Ion-exchange chromatography, capillary isoelectric focusing, and capillary zone electrophoresis coupled to mass spectrometry for charge variant analysis of monoclonal antibodies

ABSTRACT

Monoclonal antibody (mAb) charge variant analysis, which ensures the quality of biopharmaceuticals, is primarily performed by either cation exchange chromatography (CEX), capillary isoelectric focusing (CIEF), or capillary zone electrophoresis (CZE). Though selectivity is different for all methods, mAb proteoforms can be quantified on the intact level with a low risk of artifacts. Identification and characterization can be performed by mass spectrometry (MS), but coupling is hampered by the use of nonvolatile constituents. Here, we present comparative data and a thorough review of CEX-MS, CIEF-MS, and CZE-MS for the analysis of charge variants on the intact level. The separation of the proteoforms is confirmed by online UV detection with subsequent online MS coupling. Each method was tested for generic use by measuring 10 mAbs with an isoelectric point (pI) of 7.3 to 8.7 without individually adjusting the method parameters for each mAb. The methods and the identified proteoforms are compared among each other as well as to a previously published CZE-MS method on the subunit level. The lysine variant was used to compare the resolution of the different separation methods and showed a high resolving power for all separation methods tested. The different method selectivities were revealed by the monoglycosylation, which was found in the peak of the main variant with CEX and CIEF, while it was separated from the main peak with CZE. In combination with a comprehensive literature review, our results provide a complete overview of the current state of the methods with information on the advantages and limitations of each methodology.

Introduction

Monoclonal antibody (mAb) variants are critical quality attributes that must be analyzed to ensure the stability and safety of biopharmaceuticals. Given the increasing number of mAbs that are approved as therapeutics each year,¹ methods have been developed to address the challenges of mAb variant characterization. Antibody variants are post-translational modifications that are introduced during the production, purification, or storage of the mAb. These modifications often change the mass, the charge, and the isoelectric point (pI) of the mAb, which makes them accessible to analytical techniques that use these characteristics for separation, such as cation exchange chromatography (CEX), capillary zone electrophoresis (CZE), or capillary isoelectric focusing (CIEF). Currently, mAb variant identification is commonly done using mass spectrometry (MS). MS allows the identification of variants by detecting mass differences between modified and unmodified species. For large mass differences and highly abundant proteoforms (like the main mAb glycosylation forms,² MS can distinguish these species without any proteoform pre-separation. Low abundant proteoforms and/or proteoforms with small mass differences, however, are more challenging to characterize using MS. As large molecules (~150 kDa), mAbs have broad isotopic distributions (with a full peak width at half maximum of around 24 Da), which requires separation for detection of proteoforms with a smaller mass difference.

MAbs and mAb variants are positively charged at a pH below their pI value, which makes them accessible for CEX separation. The analysis can also be performed using anion exchange chromatography, but CEX is commonly used due to the more efficient ionization in subsequent electrospray ionization (ESI) in positive mode and the generally basic pI value for most mAbs. CEX-UV separation systems often use MS-incompatible salts or buffers (e.g., sodium hydrochloride or 2-(N-morpholino) ethane sulfonic acid (MES)).^3,4 For MS coupling, ammonium acetate, and ammonium formate-based eluents, which provide good variant separation and identification of mAb variants, are typically used.^5–7 CEX-UV and CEX-MS methods have a limited generic application due to the different pIs of different mAbs in combination with the pH and ionic strength of the buffer system. Thus, individual method optimizations for each mAb can improve the separation significantly.⁸ An ideally linear or at least precisely controlled pH gradient is highly desired to achieve reproducible and high-resolution separations. This can be easily achieved by involatile buffer components,^9–11 but, for MS-compatible ammonium acetate, the pH response as a function of the base content is not linear^3,12 and column-dependent.¹² A method well suited for the separation of mAb variants of one mAb might therefore not be suitable/ideal for another mAb with a different pI value. CEX-MS can be performed with standard LC-MS equipment, but the nanoESI conditions and harsh transfer parameters in MS are required to decluster the ammonia adducts and a high m/z range is needed due to native-like ESI conditions.

The different pI values of mAb variants enable separation by CIEF. CIEF, in particular imaged CIEF (iCIEF), is increasingly used for charge variant analysis of mAbs for quality control and batch release.^13–15 If not specifically stated otherwise, CIEF covers CIEF and iCIEF and both techniques are very well established using UV detection. However, generic CIEF-UV methods contain MS-incompatible substances such as ampholytes, methylcellulose, urea, or sodium hydroxide, which is why CIEF has rarely been coupled with MS until recently.^16–18 CIEF-MS coupling is only possible if MS interfering components are removed or substituted to the greatest possible extent and when focusing and ionization can still be done with the capillary outlet installed in the MS interface.^19,20 Designated commercially available CIEF-MS systems for fast, reliable, and robust mAb variant analysis have been introduced recently.^21,22 Nevertheless, the separation from focusing is generally reduced or even lost by the necessary mobilization process to transport the focused proteoforms to the MS.^19,23 Even chip-based systems, which have short separation canals, suffer from peak broadening during mobilization^22,24

CZE separates mAb variants based on their charge and size in the liquid phase. This principle has been applied to separate mAb variants, typically with a CZE-UV-based system using 6-aminocaproic acid (EACA), triethylenetetramine (TETA), and hydroxypropyl cellulose (HPC) as first published by He et al.²⁵ This generic and widespread method was evaluated in an intercompany study using 23 mAb samples with a pI between 7.4 and 9.5.²⁶ However, as previously mentioned for CEX and CIEF, many CZE-UV methods, including the EACA-UV method, cannot be coupled to MS due to incompatible substances like dynamic coating agents or polymers.^27–29 MS-compatible CZE methods either use native (ammonium acetate-based) or denatured (acetic or formic acid-based) separation systems.^30,31 Commercially available native microchip-based systems provide very good mAb variant separation using native-like separation conditions,^28,32,33 but these systems can suffer from reduced signal intensities and often lack flexibility, similar to CEX-MS methods.

In many studies and applications, one single technology is used. Thus, their performance with respect to resolution, sensitivity, instrumental aspects, and handling issues is difficult to evaluate. In this study, we present data and review the literature for CEX-MS, CIEF-MS, and CZE-MS methods for mAb variant separation on the intact level to evaluate their analytical performance along with practical concerns. Ten mAb samples covering a pI range from 7.3 to 8.7 were analyzed regarding acidic (A) and basic (B) proteoforms by CEX-MS, CIEF-MS, and CZE-MS with one generic method for each technique. Additionally, variant separation was online UV monitored, which proved to be a valuable tool for further method evaluation. The results for each method are discussed individually, compared to the other techniques, and put into perspective with results from the literature. The results from the intact analysis were also compared to a CZE-MS method on the subunit level.^34,35

Results

Intact mAb variant analysis was executed with all three separation techniques, CEX-MS, CIEF-MS, and CZE-MS, as mentioned in the method section. All mentioned proteoforms in this section are shown in the supplementary Table S2-S4 sorted by method and mAb.

CEX-UV-MS for mAb variant analysis

The initial eluents for the CEX-UV-based separation, 50 mM ammonium acetate + 2% acetonitrile (ACN) pH 5.0 as eluent A and 160 mM ammonium acetate + 2% ACN pH 8.5 as eluent B, are based on the report by Murisier et al.¹² Here, a linear gradient from 10% [B] to 80% [B] over 13 min with a flow rate of 0.3 mL/min and a detection wavelength of 280 nm was utilized. As shown by the blue line in supplementary Figure S1, a slight separation of acidic and basic species was achieved for mAb1, but an improvement of the separation of acidic and basic variants from the main form was attempted. Therefore, eluent B starting conditions were successively increased (compare supplementary Figure S1), which resulted in resolved acidic and basic variants, but also massive peak broadening if values above 55% B were used as a starting condition. For MS coupling, the amount of ammonium acetate in eluent B was decreased to 100 mM, resulting in later elution and a decreased separation efficiency (compare Figure 1a). The method length was therefore increased to 40 min after reducing the flow rate to 0.1 mL/min, and the gradient was adapted to the prolonged analysis time rising linearly after a 2 min hold at 55% [B] to 85% [B] over 25 min. The final gradient resulted in baseline-separated acidic and basic peaks, respectively, as shown in Figure 1b.

Figure 1.

Method development (combined salt and pH gradient) using 55%B starting conditions for charge variant separation of mAb1. a: reduction of ammonium acetate in eluent B from 160 mM to 100 mM. Black: 160 mM, red: 140 mM, orange: 120 mM, blue: 100 mM, the gradient is shown as a gray line. b: mAb1 variant separation using a final gradient (gray line) with 100 mM ammonium acetate in eluent B, a flow rate of 0.1 mL/min, and an adapted gradient for a 40-minute run time.

After UV-based gradient optimization, we coupled the method to the MS (including online UV detection) and measured all 10 mAb samples (2 mg/mL). The online UV chromatograms, the extracted ion chromatograms (EICs, m/z: 4500–6500), and the deconvoluted masses of the main peak are shown in Figure 2. All mAbs show retention on the column and therefore variant separation no matter their respective pI values. All mAbs contain basic proteoforms, which elute later than the main peak, and acidic proteoforms, which elute earlier than the main peak. Even for highly retained mAbs like NIST mAb and mAb2 (pI 8.6 and 8.7, respectively) acidic and basic variants are separated. The method is expected to be generic for the tested pI range (pI: 7.3–8.7). Comparing the resolution (R) for mAbs with at least one lysine variant, we calculated values between the main variant (M) and the basic variant (B1) of R = 0.85 for USPmAb003, R = 1.38 for NIST mAb, and R = 2.66 for infliximab. The separation is maintained toward the MS, which is evident when comparing the UV and MS data in Figure 2, with average extracted ion electropherogram (EIE) signal intensities in the 5E5 range. mAbs with high heterogeneities (cetuximab and pH-stressed mAb1) are challenging since signal intensities are decreased by the heterogeneity when the same concentration of mAb is used for all mAb samples. The generally low spectral quality is also a challenge for all detected acidic or basic species, especially if they are of low abundance. A +2 Da mass shift was detected only for mAb1, with its prominent acidic variant (A1). For all other, low abundant acidic peaks, no statistically significant mass shift was observed. Nevertheless, the method can be used to gain mAb information. After the deconvolution of the main peak (compare right column in Figure 2), the glycosylation pattern was analyzed and intense glycoforms (G0F/G0F, G0F/G1F, G1F/G1F) were annotated for all mAbs. There is a slight separation depending on size between the different glycosylation patterns and mAbs with smaller glycans elute slightly later than mAbs with larger glycans (compare supplementary Figure S2). In addition to that, for some mAbs, even minor glycoforms like G0F/G0F-GlcNAc, G0F/G0, or G2F/G2F+Hex (possible glycation of NISTmAb) were detected (compare right column in Figure 2). Monoglycosylated antibodies were also detected in the main form peak for USPmAb003, mAb1, mAb3, and mAb4. Basic proteoforms like the lysine variants could be deconvoluted and assigned for USPmAb003 (B1, B2), NISTmAb (B1, B2), and infliximab (B1, B3).

Figure 2.

CEX-UV-MS separation and identification of 10 mAb samples all measured on the Bruker MaXis TOF instrument. Left column: online UV traces (214 nm); middle column: extracted ion chromatograms (EICs, 4500–6500 m/z); right column: deconvoluted mass spectrum of the main form. A: acidic variant, M: main variant, B: basic variant.

