Protein A/L resins for capture of multispecific antibodies

By Bastian Franke¹, Julien Québatte¹, Sebastian Wolniak¹, Cheryl Erne¹, Ronnie Palmgren², Stefan Warmuth¹

¹Numab Therapeutics AG, Bachtobelstrasse 5, 8810 Horgen, Switzerland. ²Cytiva, Uppsala, Sweden

In this study, Numab Therapeutics AG evaluates the optimal affinity resin for the capture of two multispecific antibody-based molecules.

Binding capabilities were assessed for the highly specific, high-capacity MabSelect™ VH3 and MabSelect™ VL affinity resins.
Dynamic binding capacities were compared to MabSelect PrismA™ resin and Capto™ L resin, respectively.
The performance of the different resins in bind and elute mode was evaluated for two multispecific molecules and elution profiles, pH, and monomeric recovery.
The monomeric separation from high- and low molecular weight impurities was determined and compared at 80% of QB_10%.

Introduction

Affinity chromatography (AC) is one of the most widely used methods in monoclonal antibody (mAb) and multispecific antibody (msAb) therapeutic processing. Multispecific antibodies present challenges in purification due to their complexity with product-related impurities in addition to the common process-related impurities. Given the complexity of msAbs, there is a growing need for novel affinity ligands with affinity for specific binding sites. This requires thorough screening of resins to determine the optimal resin for a specific msAb.

Numab Therapeutics AG (Numab, numab.com) develops multispecific antibody-based immunotherapies for inflammation and cancer. Here we present how Numab has evaluated capture affinity resins for two of their antibody-based molecules.

Cytiva is continuously developing the resin portfolio to meet the needs for the diversified pipeline of antibody-based therapies. The MabSelect™ VH3 resin (see the data file ) possesses the same base properties as MabSelect PrismA™ resin . MabSelect™ VH3 diverges however in its binding specificity, exhibiting high affinity and enhanced binding for the variable region of the heavy chain (VH3), whereas MabSelect PrismA™ resin shows high affinity for both the Fc and VH3 regions. The novel MabSelect VL™ resin (see the data file ) is a redesigned protein L resin with higher binding capacity and improved alkaline stability compared to its predecessor, Capto™ L resin. Protein L resins bind to the variable region of the κ-type light chain. Capto™ L resin uses a larger resin bead compared to the other three resins.

Affinity resins were evaluated for the capture of two msAbs. The dynamic binding capacity (DBC) of MabSelect™ VH3, MabSelect™ VL, MabSelect PrismA™, and Capto™ L resins were determined at 10% breakthrough (QB_10%) and at 1.5 min and 4.0 min residence times (RT). Bind and elute runs at 80% of QB_10% were performed to obtain elution profiles, determine pH, and monomeric recovery of elution pools and to investigate the monomeric separation from the high- and low-molecular weight species impurities (HMWS and LMWS) for each molecule (Fig 1).

Fig 1. Overview of evaluated protein A and L resins. The MabSelect™ VH3 and MabSelect™ VL resins were tested. DBC and bind and elute runs at 80% of QB_10% were performed at two RT during sample application and compared to MabSelect PrismA™ and Capto™ L resins.

Studies performed with two msAbs

The DBC and bind and elute studies were performed on two msAbs. The first molecule is a variable-fragment only molecule (Fig 2A), comprising of three variable fragments connected by flexible linkers in a single amino acid chain (scMATCH3). The second molecule is an Fc-based molecule (Fig 2B), with each C-terminus of the heavy chain of the IgG backbone linked to a single-chain variable fragment (scFv), IgG-scFv₂. All variable fragments and scFvs used in the two molecules consist of a VH3 heavy chain and a λ-capped κ-type light chain family (no λ-cap is present in the light chain variable domain of the antigen binding fragment [Fab]), (1). The molecular weight of the scMATCH3 molecule is approximately M_r 75 000 (75 kDa), while the molecular weight of the IgG-scFv₂ molecule is approximately M_r 200 000 (200 kDa).

Fig 2. Overview of molecules used in this study. scMATCH3 molecule consisting of three variable fragments connected by flexible linkers (A). IgG-scFv₂ molecule consisting of an IgG backbone linking each C-terminus of the heavy chain to one scFv (B).

