Buffer optimization for bsAb separation

Introduction

Protein A affinity resins have been the first choice when purifying antibodies in manufacturing processes. Due to changes in the biotherapeutic pipeline, with increasing diversity of antibody variants such as bispecific antibodies (bsAb) and fragments, the interest in using protein L affinity resins in the capture step in purification processes has increased. Protein L resins bind to the antibody kappa light chain and can be used for purifying, for example, domain antibodies (dAbs) and antigen-binding fragments (Fabs), as well as to remove product-related impurities from bsAbs at capture. Unlike protein A resins, protein L resins bind to the free antibody light chain, which is then co-eluted with the antibody. This can be seen as a disadvantage as the light chain is often overexpressed during cell culture. Elution from a protein L resin traditionally requires a low elution pH often using 50 mM citrate pH 2 .5 to 3.0, as opposed to protein A resins where the antibody elutes at pH 3.5 or higher. Antibodies and bsAbs can be unstable (e.g., may form aggregates) at low pH and it is therefore essential to find a replacement for 50 mM citrate buffer, pH 2.5.

In this study, we performed buffer screening to find alternative elution buffers for MabSelect™ VL resin. Our focus was to increase elution pH and resolution between bsAb and its product-related impurities. The separation of a mAb and its overexpressed light chain will also be shown to demonstrate the selectivity.

First, a buffer screening was performed to identify elution buffer candidates in a capture step using MabSelect™ VL resin. We used trastuzumab, containing, in addition to the mAb, about 18% of overexpressed antibody light chain. This trastuzumab was used to evaluate both elution pH as well as the resolution between mAb and antibody light chain. We continued to evaluate the most promising buffer substances and concentrations to study the resolution between hetero- and homodimers for asymmetric bsAbs in a capture step using MabSelect™ VL resin.

Watch this video to learn more about buffer optimization for separation of bsAb.

Results

Buffer screening

Evaluation of several elution buffers was undertaken to study the effect on elution pH and resolution for mAb and antibody light chain. Mono-, di-, and tricarboxylic acids at different concentrations were evaluated with 50 mM citrate used as reference buffer. The elution pH was determined by a linear gradient elution from pH 5.0 to 3.0 with a load of 5 mg/mL resin of mAb using elution buffers specified in Table 1.

Table 1. Elution buffers and concentrations

¹ US 10,844,112 B2

The elution pH results are shown in Figure 1.

Gradient elution of mAb sample

Column:                     Tricorn™ 5/50
Resin:                         MabSelect™ VL
Sample load:            5 mg/mL resin (low load) or 45 mg/mL resin (high load) of mAb at 6
                                     min residence time (RT) via a Superloop™ sample loop connector
Equilibration:           20 mM sodium phosphate, 150 mM NaCl, pH 7.2, at 4 min RT
Wash:                        Investigated elution buffer at pH 5.0, 7 CV at 4 min RT
Gradient elution:   0% to 100% of investigated elution buffer, pH 5.0 to pH 3.0 for 20 CV
                                     at 6 min RT, followed by 100% for 5 CV
Strip:                          3 CV of 500 mM acetic acid at 4 min RT
CIP:                            3 CV of 100 mM NaOH at 5 min RT
System:                     ÄKTA pure™

Fig 1. Elution pH of mAb for the investigated elution buffers.

The buffer screening showed a large variation in elution pH of the mAb. Nearly all the elution pH values presented in Figure 1 are significantly higher than 3.2, which is the elution pH established with 50 mM citrate using a pH gradient. Most of the investigated buffers had an elution pH of 3.5 to 4.0 which is about the elution pH seen in a protein A resin step.

The highest elution pH was established using 50 mM propionate and 15 mM succinate, both resulting in an elution pH of 4.0 at a load of 5 mg/mL resin. These buffer conditions also showed a significant improvement in resolution between mAb and light chain (Fig 2). The mAb is separated from the light chain with 15 mM succinate and 50 mM propionate elution buffers.

Fig 2. Overlay of chromatograms from mAb purifications using pH gradient elution.

Gradient elution: separation of mAb and light chain (low vs high load)

By increasing the load to 70% of the dynamic binding capacity of the resin at 10% breakthrough (QB_10% = 65 mg/mL resin) the elution pH of mAb cell culture harvest showed an even higher elution pH. In a gradient elution using 15 mM succinate with low load (5 mg/mL resin) the elution pH is 4.0 and at high load (45 mg/mL resin) the elution pH increased to 4.4 (Fig 3).

Fig 3. Separation of mAb and light chain (low vs high load).

Our results from the pH gradient experiments highlight the possibility of separating mAb from the light chain using step elution, which is the more common approach used at manufacturing scale.

By obtaining high resolution during linear gradient purification, there is a strong likelihood of effectively separating the mAb through a subsequent step-elution process at an optimized pH. In a second elution step or strip, the removal of product-related impurities should be possible.

Step elution: separation of mAb and light chain (high load)

A step elution was performed to separate mAb (in step 1) from the light chain (in step 2) by selecting an appropriate pH based on the result in Figure 3.

