Ion Exchange Chromatography

Ion Exchange Chromatography

Chapter 18 Ion Exchange Chromatography Anna Grönberg GE Healthcare Life Sciences, Uppsala, Sweden ABBREVIATIONS AEC anion exchange chromatography CA...

812KB Sizes 1 Downloads 85 Views

Chapter 18

Ion Exchange Chromatography Anna Grönberg GE Healthcare Life Sciences, Uppsala, Sweden

ABBREVIATIONS AEC anion exchange chromatography CAPEX capital expenditure CIEC cation exchange chromatography COG’s cost of goods CPP critical process parameter CQA critical quality attribute CV’s column volumes DBC dynamic binding capacity dsDNA double stranded DNA FT flow through Gua-HCl guanidinium hydrochloride HCPs host cell proteins IB’s inclusion bodies IEC ion exchange chromatography KPP key process parameter mAb monoclonal antibody MM multimodal MW molecular weight pI isoelectric point QB10% dynamic binding capacity at 10% breakthrough SDM stoichiometric displacement model SEC size exclusion chromatography SMA steric mass action SPR surface plasmon resonance WFI water for injection

18.1 INTRODUCTION Ion exchange chromatography (IEC) is based on electrostatic interactions between charged patches on the surface of biomolecules and oppositely charged functional groups attached to a stationary phase via a spacer arm. The interactions are strongest when the ionic strength of the surrounding buffer is low, and binding can be modulated by changes in ionic strength and pH. Charges on molecules in solution and on the IEC resin are balanced by counter-ions, for example, salt and buffer ions, which are displaced when the target molecule binds to the charged functional group on the resin. The net binding charge of the protein will be the same as that of the counter-ions displaced from the resin, hence the term “ion exchange” [1]. Desorption of the adsorbed molecule is commonly executed by increasing the ionic strength of the buffer, and thereby eluting the protein by ionic competition. Alternatively, elution can be done by changing pH, which will change the net charge of the protein. For a more detailed description of the IEC technique, see Ref. [2]. IEC has been employed for several decades for the separation of small inorganic ions. However, it was not until hydrophilic materials of large pore size were introduced in the late 1950s that IEC of biological macromolecules became a useful separation tool [3]. IEC has since been successfully employed for protein purification and is today one of the most

Biopharmaceutical Processing. https://doi.org/10.1016/B978-0-08-100623-8.00018-9 © 2018 Elsevier Ltd. All rights reserved.

379

380  SECTION | IV  Purification Processes, Principles and Methods

c­ ommonly used chromatographic separation modes for purification of pharmaceutical proteins and peptides. Most industrial purification processes comprise one or several IEC steps. The reason for its success is that it is considered a robust method and the principles are well characterized and well understood. The IEC resins typically have high binding capacities and offer good, and controllable, selectivity. Another advantage is that elution is done under mild conditions in which biomolecules maintain their native structures.

18.2  APPLICATION AREAS There are basically two different types of resins used for IEC. In cation exchange chromatography (CIEC), negativelycharged ligands bind positively-charged molecules, whereas the opposite is true for anion exchange chromatography (AIEC). Overall, AIEC is the most frequently used type due to the fact that many recombinant proteins are acidic, and thus negatively charged at neutral pH. Furthermore, AIEC resins bind polynucleotides, DNA, and RNA (due to their negatively charged phosphate groups). AIEC is used for endotoxin and virus removal, and it is also a common method for virus and vaccine purification. IEC is very versatile for downstream processing of recombinant proteins, and can be used for capture as well as intermediate purification and polishing. In capture, the requirement for capacity and quantitative yield is predominant, whereas further downstream in intermediate purification and especially polishing, yield is often compromised by selecting conditions that deliver enhanced purity. Early in the purification process, the target molecule concentration is often low and IEC can be a very efficient concentration step, in addition to purifying the product. In most monoclonal antibody (mAb) purification platform processes, IEC contributes to at least one of two polishing steps following the Protein A affinity step [4,5]. For mAb polishing, CIEC is commonly used in bind-and-elute mode for the target molecule, which allows negatively charged impurities such as residual DNA, RNA, some host cell proteins (HCPs), leached Protein A, and endotoxin to be removed during loading or in the wash fraction. CIEC also has the power to separate antibody charge variants and aggregates during elution. Most mAbs are relatively basic molecules that enable the use of AIEC in flow-through (FT) mode for the product under conditions where viruses, DNA, leached Protein A, and acidic HCP bind. This can be done either using pure FT mode or under conditions where the target protein interacts weakly with the resin [6]. Using CIEC for capture of mAb as an alternative to Protein A has been evaluated [7,8]. CIEC offers high binding capacities and efficient removal of impurities but it is less robust than Protein A in terms of loading conditions. Additional up-front development work, including optimization of pH, conductivity, and other operating conditions for each mAb and feed often rule out CIEC capture despite the benefits of high binding capacities, low buffer costs, and relatively low resin costs. A plethora of IEC resins with different ligands, ligand densities, coupling chemistries, base matrix chemistries, and particle and pore sizes is commercially available. A comprehensive comparison can be found in [9]. Capture resins should have fairly large particle size and high binding capacity, making it possible to concentrate targets from crude samples at high flow rates without clogging the column [10]. Resins used further downstream in intermediate/removal purification and polishing steps (see Chapter 4) should have smaller particles, resulting in sharp peaks and potential separation of target protein from closely related impurities [10]. Staby et al. [10–15] have characterized many IEC resins by looking at the effect of pH on retention, binding strength, and binding capacity using new and standard test proteins and peptides covering a broad range of isoelectric point (pI) and molecular weight (MW). Furthermore, pH titration curves and column efficiency were studied. For proteins with low pI’s the impact of pH on retention varied from low to high for different strong AIEC resins [10,14]. For weak AIEC resins, an increasing retention was seen as a function of increasing pH for proteins with low pI’s [13]. In general, proteins with high pI’s still bound to strong AIEC resins at pH 9, whereas for some weak AIEC resins, a decrease in binding/retention was seen at pH 9—most probably due to deprotonation of the weak AIEC ligand [10,14]. For proteins with high pI’s, the retention on strong CIEC resins is relatively independent of pH [11]. For both AIEC and CIEC resins, the effect of ionic strength on binding/retention and dynamic binding capacity is protein dependent [10–15]. Comparison of static and dynamic experiments showed that for most of the resins, 50%–80% of the total capacity is utilized during chromatographic operation, depending on flow rate, but the utilization could vary as much as 25%–90% [11,14,15]. The authors concluded that results from these studies could be used for selection of resins for further testing and process development, with a focus on selectivity and resolution of target and impurities.

18.3 EXAMPLES Three examples are described as follows, illustrating use of IEC for capture of a biosimilar interferon α-2a, for polishing of a mAb, and for intermediate purification and polishing of adenovirus.

Ion Exchange Chromatography Chapter | 18  381



18.3.1  Capture of Interferon α-2a Using a High-Capacity CIEC Resin [16] A biosimilar molecule, interferon α-2a, was expressed as inclusion bodies (IB’s) in E. coli. After cell lysis, IB isolation, solubilization, and refolding, interferon solution was concentrated and then buffer exchanged prior to loading on the ion exchanger used for capture. The objective with the capture step was to further concentrate the interferon, and to remove the bulk of E. coli HCP while retaining high yield of the target molecule. A high- capacity CIEC resin, Capto S, was used. The calculated pI of interferon α-2a was 6.0, so a loading pH of 5.0, where the molecule is positively charged, was chosen. This pH resulted in a high dynamic binding capacity (120 g/L) at 10% breakthrough (QB10%). Seventy % of the QB10% value (84 g/L) was applied on the column and the interferon was eluted from the column with a simultaneous increase in pH and salt concentration. This method resulted in 95% HCP reduction (to 440 ppm) with the yield of 92% calculated based on the active protein concentration measured by surface plasmon resonance (SPR) analysis (Fig. 18.1). An interesting detail is the choice of elution conditions to avoid buffer exchange between the capture step and the subsequent intermediate purification chromatography step with a multimodal (MM) anion exchange resin (Capto adhere ImpRes). The sample from the capture step could be applied to the next chromatography step after a simple dilution. Interferon quality over the process steps was verified by peptide mapping, western blot, and kinetic analysis by SPR [17].

18.3.2  Polishing of mAbs Using CIEC [18] This example describes a mAb polishing step using a high-capacity CIEC resin, Capto S Impact. The mAb, an IgG1 with a pI of 8.4, was produced by CHO cells and initially purified by direct capture using Protein A chromatography. In this case, no intermediate purification step was required. The polishing by CIEC was optimized for capacity and separation of mAb monomer from aggregates by varying the pH and salt concentration under binding and elution conditions. Process optimization resulted in a QB10% of 109 mg mAb/mL of resin at pH 5.0 and 50 mM NaCl in the loading buffer. Elution was performed by applying a linear salt gradient from 50 to 400 mM NaCl for 20 column volumes (CV’s). At a sample load of 76 g mAb/L resin, corresponding to 70% of QB10, the aggregate concentration was reduced from 2%–3% to 0.9%, the HCP concentration from >300 ppm to 170 ppm, and the concentration of leached Protein A from 3.6 to less than 1 ppm (i.e., below the limit of quantification), with a monomer yield of 93%. The chromatogram in Fig. 18.2 shows the good separation of mAb fragments, monomer, and aggregates. The fragments do not bind as strongly to the resin and consequently elute earlier in the gradient than the monomer. The aggregates, on the other hand, which have a higher net binding charge compared with the monomer, tend to interact more strongly with the resin and elute after the monomer in the gradient.

