trace). After the pH and temperature stress, the HMW aggregates area percentage was increased from 1.832% to 22.855% of the total area count while the main monomer peak dropped from 96.294% down to 69.510% (Table 10). With the pH and temperature stress, the adalimumab fragments increased as well, the percentage of total peak area counts increased from 1.545%, 0.328% to 3.819% and 3.815% as shown in Table 10.

Table 10. Data displaying results for unstressed and stressed adalimumab samples.

Peaks	Unstressed sample		Stressed sample
	RT (min)	% Area	RT (min)	% Area
HMW	6.73	1.83	5.07	22.86
			5.45
			6.89
Main	7.63	96.29	7.71	69.51
LMW 1	8.12	1.55	8.13	3.82
LMW 2	9.14	0.33	9.17	3.82

Figure 6. Comparison of chromatograms of unstressed adalimumab (blue trace) and stressed adalimumab (red trace) obtained on Zenix^® column at wavelength 214 nm

CONCLUSION

This application note demonstrates the complete workflow for the characterization of mAb monomers, aggregates, and fragments using SEC-UV analysis. Zenix^® and Zenix^®-C SEC columns are capable to separate the variety of mAb’s monomers from their aggregates and fragments with sufficient resolution.

This workflow includes a system suitability test using a protein standard mix, unstressed mAbs and pH/temperature stressed mAbs. It describes the experimental practices including the detection wavelength selection (214 nm vs 280 nm), column maintenance tips, and SEC mobile phase and on-column injection amount optimization. In addition, the pH and temperature stressed adalimumab samples were also characterized utilizing a Zenix^® SEC column showing the HMW formation increased in the stressed sample, interfering with the percentage of the monomer of the mAb peak.

This SEC-UV workflow can be suitable for the quantitative analysis of mAbs, with sufficient separation of mAb monomers from their aggregates and fragments.

To Place an Order or Receive Technical Assistance

In the U.S. and Canada, call toll-free 1(800)-645-5476

For other countries across Europe, call +44 (0) 115 943 0840

For other countries across Europe and the world, please visit

MerckMillipore.com/offices

For Technical Service, please visit MerckMillipore.com/techservice

References

1. K. Sandra, I. Vandenheede, P. Sandra, J. Chrom. A 1335, 81 – 103 (2014).

2. S. A. Berkowitz, J. R. Engen, J. R. Mazzeo, G. B. Jones, Nat. Rev. Drug. Discov. 11, 527 – 540 (2012).

3. J. H. Knox, High Performance Liquid Chromatography, Edinburgh University Press, 1978.

4. V. D’Atri, S. Fekete, A. Beck, M. Lauber, D. Guillarme, Anal. Chem. 89, 2086 – 2092 (2017).

5. B. Kükrer, V. Filipe, E. van Duijn, P. T. Kasper, R. J. Vreeken, A. J. R. Heck, W. Jiskoot, Pharm. Res. 27, 2197-2204 (2010).

6. C. J. Roberts, Trends Biotechnol. 32, 372-380 (2014).

7. J. Vlasak, R. Ionescu, Mabs 3, 253-263 (2011).

8. A. Eon-Duval, H. Broly, R. Gleixner, Biotechnol. Prog. 28, 608-622 (2012).

9. B. Demeule, R. Gurny, T. Arvinte, Eur. J. Pharm. Biopharm. 62, 121-130 (2006).

10. W. Jiskoot, T. W. Randolph, D. B. Volkin, C. R. Middaugh, C. Schöneich, G. Winter, W.Friess, D. J. A. Crommelin, J. F. Carpenter, J. Pharm. Sci. 101, 946-954 (2012).