Development of an Alternative Monoclonal Antibody Polishing Step

Published on: 
, , , , , ,
BioPharm International, BioPharm International-05-01-2012, Volume 25, Issue 5

MAb polishing using salt tolerant interaction membrane chromatography.

ABSTRACT

In many monoclonal antibody (mAb) purification platforms, traditional anion exchange column chromatography or, increasingly, anion exchange membrane chromatography, is used as a polishing step in a product flowthrough mode to bind trace levels of process- or product-related impurities and assure efficient viral clearance. Anion exchange chromatography is, however, limited by the requirement for low loading buffer conductivity to efficiently remove impurities, which necessitates buffer exchange or dilution of the protein A column eluate. In this study, the authors developed a mAb polishing step using salt tolerant interaction membrane chromatography. Using a 96-well high-throughput screening (HTS) approach the authors identified the initial chromatographic parameters for acceptable step recovery and product quality. The authors then confirmed these conditions using small STIC capsules. Using a combination of HTS screening and design of experiments optimization the authors developed a mAb polishing platform which demonstrated high step recovery and efficient clearance of impurities (i.e., host cell proteins, high molecular weight species, host DNA, and leached protein A) for multiple antibodies at higher loading buffer conductivity. This simple and efficient polishing step can be easily integrated into most current mAb purification platforms, which may shorten mAb purification processes and accelerate development programs.

Monoclonal antibody (mAb) purification processes exist in different well-established platforms with extensive process performance histories for production of commercial monoclonal antibodies (1–10). These platforms, typically employing two or three chromatographic steps, are scalable and robust, and produce proteins with acceptable process yield and product quality.

In most of the two-column downstream processing platforms, the first chromatographic unit operation is protein A which binds the target mAb product directly from the harvested cell culture fluid (3, 4, 10–12). The process impurities are removed in the flowthrough and subsequent wash steps. A low pH buffer elutes the product and sets up the subsequent viral inactivation step. Anion exchange chromatography (AEX), such as Q Sepharose Fast Flow (Q FF) column chromatography (3, 13, 14) and Q membrane adsorber (6, 15–18), serves as the second chromatographic purification step. It is operated in a flowthrough mode, binding trace impurities such as host cell proteins, host DNA, endotoxins, and in some instances, high molecular weight (HMW) species while the antibody passes through. The AEX chromatography step is limited by the requirement for low loading buffer conductivity, which necessitates buffer exchange through tangential flow filtration (TFF) or dilution of the protein A column elution pool for efficient impurity clearance. However, some antibodies may have solubility issues at low ionic strength conditions. These challenges may be addressed by Sartorius Sartobind salt tolerant interaction chromatography (STIC) using a polyallylamine ligand covalently coupled to the double-porous membrane (19). The optimized base support membrane matrix combined with weak anion exchange chemistry provides a robust method for viral clearance at physiological conductivities and above (19, 20). A virus, ΦX174, used to model weak acidic contaminants, was shown to be removed (LRV >5) in the presence of 150 mM NaCl. Megta et al. demonstrated efficient viral clearance on STIC using two model viruses, MMV and MuLV (21). Furthermore, similarly to Q membrane chromatography, the STIC membrane adsorber may also provide some economic benefits as an alternative mAb polishing step (16, 22).

In this study, Sartobind STIC was evaluated as a mAb polishing platform alternative to Q column chromatography or Q membrane adsorber. Using a combination of high-throughput screening (HTS) and design of experiments (DOE) optimization, we developed a STIC mAb polishing platform which demonstrated high step recovery and efficient clearance of impurities (host cell proteins, host DNA, and leached protein A) for four antibodies at higher loading buffer conductivity. In addition, since there is no need for buffer exchange, the pre-Q column TFF step can be removed from the purification process. This polishing step, which can be easily integrated into current mAb purification platforms, offers a viable alternative to traditional AEX especially in cases where antibodies exhibit poor process performance. Furthermore, methods described here for developing STIC operating conditions can be applied to the purification process development of other membrane adsorbers.

MATERIALS AND METHODS

Purification techniques

The mAbs used for this study were fully human IgG1 produced in recombinant Chinese hamster ovary (CHO) cells grown in a serum-free medium. MabSelect SuRe protein A (GE Healthcare, Piscataway, NJ) was used to purify the antibody present in the harvested cell culture fluid (HCCF) using AKTAexplorer under the control of UNICORN 5.0 (GE Healthcare, Piscataway, NJ). Briefly, the protein A column was loaded to approximately 35 mg mAb/mL-resin. The product was eluted using 50 mM acetate buffer, pH 3.5–3.8, which was mapped out for each protein on 96-well plates. Necessary wash steps were introduced to reduce the host cell protein (HCP) level in the eluate. The protein A elution pool was held at room temperature for one hour after pH was adjusted to 3.5 using 1 M acetic acid for viral inactivation. Following low-pH treatment, the product pool was neutralized to the required pH with 2 M Tris base solution, clarified through a 0.22 µm filter (EMD Millipore, Billerica, MA), which served as the feed to STIC experiments.