Column: Sample: Sample load: Buffer A: Buffer B: Flow rate: System:

1 mL HiTrapTM Capto S Interferon α-2a material after NFF 84 mg 50 mM sodium acetate buffer pH 5.0 containing 0.8 M Urea step to 50 mM sodium phosphate buffer pH 7.8 containing 50 mM NaCl 0.4 mL/min ÄKTA avant 25 12

3000 10 8

2000 1500

6

1000

4

500

2

0 0

25

50

75

100

125

pH

A280nm (mAU)

2500

0 150

Volume (mL)

FIG. 18.1  Capture step on Capto S CIEC. The green trace shows conductivity. Courtesy: GE Healthcare Bio-Sciences AB.

382  SECTION | IV  Purification Processes, Principles and Methods

MAb in 50 mM sodium acetate, 50 mM NaCl, pH 5.0 Capto S ImpAct (B/E mode) Tricorn 5/100 76 mg MAb/mL medium (70% of QB10)

4000 MAb fragments A280 (mAU)

3000

MAb aggregates

2000

70

70

60

60

50

50

40 30

1000

20

0 0

20

40

60

40 30 20

10

10

0

0

Fragments (%)

5.4 min 50 mM sodium acetate, 50 mM Nacl, pH 5.0 5 CV of binding buffer 50 mM sodium acetate, 50 to 400 mM NaCl in 20 CV ÄKTA system

Aggregate (%)

Sample: Medium: Column: Load: Residence time: Binding buffer: Wash: Elution buffer: System:

80

Volume (mL) FIG. 18.2  Separation of mAb fragments and aggregates with Capto S Impact at high sample load. The green histogram shows the amount of fragments and the red histogram shows the amount of aggregates as the percentage of total protein amount in corresponding fraction. The light blue area under the UV trace corresponds to the pooled product fractions. Courtesy: GE Healthcare Bio-Sciences AB.

18.3.3  Purification of Type 5 Adenovirus Using AIEC [19] Adenovirus was produced by HEK 293 cells in suspension. After cell harvest and lysis, the clarified supernatantcontaining virus was treated with Benzonase (Novagen, Madison, WI), and the pH was adjusted to 8.0, prior to capture on a high-capacity AIEC resin—Q Sepharose XL. The loading buffer for the capture step was 50 mM Tris, 2 mM MgCl2, 5% sucrose pH 8.0. After loading, a wash was performed with 300 mM NaCl in binding buffer, followed by elution using a linear gradient from 300 to 750 mM NaCl in three CVs. The collected viral peak from the Q Sepharose XL column (Fig. 18.3A) was diluted to a conductivity of approximately 25 mS/cm and then further purified on a SOURCE 15Q column. In this second AIEC step, the column was washed with 300 mM NaCl in loading buffer followed by elution with a gradient from 300 to 600 mM NaCl for 3 CVs (Fig. 18.3B). The amount of free doublestranded DNA (dsDNA) was reduced from 13.0% to 1.6% over the SOURCE 15Q step. Finally, the virus was buffer exchanged using a size exclusion chromatography (SEC) column. Process evolution from gradient to step elution for the AIEC steps was also carried out. The step elution variant was scaled up to pilot production, and later used for a second adenovirus project.

18.4  IEC PRINCIPLE AND STANDARD METHODS Proteins and oligopeptides are amphoteric solutes built up of amino acid residues and containing weak acid and base groups. The net charge of a protein will therefore change gradually as the pH in the surrounding solution changes. At a pH defined as the isoelectric point (pI), the net charge of the protein will be zero, and, in theory, it should not interact with a charged resin. However, each protein will still possess a surface with exposed charge groups, some of which might be arranged in groups or patches. At a pH below the pI, the protein has a net positive charge, and at a pH above the pI, the protein has a net negative charge. Each protein has its own unique net charge versus pH relationship, which can be visualized as a titration curve. Separation of proteins in IEC utilizes the differences in the net surface charge between proteins at a given pH. In AIEC, positively charged (often amino) groups are attached to the chromatography matrix. This type of chromatography resin will retain polynucleotides (due to negatively charged phosphate groups) as well as proteins, and peptides at pH’s above their pI, where these solutes carry a net negative charge. If the charged groups on the AIEC resin are titratable

Ion Exchange Chromatography Chapter | 18  383



100 %B 80 60 40 20 0

(A)

0

10

20

30

40

50

60

70

Min.

mAU 3000 2500 2000 1500 1000 500

(B)

500

600

700

800

ml

FIG. 18.3  (A) After pH adjustment with 1 M TRIS, the virus in the lysis buffer was loaded directly onto the column at 113 cm/h. Immediately following the load, the column was washed with 30% buffer B for at least one column volume or until a stable baseline was achieved. The virus (arrow) was eluted from the column with the use of a linear gradient to 75% buffer B in three column volumes. (B) The diluted viral peak from the Q Sepharose XL column was then loaded onto a Source15Q column at a flow rate of 152 cm/h. Immediately following the load, the column was washed with 30% buffer B until a stable baseline was achieved. The virus was eluted from the column with the use of a linear gradient from 30% to 60% buffer B in three column volumes. Reproduced from Jendrek et al. Development of a production and purification method for type 5 adenovirus. BioProcess J. 5 (1) (2006) 37–42 with permission.

(e.g., they are secondary or tertiary amines), as with diethylaminoethyl (DEAE), the resin is said to be ‘weak,’ whereas if the groups carry their positive charge independent of pH over the range commonly used, as with quaternary aminoethyl (QAE/Q), the AIEC resin is said to be ‘strong.’ Thus, being classified as weak or strong does not describe the strength of the interaction between solute and resin. In the other type of IEC resin, the CIEC, negative charges will attract proteins and peptides below their pI. Common charged groups for CIEC are carboxymethyl (CM), classed as a weak ion exchanger, and sulphopropyl (SP), classed as a strong ion exchanger. Common functional groups in IEC resins are listed in Table 18.1. The ion exchange groups are often attached to the matrix with spacer arms to enhance the availability for large macromolecules. It should be noted that charged groups and spacer arms can exhibit additional interactions, such as hydrogen bonding. Because pH will affect the characteristic charge of a protein, it will have a large influence on retention and selectivity in IEC. Acidic proteins (with low pI’s) will bind to an anion exchanger at neutral pH whereas retention will be low on a cation exchanger. Retention of proteins on an anion exchanger will tend to increase with increasing pH. Below their pI’s, basic proteins will tend to bind to a cation exchanger and retention will tend to decrease with an increase in pH. The magnitude of these changes will be dictated by the slope of each protein’s titration curve (a plot of net charge versus pH). If the slopes are similar (which is often the case) then all proteins will be affected in a similar way by a change in pH without much change in selectivity, unless the pH passes the pI of one protein or if the charge distribution (exposed versus concealed) of the proteins is very different. The effect of pH on selectivity in IEC is schematically explained in Fig. 18.4, which shows how the order of elution of three imaginary proteins changes depending on pH and the type of IEC used. Desorption of bound proteins is commonly executed by increasing the ionic strength/conductivity of a buffer, either continuously or step-wise (continuous or step gradient), so as to elute the protein by ionic competition. Increase in c­ onductivity is achieved either by changing from binding buffer to the binding buffer including a non-buffering salt (e.g., NaCl), or by

384  SECTION | IV  Purification Processes, Principles and Methods

TABLE 18.1  Functional Groups Used in Ion Exchangers Name

Designation

pK

Structure

Diethylaminoethyl

DEAE

9.0–9.5

–OCH2N+H(C2H5)2

Trimethylaminoethyl

TMAE



–OCH2CH2N+(CH3)3

Dimethylaminoethyl

DMAE

Ca. 10

–OCH2CH2N+H (CH3)2

Trimethylhydroxypropyl

QA

–OCH2 CH(OH) N+H(C2H5)2

Quaternary amino ethyl

QAE

–OCH2CH2N+(C2H5)2 CH2 CH(OH) CH3

Quaternary amine

Q

–OCH2N+(CH3)3

Triethyl amine

TEAE

Anion exchangers

9.5a

–OCH2N+(C2H5)3

6.5

CH2CH(CH3)COOH

Cation exchangers Methacrylate Carboxymethyl

CM

3.5–4

–OCH2COOH

Orthophosphate

P

3 and 6

–OPO3H2

Sulfoxyethyl

SE

2

–OCH2CH2SO3H

Sulfopropyl

SP

2–2.5

–OCH2CH2CH2SO3H

Sulfonate

S

2

–OCH2SO3H

a

The pK value does not refer to quaternary groups. Data have been compiled from manufacturer’s booklets. The pK values mostly refer to an ionic strength of 0.1. Table adopted from J.-C. Janson, L. Rydén, Protein purification. Principles, High Resolution Methods and Applications, second ed., 1998, John Wiley & Sons; New York.