CIEF-UV-MS for mAb variant analysis

The initial CIEF conditions were taken from Naghdi et al..²⁰ The master mix for mAb A (pI:8.66) variant separation was composed of 1.25% of Pharmalyte pH 8 − 10.5, 0.62% of Pharmalyte pH 3 − 10, 0.3% of Pharmalyte pH 5 − 8 (total ampholyte concentration of 2.17%), 2.5 mM arginine, 0.5 mM IDA, 30% formamide and 0.5 mg/mL mAb A. CIEF was hyphenated to the MS with the new chemical mobilization methodology for charge variants separation of mAb A, demonstrating the feasibility of the CIEF−ESI-MS.²⁰ We briefly tested in a UV-based setup whether mAb1 (pI: 8.6) can be separated with the same CIEF system and if reduced concentrations of ampholyte and formamide in the master mix affect mAb1 variant separation. First, the ampholyte concentration in the master mix was reduced from 2.1% to 1%, which resulted in reduced analysis time without losing the separation of the mAb variants significantly (compare supplementary Figure S3A). Following this, the amount of formamide was successively reduced in 5% steps from 30% to 15% without losing the separation of the mAb variants (compare supplementary Figure S3B). Arginine, IDA, and the final sample concentration remained unchanged and used as mentioned by Naghdi et al.²⁰ at concentrations of 2.5 mM, 0.5 mM, and 0.5 mg/mL respectively. Ampholyte and formamide were used at concentrations of 1% ampholyte and 15% formamide as mentioned above.

This system was coupled to the MS with the nanoCEasy interface, which allowed chemical mobilization within the interface by exchanging the sheath liquid from basic conditions during focusing to acidic conditions for chemical mobilization along with acidic ESI conditions. The system was equipped with an external UV detector to obtain online UV data during mobilization before entering the MS. The method used for mAb1 variant separation was applied to all nine mAb samples (except for stressed mAb1). The online UV results, the MS spectra (m/z: 4500–6500), and the deconvoluted masses for the main peak are shown in Figure 3. For all mAbs, a separation of several acidic and basic variants in the online UV detection (see CIEF-UV column in Figure 3) is obtained, independently from their respective pI values (pI: 7.3–8.7). All mAbs contain basic proteoforms that migrate earlier than the main peak and acidic proteoforms that migrate later than the main peak. For the tested pI range, the method can be seen as generic and MS signal intensities are on average in the 1E6 range (see CIEF-MS column in Figure 3). The resolution (R) for mAbs with at least one lysine variant was calculated based on the UV signal and values between the main variant (M) and the basic variant (B1) of R = 1.50 for USPmAb003, R = 1.27 for NIST mAb, and R = 1.68 for infliximab were obtained. However, while online UV can detect all mAbs, MS detection seems to be highly antibody-dependent. USPmAb003 (pI 8.1), infliximab (pI 7.3), and mAb3 (pI 7.3) were not detected via MS, possibly due to solubility issues. The other mAbs can be detected via MS, but in most cases, separation is lost. The overall glycosylation pattern of G0F/G0F, G0F/G1F, G1F/G1F, G1F/G2F, and G2F/G2F could be detected in the main peak for all mAbs (compare right column in Figure 3). In addition, some minor glycosylation species like G0F/G0, G0F/G0F-GlcNac, and G2F/G2F+Hex (possible glycation of NIST) could be detected for several mAbs, as well as monoglycosylated species of G0F and G1F for mAb1, mAb2, and mAb4. For cetuximab, glycosylation seems to increase for the acidic species (compare supplementary Figure S4). All mAbs show some acidic and basic variants in the UV detector that could not be identified by MS. This indicates that further improvement is needed until the CIEF-MS system is robust regarding handling and can be used universally for a more detailed mAb variant analysis.

Figure 3.

CIEF-UV-MS separation and identification of nine mAb samples all measured on the Bruker MaXis TOF instrument. Left column: online UV traces (280 nm); middle column: extracted ion electropherograms (EIEs, 2000–4000 m/z, smoothed: gaussian 5); right column: deconvoluted mass spectrum of the main form. A: acidic variant, M: main variant, B: basic variant.

CZE-UV and CZE-MS for mAb variant analysis

All 10 mAb samples were analyzed with the CZE-UV method from He et al.²⁵ that was slightly modified regarding the flushing pressure due to the much longer capillary used in our setup. As shown by the CZE-UV results (Figure 4, left column), all mAbs were analyzed successfully and showed different amounts of variant separation in the 30-minute separation window. All mAbs contain basic proteoforms that migrate earlier and acidic proteoforms that migrate later than the main peak. Some mAbs like infliximab, cetuximab, and stressed mAb1 show very prominent and intense variants while other mAbs like mAb3–5 only show some minor variants. The resolution (R) for mAbs with at least one lysine variant was calculated and values between the main variant (M) and the basic variant (B1) of R = 5.10 for USPmAb003, R = 3.92 for NIST mAb, and R = 8.30 for infliximab were obtained. A slightly longer analysis time might be used for cetuximab, as the separation seems not to be completed after 30 minutes in a 60 cm capillary. The CZE-UV method shows a very good reproducibility regarding mAb1 main variant migration time (18.05 ± 0.11 min, n = 18) and is generic regarding variant separation over the tested pI range (pI: 7.3–8.7). However, the background electrolyte (BGE) of the CZE-UV method is not MS-compatible.

Figure 4.

CZE-UV and CZE-MS separation and identification of 10 mAb samples. MS data were acquired using the Orbitrap fusion Lumos. Left column: CZE-UV traces (214 nm); middle column: base peak electropherograms (BPEs, m/z: 2000–4000, smoothed gaussian 5); right column: deconvoluted mass spectrum of the main form. A: acidic variant, M: main variant, B: basic variant.

CZE-MS method development and optimization were published by Höchsmann et al.³⁰ and their final method was used here to analyze all 10 mAb samples. The method contains a BGE of 4 M aqueous acetic acid (HAc) and a cationic 5-layer DEAED-PSS successive multiple ionic-polymer layers (SMIL)-coated capillary to generate a reversed electroosmotic flow (EOF) of median velocity. The base peak electropherograms (BPEs, m/z: 2000–4000, smoothed Gaussian 5), and the deconvoluted result of the main peak are shown in Figure 4. As shown by the CZE-MS results in Figure 4, separation of acidic and basic variants can be achieved for eight of the 10 mAb samples with an average signal intensity in the 1E6 range. Compared to the CZE-UV method, in this CZE-MS method, the migration order is reversed, e.g., basic proteoforms are detected later and acidic proteoforms are detected earlier than the main peak. For cetuximab and pH-stressed mAb1, no variant separation is observed due to the increased heterogeneity from the complex glycosylation of cetuximab and the presumably increased amount of deamidation in the high pH-stressed mAb1 sample. Deamidation for pH-stressed mAb1 is assumed since the deconvoluted mass is 2–4 Da larger over the complete signal compared to the main peak of the non-stressed mAb sample (data not shown). Nevertheless, the CZE-MS system still is generic regarding variant separation over the tested pI range (pI: 7.3–8.7) if mAb heterogeneity is not too complex.

The main form of each mAb contains the different glycosylation species (compare the right column in Figure 4). For all eight mAbs, the main glycosylation species G0F/G0F, G0F/G1F, G1F/G1F (or G0F/G2F), G1F/G2F, and G2F/G2F were identified and detailed masses can be found in supplementary table S2-S4. In addition, some mAbs showed minor glycosylation species like G0F/G0, G0F/G0F-GlcNac, and G2F/G2F+Hex (or + 2Hex, possible glycation of NIST mAb and mAb1) that were also detected. The EIEs of the different glycosylation species revealed a small shift in migration time depending on the size of the glycans (data not shown, compare Höchsmann et al..³⁰ For USPmAb003 (B1 and B2), NIST mAb (B1 and B2), mAb3 (B1), and infliximab (B1 and B2), the later migrating, basic proteoforms were identified as lysine variants (1x and 2x lysine remaining at the mAb; detailed masses can be found in supplementary table S2-S4). The resolution (R) for these mAbs with at least one lysine variant was calculated and values between the main variant (M) and the basic variant (B1) of R = 1.66 for USPmAb003, R = 1.52 for NIST mAb, and R = 0.94 for infliximab were obtained. For mAb1 (B2) and mAb2 (B1), the basic proteoform could be a loss of water or ammonia (detailed masses can be found in supplementary table S2-S4). Additionally, the basic variants of USPmAb003 (B1), NIST mAb (B2), mAb1 (B1), mAb2 (B1), mAb3 (B1), mAb4 (B1), and mAb5 (B1) contain monoglycosylated mAb species. Aglycosylation was determined for mAb2 (B2) and mAb4 (B2). Other later migrating species are presumably mAb fragments (e.g., mAb with loss of a light chain) since they show the characteristic 162 Da pattern specific for glycosylation. These fragments (detailed masses can be found in the supplementary table S2-S4) were detected for USPmAb003 (123 kDa and 101 kDa), mAb2 (148 kDa, 126 kDa, and 103 kDa), mAb3 (125 kDa), mAb4 (126 kDa), and mAb5 (125 kDa).

For USPmAb003 (A1) and mAb1 (A1), earlier migrating, acidic species are mAb proteoforms that carry glycans with sialic acids (detailed masses can be found in the supplementary table S2-S4). Additionally, all eight mAbs showed acidic species with a slight shift toward lower m/z values in the charge envelope distribution compared to the main peak. This shift in m/z value in combination with a deconvoluted mass shift of around 2 Da indicates mAb variants with an open disulfide bridge.

CZE-MS for mAb variant analysis on the subunit level

In addition to all intact approaches, the mAbs were analyzed via CZE-MS (MS/MS) on the subunit level as described elsewhere.³⁵ For the 10 different mAb samples, different proteoforms like oxidation, glycation, lysin variants, sialic acids, pyroGlu, water or ammonium losses, and aglycosylation were separated and identified (compare supplementary Figure S8). Some of them were not determined on the intact level. Further information regarding the subunit approach and a detailed list of the identified variants of each antibody can be found in our previous publication and supporting information.³⁵

Discussion

CEX-UV and CEX-UV-MS for mAb variant analysis

Many systematic studies involving UV detection of CEX-separated mAb variants are available. Generally, three approaches are used to elute proteins from the CEX column: 1) increasing the pH, which reduces the charge of the protein,^11,36,37 2) adding salts that displace the proteins from the column^4,8,38–40 or 3) a combination of both approaches.^10,12 The individual methods were compared in several studies and both showed good separation of mAb variants.^3,41,42 The composition and pH of the eluents, rather than the method, seem to be more important for proper variant separation. Farsang et al.³ compared different salt and pH gradients with the result that a pure pH gradient with ammonium acetate resulted in no separation of charge variants while a pH gradient containing MES showed separated mAb variants. Still, gradients are only partly transferable and cannot be applied automatically to other mAbs with the same separation power. Goyon et al. showed that the separation power can be optimized for infliximab when adapting the gradient compared to a generic method.⁸ Zhang et al.¹⁰ proposed what is probably the most generic CEX-UV method, in which a salt-mediated pH gradient provided mAb variant separation over a tested pI range from 6.2 to 9.4. All in all, CEX-UV methods generally separate mAb variants very efficiently, but they are mostly composed of buffers that contain MES-buffer,^{4,8,38,41–43} sodium hydrochloride,^10,39,40,42 or Tris-buffer,^11,37 which are not compatible to online ESI-MS. Potentially information can be extracted from these separation systems by collecting fractions and determining the mAb variants of each fraction by an MS-compatible method^44–49 or by using 2D and 4D setups.^50,51 However, this requires extra-analysis time, as well as the risk of inducing variants during sample handling. Automatic, multidimensional methods where CEX is used in the first dimension can reduce the risk of artificial variants, but the methods are still time-consuming, and multidimensional systems are elaborative and hardly used for routine analysis applications.^50–52 The development of MS-compatible separation systems is therefore of utmost importance.

Online CEX-MS studies are listed in Table 1. MS-coupled CEX separation methods mainly work with ammonium acetate or ammonium formate-based eluents, which are either used as pure pH gradients with weak cation exchangers^5,6,48 or as a combination of salt and pH gradient with strong cation exchangers.^{7,12,64,66,68,69,71} Another common eluent system uses bicarbonate and acetic acid as eluent A and ammonium hydroxide as eluent B with strong cation exchangers.^21,42,59,63 CEX-MS methods have limited generic application due to the different pIs of mAbs in combination with the pH and ionic strength of the buffer system. As can be seen from Table 1, around half of the methods were developed specifically for just one mAb, or in the case of several mAbs, the gradient was adapted according to the respective mAb. A linear pH gradient profile is highly desired to achieve a generic method for mAbs of varying pI values and can be generated using a set of involatile buffer components.^9–11 Using ammonium acetate, the pH response as a function of the base content is nonlinear^3,12 and column dependent.¹²

Table 1.