Column:             Tricorn™ 5/50
Resin:                 MabSelect™ VL
Sample:             45 mg/mL resin of clarified mAb cell harvest at 6 min RT via a Superloop™
                              sample loop connector
Equilibration:   20 mM sodium phosphate, 150 mM NaCl, pH 7.2, at 4 min RT
Salt wash:         20 mM sodium phosphate, 500 mM NaCl, pH 7.2 at 4 min RT
Wash:                 Investigated elution buffer at pH 6.0, 7 CV at 4 min RT
Elution:              Step gradient with 15 mM succinate pH 3.9 to 4.2 in step 1, pH 3.4 in step 2 or
                             with 50 mM propionate pH 4.0 to 4.2 in step 1, pH 3.2 in step 2 or with
                             20 mM citrate pH 3.5 or 3.6 in step 1, pH 3.0 in step 2, all at 6 min RT
Strip:                   3 CV of 500 mM acetic acid at 4 min RT
CIP:                     3 CV of 100 mM NaOH at 5 min RT
System:             ÄKTA pure™

Fig 4. Step elution using 15 mM succinate with different pH in step 1: (A) pH 3.9, (B) pH 4.0, (C) pH 4.1, (D) pH 4.2. We performed step 2 at pH 3.4.

Optimization is needed to obtain the desired properties of the target molecule in terms of purity and yield. Since MabSelect™ VL resin is used in a capture step, yield is often the most important parameter while purity can be secured in the following polishing steps.

In an optimization of the pH in the first elution step using 15 mM succinate (Fig 4), our analysis indicated significant differences in yield and purity in the elution fraction containing mAb (see Table 2 below).

Similar optimization of the first elution step was also performed with 50 mM propionate. We evaluated the settings of pH 4.0 to 4.2 in step 1 and pH 3.2 in step 2 for yield and purity (Table 2). The optimal separation was achieved with pH 4.2 in step 1 (Fig 5), where mAb was well separated from light chain resulting in high yield and purity.

Fig 5. Step elution using 50 mM propionate with pH 4.2 in step 1, pH 3.2 in step 2.

We found it more difficult to find an optimal condition for 20 mM citrate, since separation of the mAb and the light chain were not achieved in the pH gradient. Optimal results were obtained with pH 3.6 in the first elution step, where the light chain was only partly removed from the mAb resulting in a high mAb yield but with low purity (Fig 6 and Table 2 below).

Fig 6. Step elution using 20 mM citrate with pH 3.6 in step 1 (to elute mAb), pH 3.0 in step 2 (to elute the light chain)

It is important to mention that we collected peak fractions with a 100 mAU threshold. Optimization was performed by varying the pH in the first elution step. Another way to optimize purity and yield is to change how to collect peak fractions, that is, from 100 mAU in the peak front to 500 mAU in the peak tail.

Table 2. Pool volume, mAb yield, and purity results from optimization of pH in the first elution step (the results in bold are the most promising)

Table 3 highlights the optimal separation of the monomer and light chain, resulting in the highest purity and yield for each buffer. We also analyzed fractions for CHO host cell protein (HCP) and ligand leakage content and our results are also presented in Table 3.

Table 3. Summary of results with monomer yield, purity, HCP, and ligand leakage

* data not available.

^† results with 50 mM citrate pH 2.5 as step elution buffer used as reference.

In addition to higher elution pH with propionate and succinate, the purity for the mAb peak increased from 80% with 50 mM citrate and 85% with 20 mM citrate to 96% with propionate and 97% with succinate.

The HCP reduction and level of ligand leakage were comparable for all results.

Gradient elution: separation of bsAb and homodimers (low load)

We determined the elution pH for the bispecific kappa-lambda heterodimer and the mispaired kappa and lambda homodimers with a gradient elution from pH 5.0 to 3.0 by loading 5 mg/mL of the bsAb in cell culture harvest on a Tricorn™ 5/50 column packed with MabSelect™ VL resin. Three elution buffers were evaluated to determine the elution pH and resolution: 50 mM propionate, 15 mM succinate, and 50 mM citrate.

Fractions of 0.5 CV were collected throughout the elution gradients and analyzed with LC-MS to confirm the separation of the different entities.

As seen in Figure 7, all three elution buffer conditions showed good separation of bsAb antibody (first peak in elution phase) from mispaired kappa homodimer (second peak in elution phase). The desired bsAb is separated from the mispaired homodimers at all conditions, however the elution pH differs.

Column:                    Tricorn™ 5/50
Resin:                        MabSelect™ VL
Sample:                    5 mg/mL resin (low load) of bsAb in cell culture harvest at 6 min RT
                                    via a Superloop™ sample loop connector
Equilibration:         20 mM sodium phosphate, 150 mM NaCl, pH 7.2 at 4 min RT
Wash:                       Investigated elution buffer at pH 5.0, 7 CV at 4 min RT
Gradient elution:  0 to 100% of investigated elution buffer, pH 3 for 20 CV at
                                  6 min RT, followed by 100% for 5 CV
Strip:                        3 CV of 500 mM acetic acid at 4 min RT
CIP:                           3 CV of 100 mM NaOH at 5 min RT
System:                   ÄKTA pure™

Fig 7. Overlay of chromatograms from gradient pH elution using 15 mM succinate, 50 mM propionate, or 50 mM citrate.