increasing the concentration of the buffering ion. Elution can also be achieved by changing pH, which will titrate the amino acid residues of the protein and thus change the surface charge. In other words, to elute a protein from an IEC, lower pH in the case of AIEC and higher pH for CIEC. The use of pH gradients for preparative chromatography is not common, but has been described [20,21]. An increase in ionic strength combined with a simultaneous change in pH to decrease retention is used more frequently (Example 1 in Section 18.2 and Refs. [22,23]). Due to the presence of protein surface regions or patches that are enriched with anionic or cationic charged groups (and therefore not neutral at the pI for the entire protein), proteins can be retained on an IEC resin if their net charge is neutral or even has the same charge as the IEC resin. Lysozyme and lactoferrin are two proteins with such charge asymmetry that can, for example, give rise to interactions with CIEC resins above their pI’s [24,25]. For lysozyme it was suggested that the protein can reorient to interact with a second interaction site when pH was increased [25]. Kopaciewitcz et al. reported retention of numerous proteins on a Q-resin up to one pH unit below their pI’s [26]. With that in mind it can be useful to characterize protein binding to IEC resins over a broad pH range, and for that purpose, high-throughput process development tools such as 96-well filter screening plates are useful.

18.5 BUFFERS Given the pH dependence of the charge of proteins, peptides, and weak IEC resins, controlling the pH is crucial for intended functionality during IEC chromatography. Furthermore, protein stability is an important consideration during purification because some proteins are only stable over a limited pH range. pH fluctuations and excursions can occur in the course of IEC for several reasons. Due to the Donnan effect, protons are repelled or attracted by the charges of the IEC resin. For a CIEC resin, the pH in the microenvironment close to the matrix is usually about one unit lower than the surrounding buffer, and for an AIEC resin, it can be one unit higher than the surrounding buffer [1]. This can result in unexpected stability issues for sensitive proteins [1,27]. pH fluctuations may also occur when using buffering ions of opposite charge to the IEC resin, when loading a protein at high concentration [1], or when changing from a buffer without NaCl to the same buffer system including NaCl [28]. To minimize pH fluctuations, the use of buffers with high buffering capacity combined with careful equilibration of the IEC resin is necessary.

Ion Exchange Chromatography Chapter | 18  385



Most acidic pH: all three proteins are below their isoelectric point, positively charged, and bind only to a cation exchanger. Proteins are eluted in the order of their net charge.

Abs

Most alkaline pH: all three proteins are above their isoelectric point, negatively charged, and bind only to the anion exchanger. Proteins are eluted in the order of their net charge.

Abs

Abs

Abs

V

V

V

V

Surface net charge

+ Cation

0

pH Anion

− Abs

Abs

Abs

Abs

V

V

V

Less acidic pH: blue protein is above its isoelectric point, negatively charged, other proteins are still positively charged. Blue protein binds to an anion exchanger and can be separated from the other proteins which wash through. Alternatively, red and green proteins can be separated on a cation exchanger and the blue protein washes through.

V

Less alkali pH: red protein below its isoelectric point, positively charged. Red protein binds to cation exchanger and can be separated from the other proteins which wash through. Alternatively, blue and green proteins can be separated on an anion exchanger and the red protein washes through.

FIG. 18.4  Effect of pH and proteins binding and elution patterns. Courtesy: GE Healthcare Bio-Sciences AB.

The IEC resin has to be washed with the binding buffer to ensure that the resin- charged groups are in equilibrium with the counter-ions of the binding buffer [27]. Generally, this is done by applying the binding buffer until pH and conductivity in the column effluent are the same as in the in-going buffer. Incomplete column equilibration can cause transient fluctuations of pH during sample application, impairing adsorption. The duration and volume of the equilibration stage of an ion exchange run can be reduced by pre-equilibration with a buffer with several-fold higher concentration of buffering salts prior to equilibration with the dilute buffer [28]. Examples found in the literature are: (1) Pre-equilibration with 3 CVs with 2 M NaCl in 250 mM citrate, pH 5.5 before equilibration with 2 CVs 25 mM citrate, pH 5.5 [28]. (2) Pre-equilibration with 3 CVs of 100 mM sodium acetate containing 1 M NaCl at a specified pH and then equilibration with 3 CVs of 25 mM sodium acetate at the same pH [22]. To enhance retention during adsorption it is recommended that one apply the sample in a low ionic-strength buffer. Usually the concentration of buffer salts during adsorption is 0.01–0.05 M. Due to the low buffer concentration, it is

386  SECTION | IV  Purification Processes, Principles and Methods

i­mportant to select a buffer with a pKa close to the starting pH to ensure sufficient buffer capacity and to eliminate pH disturbances that might result from an incompletely titrated adsorbent or from the adsorption process itself [29]. To ensure a high buffering capacity, the pKa of the buffering species should generally be within 0.5 units of the working pH. It is recommended that the charge of the buffering ions should have the same sign as the functional groups on the IEC resin or be uncharged, otherwise buffering ions can act as counter-ions and result in fluctuations in pH in the microenvironment close to the adsorbed molecule during binding and elution. One noteworthy exception to this is the frequent, successful use of phosphate buffers in anion exchange chromatography [1]. To ensure adsorption of a weakly binding protein, one of the buffering species should be uncharged so as to not contribute to the ionic strength. An example of a suitable buffer for AIEC is Tris-chloride (pKa of 8.07 at 25°C) in which the buffering species are HTris+ (non-interactive), Tris (neutral) and the counter-ion is Cl−. For CIEC, the buffering ions should be negative, such as phosphate (pKa 4.75) and acetate (pKa 7.2) and the counter-ion is most likely Na+. Examples of buffers appropriate for AIEC and CIEC and their corresponding pKa values [30] are shown in Tables 18.2 and 18.3 [31]. High protein concentrations during loading can lead to pH and conductivity fluctuations. When using Tris as a buffer system in AIEC, the counter-ion is Cl-. When the protein displaces Cl- during binding, Tris-Cl is formed, which is the acidic form of Tris. If the displacement occurs quickly, it can lower the pH and increase conductivity in the column. In such conditions, the pH of the effluent can be seen to be lower than the preceding buffer for AIEC, and higher for CIEC, which in turn can impact the binding capacity negatively [1]. To avoid these types of pH and conductivity changes, the protein concentration during load can be decreased by simple dilution. Elution is commonly performed by increasing the ionic strength of the buffer with a non-buffering salt like NaCl, and thereby increasing the ionic competition for the ion exchange groups and weakening the strength of electrostatic interactions. This can be made by a gradual change over 5–10 CV’s, or by a step-wise change from the binding buffer to the binding buffer including the non-buffering salt. The conductivity increase can also be obtained by increasing the concentration of the eluent buffer. Elution can also be performed by an appropriate change in pH toward decreased retention [32], but this is not common for preparative chromatography, as discussed in Section 18.4. More frequent is a concomitant increase in ionic strength and change in pH toward decreased retention. An increase in pH for CIEC, for example, will allow lower ionic strengths for desorption. The influence of the type of counter ions on selectivity has been discussed (e.g., see Ref. [33]). However, it has been shown that specific effects of the salt used are not to be expected [34]. The governing parameter is the elution strength of the salt, and the effect of one type of salt may be obtained by another type of salt by adjusting the concentration. The relative elution strengths of different ions are listed in Table 18.4. Other ion-specific forces that may be important for the selectivity during elution are dispersion forces that depend on the polarizability of the ion. These forces become more dominant at high salt concentrations, where the electrostatic forces are shielded, or close to the pI of the protein [35]. Unintentional pH transitions can also be observed when a CIEC column that has been equilibrated with one buffer receives a buffer of the same buffer system and pH, but at a different salt concentration [28]. A possible explanation is that the increased Na+ concentration in the mobile phase displaces residual H+ ions on the stationary phase and causes a pH dip. In the study, the pH dip was more pronounced when using a cation exchanger with a polymethacrylate backbone compared with agarose-based and polystyrene divinyl benzene backbones, and was also more severe when evaluated for weak cation exchangers, probably due to the higher density of functional groups and a higher degree of H+ bound to those. The magnitude of the pH dip could be decreased by using buffer with more than one pKa, so-called multiprotic buffers, like citrate and phosphate with a broader pKa range compared with the monoprotic MES buffer. In addition, the ionization strength of the eluting salt was important, so sodium chloride was changed to sodium citrate, which has a lower effective concentration of Na+ [28].