Direct and online CEX-MS methods for intact mAb analysis (excludes ADCs, fraction collection, or multidimensional methods). mAbs 1–5 included in this table do not correlate with mAbs 1–5 mentioned in the results section.

Sample(s)	Column	Eluent A	Eluent B	Gradient	Interface	MS	Aim of analysis	Ref.
mAb1, 2, and 3	ProSwiftTM SCX-1S AND ProSwiftTM WCX-1S (4.6 mm × 50 mm)	5 mM AMH buffer containing 20% (v/v) methanol at pH 9.5	5 mM AMH buffer containing 20% (v/v) methanol at pH 10.5	A linear gradient of eluent A to eluent B in 20 min. The flow rate was 1 mL/min.	Agilent G1385A microflow nebulizer	TOF	Resolve charge variants of mAbs with the aid of CEX monolithic columns	Talebi et al.53
Trastuzumab	ProPac WCX-10 column (2.0 mm × 250 mm, 5 μm particle size)	50 mM ammonium acetate pH 6.6	50 mM ammonium acetate, pH 10.1	20 min gradient of 1–8% B and flow rate of 0.3 mL/min	Not specified	Orbitrap	Native intact mass and charge variant analysis	Bailey et al.5
Trastuzumab Adalimumab Infliximab Bevacizumab Cetuximab	MAbPac SCX-10 RS (2.1 mm × 50 mm, 5 μm particle size)	25 mM ammonium bicarbonate and 30 mM acetic acid (pH 5.3)	10 mM ammonium hydroxide in 2 mM acetic acid (pH 10.18)	mAb dependent compare Table 1 in publication	Heated electrospray ionization-II (HESI-II) Ion Max ion source	Orbitrap	Low ionic strength pH gradient elution for characterization of mAb charge variants	Füssl et al.54
mAb	ProPac™ WCX-10 (4 mm × 250 mm, 10 μm particle size)	20 mM ammonium formate pH 6.5	20 mM ammonium formate pH 8.6	0% B for equilibration; change in the composition of B to 17% for 0.00–10 min, 38% B for 10.00–20.00 min, then linear increase to 100% B for 20.00–50.00 min and 50.00–57.00 min, and re-equilibration by 0% B for 57.00–58.00 and 58.00–65.00 min in the end. Used a constant flow rate of 0.8 mL/min	Not specified	TOF	Analysis and characterization of mAb charge variant heterogeneities	Sankaran et al.6
deglycosylated NISTmAb and mAb1–5	YMC-BioPro SP-F (100 mm × 4.6 mm, 5 μm particle size)	20 mM ammonium acetate, pH 5.6	140 mM ammonium acetate, 10 mM ammonium bicarbonate, pH 7.4	100% mobile phase A for 2 min; linear increase to 100% mobile phase B in 16 min, hold at 100% mobile phase B for 4 min, and then return to 100% mobile phase A to recondition the column for 7 min. Used a flow rate of 0.4 mL/min.	Nanospray Flex Ion Source	Orbitrap	In-depth charge heterogeneity characterization of therapeutic mAbs	Yan et al.55
Trastuzumab Adalimumab Infliximab Bevacizumab Cetuximab NIST mAb Rituximab	MAbPac™ SCX-10 RS column (2.1 mm × 50 mm, 5 μm particle size)	25 mM ammonium bicarbonate and 30 mM acetic acid in water (pH 5.3)	10 mM ammonium hydroxide in water (pH 10.9)	mAb dependent compare Table 1 in publication	Heated electrospray ionization-II (HESI-II) Ion Max ion source	Orbitrap	Charge variant analysis of mAbs (Adalimumab in detail)	Füssl et al.56
Trastuzumab Infliximab NIST mAb Rituximab	BioResolve SCX mAb, (2.1 mm × 50 mm, 3 μm particle size)	50 mM ammonium acetate, pH 5.0, 2% acetonitrile	160 mM ammonium acetate, pH 8.5, 2% acetonitrile	Start at 40% Eluent B for 1 min, linear gradient to 98% Eluent B (1-21 min), hold for 1 min, return to start conditions (1 min), and re-equilibration for 7 min. Used a flow rate of 0.1 mL/min.	Not specified	TOF	The broad applicability of the method on the BioAccord LC-MS System	Ippoliti et al.57
NISTmAb and mAb1, 2, and 3	Bioresolve SCX mAb (2.1 mm × 100 mm, 3 µm particle size)	20 mM ammonium acetate pH 5.6	200 mM ammonium acetate pH 7.0	Isocratic elution at 22% B for 2 min, linear gradient from 22% to 100% B in 16 min, column wash for 9 min at 100% B, and equilibrated during 17 min at 22% B. A flow rate of 0.4 mL/min	Not specified	TOF	A cation exchanger was evaluated for the charge variant analysis of mAbs in direct coupling with native MS	Leblanc et al.7
Bispecific mAbs Misspaired mAbs	ProPac WCX-10 column (4 mm ×250 mm, 10 µm particle size)	50 mM ammonium acetate, pH 7	50 mM ammonium acetate plus 0.8% ammonium hydroxide, pH 10	1%-8% solvent B over 40 min. Step-change to 100% B for 5 min. Step-changed to 1% B for 10 min. Used a flow rate of 300 mL/min	Not specified	Orbitrap	Optimized native charge variant mass spectrometry application for bispecific and mispaired IgG separation	Phung et al.58
Cetuximab	Not specified	6.25 mM ammonium bicarbonate and 7.5 mM acetic acid in water (pH 5.40)	5 mM ammonium hydroxide in water (pH 9.85)	A linear gradient of 20 − 100% buffer B in 15 min followed by 3 min of column flushing at 100% buffer B, 3 min equilibration to 0% buffer B, and 12 min equilibration to the starting conditions of 20% buffer B. Flow rate not specified.	Not specified	Orbitrap	In-depth analysis in terms of charge variant and glycoform heterogeneity using two orthogonal approaches (CEX and CZE)	Füssl et al.59
Trastuzumab (stressed) Infliximab NIST mAb mAb1–4	MABPac SCX-10 RS (2.1 mm × 50 mm column, 5 µm particle size)	20 mM ammonium acetate (pH 5.5)	140 mM ammonium acetate, 10 mM ammonium bicarbonate, pH 7.4	100% A or 60% A for 2 min, followed by a linear curve to 100% B in 12 min. The gradient stayed at 100% B for 3 min for column flushing and returned to the initial gradient. It was then re-equilibrated for 6 min. Three different flow rates (0.1 mL/min, 0.2 mL/min, 0.4 mL/min)	Not specified	Orbitrap	pH and salt gradients were used for the universal separation of mAbs (wide range of pI values)	Ma et al.60
Roledumab	MAbPac SCX-10 column (4 mm ×150 mm)	50 mM ammonium formate buffered with formic acid at pH 3.9	500 mM ammonium acetate at pH 7.4	After an isocratic elution at 15% of B for 5 min, a linear gradient was applied from 15% to 31% in 40 min thus to 85% of B in 30 min. The column was then washed for 5 min at 90% of B and further equilibrated for 15 min at 15% of B. Using a flow rate of 0.4 mL/min	Not specified	TOF	Identification of the charge variants of purified mAb dimers.	Rouby et al.61
mAb	ProPac WCX-10 column (2.0 mm × 250 mm, 5 µm particle size)	Ammonium acetate, pH 5.0	Ammonium acetate, pH 9.5	100% A for 1 min followed by an increase to 16% B in 2 min. Increase to 18% B linearly in 40 min. Further, increase to 100% B in 2 min and hold for 4 min. Return to 100% A in 2 min and hold at 100% phase A for 10 min. Three different flow rates (0.1 mL/min, 0.2 mL/min, 0.4 mL/min)	Not specified	Orbitrap	Addition of top-down to CEX-MS to achieve site-specific structural information of charge modifications	Shi et al.62
MabThera® Reditux™	MAbPac SCX-10 RS (2.1 mm × 50 mm column, 5 µm particle size)	25 mM ammonium bicarbonate and 30 mM acetic acid in water (pH 5.3)	10 mM ammonium hydroxide in water (pH 10.8)	75% to 99% mobile-phase B in 15 min followed by a flushing step with 99% B for 5 min an equilibration step with 10% B for 10 min and 75% B for 25 min. Using a flowrate of 400 µL min − 1	Not specified	Orbitrap	Separation of charge variants of MabThera® and Reditux™ under native conditions using volatile salts	Di Marco et al.63
TCB Mab	ProPac WCX-10 HPLC column (4 mm × 250 mm, 10 μm particle size)	50 mM ammonium acetate, pH 6.0	50 mM ammonium acetate, pH 10.7	1% to 8% solvent B in 25 min, from 8% to 18% solvent B in 15 min, and from 18% to 100% solvent B in 2 min washing step from 100% to 1% solvent B in 2 min and an equilibration step with 1% solvent B for 6 min. Using a flow rate of 300 μL/min	Nanospray Flex ion source	Orbitrap	Characterization of the TCB Mab charge variants including native CEC-UV/MS method development	Haberger et al.48
NIST mAb, Adalimumab, Dalotuzumab, Elotuzumab, Infliximab Trastuzumab	BioResolve SCX mAb (2.1 mm × 50 mm, 3 μm particle size) AND MabPac SCX-10 RS (2.1 mm × 50 mm, 5 μm particle size)	50 mM ammonium acetate with 2% acetonitrile, pH 5.0	160 mM ammonium acetate with 2% acetonitrile, pH 8.5	mAb dependent: compare method section in the publication	Not specified	TOF	CEX-MS method development for the analytical characterization of various biopharmaceutical products	Murisier et al.12
Emicizumab	BioResolve SCX mAb (2.1 mm × 50 mm, 3 μm particle size)	50 mM ammonium acetate in 2% acetonitrile (pH 5.0)	160 mM ammonium acetate in 2% acetonitrile (pH 8.5)	Isocratic step at 50%B for 1 min followed by 50%–98%B in 16.5 min. 98%B was kept for 1.5 min followed by a column re-equilibration step at 50%B for 7 min. The flow rate was 0.1 mL/min.	Not specified	TOF	Comprehensive characterization of emicizumab by using orthogonal chromatographic methods coupled to MS	Duivelshof et al.64
NIST mAb IgG1, 2, 3, 4, 5, 6	YMC-BioPro QA-F SAX column (4.6 mm × 100 mm, 5 μm particle size) AND YMC BioPro SP-F SCX column (4.6 mm × 100 mm, 5 μm particle size)	10 mM ammonium acetate, pH 6.7/20 mM ammonium acetate, pH 5.6	300 mM ammonium acetate, pH 6.8/150 mM ammonium acetate, pH 6.8	100% mobile phase A for 2 min followed by a linear increase to 100% mobile phase B over 16 min. The gradient was then held at 100% mobile phase B for 4 min before returning to 100% mobile phase A. Using a flow rate of 0.4 mL/min.	Nanoelectrospray ionization (NSI) using a Microfabricated Monolithic Multinozzle emitter	Orbitrap	Development and optimization of a native AEX-MS method for charge variant analysis of mAbs.	Liu et al.65
Bevacizumab NISTmAb	BioResolve SCX mAb (2.1 mm × 100 mm, 3 μm particle size)	50 mM ammonium acetate with 20% acetonitrile, pH 5.5	150 mM ammonium acetate with 20% acetonitrile, pH 8.5.	Linear gradient: 45%–68% B in 25 min for bevacizumab and 40%–98% B in 20 min for NISTmAb. The gradient flow rate was 0.1 mL/min.	Jet stream electrospray ionization ion source	TOF	In-depth analysis of unstressed therapeutic mAbs and detailed protein heterogeneity	Dai et al.66
Adalimumab and biosimilars	ProPac 3 R SCX (2.1 mm × 50 mm, 3 µm particle size)	25 mM ammonium bicarbonate and 30 mM acetic acid water, pH 5.3	10 mM ammonium hydroxide in water pH 10.9	A linear gradient from 20–60% B in 20 min. The gradient time was followed by a 5 min column regeneration at 100% B and 10 min re-equilibration. The flow rate was 0.2 mL/min.	Not specified	Orbitrap	Characterization of a monodisperse SCX column format for monoclonal antibody charge variant separation	Füssl et al.42
Eculizumab Trastuzumab Pembrolizumab	BioResolve SCX column (2.1 mm × 100 mm, 3 μm particle size)	50 mM ammonium acetate at pH 5.0	160 mM ammonium acetate at pH 8.6	For intact trastuzumab, a linear gradient from 50% to 70% B in 10 min the mixture of mAbs was separated by first maintaining 0% B for 1 min followed by a linear gradient from 30% to 70% B in 9 min. The flow rate was 0.1 mL/min.	Z-Spray source	TOF	Online CEX−CIU method for the in-depth characterization of different mAb populations in their native state	van Schaick et al.67
mAb1	MabPac SCX-10, (2 mm × 250 mm, 10 μm particle size)	20 mM ammonium acetate, pH 5.6	150 mM ammonium acetate pH 6.8	According to a paper “similar to” Ma et al.60 and Yan et al.55 mentioned in this table here	Heated electrospray ionization source.	Orbitrap	Evaluation of succinimide formation in the CDR of mAb1	VanAernum et al.68
Infliximab Pembrolizumab Adalimumab Bevacizumab Daratumumab Atezolizumab Denosumab Guselkumab Rituximab	SCX: Not further specified	25 mM ammonium bicarbonate, pH 5.3	10 mM ammonia, pH 10.9	mAb dependent compares Table 2 in publication. The flowing rate was set at 0.3 mL/min	Ion Max ESI ion source, and a 34-gauge needle	Orbitrap	Comparison of CIEF and SCX regarding protein charge variant separation resolution, sensitivity, carryover effects, accuracy, and throughput	Wu et al.21
Bispecific mAb	bio pro-SPF-IEX (100 mm × 4.6 mm, 5 μm particle size)	50 mM ammonium acetate, pH 5.5	150 mM ammonium acetate, pH 10.2	100% A for 3 min, a 2% B/min gradient applied for 35 min, then washed with 100% buffer B for 5 min and equilibrated with 100% A for 5 min. Using a flow rate of 0.5 mL/min.	Not specified	Orbitrap	Characterization of the charge heterogeneity associated with bispecific antigen-binding proteins	Shah et al.69
Pembrolizumab Cetuximab Trastuzumab	Self-packed SCX columns (150 mm × 75 or 100 μm I.D.)	50 mM ammonium acetate, pH 5.0	250 mM ammonium acetate, pH 8.5	100% MPA went to 100% MPB in 22.5 min with directly coupled nanoflow (250 or 500 nL min − 1)	Nanospray-flex-series ion-source platform	Orbitrap	Development of an online nanoflow strong cation exchange chromatography with native MS detection for mAb separation	Zhai et al.70