Materials and methods

Column overview

MabSelect™ VH3 resin was obtained from Cytiva as a prototypic resin from development and has similar performance as the final product. MabSelect PrismA™ resin was purchased from Cytiva.

MabSelect PrismA™ and MabSelect™ VH3 resins were packed in Tricorn™ 10/50 columns to a final column volume of 0.51 mL and bed height of 0.65 cm. MabSelect™ VL and Capto™ L resins were purchased from Cytiva in the 1 mL HiTrap™ column format.

Sample preparation

scMATCH3 and IgG-scFv₂ molecules were expressed in CHO cells and pre-purified on protein A (data not shown) to be used for DBC determination and bind and elute experiments. Both molecules were selected to exhibit a monomeric purity below 90% to evaluate resin efficiency in terms of HMWS and monomer separation and purity increase.

The protein concentration and pH of the eluate were adjusted to approximately 2.1 mg/mL and pH 7.0, respectively. Aliquots were frozen and stored at -80°C for further use.

Determination of DBC

The protein sample was loaded via the sample pump onto the column until > 10% breakthrough. The DBC was determined by frontal analysis at 10% breakthrough for each resin at 4.0 min and 1.5 min RT, respectively. The corresponding flow rates are outlined in Table 1.

All DBC experiments were performed on ÄKTA™ chromatographic systems operated by UNICORN™ software.

Table 1. RT and corresponding flow rates during sample application

Resins and columns	RT	Column/resin volume (mL)	Flow rate (mL/min)
MabSelect PrismA™ resin in Tricorn™ column MabSelect™ VH3 resin in Tricorn™ column	4.0 min	0.51	0.128
	1.5 min	0.51	0.341
HiTrap™ Capto™ L prepacked column HiTrap™ MabSelect™ VL prepacked column	4.0 min	0.96	0.241
	1.5 min	0.96	0.641

Bind and elute runs at 80% of QB_10%

The protein sample with a volume containing a protein amount corresponding to 80% of QB_10% was loaded via the sample pump onto the column at 4.0 min and 1.5 min RT, respectively. The elution was performed in a linear gradient from 41% of buffer B to 100% B (Table 2).

Table 2. Purification parameters

Resin formats

Buffer system

RT sample application

Wash step

Elution gradient

MabSelect PrismA™ MabSelect™ VH3

Buffer A, pH 8.5

Buffer B, pH 2.7

4.0 min, 1.5 min

41% B

41% B to 100% B in 6 CV

Capto™ L

MabSelect™ VL

Buffer A, pH 8.5

Buffer B, pH 2.0

4.0 min, 1.5 min

41% B

41% B to 100% B in 6 CV

Protein analytics

The elution peak was fractionated and fractions were analyzed at A280 and size exclusion chromatography-HPLC to determine protein concentration and post-capture monomeric purity, respectively. After combining fractions of the elution peak, the pH was measured.

Results and discussion

Dynamic binding capacity determination at 10% breakthrough (QB_10%) of protein A and L resins

For IgG-scFv₂, MabSelect™ VH3 resin demonstrates the highest binding capacity, with 35% higher QB_10% compared to MabSelect PrismA™ resin and 14% higher QB_10% compared to MabSelect™ VL resin at 4.0 min RT (Table 3). Capacities can vary between column formats due to the degree of bed compression so comparing data from HiTrap™ and Tricorn™ columns is approximate. At 1.5 min of RT, MabSelect PrismA™ and MabSelect™ VH3 resins exhibit similar DBC, and slightly lower DBC for MabSelect™ VL is recorded.

The findings suggest that the flow rates influence the resin's capacity to capture larger target proteins like IgG-scFv₂, with lower flow rates favoring higher binding capacities.

Evaluation using the scMATCH3 molecule shows that MabSelect™ VL provides a higher DBC compared to the other tested resins, 10% to 17% higher QB_10% at 1.5 min and 4.0 min compared to MabSelect PrismA™, which is the resin with the second highest QB_10% values. RT of 1.5 min and 4.0 min have only a small impact on QB_10% and less impact than observed for the IgG-scFv₂ molecule.

MabSelect™ VH3, MabSelect PrismA™, and MabSelect™ VL resins use the same coupling chemistry and the same base matrix, resulting in significantly higher QB_10% compared to Capto™ L resin, which uses a larger bead size and a different coupling chemistry.