18.6  CLEANING IN PLACE AND SANITIZATION When developing cleaning-in-place methods for resins, it is important to consider what types of impurities and what types of interactions can be involved in the fouling. For IEC resins, it is important to break ionic interactions between fouling compounds and charged groups on the IEC resin. The obvious choice is to use a buffer with high salt (1.0–2.0 M NaCl) at low or high pH, depending on whether an AIEC or a CIEC resin is being used. This resin ‘strip’ with high salt should be followed by cleaning with 0.5 or 1.0 M NaOH, which takes care of most bound contaminants (see Chapter 33). High concentrations of NaOH will also have a sanitizing effect. Sometimes the use of NaCl and NaOH is not sufficient, especially for AIEC resins where cleaning can be a challenge. It has been shown that high concentration of the hydrogen bond disruptor, urea, and salt at low pH (8.0 M Urea, 0.1 M citric acid, 1.0 M NaCl, pH 2.5), in addition to cleaning with NaOH, was efficient for removal of discoloration of an AIEC resin [36]. Arginine and guanidinium hydrochloride (Gua-HCl), which also break hydrogen bonds, should work in a similar way (although both are expensive compared with Urea).



TABLE 18.2  Buffers for AEC pH interval 4.3–5.3 4.8–5.8 5.5–6.5 6.0–7.0 6.2–7.2; 8.6–9.6 7.3–8.3 7.6–8.6 8.0–9.0 8.0–9.0 8.4–9.4

Substance N-Methylpiperazine Piperazine L-Histidine bis-Tris bis-Tris propane Triethanolamine Tris N-Methyldiethanolamine N-Methyldiethanolamine Diethanolamine

Conc. (mM) 20 20 20 20 20 20 20 20

Counter-ion −

Cl





Cl or HCOO

pKa (25°C)a

d(pKa)/dT (°C)

4.75

−0.015

5.33

−0.015

Cl



6.04

Cl



6.48

Cl



6.65; 9.10





Cl or CH3COO Cl

− 2−

SO4 −

−0.017



7.76

−0.020

8.07

−0.028

8.52

−0.028 −0.028

Cl

8.88

−0.025

20

Cl−

8.88

−0.031

20

Cl



9.50

−0.029



9.73

−0.026

20 at pH 8.4

Cl or CH3COO

50 at pH 8.8 8.4–9.4 9.0–10.0

Propane 1,3-Diamino Ethanolamine

9.2–10.2

Piperazine

20

Cl

10.0–11.0

Propane 1,3-Diamino

20

Cl−

10.55

−0.026

10.6–11.6

Piperidine

20

Cl−

11.12

−0.031

a

Ref: Handbook of chemistry and physics, 83rd edition, CRC, 2002–2003. Courtesy: GE Healthcare Bio-Sciences AB (Ion Exchange Chromatography. Principles and Methods. GE Healthcare, 2016).

Ion Exchange Chromatography Chapter | 18  387

8.52



50

388  SECTION | IV  Purification Processes, Principles and Methods

TABLE 18.3  Buffers for CEC pH interval 1.4–2.4

Substance Maleic acid

Conc. (mM) 20

Counter-Ion +

Na

+

pKa (25°C)a 1.92

+

2.6–3.6

Methyl malonic acid

20

Na or Li

3.07

2.6–3.6

Citric acid

20

Na+

3.13

50

+

3.3–4.3

Lactic acid

Na

+

d(pKa)/dT (°C)

−0.0024

3.86 +

3.3–4.3

Formic acid

50

Na or Li

3.75

+0.0002

3.7–4.7, 5.1–6.1

Succinic acid

50

Na+

4.21, 5.64

−0.0018

4.3–5.3

Acetic acid

50

Na+ or Li+

4.75

+0.0002

5.2–6.2

Methyl malonic acid

50

Na+ or Li+

5.76

5.6–6.6

MES

50

Na+ or Li+

6.27

−0.0110

7.20

−0.0028

7.56

−0.0140

8.33

−0.0180

6.7–7.7 7.0–8.0 7.8–8.8

Phosphate HEPES BICINE

50 50 50

+

Na

+

+

Na or Li +

Na

a

Ref: Handbook of chemistry and physics, 83rd ed., CRC, 2002–2003. Courtesy: GE Healthcare Bio-Sciences AB (Ion Exchange Chromatography. Principles and Methods. GE Healthcare, 2016).

TABLE 18.4  Elution Strengths of Different Ions [3] Anion-exchange chromatography Acetate < formate < chloride < bromide < sulphate < citrate Cation-exchange chromatography Lithium < sodium < ammonium < potassium < magnesium < calcium

18.7  PROCESS DEVELOPMENT WORKFLOW The objectives with process development will vary depending on the purpose of the chromatographic step. Capacity and recovery are often given highest priority during capture steps early in the purification train. Further downstream, during intermediate purification and polishing, efforts are focused on selectivity and resolution of target molecules from impurities. Product capacity, recovery, selectivity, and resolution can vary significantly between different IEC resins due to differences in ligand type and density, base matrix, and coupling chemistry [10]. Optimal operating conditions will also be different for each resin [37]. Therefore, process development for IEC steps starts with screening of chromatography resins and buffer conditions for binding, wash, and elution. Resin screening might not be restricted to IEC resins, but could also include, for example, MM and HIC resins [16,38]. The most important buffer parameters for IEC are pH and counter-ion concentrations, but buffer and salt species and additives can also be of importance. The number of combinations of chromatography resins and conditions can be high, and therefore parallel and miniaturized high throughput tools, such as 96-well filter plates containing chromatography resins and mini-columns, are useful [37]. A protein with a pI in the range 5.5–7.5 can be purified either using strong or weak IEC resins. Advantages with strong IEC resins are that the charge of the ligand does not change with pH, which means that the interaction between solute and charged group is simpler, and that capacity is maintained even at high or low pH (AIEC or CIEC respectively). This can make development and optimization of separations with strong IEC resins easier. However, weak IEC resins might reveal a different selectivity, and could therefore be an alternative for challenging separations [31]. 96-well filter plates are commonly used to evaluate the static binding capacity for different resins at various pH’s and counter-ion concentrations, and possibly with different additives. This is done with small volumes of chromatography resin

Ion Exchange Chromatography Chapter | 18  389



> 100 mg/mL

0

6.0 5.9 5.8 5.7 5.6 5.5 5.4 5.3 5.2 5.1 5.0 4.9 4.8 4.7 4.6 4.5

50 100 150 200 250 300 350

0

Concentration (mM) 0–350 mM NaCl

Concentration (mM) 0–350 mM NaCl

6.0 5.9 5.8 5.7 5.6 5.5 5.4 5.3 5.2 5.1 5.0 4.9 4.8 4.7 4.6 4.5

pH 4.5–6.0 pH

pH 4.5–6.0 pH

50 100 150 200 250 300 350

(D)

(C)

0

6.0 5.9 5.8 5.7 5.6 5.5 5.4 5.3 5.2 5.1 5.0 4.9 4.8 4.7 4.6 4.5

50 100 150 200 250 300 350

0

(E)

50 100 150 200 250 300 350

Concentration (mM) 0–350 mM NaCl

Concentration (mM) 0–350 mM NaCl

(F) 6.0 5.9 5.8 5.7 5.6 5.5 5.4 5.3 5.2 5.1 5.0 4.9 4.8 4.7 4.6 4.5

pH 4.5–6.0 pH

pH 4.5–6.0 pH

70–40 mg/mL

(B) 6.0 5.9 5.8 5.7 5.6 5.5 5.4 5.3 5.2 5.1 5.0 4.9 4.8 4.7 4.6 4.5

pH 4.5–6.0 pH

pH 4.5–6.0 pH

(A)

100–70 mg/mL

0

50 100 150 200 250 300 350

Concentration (mM) 0–350 mM NaCl

6.0 5.9 5.8 5.7 5.6 5.5 5.4 5.3 5.2 5.1 5.0 4.9 4.8 4.7 4.6 4.5 0

50 100 150 200 250 300 350

Concentration (mM) 0–350 mM NaCl

FIG.  18.5  Contour plots for determination of static binding capacity (SBC) of (A) Capto S ImpAct, (B) Capto SP ImpRes, (C) Eschmuno CPX, (D) Fractogel SO3− (M), (E) Nuvia HR-S and (F) Poros XS for a mAb. Experiments were run in triplicates. Courtesy: GE Healthcare Bio-Sciences AB.