Abbreviations: AMH: Ammonium hydroxide; CDR: Complementarity-determining regions, CEC-UV-MS: Cation-exchange chromatography native electrospray mass spectrometry, CIEF: Capillary isoelectric focusing, CIU: Collision-induced unfolding, CZE: Capillary zone electrophoresis, HESI: Heated electrospray ionization, IEX: Ion exchange chromatography, SAX: Strong anion exchanger, MPA/MPB: Mobile phase A or B; SCX: Strong cation exchanger, WCX: Weak cation exchanger.

A buffer system consisting of 50 mM ammonium acetate with 2% of ACN at pH 5.0 (eluent A), and 160 mM ammonium acetate with 2% of ACN at pH 8.5 (eluent B) provides a good compromise between pH range, stability, and MS compatibility.¹² However, we witnessed quite high ion suppression effects when using these concentrations coupled to the MS. After method optimization, we coupled the CEX to the MS and analyzed the separated signals. Overall, we were able to detect several proteoforms, including the glycosylation pattern, monoglycosylation, and lysine variants when applicable. This corresponds to data from other studies since these variants are frequently detected with CEX-MS.^7,12,59,63 The difference in main and lysine variant resolution (ranging from R = 0.8 to R = 2.6) for the different mAbs most likely correlates with the previously mentioned challenge for linear pH gradients in ammonium acetate buffer systems. We were not able to detect minor mass proteoforms like sialylated glycoforms, succinimide, and pyroGlu modifications even though CEX-MS is generally able to detect sialylated glycans,⁶³ succinimide,⁶ and pyroGlu¹² proteoforms for other mAbs with possibly higher abundances. Our experiments, especially for cetuximab and stressed mAb1, show that the increase in heterogeneity, caused by a second glycosylation site or the induced deamidations, significantly reduces the spectral quality. The low spectral quality in combination with the native ionization and the generally low abundance of some proteoforms limited the resulting information, which can be overcome by applying nanoESI as discussed in more detail below.

Limitations regarding the information from the MS are particularly related to acidic variants containing deamidated proteoforms. A common conclusion in the literature is that the unambiguous assignment of deamidated species is not feasible due to the small mass shift of 1 Da, which is mostly not distinguishable by the MS on an intact level.^{6,12,42,63,66} Instead, often fractions of the CEX separation are collected and then additionally analyzed by peptide mapping approaches^{48,54,56,69,71} or on a subunit level.^6,71

CIEF-UV and CIEF-UV-MS for mAb variant analysis

Two general detection approaches are used when mAb variants are analyzed by CIEF. The whole capillary can be monitored (imaged, iCIEF) or the focused analytes can be mobilized toward the detector (CIEF). If not specifically stated otherwise, our discussion of CIEF covers CIEF and iCIEF. Depending on whether the mobilization is done pressure-driven or chemically, the separation previously established during the focusing step can decrease. In particular, pressure-driven mobilization can significantly reduce the peak resolution compared to a chemical mobilization,⁸⁷ which is why chemical mobilization is standard in CIEF-UV.^{17,18,39,87–89} In the case of the iCIEF methods, the focused variants do not have to be mobilized toward the UV detector, which eliminates losses based on mobilization. However, this benefit of iCIEF does not play any role when coupled with MS since mobilization is required for MS characterization. Apart from mobilization, the separation is influenced by various chemicals used to optimize the CIEF process, as shown by Cao et al.¹⁷ and Lin et al..¹⁸ Both studies showed that especially the urea, methylcellulose, carrier ampholyte, and arginine concentrations have a significant effect on the mAb variant separation and have to be carefully adjusted. Methylcellulose enhances the viscosity, which facilitates better separation and serves as a neutral hydrodynamic coating that reduces protein wall interactions.^16,17 Urea is used as a solubilizer to prevent protein aggregation,^{16–18,89–91} and arginine functions as a spacer to prevent the loss of carrier ampholytes, pI markers, or analytes.^16–18,90 The carrier ampholytes establish the pH gradient during focusing and therefore have a major contribution to mAb variant resolution.¹⁶ However methylcellulose^{16,17,39,87,88} or molar concentrations of urea in the master mix^{16–18,39,87,89,91} are not MS compatible and the use of sodium hydroxide as catholyte^{16–18,39,89} eliminates any protein signal when coupled to MS.

The process of focusing and mobilization is also a challenge when CIEF is coupled to MS since the capillary is no longer positioned in an outlet vial and cannot be exchanged in case of chemical mobilization. These were the main reasons why CIEF was not directly coupled to MS for years and, for example, 2D separations were developed.^92,93 Indirect coupling to MS via a valve was shown by Montealgre et al. and Hühner et al.^94,95 to remove MS-incompatible substances before the MS. In 2018 Dai et al. coupled CIEF directly to MS for the detection and characterization of mAb variants¹⁹ by applying a nanoSL interface (EMASS-II) that allowed focusing with an ammonium hydroxide plug (partial filling approach) in the capillary while the capillary end was installed in the interface with acidic SL. MS-interfering components were eliminated as much as possible by using a neutral-coated capillary, reducing the ampholytes concentration, and substituting urea with glycerol. Substituting MS incompatible components is the main task when CIEF is to be coupled to MS as shown in Table 2. Methylcellulose in the master mix can be avoided using capillaries with covalently bound methylcellulose^75,76 or by using other covalently bound capillary coatings like polyvinyl alcohol (PVA),^20,82,83 linear polyacrylamide (LPA),^84,96 dimethylpolysiloxane,⁷² acrylamide,^{21,74,85,86,97–99} PS1 or PS2,^{19,77,100,101} or other hydrophilic coatings.^24,73,81 Urea can be substituted by glycerol^{19,82–84,96,101} or formamide^{20,73,74,86,97,99} and the basic sodium hydroxide catholyte can be exchanged for ammonium hydroxide,^{19,20,82,83,101} ammonium bicarbonate,^84,96 ammonium acetate,⁸¹ or diethylamine.^{22,24,73,78,79} Several kits and systems, such as the CR3520 CIEF-MS reagent kit (CPM Scientific Corp),^77,100 CEInfinite (Advanced Electrophoresis Solutions Ltd),^{21,74–76,80,85,86,97–99} the Intabio ZT system (SCIEX)^22,78,79 or the recently launched BioSummit™ CVA CIEF-MS system (CMP Scientific Corp), are commercially available. Despite the substitution of all MS-interfering compounds, CIEF generic mAb application can be limited due to the different pIs of different mAbs in combination with the used ampholytes as shown in Table 2. Some systems apply a narrow pH range that is optimized for mAbs with their pI in that specific range. In cases where several mAbs are analyzed ampholytes with a broader pH range are generally chosen to provide a more generic method.

Table 2.

Direct, online (i)CIEF-MS methods for intact mAb analysis (excluding ADCs). mAbs 1–5 included in this table do not correlate with mAbs 1–5 discussed in the results section.