Table 3. QB_10% of protein A and L resins at two different RT

Resin	RT (min)	IgG-scFv₂		scMATCH3		mAb
Resin	RT (min)	QB_10% (mg/mL resin)	QB_10% × 10^-7 (mol/mL resin)	QB_10% (mg/mL resin)	QB_10% × 10^-7 (mol/mL resin)	QB_10% (mg/mL resin)
MabSelect PrismA™	4.0	48	2.4	52	6.7	65^*
MabSelect PrismA™	1.5	32	1.6	48	6.2	N/A
MabSelect™ VH3	4.0	65	3.2	41	5.3	60 ^†
MabSelect™ VH3	1.5	34	1.7	41	5.3	N/A
Capto™ L	4.0	38	1.9	33	4.3	25 ^{‡, §}
Capto™ L	1.5	22	1.1	28	3.6	N/A
MabSelect™ VL	4.0	57	2.8	61	7.9	60 ^¶
MabSelect™ VL	1.5	27	1.3	53	6.9	N/A

^* See Instructions: MabSelect PrismA™ affinity chromatography, 29262586 AE V:5 07/2020.
^† See Data file: MabSelect™ VH3 resin affinity chromatography resin, CY36703.
^‡ Fab was captured here.
^§See Instructions: Capto™ L affinity chromatography (29003349 AB V:6 06/2020).
^¶ See Data file: MabSelect™ VL affinity chromatography resin, CY26149.
N/A = not available.

Bind and elute runs at 80% of QB_10% using protein A resins

Both protein A resins demonstrate strong binding for IgG-scFv₂ and scMATCH3 (Table 4).

For IgG-scFv₂, MabSelect PrismA™ resin shows greater recovery of 95% to 97% at both RT considering elution fractions with > 80% monomeric purity than the other resins. However, no efficient separation between monomer and HMWS is visible in Figure 4, with the post-capture monomeric purity being constant across all elution fractions.

The separation efficiency between the monomer and HMWS is more pronounced on the MabSelect™ VH3 resin at 4.0 min RT, which is also visible in the lower elution recovery of 80%. The monomeric target protein elutes earlier, while the HMWS elutes later in the elution gradient.

A total of 22% of the protein content exhibits a post-capture monomeric purity of 97%, while 55% shows a post-capture monomeric purity of 93%. In contrast, when using the MabSelect PrismA™ resin, all elution fractions show a post-capture monomeric purity < 90%.

Changing the linear elution gradient to a step gradient should further improve the separation between HMWS and the monomeric target protein. At 1.5 min RT, this separation efficiency effect between monomer and HMWS is less pronounced but still better than MabSelect PrismA™ resin.

Table 4. Bind and elute runs at 80% of QB_10% on protein A resins

Molecule	Resins	RT (min)*	pH of elution pool	Post-capture monomeric purity of elution pool (%)	Recovery at > 80% monomeric purity (%)	Elution pool volume (CV)
IgG-scFv₂	MabSelect PrismA™	4.0	4.3	88.8	95	4.8
	MabSelect PrismA™	1.5	4.3	88.7	97	3.9
	MabSelect™ VH3	4.0	4.6	88.9	80	6.9
	MabSelect™ VH3	1.5	4.5	89.0	93	4.7
scMATCH3	MabSelect PrismA™	4.0	4.6	84.9	78	4.7
	MabSelect PrismA™	1.5	4.6	85.0	81	4.5
	MabSelect™ VH3	4.0	4.3	84.7	76	4.6
	MabSelect™ VH3	1.5	4.4	84.8	75	4.5

^*During sample application.

Fig 3. Overlay of MabSelect PrismA™ and MabSelect™ VH3 resins wash and elution phases. IgG-scFv₂ elutes at a lower pH from MabSelect PrismA™ resin compared to MabSelect™ VH3 resin. A partial elution in the 41% wash step is observed when MabSelect™ VH3 is used. IgG-scFv₂ is loaded in both runs at 4.0 min RT (A). The scMATCH3 molecule elutes at a slightly lower pH from MabSelect™ VH3 resin at both RT compared to MabSelect PrismA™ resin (B).