(from 2 to 50 μL) operating in overload mode. Fig. 18.5 shows how the static binding capacity varies with pH (4.5–6.0) and NaCl concentration (0–350 mM) for a protein A purified mAb using six different CIEC resins on the market. In cases where there are abundant impurities, displacement of the target protein during overloading can occur if the impurities are more strongly bound than the target. Thus, experiments under overloading conditions might not enable determination of maximum binding capacity. To overcome this problem, Heldin, E. et al. developed a method using both overloaded and non-overloaded mode for an insulin sample containing C-peptide, a highly abundant impurity [38]. Four consecutive loadings of the insulin sample at various NaCl and Ethanol concentrations were done on 96-well filter plates containing 20 μL CIEC resin that was being evaluated for intermediate purification. The unbound fractions after each loading were analyzed for insulin and C-peptide, and showed that the C-peptide was competing for binding under low salt conditions (Fig. 18.6). 96-Well filter plate binding data will also give information on potential wash and elution conditions. Conditions in which the static binding capacity is low indicate potential elution conditions. However, a careful examination of material (mass) balance should be done by evaluating different elution conditions (pH, salt concentration, etc.) because chromatography resins may show very high binding capacity under conditions that result in subsequent low yield. Kelly et al. studied

390  SECTION | IV  Purification Processes, Principles and Methods

35

Insulin bound (mg/ml)

30 25 20 15 10 100 mM NaCl No salt

5 0 0

5

10

15

20

25

30

35

C-peptide bound (mg/ml) FIG. 18.6  Repeated loading of enzymatically cleaved proinsulin at two salt concentrations (0 and 100 mM NaCl), at 10% ethanol. Amount of bound insulin plotted against bound C-peptide. Insulin binding decreases as the amount of bound C-peptide increases. Reproduced from E. Heldin, et  al., Development of an intermediate chromatography step in an insulin purification process. The use of a High Throughput Process Development approach based on selectivity parameters. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 973C (2014) 126–132 with permission. Fractogel SO3 Fractogel SE Hicap Fractoprep SO3 SP Seph FF SP Seph XL UNOsphere S Toyopearl 550C SP Toyopearl 650M SP

100

% Eluted

80

60

40

20

0 100

150

200

250

300

500

650

800

2000

[NaCl] mM

FIG.  18.7  The mAb elution profiles for the panel of cation-exchange resins are shown for the pH 5, 100 mM sodium chloride load condition. The strength of interaction between the mAb and the resins is reflected in the salt concentration necessary to elute the bound protein, which allows the resins to be rank-ordered by binding affinity. Reproduced from B.D. Kelley, et al., High-throughput screening of chromatographic separations: IV. Ion exchange. Biotechnol. Bioeng. 100 (5) (2008) 950–963 with permission.

binding and elution profiles for a mAb on eight different CIEC resins using a single 96-well filter plate [37]. The sample was loaded at a low challenge (i.e., assumed to be within the linear range of the binding isotherm), using four different pH levels and three salt levels. Elution was carried out by increasing NaCl concentration (0.1–0.8 M NaCl). The difference in binding strength for the mAb at pH 5 and 100 mM sodium chloride loading conditions is displayed in Fig. 18.7. For flow-through mode, one should search for resins and conditions that ensure weak or low binding of targets and stronger interactions with impurities. A 96-well filter plate can be used for generating a so-called Kp plot showing the effect of pH and salt concentrations on Log(Kp) values, where Kp is a partitioning coefficient defined as a ratio of the equilibrium concentration of protein bound to the resin to the concentration of protein in solution [37]. If the Kp plots are generated for both the product and a chosen impurity, or impurities, conditions under which an IEC should be operated can be easily identified [37]. Based on the initial screening, one or two good resin candidates can be selected for further evaluation and optimization, particularly to study dynamic binding capacity and selectivity. The pH and salt concentrations of interest should be based on results from the screening studies but now with narrower ranges. Optimization studies are most efficiently carried out by employing design of experiment (DoE) approaches where the effect of different parameters (loading, wash, and elution conditions) on purity and yield are studied simultaneously. Important factors for IEC are pH and conductivity, gradient slope during elution, and sample load density (target protein and impurity). From these studies, critical and key process parameters (see Chapters 32 and 48) can be determined. The dynamic binding capacity (i.e., the breakthrough capacity) over a range of pH’s and salt conditions at different flow rates/residence times can be evaluated by frontal analysis using mini-columns [39] or by using small-scale columns.

Ion Exchange Chromatography Chapter | 18  391



Selectivity can be understood by analyzing the purity of fractions from a 96-well filter plate [37] or from ‘pseudo gradients’ using mini-columns [39]. The most common way is to use a small-scale column and a salt gradient for elution in order to study the resolution of target protein and impurities at different pH’s. Preferably, loading conditions that result in both high dynamic binding capacity and good selectivity can be defined. An example of this process development workflow (i.e., use of 96-well filter plates, mini-columns, and small- scale columns) for development of a CIEC step for removal of mAb aggregates on a high-capacity CIEC resin was reported [18]. A more thorough description of optimization of parameters in IEC is found later in this chapter (Section 18.10).

18.8  CRITICAL AND KEY PROCESS PARAMETERS Critical process parameters (CPP’s) are process parameters (variables) that have an impact on a critical quality attribute (such as purity or activity of the target protein) and therefore should be monitored and controlled to ensure the process produces product with the desired quality. A key process parameter (KPP) is an adjustable process parameter that does not significantly affect product critical quality attributes (CQA), but does impact process performance and should be maintained within a defined range. IEC parameters commonly evaluated for their impact on the product quality and process performance include: ● ● ● ● ● ● ● ● ●

Protein load (g/Lbed) Conductivity, pH, flow rate, and temperature for all phases of a chromatography cycle (i.e., equilibration/load/wash/elution) Start and end collection points for elution of target proteins Gradient slope during elution Column bed height Column integrity (HETP and asymmetry) Column reuse (number of cycles) Number of CVs for wash and equilibration Resin lot-to-lot variability

Classification of process parameters into CPP’s and KPP’s will depend on the outcome of process characterization and will vary from process to process (Chapter 32). In a case study of CIEC, protein load, load/wash conductivity, elution pH, and elution stop collecting point were found to be CPP’s, and the other parameters listed herein were considered to be KPP’s (the authors used the term ‘general process parameter’) [40]. In the same study, CPP’s identified for a subsequent AIEC step included equilibrium/wash and load conductivities, load pH, operating flow rate, and protein load. Other authors have reported that gradient slope can be important as a CPP [41]. Reports have also shown that ligand density can be a CPP [42,43], or, in other cases, that it did not affect the product CQA or process performance [44]. Similarly, temperature can be either a CPP or a KPP depending on its impact on the process in question. For instance, the chemical stability of modern resins and the long-range nature of the electrostatic forces involved in IEC can mean that temperature effects on the resin and its performance are small in some cases. However, temperature can also affect mobile phase viscosity, especially if the process is to be run at low temperature, which in turn can have a significant effect on operating pressures that might be critical for performance of equipment or soft resins. Temperature also impacts mass transfer kinetics, and, even more important in specific cases, it can affect protein stability, structure, and tendency to aggregate. With respect to KPP’s, it might be anticipated that exact timing for either the start or end of peak collection would have a detrimental effect on the process yield, making it a KPP without affecting the product purity—a CQA. However, in other cases, where impurities are poorly separated from the target product, the exact starting or finishing point for the product peak collection would be a CPP. Operational flow rate or inadequate CIP contact time might result in a shortening of the resin life time leading to reduced yield.

18.9 METHODOLOGY The interaction in IEC has traditionally been described by a stoichiometric model where a solute molecule will displace a number of counter-ions from charged groups on the surface of the matrix, equal to the number of interacting sites on the target solute molecule [26,45]. This model has been questioned, and another model, in which general electrostatic interaction theory for charged surfaces is used to explain the retention, has been presented [46]. An evaluation of the two concepts, considering a weakly charged chromatography resin, gave results in favor of the electrostatic interaction theory [47]. However, it has also been suggested that the two models might describe two extremes in IEC, and further investigations must be performed before a conclusion can be made. Because the stoichiometric displacement model (SDM) currently provides the basis for ion exchange theory, it will be used in this section.

392  SECTION | IV  Purification Processes, Principles and Methods

The stoichiometric model has been refined to incorporate steric shielding of ion exchange groups by large solutes. This is called the steric mass action (SMA) model, and has been used for modelling IEC in overload mode [48]. SMA is well suited to bioprocessing conditions, and the impact of studies by Cramer et al., in this regard, is well recognized.

18.9.1  Retention in IEC The retention volume in IEC under isocratic conditions (i.e., constant salt concentration) is given by Eq. (18.1). VR = VM + k¢VM

(18.1)

where k′ is the retention factor given by Eq. (18.2), which is valid for the stoichiometric ion exchange model. In the electrostatic model, ln k′ is proportional to I–1/2, where I is the ionic strength [46]. k ¢ = k0¢ c - z Þ logk ¢ = logk0¢ - zlogc

(18.2)

where k′0 is related to the ion exchange capacity of the resin, Qy, (k′0 is proportional to Qyz) and z is the interacting, or characteristic, charge of the target molecule. The retention factor is a function of the concentration of salt in the mobile phase, c, and the properties of the target molecule and chromatographic resin [49]. As shown by Eq. (18.2) the retention factor, and thus the retention volume, varies drastically with c and z. The retention is sensitive to the ionic strength in a limited region, the so-called elution window, and keeping the ionic strength constant will widely separate molecules that differ only slightly in z. On the other hand, separation of mixtures of solutes varying substantially in z requires a continuous change in c (i.e., gradient elution). In preparative separations, large differences in z are favorable because this allows step gradient elution. Typical values for the characteristic charge of proteins in IEC are in the range of 3.6–8.2 [26], and 4.8–7.5 [33], although these values will vary with pH. As per Eq. (18.2) in IEC, in order for the sample to be retained on the column, it should be applied in a low ionicstrength buffer. For molecules that are strongly adsorbed, the mobile phase ionic strength may need to be increased considerably to desorb them. In order to reduce separation times (and excessive dilution of sample zones as they move down the column), the salt concentration in the eluent is varied either continuously or step-wise. In special cases, a combination of isocratic, step, and continuous gradient elution may be useful. It should be noted here, that in cases where the concentration of salt in the mobile phase is changed (i.e., as in gradient elution), the retention factor is also continuously changed. The apparent retention factor calculated from the retention volume in gradient elution does not have any physicochemical meaning [49].