Sample(s)	Ampholytes (final concentration)	Capillary coating	Catholyte / Anolyte	Focusing / Mobilization	Interface / Sheath liquid	MS	Aim of analysis	Ref.
NIST mAb	Pharmalyte pH 3 − 10, (5%)	DB-1	5% ammonia in 0.1% methylcellulose / 200 mM phosphoric acid (pH 1.38) in 0.1% methylcellulose	30 kV (15 min) / 0.5 psi (30 kV)	T-Junction Sheath-Flow Interface connected to nonorthogonal OptiFlow MS interface OR Open Port Interface connected to an OptiFlow Turbo V ion source / 50% methanol + 1% acetic acid pH 2.88 OR 50% acetonitrile + 2% formic acid pH 2.18	TOF	Introduction of a new interface for CZE-MS coupling (OPI) and compare its performance to that of a standard CE-MS interface (SFI) for capillary isoelectric focusing.	Beloborodov et al.72
Infliximab Trastuzumab Bevacizumab Cetuximab	Pharmalyte pH 3 − 10 (1.5%)	PS1	0.2 N ammonia with 15% glycerol / 1% formic acid with 15% glycerol	250 V/cm / 250 V/cm +10 mbar	EMASS-II ion source / 20% acetic acid with 25% acetonitrile	TOF	Development of CIEF-MS (TOF-MS detection and characterization) method for mAb charge variants correlating with iCIEF-UV profiles	Dai et al.19
mAb1, 2, 3, 4	Pharmalyte pH 3–10 (4.0%)	Not specified (hydrophilic coated)	1% diethylamine (pH 13) / 1% formic acid (pH 2.3)	1500 V (1 min) + 3000 V (1 min) + 4500 V (4.5 min) / 3000 V offset	Microfluidic chip-based iCIEF-MS / 25% acetic acid with 25% acetonitrile (pH 1.8)	Orbitrap	Charge variant analysis of mAbs using an iCIEF-MS setup	He et al.73
NISTmAb USP-mAb002 Bispecific mAb (BsAb)	NISTmAb: UR AESlyte™ 8.5–10 (2%) USP-mAb002: UH AESlyte™ 7–9 (2%) BsAb: HR AESlyte™ 8–10.5 (2%)	acrylamide derivate		1000 V (1 min) + 2000 V (1 min) + 3000 V (10 min) / 3000 V + mobilization solution (50 nL/min, 0.1% formic acid)	Ion Max ESI Ion Source with a 34-gauge needle OR Easy-Spray Ion Source / 50% acetonitrile and 0.5% (v/v) formic acid	Orbitrap	iCIEF-MS system with a nano-ESI interface to reduce flow rates, improve the ionization efficiency, lower analyte dispersion, and reduce the sample dilution.	Kwok et al.74
NISTmAb Bevacizumab Bispecific mAb (BsAb)	NISTmAb: Pharmalyte 3–10 (1%) + Pharmalyte 8–10.5 (3%) Bevacizumab: Pharmalyte 3–10 (4%) BsAb: HR AESlyte 9–12 (4%)	bi-layer polymerization, MC-coated capillary		1000 V (1 min) + 2000 V (1 min) + 3000 V (10 min) / 3000 V (15 min) + mobilization solution (50 OR 100 nL/min, 0.1% formic acid)	Ion Max ESI Ion Source with a 34-gauge needle / 50% acetonitrile and 1.0% (v/v) formic acid	Orbitrap	Development of an iCIEF method using an MC-coated capillary cartridge, to avoid the need for MC in the analysis.	Kwok et al.75
mAb-M-AT	Pharmalyte 3–10 (3.5%) + Pharmalyte 8–10.5 (0.5%)	MC-coated capillary and micro-tee integrated AD-coated cartridges		1500 V (1 min) + 3000 V (7 min) / 3000 V (15 min) + mobilization solution (50 nL/min, 0.1% formic acid)	Ion Max ESI Ion Source with a 34-gauge needle / 50% acetonitrile and 0.1% (v/v) formic acid	Orbitrap	Using mAb-M-AT for charge variant characterization on an integrated iCIEF platform	Kwok et al.76
Bevacizumab	Pharmalyte 3–10 + Pharmalyte 8–10.5 (CR3520 cIEF-MS)	PS1	Buffer B / Buffer A	18 kV / 7 mbar (18 kV)	EMASS-II ion source / Not specified (acidic SL)	TOF	Identification of a bevacizumab clipping variant through several CE-MS technologies	Li et al.77
Trastuzumab biosimilar	Pharmalyte 5–8 (1%) + Pharmalyte 8–10.5 (3%)	Not specified (hydrophilic coated)	1% diethylamine / 1% formic acid	1500 V (1 min) + 3000 V (2 min) + 4500 V (4 min) / 2300 V offset	NanoElectrospray capillary cap set / 1% formic acid, 50% acetonitrile, and 49% water.	TOF	Development and application of a microchip-based iCIEF-MS system for CQA analysis	Mack et al.24
NISTmAb	Pharmalyte 3–10 (1%) Pharmalyte 8–10.5 (2.5%)	Not specified (hydrophilic coated)	0.25% diethylamine / 1% formic acid	1500 V (1 min) + 3000 V (1 min) + 4500 V (5 min) / 5500 V (10 min)	OptiFlow interface / 25% acetic acid, 25% acetonitrile, and 50% water.	TOF	The chip-based platform shows high resolution, high throughput, and high reproducibility regarding mAb charge variants	Madda et al.78
Deglycosylated NIST mAb	Pharmalyte 3–10 (1%) Pharmalyte 8–10.5 (2.5%)	Not specified	0.25% diethylamine / 1% formic acid	1500 V (1 min) + 3000 V (1 min) + 4500 V (5 min) / 5500 V (10 min)	OptiFlow interface / 25% acetic acid, 25% acetonitrile, and 50% water.	TOF	The Intabio ZT system enables a workflow to separate charge variants by iCIEF	Madda et al.79
mAb A	Pharmalyte 8 − 10.5 (1.25%) + Pharmalyte 3–10 (0.62%) + Pharmalyte 5 − 8 (0.3%)	PVA	0.2% ammonia in isopropanol/water (50/50) / 1% formic acid	290 V/cm (25 min / 500 V/cm (25 min	nanoCEasy interface / 0.20% formic acid in Isopropanol/water (50/50)	TOF	Chemical mobilization CIEF−MS method using the nanoCEasy interface and its application for different sample sets	Naghdi et al.20
Bispecific mAb	Pharmalyte 3–10 (1%) Pharmalyte 8–10.5 (2.5%)	Not specified	1% v/v diethylamine (pH 12.5) in water / 1% v/v formic acid (pH 2.2)	1500 V (1 min) + 3000 V (1 min) + 4500 V (4.5 min) / 5500 V (6.5 min)	Not specified / 25% acetic acid and 25% acetonitrile in water.	TOF	Demonstration that the microfluidic chip-based integrated iCIEF-MS technology is a new approach to CQA and charge variant analysis	Ostrowski et al.22
Trastuzumab	2%–4% (v/v) ampholytes (Pharmalyte 3–10 and Pharmalyte 8–10.5, 1:3).	acrylamide derivative OR fluorocarbon	100 mM sodium hydroxide / 80 mM phosphoric acid	1500 V (1 min) + 3000 V (10 min) / 3000 V + mobilization solution (30 or 50 nL/min)	nanoCEasy / 50% isopropanol + 0.5% formic acid	TOF/Orbitrap	Online coupling of an iCIEF to mass spectrometry via the nanoCEasy interface for online identification of charge variants	Schlecht et al.80
SigmaMAb NIST mAb	Pharmalyte 3–10 (0.25%)	LCP OR LPA	50 mM ammonium acetate (pH 9.0) / 25 mM ammonium acetate (pH 6.8)	30 kV (5–20 min) / 50 mbar (30 kV)	EMASS-II ion source / 10 mM ammonium acetate (pH 6.8)	TOF	Novel native capillary isoelectric focusing (cIEF)-assisted CZE-MS	Shen et al.81
Infliximab	Carrier ampholytes 3–10	PVA	1.0% (w/v) ammonia in 39% water and 60% methanol / 1.0% (v/v) acetic acid	30 kV (30 min) / 34 mbar (30 kV)	Flow-through microvial interface / 2.0% formic acid in 80% methanol	Orbitrap	The progress of cIEF−MS analysis by the flow-through microvial interface is shown.	Wang et al.82
mAb Infliximab Adalimumab NIST	Fluka carrier ampholyte pH 3–10 (1.0%)	PVA	1.0% w/v ammonia in water (pH 11.51) containing 20% glycerol / 1.0% v/v acetic acid (0.175 M, pH 2.75) containing 20% v/v glycerol	30 kV (25 min) / 0.5 psi (25.5 kV)	Flow-through microvial interface / 1.0% formic acid and 80% methanol	TOF	Showing feasibility of online cIEF – MS method for charge and structural heterogeneity of mAbs using fully automated analyses and a sandwich on-capillary configuration	Wang et al.83
Infliximab Pembrolizumab Adalimumab Bevacizumab Daratumumab Atezolizumab Denosumab Guselkumab Rituximab	Compare Table 1 in Paper	acrylamide derivative	Not specified	1000 V (1 min) + 2000 V (1 min) + 3000 V (10 min) / 80 nL/min with water containing 0.1% formic acid, 15 min (3000 V)	Ion Max ESI Ion Source with a 34-gauge needle / water/acetonitrile 1:1, containing 1% formic acid	Orbitrap	Comparison of CIEF and SCX regarding protein charge variant separation resolution, sensitivity, carryover effects, molecular weight accuracy, and throughput	Wu et al.21
NIST mAb Trastuzumab Cetuximab mAb1	2.0% ampholyte mixtures (pH range of 3–10, 5–8, and 8–10.5, ratio 1: 2: 4)	LPA	10 mM ammonium bicarbonate, 15% glycerol, pH 10 / 0.1% formic acid and 15% glycerol, pH 2.7	20 kV (20 min) / 10 mbar (20 kV)	EMASS-II ion source / 25% acetonitrile and 20% acetic acid	TOF	Improve automated cIEF-MS method for characterization of intact charge variants of mAbs with high resolution and reproducibility.	Xu et al.84
Pembrolizumab	2% HR AESlyte 6–8	acrylamide derivative	Not specified	1500 V (1 min) + 3000 V (10 min) / 40 nL/min with water containing 10 mM acetic acid (3000 V)	Ion Max ESI Ion Source with a 34-gauge needle / water/acetonitrile 1:1, containing 0.1% formic acid	Orbitrap	Development of an iCIEF-MS strategy for rapid iCIEF separation and reliable HRMS identification of protein charge variants	Zhang et al.85
NIST mAb Bevacizumab Pembrolizumab	NIST mAb: 0.5% HR AESlytes 3–10 + 1.5% HR AESlytes 8–10.5% Bevacizumab: 2% HR AESlyte 3–10 Pembrolizumab: 2% HR AESlyte 6–8	acrylamide derivative	Not specified	1000 V (1 min) + 2000 V (1 min) + 3000 V (10 min) / 50 nL/min with water containing 0.1% formic acid, 15 min (3000 V) ➔ Slightly adapted depending on mAb	Ion Max ESI Ion Source with a 34-gauge needle / water: acetonitrile = 1: 1, v/v, containing 0.1% OR 0.5% formic acid	Orbitrap	Development of an iCIEF-MS strategy for rapid iCIEF separation and reliable HRMS identification of protein charge variants	Zhang et al.86

Abbreviations: CQA: Critical quality attribute, CZE: Capillary zone electrophoresis, ESI: Electrospray ionization, HRMS: High-resolution MS, LCP: Linear carbohydrate polymer, LPA: Linear polyacrylamide, MC: Methyl cellulose, PVA: Polyvinyl alcohol, TOF: Time of flight, SCX: Strong cation exchange chromatography.

In our CIEF-MS method, we reduced the amount of formamide and ampholyte in the master mix used by Naghdi et al.²⁰ to 15% and 1%, respectively, which reduces the amount of interfering ions and MS contamination. However, the CIEF-MS system is highly mAb-dependent. While a very nice separation and identification of mAbA was published by Naghdi et al.,²⁰ our experiments showed significant differences depending on the mAb samples. Infliximab, USPmAb003, and mAb3 could be detected via the online UV detector, but were never detected by the MS. This indicates a solubility problem at the emitter tip when nearly neutral mAbs are transferred to the acidic and isopropanol-containing sheath liquid. The reduced separation performance for CIEF-MS in comparison to the online UV signal is described in many publications and was also noted in our experiments even though chemical mobilization was used. This will be subject to further investigations in the future since CIEF-MS is a promising tool for mAb variant identification.

The strength of (i) CIEF-MS charge variant analysis is evident by several publications where the general glycosylation pattern,^{19,21,22,24,73,74,79,80,82–84,96,99} the presence of sialic acids,^73,84 additional glycation,^{21,73,74,79,84,86,99} incompletely clipped lysine variants,^{19–22,24,73,74,77,79,82–84,86,96,99} N-terminal pyroGlu formation,^19,22,84,96 succinimide formation,^24,80 oxidation,²¹ or deamidation^{19–22,24,74,77,79,80,82–84,86,99} is analyzed. As for CEX-MS, deamidations are often supplemented with a peptide mapping approach to support the results gained by intact CIEF-MS. Nevertheless, this variety of variants for multiple mAbs shows the high potential of CIEF-MS coupling.

CZE-UV and CZE-MS for mAb variant analysis

The most commonly applied CZE-UV method, which consists of a fused silica capillary and a BGE containing EACA, TETA, and HPC, was developed by He et al..²⁵ As demonstrated by our results, this generic method shows very good separation of acidic and basic charge variants for all 10 mAb samples in this study. This method is also commonly used for UV-based charge variant separation.^{26,28,29,89,102–106} A recent interlaboratory study was conducted because users experienced issues regarding reproducibility.²⁹ The EACA lot did not cause inconsistent results, which were instead attributed to unintended changes in the method.²⁹ Alternative dynamic coating agents, additives, and polymeric compounds are tested for similar or better charge variant separation.^{27,89,107–112} BGEs contain, for example, Bis-Tris,^27,107 MES,²⁷ 3-morpholinopropane-1-sulfonic acid,²⁷ or ammonium or sodium acetate^{108–110,112} in addition to EACA or as a substitute. HPC and TETA are substituted or supplemented by polyethylene oxide,^27,107,112 polysorbate 20,¹¹¹ Tween 20,⁸⁹ or spermine.²⁷ However, these methods cannot be coupled directly to the MS. 2D CZE systems are applied to remove the MS incompatible substances,^113–115 but the analysis time is prolonged and coupling two CZE systems is so far not suitable for routine analysis. This shows the need for CZE systems that provide proper mAb variant separation and can be directly coupled to the MS.