Figure 3 shows the elution profiles for the two msAbs with the two evaluated protein A resins. IgG-scFv₂ elutes at a lower pH from MabSelect PrismA™ resin compared to MabSelect™ VH3 resin. When lowering pH during wash, a small protein loss of 3.2% and 1.5% occurred (compared to the area of the elution peak) in the 41% buffer B wash step of the MabSelect™ VH3 resin at 4.0 min and 1.5 min, respectively. In comparison, only 0.3% and 0.2% protein loss occurred in the wash step when MabSelect PrismA™ was used (Fig 3A).

To mitigate protein loss in the wash step, future runs should consider decreasing the percentage of buffer B in the wash step resulting in a higher wash pH. Due to its lower binding capacity, the elution pool volume for IgG-scFv₂ from MabSelect PrismA™ resin is smaller than that from MabSelect™ VH3 resin at both RT. Post-capture monomeric purities of the elution pools remain comparable across all protein A resins and RT (Table 4).

For scMATCH3, the elution profiles of MabSelect PrismA™ and MabSelect™ VH3 resins are almost identical at both RT. This is a result of the very similar QB_10% values obtained on both protein A resins (Fig 3B).

MabSelect™ VH3 resin shows a slightly lower pH in the elution pool compared to MabSelect PrismA™ resin. Post-capture monomeric purities and elution pool volumes remain comparable across all protein A resins and RT for scMATCH3 (Table 4).

The elution recovery is slightly lower for MabSelect™ VH3 resin, although no clear separation is visible in Figure 5. When lowering pH during wash a small protein loss ranging from 1.6% to 2.2% for both resins and RT is observed.

No difference in separation efficiency between the monomer and HMWS or LMWS can be observed for both resins at both RT, with the percentage of HMWS, monomer, and LMWS as well as the post-capture monomeric purity in the fractions being very similar (Fig 5). Therefore, the purification performance of MabSelect™ VH3 resin is very consistent for the small variable-fragment scMATCH3 molecule and comparable to MabSelect PrismA™ resin.

Fig 4. Separation efficiency of IgG-scFv₂ eluted from MabSelect PrismA™ and MabSelect™ VH3 resins. The percentage of HMWS, monomer, and LMWS in each elution fraction from its corresponding total content value is shown on the y-axis. Only MabSelect™ VH3 shows an increased separation efficiency between monomer and HMWS, which can be further exploited in a step gradient. Percentages above the monomer bars correspond to the post-capture monomeric purity in that elution fraction.

Fig 5. Separation efficiency of scMATCH3 eluted from MabSelect PrismA™ and MabSelect™ VH3 resins. The percentage of HMWS, monomer, and LMWS in each elution fraction from its corresponding total content value is shown on the y-axis. MabSelect™ VH3 resin shows a consistent separation efficiency to MabSelect PrismA™ resin at both RT. Percentages above the monomer bars correspond to the post-capture monomeric purity in that elution fraction.

Bind and elute runs at 80% of QB_10% using protein L resins

Both protein L resins show strong binding affinities for IgG-scFv₂ and scMATCH3 (Table 5).

The post-capture monomeric purities of the elution pools are comparable across all protein L resins and RT. Due to its higher binding capacity, the elution pool volumes from MabSelect™ VL are larger than those from Capto™ L resin (Table 5, Fig 6).

MabSelect™ VL resin shows similar pH values in the elution pool compared to Capto™ L resin for IgG-scFv₂ and slightly lower pH values for scMATCH3.

Table 5. Bind and elute runs at 80% of QB_10% on protein L resins

Molecule	Resins	RT (min)*	pH of elution pool	Post-capture monomeric purity of elution pool (%)	Recovery at > 80% monomeric purity (%)	Elution pool volume (CV)
IgG-scFv₂	Capto™ L	4	3.6	87.3	94	4.1
	Capto™ L	1.5	3.4	87.9	92	3.3
	MabSelect™ VL	4	3.5	85.7	101	5.3
	MabSelect™ VL	1.5	3.4	87.2	108	3.9
scMATCH3	Capto™ L	4	3.0	85.8	96	5.4
	Capto™ L	1.5	2.9	85.6	99	4.8
	MabSelect™ VL	4	2.7	85.3	86	6.8
	MabSelect™ VL	1.5	2.7	85.7	91	6.2

^*During sample application.