18.9.2  Zone Broadening in IEC Zone broadening in isocratic elution will be affected by factors discussed in Chapter 16. For instance, if the adsorption– desorption reaction is slow, this factor will also contribute to zone broadening. However, in gradient elution, a sharpening effect from the gradient is obtained (i.e., molecules at the front of the zone sense a lower ionic strength and thus a higher retention factor than molecules at the rear of the zone). This means that a steady state regarding zone broadening will be reached, and furthermore, all sample zones will have the same (narrow) width on the column, provided the elution conditions (e.g., column length and gradient) are sufficient to promote this steady state. The degree of zone sharpening will depend upon the slope of the gradient (a steeper gradient gives a higher degree of sharpening), and also on the relationship between k′ and ionic strength for the molecule in question. This sharpening effect is, together with the possibility to regulate k′, the main advantage of gradient elution.

18.10 OPTIMIZATION The dynamic binding capacity (DBC) is dependent on several factors (e.g., charge and size of target molecule, pore and resin particle size, charge of the chromatography resin, and ionic strength of the solvent). The highest DBC for a target protein in IEC is expected to occur at low conductivity and at a pH far away from its pI, where the net charge is high (Fig.  18.8A). This can be referred to as “traditional behavior” where capacity is a function of the strength of proteinresin interaction. However, under some conditions it has been found that unexpected behavior occurs, with capacity going through a maximum with increasing conductivity and also decreasing protein charge (Fig. 18.8B) [50,51]. Such behavior is referred to as non-traditional and appears to be related to protein-protein and protein-surface interactions that affect mass transfer rates (Chapter 16). In cases where the IEC ligand is coupled by a surface extender to increase capacity, an increase in conductivity is also believed to increase the flexibility of the extender and thereby increase the mass transfer. Careful

Ion Exchange Chromatography Chapter | 18  393



α-Chymotrypsin pH 4.25 pH 4.75 pH 5.25

250

DBC

200 150 100 50 0 0

(A)

5

10 15 Conductivity (mS/cm)

20

Conalbumin 150

pH 4.25 pH 4.75 pH 5.25

DBC

100

50

0

(B)

0

5

10 15 Conductivity (mS/cm)

20

FIG. 18.8  (A) DBC vs. conductivity curves for α-chymotrypsin at different pH values. α-chymotrypsin behaves in a traditional manner where DBC decreases with increasing conductivity. Courtesy: GE Healthcare Bio-Sciences AB. (B) DBC vs. conductivity curves for conalbumin at different pH values. Conalbumin behaves in a non-traditional manner where DBC goes through a maximum with increasing conductivity. Courtesy: GE Healthcare Bio-Sciences AB.

investigation of the effect of pH and salt concentration on DBC should be undertaken in order to reveal such behavior and locate conditions for the highest binding capacity. Sequential and design of experiments (DoE) approaches are described in Ref. [51]. Evaluation could also be done using high-throughput screening as described in Ref. [52]. The easiest way to influence selectivity is to change the charge of the target molecule (protein), i.e., by varying pH. However, there is no simple relationship between the net charge and the retention time in IEC of proteins, due to the complex distribution of charges over the molecular surface, and accurate prediction of the impact of pH on resolution of sample components is challenging. Selectivity between target molecule and impurities can be mapped out by running a series of experiments eluting with a salt gradient at different pH’s. Fig. 18.9A shows a comparison of selectivity for intact anti-HER2 mAb and product related impurities during elution with a salt gradient from a CIEC resin at three different pH values. The mAb was expressed in glycoengineered Pichia pastoris and contained misassembled and cleaved mAb which could be separated at pH 4.5 and 5.0 but not at pH 6.0 [53]. Altering the gradient is one of the most common parameters used for regulating the separation in adsorption chromatography. Best results in IEC are usually found in the range of 5–20 CV’s for a gradient from 0 to 1 M NaCl. One way to improve a separation is to decrease the gradient slope, i.e., increase the number of CV’s in the gradient and/or reduce the gradient range. Historically, industrial-scale purifications by IEC have been based upon step changes in salt concentration to elute the target solute separated from impurities, partly due to robustness considerations (continuous gradient development at large scale having been regarded as a challenge). However, since continuous gradient elution generally offers higher resolution of solutes with similar affinities for the adsorbent, this elution principle has gained popularity, especially as equipment for reliable large-scale gradient elution is now available (see Chapters 25 and 27).

394  SECTION | IV  Purification Processes, Principles and Methods

mAU

2000

1500

pH 6.0

1000

pH 5.0

500

pH 4.5

0

(A)

0

50

100

150

200

250

300

ml

2

mAU 600 1

400 200 0

(B)

3 0

50

100

150

200

250

300

350 ml

FIG.  18.9  Anti-HER2 mAb cation exchange chromatography. The column (1 cm i.d. × 19 cm) was equilibrated with 25 mM sodium acetate, pH 4.5 (5 CV), loaded with 45 mg of Protein A purified anti-HER2 mAb at pH 4.5 and washed with 25 mM sodium acetate, pH 5.0 (2 CV) before the following chromatographic separations. (A) Chromatograms at variant pH. SOURCE 30S runs were compared in a linear gradient from 0 to 300 mM NaCl (10 CV) in 25 mM sodium acetate at pH 4.5 and 5.0 respectively, and in 12.5 mM sodium acetate, 12.5 sodium phosphate at pH 6.0. Five hundred mM NaCl (3 CV) in the same buffers followed after the linear gradient. (B) NaCl stepwise chromatogram of SOURCE 30S. It was performed with NaCl at 100 mM (3 CV), 125 mM (3 CV), 150 mM (3 CV), 175 mM (3 CV), 300 mM (2 CV) and 500 mM (2 CV) in 25 mM sodium acetate pH 5.0. Reproduced from Y. Jiang, et al., Purification process development of a recombinant monoclonal antibody expressed in glycoengineered Pichia pastoris, Protein Expr. Purif. 76 (1) (2011) 7–14 with permission.

If step elution is required, the optimized gradient conditions are converted to a series of steps that first elute less-retained solutes, then the product, and finally desorb all tightly bound solutes in a wash step (Fig. 18.9B). The resolution of complex samples cannot be expected to be as good in a step elution as in a gradient run, but may be sufficient to yield a product of required purity. This is especially true of the latest IEC resins in which particles, pore sizes, and ligands have been optimized with regard to capacity, selectivity, and eluted peak shape. After study of binding capacity and conditions for best selectivity, the resolution during elution can be utilized to increase the sample load up to the resolution limit that is acceptable with respect to product purity and yield, and process robustness. In early purification, in which capacity is often a major challenge and achieving full purity is not prioritized, applying a sample close to the maximum load is possible, but limited by feed-stream variability. As a practical rule, keeping the total sample loading below 80% of the dynamic-binding capacity of the ion exchanger is recommended. However, at high concentrations, the system is operating in the non-linear region of the binding isotherm, and severe tailing of peaks will be observed—something that may decrease the purity and/or yield of product. The risk of losing material due to aggregation or precipitation at high concentrations must also be considered. A different approach applies when quantitative recovery and high purity are the objectives. In this case, the load may have to be significantly lower.

Ion Exchange Chromatography Chapter | 18  395



18.11  PRODUCTIVITY AND ECONOMY Productivity, expressed as the amount of product per unit time and chromatography resin volume, is determined by the load on the IEC resin, the recovery (yield), of active product, and the cycle time (Chapter 16). Regardless of the purpose of the purification step (capture, intermediate purification, or polishing), the use of highcapacity resins will always have a positive effect on productivity. To be able to utilize the high capacity of modern chromatography resins, investigation of binding conditions—pH and conductivity for optimal binding—is a prerequisite. The cycle time can be reduced by selecting modern chromatography resins that can be operated under high volumetric flow rates. Even though the binding phase may be constrained by a need for a certain residence time, due to mass transfer limitations, equilibration, wash, and elution can be carried out at maximum allowed operating velocity. This is exemplified in Fig. 18.10, which shows the effect of residence time in the loading phase on binding capacities and productivities for three CIEC resins: high-flow resin (Capto S), and two softer resins (SP Sepharose XL and SP Sepharose FF). Comparing the results for the softer resins, it can be seen that the productivity gain for XL resin as compared with the FF variant was directly related to the improved binding capacities for the XL version (both resins can be operated only at residence times not lower than 6 minutes). On the other hand, results obtained with the high-flow resin showed that additional productivity gains (200%) can be realized because this resin can be operated at three-fold higher flow rates (three times shorter residence times) as compared with the softer resins without the deteriorating effect of flow rate on the packed bed [54]. The preceding

QB10% (mg/ml) 200

SP Sepharose XL

+

150

+

+

+

+

+

Capto S

+

+

SP Sepharose Fast Flow

+

+

100

50

0 2 4 6 8 Loading residence time (min)

0

(A)

10

Productivity (kg/h, m3) 60

SP Sepharose Fast Flow SP Sepharose XL Capto S

50 40 30 20 10 0 0

(B)

2

4

6

8

10

12

Loading residence time (min)

FIG.  18.10  (A) Dynamic binding capacity (QB10%) at different loading residence times in Tricorn 5/100 columns. Sample was α-chymotrypsin (4 mg/ml) in E. coli homogenate in equilibration buffer. (B) Productivity versus loading residence time. The superior performance and wider operating window of Capto S are clearly evident. Courtesy: GE Healthcare Bio-Sciences AB.