To couple CZE to MS, all interfering components like coating materials or polymeric compounds must be removed from the BGE, as shown in Table 3. Capillaries are covalently coated with either neutral-coating agents like linear polyacrylamide,^{81,116,117,127} HP(M)C,^31,139 PVA,^2,134 or other neutral agents,^{32,33,59,77,118–120,125,126,128,129,131,133,135,136} or they are cationic coated with M7C4I,¹¹⁶ polyethyleneimine,^121–124 or other cationic polymers in a multilayer approach.^30,130 Antibody variants can be separated in an acidic BGE or a less acidic BGE typically based on ammonium acetate. The latter is here called “native,” though it is not clear whether the acidic BGE is really denaturing and the better separation observed in ammonium acetate-based BGEs is due to native conditions or the fact that the mAb and the variants are less positively charged, resulting in a better separation as shown by Shi et al..¹¹² The system developed by Höchsmann et al. allows MS detection in a mass range of 2500–4000 m/z and a detailed analysis as shown by the evaluation of USPmAb003 variants using 10 kV separation voltage.³⁰ The method was applied to the 10 mAb samples and is universally applicable if the heterogeneity of the mAb is not too complex (cetuximab or pH-stressed mAb1). However, the variant separation we achieved can most likely be increased if the separation voltage is decreased from −20 kV to −10 kV or even smaller values as shown by Höchsmann et al..³⁰ Applying a voltage of −20 kV for the separation allowed the analysis to be done in 45 min and the separation of proteoforms like the mAb glycosylation including sialic acid proteoforms, monoglycosylation, aglycosylation, lysine variants, water or ammonia losses, reduced disulfides, and the presumed deamidation of pH stressed mAb1. Such a good separation has not been achieved by any acidic separation system so far. Redman et al. showed good separation of lysine variants of infliximab using BGE of 10% isopropanol/0.2% HAc (pH 3.17).¹³¹ Giorgetti et al. also show some separation of variants using a 3% HAc BGE, but the deconvoluted masses for acidic, main, and basic proteoforms were similar and the author thus concluded that no specific proteoforms can be assigned.^121,122 In other cases, denatured systems only provide one signal without any separation of charge variants^2,127 or only a slight separation of the glycosylation species.¹¹⁶ This makes the separation system applied here³⁰ very valuable in the field of denatured mAb variant analysis. A better separation is only possible at higher pH (native separation conditions, as listed also in Table 3). These CZE-MS systems provide very good charge variant separation and identification as recently shown by van der Zon et al.³¹ and multiple chip-based methods.^{32,33,59,118,119,133,135} However, native systems require an MS instrument that covers a high mass range (4500–7500 m/z depending on the mAb) and provides proper declustering.

Table 3.

Direct, online CZE-MS methods for intact mAb analysis (including MCE-MS methods but not CE-MALDI or the analysis of ADCs). mAbs 1–5 included in this table do not correlate with mAbs 1–5 discussed in the results section.

Sample(s)	Capillary coating	Background electrolyte	Sheath liquid	Interface	MS	Aim of analysis	Ref.
mAb	M7C4I	10% isopropanol, 0.2% formic acid	–	CESI (sheathless)	Orbitrap	Analysis of mAb on intact, native, and subunit levels for detailed variant analysis	Belov et al.116
mAb	Polyacrylamide	40 mM ammonium acetate, pH 7.5		CESI (sheathless)	Orbitrap	Analysis of mAb on intact, native, and subunit levels for detailed variant analysis	Belov et al.116
Trastuzumab	Polyacrylamide	20 mM ammonium acetate, pH 8.0	–	CESI (sheathless)	Orbitrap	Demonstration that native CZE-MS can serve as an orthogonal and complementary technique to conventional native MS methodologies	Belov et al.117
mAb and biosimilars	ZipChip HRN chip	Native Antibody BGE (908 Devices) + 4% DMSO	–	ZipChip Ti interface	Orbitrap	Investigation if MCE-MS is suitable as a tool for rapidly assessing monoclonal cell line candidates	Cageling et al.33
IgG4 mAb IgG1 mAb	ZipChip HRN chip	Native Antibody BGE (908 Devices)	–	ZipChip interface	Orbitrap	mAbs were analyzed by peptide mapping and native intact methods using a single microfluidic CZE-MS system	Cao et al.118
Rituximab, Bevacizumab Trastuzumab	ZipChip HRN chip	Native Antibody BGE (908 Devices) + 4% DMSO	–	ZipChip interface	Orbitrap	Confident identification of mAb variants	Carillo et al.119
NIST mAb	ZipChip HR chip	0.2% acetic acid and 10% 2-propanol (pH 3.17)	–	ZipChipTM CE interface	Orbitrap	Intact mAb characterization, and quantitative results on protein variants	Chen et al.120
Panitumumab	ZipChip HRN chip	Native Antibody BGE (908 Devices) + 4% DMSO	–	ZipChip interface	Orbitrap	The heterogeneity of panitumumab was investigated using SEC-MS and microchip CZE-MS.	Füssl et al.32
Cetuximab	ZipChip HRN chip	Native Antibody BGE (908 Devices)	–	ZipChip interface	Orbitrap	In-depth analysis in terms of charge variant and glycoform heterogeneity using two orthogonal approaches (CEX and CZE)	Füssl et al.59
Adalimumab Natalizumab Nivolumab Palivizumab Infliximab Rituximab Trastuzumab	PEI	3% acetic acid	–	Nano-ESI source (sheathless)	TimsTOF Pro (tims was switched off)	Development of a multi-level combined approach for complete structural characterization of mAbs based on a CE-ESI-MS coupling.	Giorgetti et al.121
Trastuzumab Rituximab Palivizumab	PEI	3% acetic acid	–	sheathless interface	TOF	Development of a CZE-ESI-MS method for the analysis of three well-known approved mAbs at the intact level	Giorgetti et al.122
bispecific mAbs	PEI	3% acetic acid	–	CESI 8000 (sheathless)	TOF	Analysis of the intact BsAbs for monitoring incomplete assemblies and free chains.	Gstöttner et al.123
bispecific mAbs	PEI	10% acetic acid (pH 2.1)	–	CESI 8000 (sheathless)	TOF	Potential evaluation of sheathless CE-MS to monitor exchange efficiency and stability of BsAbs.	Gstöttner et al.124
mAb-A	Neutrally coated OptiMS capillary	50 mM ammonium acetate pH 6.8 containing the FcgRIIIa V158 or F158 at a concentration of 0.25 mM.	–	CESI 8000 (sheathless)	Orbitrap	ACE-MS is a powerful platform for glycoform-resolved interaction studies between IgGs and different FcgRs.	Gstöttner et al.125
forced oxidation mAbs	OptiMS neutrally coated capillaries	50 mM Ammonium acetate solution was adjusted to pH 6.0	–	CESI 8000 (sheathless)	Orbitrap	Developed an approach based on mobility-shift ACE−MS	Gstöttner et al.126
Deglycosylated IgG1	LPA	30% acetic acid	0.5% formic acid, 50% methanol	EMASS-II interface	TOF	Proof of concept that the CZE-TOF platform can be used to measure the intact mass and fragments of mAbs	Han et al.127
Trastuzumab Infliximab Ustekinumab	Neutrally coated OptiMS capillary	10% acetic acid	–	CESI 8000 (sheathless)	TOF	Exploration of the capabilities of low-flow sheathless CZE-MS using acidic BGEs for antibodies	Haselberg et al.128
USPmAb003 Infliximab Adalimumab Trastuzumab NISTmAb	SMIL	4 M acetic acid	Isopropanol: water (50:50) + 0.5% FA	nanoCEasy	Orbitrap	Develop a generic CZE-MS method for charge variant analysis and increase separation under acidic conditions using a new cationic capillary coating	Höchsmann et al.30
NIST mAb	Neutrally coated OptiMS capillary	40 mM ammonium acetate	–	Nanospray FlexTM ion source (sheathless)	Orbitrap	Establishing and discussing standard procedures for nCZE-TDMS using sheathless ionization	Jooß et al.129
Infliximab	Triple layer cationic coating based on PB-DS-PB	40 mM ammonium acetate pH 6.0	Water-isopropanol-acetic acid 50:50:0.1	Coaxial SL electrospray interface	TOF	Investigated online CZE-native MS with an SL interface to analyze stressed and unstressed mAb	Le-Minh et al.130
Bevacizumab	PS2	CZE sample buffer (P/N SB1121)	Not specified (acidic SL)	EMASS-II ion source	TOF	Identification of a bevacizumab clipping variant through several CE-MS technologies	Li et al.77
mAb1–3	PVA	2 M acetic acid	Isopropanol, water, and formic acid (50/49.9/0.1, v/v/v)	nanoCEsay / triple-tube SL interface	Orbitrap / TOF	Presentation of an optimized workflow for high throughput glycosylation analysis (RPLC-MS) Evaluation of separation techniques, RPLC and CZE, and data processing approaches	Naumann et al.2
Infliximab, mAb A and B	NHS-PEG450	10% 2-propanol/0.2% acetic acid (pH 3.17)	–	Laboratory setup	TOF	New capillary coating and interface for mAb variant separation	Redman et al.131
SigmaMAb NIST mAb	LCP OR LPA	25 mM ammonium acetate (pH 6.8)	10 mM ammonium acetate (pH 6.8)	EMASS-II ion source	TOF	Novel native capillary isoelectric focusing (cIEF)-assisted CZE-MS	Shen et al.81
Sigma mAb NIST mAb Herceptin	Interlaboratory study: different systems and approaches were used. Compare Table S8					Highlights the current state of TD/MD MS to address biotechnology medicinal products and their limitations	Srzentić132
mAb	ZipChip HRN chip	Native Antibody BGE (908 Devices) + 4.0% DMSO, pH ~ 5.5	–	ZipChip interface	Orbitrap	CE-MS was used to identify mAb charge variants. CE-MS was compared to a CEX-UV separation	Sun et al.133
Innovator and biosimilar rituximab	PVA	2% acetic acid	0.5% acetic acid in 50% methanol	Not specified	TOF	CZE was used for monitoring the charge variants of innovator and biosimilar rituximab	Elektroforéza134
Pembrolizumab, Rituximab, Cetuximab, Pembrolizumab, NISTmAb	HPMC	50 mM acetic acid adjusted to pH 5.0	0.5% formic acid in 50/50 iso- propanol/water	nanoCEasy	Orbitrap	Closing the gap between a flexible CZE method that enables mAb variant quantification at pH 5–6 via UV and offering MS coupling for identification	van der Zon et al.31
deglycosylated NIST mAb	ZipChip HRN chip	Native Antibody BGE (908 Devices) + 3.8% DMSO, pH ~ 5.5	–	ZipChip interface	Orbitrap	Evaluation of a highly sensitive, high-resolution microfluidic nCE-MS method for antibody charge variant and impurity analysis	Wu et al.135
NIST mAb, mAb1 bispecific mAb1	ZipChip HRN chip	Native Antibody BGE (908 Devices) + 3.8% DMSO, pH ~ 5.5	–	ZipChip interface	Orbitrap	Applied ZipChip nCE-MS to investigate therapeutic antibodies under various stress conditions	Wu et al.136
mAb1, 2, 3	LPA	5% acetic acid	2% formic acid and 10% methanol	EMASS-II interface	TOF	Good performance of CZE-MS techniques for separating mAbs, half-assembled mAbs as well as free LC fragments	Xu et al.137
IgG1, IgG2	PVA	50 mM acetic acid	0.5% acetic acid in 50% methanol	Orthogonal triple tube sheath liquid interface (G1607B)	TOF	Analysis of two mAbs on an intact and reduced level.	Suresh Babu and Gudihal138

Abbreviations: ACE: Affinity capillary electrophoresis, BGE: Background electrolyte, CESI: Commercial sheathless interface, DMSO: Dimethyl sulfoxide, ESI: Electrospray ionization, HPMC: Hydroxypropyl methylcellulose, HRN: High resolution native, LCP: Linear carbohydrate polymer, LPA: Linear polyacrylamide, MCE: Microfluidic capillary electrophoresis, PEI: Polyethylenimine, PVA: Polyvinyl alcohol, SEC: Size exclusion electrophoresis, SL: sheath liquid, SMIL: Successive multiple ionic-polymer layers, TOF: Time of flight.