All elution pH values of the bsAb are almost one pH unit higher for 15 mM succinate and 50 mM propionate elution buffers compared with 50 mM citrate
(Table 4). Prior to the gradient (at pH 5.0 to 3.0), a wash step is performed at pH 5.0. During this step, the lambda (subclass 2) homodimer is washed out.

The resolution between the bsAb and kappa homodimer is also increased with elution buffers containing 50 mM propionate or 15 mM succinate when compared with 50 mM citrate.

Table 4. Elution pH found in the gradient elution pH 5.0 to 3.0 with bsAb at low load

Gradient elution: separation of bsAb and homodimers (high load)

Similar for the result seen with mAb, a higher load of the bsAb sample leads to an increased elution pH. 

We purified a high load of bsAb cell culture harvest on MabSelect™ VL using both 50 mM propionate and 15 mM succinate in pH gradient from 6.0 to approximately 3.0. To save on materials, a smaller column, Tricorn™ 5/20 (0.45 mL CV) was used, still with the same good resolution as a result (Fig 8).

Column:                    Tricorn™ 5/20
Resin:                        MabSelect™ VL
Sample:                    31 mg/mL resin (high load) of bsAb in cell culture harvest at 6 min RT
                                    via a Superloop™ sample loop connector
Equilibration:          20 mM sodium phosphate, 150 mM NaCl, pH 7.2, at 4 min RT
Wash:                       Investigated elution buffer at pH 6.0, 7 CV at 4 min RT
Gradient elution:   0% to 100% of investigated elution buffer, pH 3.0 for 20 CV at
                                   6 min RT, followed by 100% for 5 CV
Strip:                         3 CV of 500 mM acetic acid at 4 min RT
CIP:                           3 CV of 100 mM NaOH at 5 min RT
System:                   ÄKTA pure™

Fig 8. Gradient elution with 50 mM propionate pH 6 to 3.2 to separate bsAb from the mispaired homodimers.

Using a pH gradient from 6.0 to ~ 3.0 the lambda homodimer will elute in the gradient together with the bsAb and the kappa homodimer, still with a good resolution in between the three species.

Gradient elution using 15 mM succinate (chromatogram not shown) and 50 mM propionate (Fig 8) resulted in separation of the bsAb and the product-related impurities. Peak 1, at the highest elution pH, is identified as the lambda light chain subclass 2 homodimer. Peak 2, at the second highest elution pH, corresponds to the heterodimeric bsAb having one kappa light chain subclass 1 chain and one lambda light chain, which is the bsAb that is the target for the purification and separation. Peak 3 corresponds to a homodimer species having two kappa light chains.

The elution pH of the different entities at the three buffer conditions are presented in Table 5.

Table 5. Elution pH found in the gradient elution pH 6.0 to 3.0 with bsAb at high load

Based on the pH gradient results, step elution was performed to separate the three entities. Step elution with 50 mM propionate was performed at pH 4.4 and 3.5, which enabled a complete separation of the bsAb and the kappa and lambda homodimers (Fig 9). The lambda (subclass 2) homodimer is weakly bound to MabSelect™ VL resin and could easily be washed out at pH 5.0.

Fig 9. Step elution with 50 mM propionate pH 4.4 in step 1 and 3.5 in step 2 to separate bsAb from the mispaired homodimer.

In the purification described in this example, we eluted the bsAb and the kappa homodimer in two different pH steps followed by a strip step. However, in a normal process, the bsAb is eluted at one pH, and the remaining substances still bound to the resin are stripped off at a lower pH (including the homodimer).

LC-MS analysis of the collected peak fractions was performed to confirm the separation of the bsAb and the product-related impurities. LC-MS data from the step elution runs confirmed the identity of the molecules in respective fraction (data not shown).

Conclusions

Propionate and succinate buffers had a major impact on the resolution and the elution pH, showing improved performance compared to the reference protocol using citrate buffer.

The results presented in this study show an increase of elution pH of trastuzumab from pH 3.2, obtained with 50 mM citrate, to pH 4.0 with either 50 mM propionate or 15 mM succinate buffers. Besides the increase of elution pH, the resolution between mAb and light chain was improved during the gradient elution, which enabled complete separation of mAb and light chain using step elution.

MabSelect™ VL can separate a kappa-lambda heterodimer bsAb from impurities (homodimers of kappa and lambda) at a higher elution pH using propionate or succinate as elution buffers – with an elution pH about 0.9 units above the elution pH achieved with the reference buffer, 50 mM citrate.

Learn more about MabSelect™ VL protein L resin.

CY43331