396  SECTION | IV  Purification Processes, Principles and Methods

shows a practical example of what was discussed in Chapter 16, that a reduction in the overall cycle time, by shortening the time required for various stages of a chromatography cycle (because of higher flow rates allowed), is more important for the productivity as compared with the increase in the binding. Another mean of reducing cycle time is to include a pre-equilibration step with buffers of higher concentrations as described herein (see Section 18.5). Replacing a continuous gradient with a step change will have the same effect (shortening time) on the elution phase, sometimes even a few folds [22]. Although not directly related to the productivity levels attained in a bind-elute step, optimization of the elution conditions to better match the operating pH and conductivity conditions for the subsequent purification step will eliminate the need for buffer exchange steps and minimize conditioning time in between steps, which will have a positive effect on the overall productivity of the downstream process. Naturally, the use of sodium chloride as the eluting component without changing the binding pH results in high conductivity in the elution pool. If the following step requires low operating conductivity conditions, either dilution or buffer exchange will be needed. Instead, a combination of pH and conductivity change could be useful for elution from the ion exchanger [22,23] as shown in example 1 (see Section 18.3). A change in pH that weakens the binding between the target and the resin requires a lower concomitant increase in conductivity for effective elution of the protein and simpler adjustment prior to the next downstream process operation. Typically, a flow-through step on IEC has significantly higher productivities as compared with a bind-elute step for the target molecule. In the FT mode, sample load can be significantly increased if the impurities are present at low concentrations, which has a direct impact on the step productivity. Similarly, if an IEC step is operated in a so-called ‘isocratic overload mode’ where impurities bind stronger to the resin than the target protein and thus displace bound target protein during the loading phase, sample load can be much higher as compared with a bind-elute operation performed under the same conditions of pH and conductivity. For instance, the use of CIEC for mAb polishing is usually performed in bind elute mode and the dynamic binding capacity is typically <100 g/L. However, in the ‘isocratic overload mode,’ loads of up to 1000 g/L with good results for removal of impurities were reported [55]. Additional benefits of using this mode of operation are that smaller columns and thus smaller buffer volumes are needed, which has a direct impact on the economy of the IEC step. IEC resins are relatively inexpensive compared with, for example, affinity resins with protein ligands. IEC resins can also withstand harsh and efficient cleaning conditions that make it possible to use them for hundreds of cycles. The cost of goods (COG’s) of an IEC step can be reduced by selecting less costly buffers and cleaning agents and by optimizing the process in order to reduce the buffer volumes (see Section 18.10). By using high-capacity resins, the buffer consumption will also be reduced due to smaller columns or fewer cycles that need to be performed. Even though the COG’s for modern IEC resins are higher than for older products, the total production cost will usually be lower due to lower liquid costs (buffer salts/WFI) and shorter process times (also contributing to lower labor costs) [37,54] (Fig. 18.11).

Separation cost (USD/g) 0.4

Capital

Medium

Labor

Liquids

0.3

0.2

0.1

0

SP Sepharose fast flow

SP Sepharose XL

Capto S

FIG. 18.11  Cost distribution per medium. Total separation costs are lowest for Capto S. Courtesy: GE Healthcare Bio-Sciences AB.



Ion Exchange Chromatography Chapter | 18  397

18.12 FUTURE Current and future needs in the biopharmaceutical industry are to produce more product in shorter time, at a lower cost, without compromising product quality and yield. Advances in IEC can be related not only to development of new IEC resins with higher capacities and withstanding higher flow rates, but also to innovative modes of operation for IEC products already on the market. The capacities of modern IEC resins are high, in the range of 100–150 g/L. Development of new chromatography resins will probably only lead to incremental improvements. Traditional IEC ligands are well established and advances in capacity will likely originate from developments in base matrices. Existing technologies can also be operated in innovative ways. As discussed herein, both AIEC and CIEC can be operated in flow-through or overloaded mode, that is, beyond the breakthrough of the target molecule while still binding impurities [6,55], which can result in a 10-fold increase in load. Matching loading conditions in two polishing steps (AIEC and CIEC) would enable connecting two columns in a series without pool tanks and conditioning steps in between. This mode of operation can be facilitated by the use of ‘salt tolerant’ IEC resins or multimodals (MM) (Chapter 20). The salt tolerance observed for MM resins can also make them more suitable for capture conditions because the feed from upstream operations typically has a high (physiological) salt concentration [56]. MM ligands will also offer different selectivity compared with IEC ligands. It is quite possible that the use of MM chromatography will compete even more with IEC in the future, especially for challenging separations. Challenging separations of impurities closely related to the target molecule put increasing demand on both selectivity and resolution. Selectivity and resolution depend not only on the ligand and the base matrix, but also on the spatial distribution of a ligand within the base matrix. It has been shown that clustering of charges may be important for IEC adsorption/ desorption and engineered clusters may be more efficient than the random clustering that is normally present in IEC [57]. In some cases, charged membrane chromatography could replace traditional IEC in packed columns, offering dynamic binding capacities theoretically independent of flow rate because of the convective mass transfer. IEC exchange membranes are used successfully today in many flow-through polishing applications, in which low levels of contaminants are bound and the requirements for capacity and resolution are low because the target product is not bound. In these cases, the load of the target protein can be very high, in the order of several kg/L. In addition to flow-through applications, there may be niche applications in which charged membrane chromatography could perform better compared with packed bed chromatography for capture or a combination of clarification and capture steps [58,59]. Another application area of interest for membrane chromatography is purification of large molecules such as viruses and plasmid DNA, in which the capacity of charge membranes outperforms that of the chromatographic beads [60,61]. However, for the bind-elute applications, the cost of charge membranes and a much higher buffer consumption so far has prevented this technology from competing with IEC resins. More information on the use charge membranes in the field of bioprocessing can be found elsewhere [62].

REFERENCES [1] R.K. Scopes, Protein Purification, Springer-Verlag, New York, 1994. [2] P.R. Haddad, P.E. Jackson, Ion Chromatography: Principles and Applications, Elsevier, Amsterdam, 1990. [3] S. Yamamoto, K. Nakanishi, R. Matsuno, Ion-Exchange Chromatography of Proteins, second ed., Chromatographic Science Series (Book 43), CRC Press, Boca Raton, FL, 2012. [4] A.A. Shukla, et al., Downstream processing of monoclonal antibodies--application of platform approaches, J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 848 (1) (2007) 28–39. [5] P.A. Marichal-Gallardo, M.M. Alvarez, State-of-the-art in downstream processing of monoclonal antibodies: process trends in design and validation, Biotechnol. Prog. 28 (4) (2012) 899–916. [6] B.D. Kelley, et al., Weak partitioning chromatography for anion exchange purification of monoclonal antibodies, Biotechnol. Bioeng. 101 (3) (2008) 553–566. [7] Y. Tao, et al., Evaluation of high-capacity cation exchange chromatography for direct capture of monoclonal antibodies from high-titer cell culture processes, Biotechnol. Bioeng. 111 (7) (2014) 1354–1364. [8] G.R. Miesegaes, et al., Monoclonal antibody capture and viral clearance by cation exchange chromatography, Biotechnol. Bioeng. 109 (8) (2012) 2048–2058. [9] D.C. Nash, H.A. Chase, Comparison of diffusion and diffusion-convection matrices for use in ion-exchange separations of proteins, J. Chromatogr. A 807 (2) (1998) 185–207. [10] A. Staby, I.H. Jensen, I. Mollerup, Comparison of chromatographic ion-exchange resins. I. Strong anion-exchange resins, J. Chromatogr. A 897 (1–2) (2000) 99–111. [11] A. Staby, et al., Comparison of chromatographic ion-exchange resins: III. Strong cation-exchange resins, J. Chromatogr. A 1034 (1–2) (2004) 85–97. [12] A. Staby, et al., Comparison of chromatographic ion-exchange resins IV. Strong and weak cation-exchange resins and heparin resins, J. Chromatogr. A 1069 (1) (2005) 65–77. [13] A. Staby, et al., Comparison of chromatographic ion-exchange resins VI. Weak anion-exchange resins, J. Chromatogr. A 1164 (1–2) (2007) 82–94.