Comparison of CEX, CIEF, and CZE on an intact and subunit level

To compare all tested methods, the variant separation and the deconvoluted mass spectra of mAb1 as obtained by CEX-MS, CIEF-MS, and CZE-MS are shown in Figure 5 (Supplementary Figure S5 – S7 show similar data for infliximab, NIST mAb, and mAb2, respectively). The acidic and basic variant assignment was highly method-dependent. In particular, the CEX-separated variants suffered from low spectral quality when introduced into the MS. This is due to the native ionization and still relatively high ammonium acetate amounts that are necessary for variant separation. As a result, our CEX-MS method is only applicable to abundant variants or variants that change the mAb mass to a greater extent. Our CIEF-MS method provides relatively good spectral quality (compare CIEF-MS in Figure 5), even though some mAbs cannot be detected using MS. With our CZE-MS method, low abundant mass and charge variants were detected (compare B2 in the CZE-MS column in Figure 5). Even though our CIEF and CEX spectral quality may seem significantly lower compared to the CZE-MS experiments, it needs to be kept in mind that different mass spectrometers, as well as different methods using the same mass spectrometer, might lead to changing declustering abilities and, thus, different sensitivity. Schwenzer et al. provide a more detailed discussion on this subject.¹⁴⁰

Figure 5.

Comparison of CEX-MS, CIEF-MS, and CZE-MS separation and identification of mAb1 variants. For CZE-MS, data were acquired using the Orbitrap fusion Lumos while CEX-MS and CIEF-MS, data were acquired using the MaXis instrument.

All separation techniques used for intact mAb analysis selectively separate acidic and basic mAb variants. The lysine variant of three mAbs was used to compare the resolution of the different separation techniques. The resolution calculated between the main variant and the lysine variant of infliximab, USP mAb003, and NIST mAb can be found in supplementary Table S1. The MS incompatible CZE-EACA-UV method provided the highest resolution for all mAbs. Upon the MS-coupled methods, CEX separated the main and the lysine variant of infliximab best, followed by CIEF and CZE. However, for USP mAb003 and NIST mAb, CZE-MS showed the best resolution, possibly influenced by the mass spectra quality of these low-intensity lysine variants. The different selectivity is also evident for monoglycosylation, which is found in the main variant peak using CEX and CIEF and is separated from the main variant peak using the CZE. This CZE size-based selectivity enabled a more detailed analysis of the mAb monoglycosylation and the detection of low-abundant monoglycosylated and aglycosylated species. Additionally, the different selectivity is evident for highly heterogeneous mAbs like cetuximab, where variants were separated by CIEF-UV and CEX-UV but not by CZE-MS (compare cetuximab in Figures 2, 3 , and 4). As shown in Table 4, many variants (marked green) can be unequivocally identified due to these selective separation approaches.

Table 4.

Overview of charge variants analyzed with the different methods for different antibodies. Compared to our approach, for literature assignment other mAbs and stressed mAbs were taken into consideration as well as native CZE-MS approaches. Exemplary literature is given as a reference. Green: separated and identified by MS, orange: only MS detected w/o separation, red: not detected/no information available.

*typically further analysis is needed for unequivocal determination; # only separated and not MS detected.

In the literature, a broad mAb variety was analyzed with different methods. Almost every variant was selectively separated and identified for at least one mAb. This study only analyzed a selection of 10 mAbs, but still many variants were selectively separated and identified. Selective separation is the first part of successful mAb variant identification. Sensitive MS detection is a crucial part of the analysis of mAb variants as well. The orange-marked proteoforms in Table 4 were not separated by the respective separation technique and were only identified by sensitive MS detection. Further confirmation by applying a middle-down or peptide mapping method was sometimes needed (*marked proteoforms in Table 4).

All the methods described in this work have an analysis time of around 60 min to provide selective mAb variant separation. In the literature, a similar duration of the methods is described for CEX-MS,⁶ CZE-MS,³¹ and CIEF-MS¹⁹ to achieve the best possible separation. The method duration can be shortened to some minutes for all techniques,^24,54,121 which proves that measuring time is not the most relevant criterion for the choice of the separation technique of mAb variants on the intact level.

After the mAb variants were analyzed on the intact level, the results can be compared to data obtained on the subunit level. The intact separation and identification of mAb variants is suitable to determine the overall mAb composition. This information is partly lost using the subunit approach if the variants are not on the same subunit moiety. The subunit approach, however, enables a more detailed analysis of the location of the variant within the mAb ¹⁴¹. Detailed information about the mAb variant is obtained because small mass changes can be better detected on the subunit level compared to the intact level due to a narrower isotopic pattern and possible isotopic resolution using modern time of flight (TOF) and Orbitrap MS instrumentation. Additionally, the modification site can be narrowed down to the subunit moiety and even localized by MS/MS experiments.

With our CZE-MS subunit method (IdeS digested and reduced subunits) sialic acids, glycoforms, glycation, lysine variants, multiple oxidations, water and ammonia losses, and incomplete sample reduction, as well as enzymatic misscleavage, were determined.^34,35 Mono- and aglycosylation of the mAb could be determined based on the detection of aglycosylated Fc/2 fragments. MS/MS experiments were of utmost importance for the determination of the deamidation of pH-stressed mAb1 and the determination of sample preparation-induced modifications. The information gained by fragmentation was only limited when positional isomers were not properly separated and several possible modification sites were available within the subunit. Charge variant analysis of digested and reduced subunits (Fd, light chain, Fc/2) may not be performed selectively with just any separation technique due to, for example, a large change in pI of one or two of the subunits compared to the intact mAb. Subunit analysis on the CEX is mostly done with IdeS enzymatically digested mAbs,^{7,55,60,71,142} which leaves a large 100 kDa F(ab’)₂ fragment. Digested and reduced subunits are not analyzed since the chaotropic salt mandatory in the reduction process interferes with the column and removal of the chaotropic salt might cause disulfide scrambling if the disulfides are not alkylated.

CIEF was used for subunit analysis of cetuximab,¹⁰¹ but apart from that, CIEF is commonly used for the intact analysis of mAbs. This may be attributed to the fact that different subunit moieties exhibit different pI values. To analyze a wide pI range, a wide pH range is needed during focusing, which potentially reduces the separation performance of charge variants and hamper the generic application of CIEF for different mAbs. CZE is the only method that can be performed independent of differences in pI of the subunits (by use of a sufficiently acidic BGE) or the chaotropic salt needed for reduction, though different resolution of charge variants is still obtained.

Finally, besides all the previously discussed parameters, instrumental parameters and sample-specific requirements have an enormous impact on the selective and sensitive separation and identification of mAb variants. This is especially evident in our CIEF-UV-MS method and CEX-UV-MS method. For both methods, the online UV detection hints at a selective separation of mAb variants, but MS detection and identification of these variants could not be fully achieved because of the previously discussed solubility problem in CIEF and the native ionization in CEX. Our CEX method most likely suffers from the application of a standard LC sprayer in combination with the high flow rate toward the MS. The literature applications shown in Table 1 either attempt to minimize the amount of eluent by using a post-column flow splitter^55,143 or nanoflow-CEX⁷⁰ to reduce the amount of salt in the eluents,^21,42,48 or to enhance the ionization efficiency by using special nanoflow and microspray ionization sources.^{21,48,53,55,65} Our CIEF most likely suffers from handling and robustness issues in combination with mAb-specific requirements because the CIEF-nanoCEasy-MS was previously successfully applied for the analysis of mAb A variants.²⁰ For the literature applications shown in Table 2, either a small cartridge^21,74,80,86 or chip^24,73 was used or standard CIEF was coupled to the MS by a nanospray source.^19,20,72,83 All these approaches provide quite high mass spectra quality, even though separation is always impaired in the mobilization step.^19,23 Similar to CIEF, CZE may also benefit from microchip applications,^119,120,133 which provide a fast and efficient separation of mAb variants.^116,121,122 When a standard CZE system is coupled to the MS, the interface is a critical part as shown by Höcker et al..¹⁴⁴ Sample dilution due to excessive SL consumption can significantly reduce the signal intensity.

All aspects mentioned here prove that the method and instrumental setup for mAb variant analysis must be chosen carefully. With the increase in heterogeneity due to new mAb products like antibody-drug-conjugates (ADCs) and bispecific antibodies, the requirements for selective and sensitive variant characterization increase. Compared to mAbs, ADCs are composed of antibodies, linkers and payloads, which substantially increase heterogeneity.^98,145 Depending on the type and linkage of the conjugate, the methods presented here can be applied to separate ADCs, as shown for CEX-MS and icIEF-MS.⁹⁸

In conclusion, no matter if executed on the intact or subunit level, all tested methods have their advantages and disadvantages, but, nevertheless, all of them are valuable for mAb variant characterization. Some methods are better platform methods (CZE and CEX to a certain extent) while others (CIEF) need further method development especially when considered generic methods. A combination of several methods should always be considered when analyzing mAb variants to get the overall mAb composition.

Materials and methods

Materials

Infliximab (10 mg/mL, charge: 8KMKA90603) and cetuximab (5 mg/mL, charge: 248033), were purchased from Evidentic GmbH (Berlin, Germany). NIST mAb (10 mg/mL, Reference Material 8671, lot: 14HB-D-002) was purchased from the National Institute of Standards and Technology (NIST, Gaithersburg, MD, USA). Protein A purified research antibodies mAb1 (20 mg/mL), mAb2 (16.8 mg/mL), mAb3 (18.3 mg/mL) mAb4 (21.7 mg/mL), and mAb5 (18.5 mg/mL) were kindly provided by Rentschler Biopharma SE (Laupheim, Germany). USP mAb003 (Cat. No. 1445595, LOT: F12980, 10 mg/mL), iminodiacetic acid (IDA), capillary coating material poly(sodium styrene sulfonate) (PSS, polyanion, Mw 70,000 g/mol), 4-(2 hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), hydroxypropyl cellulose (HPC, Mw: 100.000), triethylenetetramine (TETA, ≥ 97%), sodium hydroxide, 3-(N-morpholino) propane sulfonic acid (MOPS)-buffer, 1,4-dithiothreitol (DTT), and L-arginine were purchased from Sigma-Aldrich (Steinheim, Germany). Hydrofluoric acid (40%(v/v), HF) and hydrochloric acid (1 M, HCl) were purchased from Merck (Darmstadt, Germany). Isopropanol (IPA, LC-MS grade), ammonia solution (ROTIPURAN p.a., 30%), ammonium acetate (LC-MS grade, ≥98%), ACN (ROTISOLV ≥ 99.95%), acetic acid (HAc, ROTIPURAN 100%), and formic acid (FA, ≥98%) were purchased from Carl Roth GmbH & Co. KG (Karlsruhe, Germany). Urea (Ultrapure) was purchased from ThermoFisher Scientific (Dreieich, Germany). Formamide, polyethylene oxide (PEO, Mw:1.000.000), and 6-aminocaproic acid (EACA, ≥99%) were purchased from Alfa Aesar (Kandel, Germany). Pharmalyte pH 3–10, Pharmalyte pH 8–10.5, and Pharmalyte pH 5–8 were obtained from GE Healthcare Bio-Sciences AB (Waukesha, WI, USA). Capillary coating material diethyl aminoethyl-dextran (DEAED, Mw = 500,000 g/mol, Batch 21,046) was from TdBLabs (Uppsala, Sweden). IdeS protease was purchased from GENOVIS (FabRICATOR, 5000 units, Lund, Sweden).