Fig 6. Overlay of Capto™ L and MabSelect™ VL resin elution phases. The IgG-scFv₂ molecule elutes at a similar pH with varying elution pool volumes due to the difference in the binding capacity of the resins and the influence of the RT on the binding (A). The scMATCH3 molecule elutes at a slightly lower pH from MabSelect™ VL resin at both RT compared to Capto™ L (B).

For lgG-scFv₂, the elution recovery (> 80% monomeric purity) is slightly lower for Capto™ L resin, although no clear separation is visible in Figure 7. No difference in separation efficiency between the monomer and HMWS or LMWS is observed for IgG-scFv₂ (Fig 7).

Although the purification performance of MabSelect™ VL is similar for the IgG-scFv₂ molecule and comparable to its predecessor Capto™ L, the binding capacity increases by approximately 50%, making MabSelect™ VL a more cost-efficient resin as more protein can be bound and processed per milliliter of resin.

In the case of scMATCH3, the elution profiles of Capto™ L and MabSelect™ VL resins are almost identical at both RT (Fig 6B). This is a result of the very similar QB_10% values obtained on the different resins. MabSelect™ VL resin shows a slightly lower pH in the elution pool compared to Capto™ L resin.

No difference in separation efficiency between the monomer and HMWS or LMWS can be observed for the scMATCH3 molecule, highlighting a similar performance of Capto™ L and MabSelect™ VL resins (Fig 8). The binding capacity of MabSelect™ VL resin increases by approximately 85% making it more economical than Capto™ L resin.

Fig 7. Separation efficiency of IgG-scFv₂ eluted from Capto™ L and MabSelect™ VL resins. The percentage of HMWS, monomer, and LMWS in each elution fraction from its corresponding total content value is shown on the y-axis. MabSelect™ VL resin shows consistent separation efficiency to its predecessor, Capto™ L resin, at both RT. Percentages above the monomer bars correspond to the post-capture monomeric purity in that elution fraction.

Fig 8. Separation efficiency of scMATCH3 eluted from Capto™ L and MabSelect™ VL resins. The percentage of HMWS, monomer and LMWS in each elution fraction from its corresponding total content value is shown on the y-axis. MabSelect™ VL resin shows a consistent separation efficiency to its predecessor, Capto™ L resin, at both RT. Percentages above the monomer bars correspond to the post-capture monomeric purity in that elution fraction.

Conclusions

Due to the complexity of msAbs, there is a growing need for thorough evaluation of capture resins to determine the optimal resin for a specific msAb. This study evaluates the binding capacity (QB_10%) of MabSelect™ VH3, MabSelect PrismA™, MabSelect™ VL, and Capto™ L resins for Numab’s two msAbs (IgG-scFv₂ and scMATCH3).

For the IgG-scFv₂ molecule, MabSelect™ VH3 resin showed the highest binding capacity, 35% higher compared to MabSelect PrismA™ resin and 14% higher binding capacity compared to MabSelect™ VL resin at 4.0 min RT. The post-capture monomeric purity of IgG-scFv₂ could be increased from 89% to over 96% for early eluting fractions by applying a simple elution gradient for the MabSelect™ VH3 resin. This highlights the potential, after optimization, to achieve highly pure msAb in the capture step. This indication of higher monomeric purity must be established during process development to evaluate the suitability of MabSelect™ VH3 over MabSelect PrismA™ for the IgG-scFv₂ molecule.

For the scMATCH3 molecule, MabSelect™ VL provided the highest binding capacity compared to the other evaluated resins. The enhanced binding capacity for MabSelect™ VL would allow for a higher throughput per purification cycle, boosting the productivity of current chromatography setups without requiring expensive capital investments. The observed purity and recovery were similar for MabSelect™ VL resin, compared to the other tested resins, and therefore MabSelect™ VL would be the optimal resin for the scMATCH3 molecule.

Learn more about applications for protein A and protein L chromatography

Get guidance for purifying monoclonal antibodies and fragments with protein A and protein L chromatography.

Read now

Reference

1. Egan TJ, Diem D, Weldon R, Neumann T, Meyer S, Urech DM. Novel multispecific heterodimeric antibody format allowing modular assembly of variable domain fragments. MAbs. 2017 Jan;9(1):68-84. doi: 10.1080/19420862.2016.1248012.

CY43332

Protein A and L resins for purification of multispecific antibody therapeutics