398  SECTION | IV  Purification Processes, Principles and Methods

[14] A. Staby, I.H. Jensen, Comparison of chromatographic ion-exchange resins. II. More strong anion-exchange resins, J. Chromatogr. A 908 (1–2) (2001) 149–161. [15] A. Staby, et al., Comparison of chromatographic ion-exchange resins V. Strong and weak cation-exchange resins, J. Chromatogr. A 1118 (2) (2006) 168–179. [16] Application note 29-1353-93 AA: Development of a production process for biosimilar interferon α-2a. GE Healthcare, 2015. [17] Application note 29-1154-78 AC: Accurate comparability assessment of a biosimilar interferon in process development. GE Healthcare, 2014. [18] Application note 29-1450-68 AA: Optimization of dynamic binding capacity and aggregate clearance in a monoclonal antibody polishing step. GE Healthcare, 2015. [19] S. Jendrek, D. Ekstrom, D. Stoughton, S. Ishikawa, D. Poon, W. Cheng, S. Giardina, D. Mallard, Development of a production and purification method for type 5 adenovirus, BioProcess. J. 5 (1) (2006) 37–42. [20] A.A. Shukla, M.R. Etzel, S. Gadam (Eds.), Process Scale Bioseparations for the Biopharmaceutical Industry, CRC Press, Boca Raton, FL, 2007. [21] K.A. Logan, I. Lagerlund, S.M. Chamow, A simple, two-component buffer enhances use of chromatofocusing for processing of therapeutic proteins, Biotechnol. Bioeng. 62 (2) (1999) 208–215. [22] J.X.  Zhou, et  al., pH-conductivity hybrid gradient cation-exchange chromatography for process-scale monoclonal antibody purification, J. Chromatogr. A 1175 (1) (2007) 69–80. [23] A. Grönberg, et al., A strategy for developing a monoclonal antibody purification platform, BioProcess Int. 5 (1) (2007) 48–56. [24] G. Malmquist, et al., Electrostatic calculations and quantitative protein retention models for ion exchange chromatography, J. Chromatogr. A 1115 (1–2) (2006) 164–186. [25] F. Dismer, M. Petzold, J. Hubbuch, Effects of ionic strength and mobile phase pH on the binding orientation of lysozyme on different ion-exchange adsorbents, J. Chromatogr. A 1194 (1) (2008) 11–21. [26] W. Kopaciewicz, et al., Retention model for high-performance ion-exchange chromatography, J. Chromatogr. A 266 (1983) 3–21. [27] J.-C. Janson, L. Rydén, Protein purification, Principles, High Resolution Methods and Applications, second ed., John Wiley & Sons, New York, 1998. [28] S. Ghose, T.M. McNerney, B. Hubbard, pH Transitions in ion-exchange systems: role in the development of a cation-exchange process for a recombinant protein, Biotechnol. Prog. 18 (3) (2002) 530–537. [29] E.A. Peterson, H.A. Sober, Column chromatography of proteins: Substituted celluloses, in: P.C. Sidney, O.K. Nathan (Eds.), Methods in Enzymology, Academic Press, New York, 1962, pp. 3–27. [30] D.R. Lide, CRC Handbook of Chemistry and Physics, 85th ed., Taylor & Francis, Boca Raton, FL, 2004. [31] Ion Exchange Chromatography. Principles and Methods. GE Healthcare, 2016. [32] T. Ahamed, et al., pH-gradient ion-exchange chromatography: an analytical tool for design and optimization of protein separations, J. Chromatogr. A 1164 (1–2) (2007) 181–188. [33] S.D. Gadam, G. Jayaraman, S.M. Cramer, Characterization of non-linear adsorption properties of dextran-based polyelectrolyte displacers in ionexchange systems, J. Chromatogr. A 630 (1–2) (1993) 37–52. [34] G. Malmquist, N. Lundell, Characterization of the influence of displacing salts on retention in gradient elution ion-exchange chromatography of proteins and peptides, J. Chromatogr. 627 (1–2) (1992) 107–124. [35] L.A. Moreira, et al., Hofmeister effects: Why protein charge, pH titration and protein precipitation depend on the choice of background salt solution, Colloids Surf. A Physicochem. Eng. Asp. 282–283 (2006) 457–463. [36] Cleaning-in-Place of Chromatographic Media, IPCOM000188742D, Editor, 2009. [37] B.D. Kelley, et al., High-throughput screening of chromatographic separations: IV. Ion-exchange, Biotechnol. Bioeng. 100 (5) (2008) 950–963. [38] E. Heldin, et al., Development of an intermediate chromatography step in an insulin purification process. The use of a high throughput process development approach based on selectivity parameters, J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 973C (2014) 126–132. [39] A. Susanto, et al., High throughput screening for the design and optimization of chromatographic processes: automated optimization of chromatographic phase systems, Chem. Eng. Technol. 32 (1) (2009) 140–154. [40] CMC Biotech Working Group, A-Mab: A Case Study in Bioprocess Development, CMC Biotech Working Group, 2009. [41] O.  Kaltenbrunner, et  al., Application of chromatographic theory for process characterization towards validation of an ion-exchange operation, Biotechnol. Bioeng. 98 (1) (2007) 201–210. [42] H. Aono, et al., Mitigation of chromatography adsorbent lot performance variability through control of buffer solution design space, J. Chromatogr. A 1318 (2013) 198–206. [43] J.T.  McCue, P.  Engel, J.  Thömmes, Effect of phenyl sepharose ligand density on protein monomer/aggregate purification and separation using ­hydrophobic interaction chromatography, J. Chromatogr. A 1216 (6) (2009) 902–909. [44] J. Fogle, J. Persson, Effects of resin ligand density on yield and impurity clearance in preparative cation exchange chromatography. II. Process characterization, J. Chromatogr. A 1225 (2012) 70–78. [45] N.K. Boardman, S.M. Partridge, Separation of neutral proteins on ion-exchange resins, Biochem. J. 59 (4) (1955) 543–552. [46] J. Stahlberg, B. Jonsson, C. Horvath, Theory for electrostatic interaction chromatography of proteins, Anal. Chem. 63 (17) (1991) 1867–1874. [47] C.-h. Cai, V.A. Romano, P.L. Dubin, Ionic strength dependence of protein retention on Superose 12 in SEC-IEC mixed mode chromatography, J. Chromatogr. A 693 (2) (1995) 251–261. [48] S.R. Gallant, A. Kundu, S.M. Cramer, Modeling non-linear elution of proteins in ion-exchange chromatography, J. Chromatogr. A 702 (1–2) (1995) 125–142. [49] Gradient elution in column liquid chromatography, theory and practice, in: J.C.P. Jandera (Ed.), Journal of Chromatography Library, 31, Elsevier, Amsterdam, 1985.



Ion Exchange Chromatography Chapter | 18  399

[50] C. Harinarayan, et al., An exclusion mechanism in ion exchange chromatography, Biotechnol. Bioeng. 95 (5) (2006) 775–787. [51] Application Note 28-4078-16 AA: Screening and optimization of the loading conditions on Capto S., 2006. [52] E.M. Johnson, et al., Diffusion and Partitioning of proteins in charged agarose gels, Biophys. J. 68 (1995) 1561–1568. [53] Y. Jiang, et al., Purification process development of a recombinant monoclonal antibody expressed in glycoengineered Pichia pastoris, Protein Expr. Purif. 76 (1) (2011) 7–14. [54] Application note 28-4078-15 AA: High productivity capture of α-chymotrypsin on Capto S cation exchanger. GE Healthcare, 2006. [55] H.F.  Liu, et  al., Exploration of overloaded cation exchange chromatography for monoclonal antibody purification, J. Chromatogr. A 1218 (39) (2011) 6943–6952. [56] T. Yang, et al., Evaluation of multi-modal high salt binding ion exchange materials, J. Chromatogr. A 1157 (1–2) (2007) 171–177. [57] L. Kisley, et al., Unified superresolution experiments and stochastic theory provide mechanistic insight into protein ion-exchange adsorptive separations, Proc. Natl. Acad. Sci. U. S. A. 111 (6) (2014) 2075–2080. [58] B.V. Bhut, K.A. Christensen, S.M. Husson, Membrane chromatography: Protein purification from E. coli lysate using newly designed and commercial anion-exchange stationary phases, J. Chromatogr. A 1217 (30) (2010) 4946–4957. [59] V. Orr, et al., Simultaneous clarification of Escherichia coli culture and purification of extracellularly produced penicillin G acylase using tangential flow filtration and anion-exchange membrane chromatography (TFF-AEMC), J. Chromatogr. B 900 (2012) 71–78. [60] S.R. Wickramasinghe, et al., Characterizing solute binding to macroporous ion exchange membrane adsorbers using confocal laser scanning microscopy, J. Membr. Sci. 281 (1–2) (2006) 609–618. [61] M.A. Teeters, et al., Adsorptive membrane chromatography for purification of plasmid DNA, J. Chromatogr. A 989 (1) (2003) 165–173. [62] V. Orr, et al., Recent advances in bioprocessing application of membrane chromatography, Biotechnol. Adv. 31 (4) (2013) 450–465.