If not otherwise stated samples and chemicals were diluted and dissolved with ultrapure water (UPW) provided by an Ultra Clear TWF UV water system from Siemens (Munich, Germany).

Sample preparation

mAb1 was pH stressed by pipetting 0.5 µL concentrated ammonia solution to 10 µL undiluted mAb1. After 7 days in the dark at room temperature, the sample was directly used for further sample preparation. For intact (native) analysis with CEX, antibodies were diluted with UPW to a final concentration of 2 mg/mL and stored at −20°C until measurement. For CIEF, mAbs were first diluted to a concentration of 1 mg/mL with UPW. Then, each mAb was mixed 1:1 with CIEF master mix, resulting in a final concentration of 0.5 mg/mL mAb in 15% formamide, 1% ampholyte-mix (1:4:2 of Pharmalyte pH 5–8, Pharmalyte pH 8–10.5 and Pharmalyte pH 3–10), 2.5 mM arginine and 0.5 mM IDA. For intact CZE-UV analysis, mAbs were diluted to 1 mg/mL with UPW and for CZE-MS mAbs were diluted to 1 mg/mL in 4 M HAc (also serves as BGE). For subunit analysis, mAbs were first digested using the FabRICATOR digestion protocol of GENOVIS. Two hundred microgram mAb were digested using 200 U of IdeS in 100 mM MOPS buffer at pH 7.2 for 30 min at 37°C and 500 rpm. Following this, mAbs were further reduced in 4 M urea using DTT for 60 min at 37°C and 500 rpm. Detailed sample preparation parameters can be found elsewhere.³⁴

Separation approaches

All CIEF and CZE experiments were conducted on an Agilent 7100 CE instrument (Agilent Technologies GmbH, Waldbronn, Germany). If not otherwise stated fused silica capillaries (separation capillary: 50 μm ID, 365 μm; sheath liquid capillary: OD 100 μm ID, 240 μm OD) were purchased from Polymicro Technologies (Phoenix, AZ, USA). CIEF-MS and CZE-MS using the nanoCEasy interface require the etching of the separation capillary at one end as described in detail elsewhere.³⁴ Briefly, for capillary etching, the polyimide was removed from the capillary end and then dipped into concentrated HF for 60 min. Detailed information on the nanoCEasy interface can be found in the work of Schlecht et al.¹⁴⁶

Ion exchange chromatography of intact mAbs

All samples were analyzed on an UltiMate™3000 HPLC system from Dionex (Germering, Germany), equipped with a variable wavelength detector VWD-3400 and directly coupled to the MS. The column used for all separations was a Proteomix® SCX-NP3 (2.1 mm × 100 mm) from Sepax Technologies (Delaware, USA). A 100 µL/min flow rate was applied, and the column temperature was held at 40°C. The injection volume was 12.5 µL, resulting in 25 µg of antibody on the column. The mAb variants were separated with a combined salt and pH gradient. Mobile phase A consists of 50 mM ammonium acetate + 2% ACN, at pH 5.0 (adjusted with acetic acid), and mobile phase B consists of 100 mM ammonium acetate + 2% ACN, at pH 8.5 (adjusted with ammonia solution). Subsequently, the eluents were degassed using an ultrasonic bath. The gradient started at 55% [B] for 2 min and was raised over 25 min to 85% [B]. The amount of [B] was quickly increased to 100% [B] over 0.1 min and held for 2 min. Afterward, the column was brought back to starting conditions over 0.1 min and re-equilibrated for 11 min. The total run time was 40 min. Online UV detection was performed at 280 nm.

Capillary isoelectric focusing of intact mAbs

PVA-coated separation capillaries (50 μm ID, 365 μm OD, 60 cm total length, 49 cm effective length) were kindly provided by Agilent Technologies (Waldbronn, Germany). Before each analysis, the PVA capillary was flushed (950 mbar) with UPW, 4 M urea, and UPW for 5 min each. The sample was injected for 120 seconds at 950 mbar to guarantee complete capillary filling. The ends of the capillary were then dipped in anolyte (1% FA, CZE inlet side) and catholyte (0.2% ammonium hydroxide; CZE outlet side), and a focusing voltage of 20 kV was applied for 20 min. After focusing, the voltage was turned off to switch to 0.5% FA for mobilization. Mobilization was done at 20 kV until the mAb was detected with the internal UV detector of the CZE instrument at 280 nm.

Capillary zone electrophoresis UV-system of intact mAbs

The separation system for UV detection was adapted from He et al..²⁵ A 60 cm (effective length: 52 cm) fused silica capillary (50 µm ID/365 µm OD) was used for the separation. A mixture of 400 mM EACA, 0.05% m/v HPC, 2 mM TETA, at pH 5.7 was used as BGE, and samples were diluted to 1 mg/mL in UPW. The capillary was flushed (950 mbar) previous to the first separation with 0.1 M HCl for 5 min and BGE for 10 min, followed by the application of 30 kV for 10 min. Samples were injected at 50 mbar for 8 seconds. The separation voltage was 30 kV and separated variants were detected at 214 nm with the internal UV detector of the CZE instrument. Between runs, the capillary was flushed (950 mbar) with 1 M HCl for 1 min followed by BGE for 3 min.

Capillary zone electrophoresis MS-system of intact mAbs

The etched fused silica separation capillary had a length of 60 cm and was coated with SMIL according to Dhellemmes et al.¹⁴⁷ and Höchsmann et al..³⁰ Briefly, the capillary is prepared by flushing with 1 M sodium hydroxide (10 min), UPW (5 min), and 20 mM aqueous HEPES solution (pH 7.4; 10 min). The coating is established by alternate flushing of DEAED and PSS solution (3 g/L in 20 mM HEPES solution; pH 7.4; 7 min), including a rinsing step with HEPES solution for 3 min after each polymer layer. After five layers have been applied a waiting step (5 min), a flushing step with UPW (3 min), and a final flushing step with BGE (4 M HAc;10 min) complete the coating procedure. All coating steps were performed at 2 bar external pressure. For measurement samples were hydrodynamically injected (40 mbar,5 seconds) and the separation was executed in 4 M HAc at −20 kV for 60 min. BGE was exchanged after two measurements to prevent contamination and capillaries were stored overnight in BGE.

Capillary zone electrophoresis MS-system of subunit mAbs

The detailed method can be found elsewhere.³⁴ Briefly, an etched 60 cm PEO-coated capillary was used. First, capillaries were etched with HF (40% (v/v)) for 1 hour followed by PEO coating of the capillary. For the PEO coating, the capillary was prepared with 1 M NaOH, water, and 1 M HCl for 5 min, respectively, followed by the PEO coating solution for 10 min, water, and BGE for 5 min, respectively. The solutions were flushed using an external pressure of three bar. The BGE consisted of 10% IPA in water containing 1 M FA. Sample injection was done hydrodynamically using 50 mbar pressure for 10 seconds. A 20 kV voltage was applied for the separation.

MS instrumentation

CEX-MS coupling

Native CEX was coupled to a maXis Q-TOF from Bruker Daltonics (Bremen, Germany) using an ESI source. The integrated valve at the MS was used to direct the flow to the source or the waste. The MS method consisted of three segments adjusted to the mAb elution time. In segments one and three, the valve is in the waste position and the ES voltage is turned off, while in segment two the valve is in the source position and the ES voltage is turned on. After calibrating the MS in the high mass range, measurements were done using a capillary voltage of 4500 V, an endplate offset of 500 V, a nebulizer pressure of 2.5 bar, and a dry gas flow and temperature of 8 L/min and 220°C, respectively. The acquisition was done from 1500 to 8000 m/z.

CIEF-MS coupling

The CIEF was coupled to a maXis Q-TOF from Bruker Daltonics (Bremen, Germany) using the nanoCEasy interface, which was supervised using a digital microscope (Dino-Lite, Almere, The Netherlands). Instead of an outlet CIEF vial the end of the separation capillary is introduced to a glass emitter that was purchased from BioMedical Instruments (Zöllnitz, Germany). The emitter had a 30 µm tip opening and a tip length of 4 mm and was placed 3 mm in front of the MS orifice. The integrated valve at the MS and two syringe pumps were used to switch from catholyte (0.2% ammonium hydroxide) during focusing to sheath liquid (ACN: UPW 1:1 + 0.5% FA) during mobilization. The MS method consisted of two segments. In segment one, during focusing, the ES voltage is turned off, the catholyte is flushed to the emitter and the separation capillary is positioned behind the sheath liquid capillary. In segment two, during mobilization the ES voltage is turned on, the sheath liquid (SL) is flushed to the emitter, and the separation capillary is placed close behind the emitter tip in front of the SL capillary. Here, the 20 kV CIEF voltage can be maintained during focusing and mobilization. Online UV detection was done at 280 nm using an external ECD2600 EX UV detector (ECOM spol. S r.o., Prague, Czech Republic). MS acquisition was done using a capillary voltage of 2000 V, a dry gas flow of 3 L/min, and a dry gas temperature of 200°C in the mass range from 250 to 4500 m/z.

CZE-MS coupling

The CZE was coupled to an Orbitrap Fusion Lumos (Thermo Fisher Scientific, SanJose, CA, USA) using the nanoCEasy interface and a digital microscope (Dino-Lite, Almere, The Netherlands). The same emitters as for the CIEF application were used. IPA: UPW (50:50) + 0.5% FA was utilized as SL. MS acquisition was performed using a capillary voltage of 2000 V, sweep gas of 3 Arb, and 300°C ion transfer tube temperature. The MS was operated in positive ionization mode using a mass range of 700–4000 m/z and a resolution of 30,000. Further MS parameters can be found in the publications of Höchsmann et al.³⁰ and Schairer et al.,³⁴ respectively

Data evaluation

Theoretical mAb masses and pI values were calculated sequence-based with ProtPi (https://www.protpi.ch). CEX-UV data were processed with Chromeleon™ 7 (Thermo Fisher Scientific) analysis software. CIEF-UV data were processed with ECOMAC – ECOM Acquisition and Control Version 2.9.5.0 software and CZE-UV was processed with CEval 0.6h9 (available at: https://echmet.natur.cuni.cz/). For variant analysis, the CEX-MS, CIEF-MS, and CZE-MS data were 0.2 min time-sliced and processed using Intact Mass 3.4 (Protein Metrics). Additionally, CEX-MS and CIEF-MS data were evaluated using Data Analysis 6.0 (Bruker Daltonics). CZE-MS data were additionally processed using Freestyle 1.8 (Thermo Fisher Scientific).

Abbreviations

A

Acidic variant

ACN

Acetonitrile

ADCs

Antibody-drug-conjugates

B

Basic variant

BGE

Background electrolyte

BPEs

Base peak electropherograms

CEX

Cation exchange chromatography

(i)CIEF

(imaged) Capillary isoelectric focusing

CZE

Capillary zone electrophoresis

DEAED

Diethylaminoethyl-dextran

EACA

6-aminocaproic acid

EICs

Extracted ion chromatograms

EIEs

Extracted ion electropherograms

EOF

Electroosmotic flow

ES

Electrospray

ESI

Electrospray ionization

FA

Formic acid

HAc

Acetic acid

HCl

Hydrochloric acid

HEPES

4-(2hydroxyethyl)-1-piperazineethanesulfonic acid

HF

Hydrofluoric acid

HPC

Hydroxypropyl cellulose

ID

Inner diameter

IDA

Iminodiacetic acid

IPA

Isopropanol, LPA: Linear polyacrylamide

M

Main variant

mAb

Monoclonal antibody

MES

2-(N-morpholino)ethanesulfonic acid

MS

Mass spectrometry

OD

Outer diameter

pI

Isoelectric point

PSS

Poly(sodium styrene sulfonate)

PVA

Polyvinyl alcohol

R

Resolution

SL

Sheath liquid

SMIL

Successive multiple ionic-polymer layers

TETA

Triethylenetetramine

UPW

Ultrapure water

Funding Statement

The author(s) reported that there is no funding associated with the work featured in this article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from the corresponding author, [CN], upon reasonable request.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19420862.2025.2537116