The STIC equilibration buffer conditions were first screened using 96-well plates (gifts of Sartorius Stedim Biotech, Bohemia, NY) with a full factorial design of experiments. The buffer condition was evaluated at 5 pH levels of 6.5, 7.0, 7.5, 8.0, and 8.5, and 6 NaCl concentrations of 0, 25, 50, 75, 100, and 150 mM. Before loading into each well of STIC plates, the protein A eluate was adjusted to the appropriate pH using 2 M Tris base solution and to the target salt concentration using 5 M NaCl stock solution. The flowthrough/subsequent wash from each well was collected as the product. Response parameters, process yield, HCP, and HMW were determined for each experimental run. The response surfaces were defined in a group of optimization experiments using a 96-well STIC plate through a central composite design with 4 center points. All experimental design and data processing were performed using JMP version 8.0 software (SAS Institute, Cary, NC).

Antibody dynamic loading capacity (DLC) on Sartobind STIC was determined at the optimized buffer pH and NaCl conditions using 1 mL STIC Nano capsule (Sartorius Stedim Biotech, NY) at 10 membrane volume (MV)/min. We collected different flowthrough fractions and determined the HCP level in each fraction. The DLC value was the antibody amount applied to the membrane adsorber when HCP in the flowthrough reached 10 ppm or 20 ppm when applicable. The bound materials in the case of Mab-T were eluted using 50 mM Tris, pH 7.2, 2.0 M NaCl and analyzed for the level of HMW species. The process and product related impurities in the STIC purified products were determined using different analytical techniques.

Analytical techniques

Antibody concentrations in purified solutions were determined by the absorbance at 280 nm, using the NanoDrop spectrophotometer (Thermo Fisher Scientific, Wilmington, DE). Size exclusion high performance liquid chromatography (SE–HPLC) was used to monitor the size heterogeneity of mAbs under native conditions on Agilent HPLC system using ChemStation as the controlling software (Santa Clara, CA). A TSK-Gel G3000SWXL column (Tosoh Bioscience, Montgomeryville, PA) was utilized to separate HMW species, monomers, and fragments. The mobile phase was phosphate buffer saline (without Ca2+ and Mg2+ ), pH 7.2 (Life Technologies, Carlsbad, CA).

A CHO host cell protein (CHOP) kit (Cygnus Technologies, Southport, NC) was used to determine the residual HCP level in purification in-process samples and purified mAb product during the screening stage of experiments according to the manufacturer's protocol. The HCP level in antibodies purified on STIC Nano was also measured by electrochemiluminescence (ECL) technology (Meso Scale Discovery or MSD, Gaithersburg, MD) developed at ImClone. Briefly, 25 µL of 3 µg/mL in-house purified anti-CHOP capturing antibodies were immobilized overnight on a 96-well MSD plate. The plate was blocked for 1 h with 3% BSA at room temperature. 25 µL of mAbs in 2-fold serial dilutions and HCP standards were added into the plate and incubated for 2 h at room temperature. The bound HCPs were detected by addition of 25 µL of biotinylated anti-CHOP probe at 3 µg/mL, which was then detected by the addition of 25 µL of streptavidin conjugated sulfo-Tag at 3 µg/mL. After the completion of reaction, 150 µL of MSD buffer was added and the plate was read with MSD SECTOR Imager 2400 for relative electrochemiluminescence units (ECLU). The intensity of the ECLU was proportional to the amount of residual HCP present in antibodies by extrapolation from the standard curve with a quantification limit of 16 ng/mL. All HCP results were normalized to the in-house CHOP standards.

The leached MabSelect SuRe ligand in antibodies was determined using the RepliGen's protein A ELISA kit (Waltham, MA) with a detection limit of 0.1 ng/mL according to the manufacturer's protocol. Residual CHO DNA in antibodies was measured by quantitative PCR (qPCR) using the resDNASEQ quantitative CHO Kit (Life Technologies, Carlsbad, CA), combining high-recovery PrepSEQ sample preparation and TaqMan based-quantitation. The assay was developed at ImClone using in-house CHO DNA standards. The quantification limit of the assay was 0.1 pg/mL.

RESULTS AND DISCUSSION

Condition screening and optimization using 96-well plates

Using protein A column chromatography under our platform operating conditions, we first prepared four antibodies, which served as model proteins to evaluate STIC as an alternative antibody polishing platform to AEX chromatography. These partially purified proteins and their properties are shown in Table I. Among them, Mab-D and Mab-S showed poor solubility at low ionic strength solution conditions (< 5 mS/cm), which posed challenges to our current purification platform process. Mab-T was considered as the worst-case scenario material in terms of levels of residual HCP and HMW impurities. Thus it was used here to illustrate the procedure of condition screening and optimization. Process yield, HCP, and HMW were evaluated during the condition screening and optimization. Although the study described here focused on HCP and HMW, a similar method could be applied for other impurities.

Table I

The STIC equilibration buffer conditions were first screened using a Sartorius Sartobind STIC 96-well plate in a full factorial experimental design, as described in the Materials and Methods section. The load eluate (or flowthrough) and wash from each well, representing the purified product from an experimental run, were collected and evaluated for yield, HCP, and HMW. More than 90% process yield was achieved in all 30 experimental runs. The residual HCP and HMW levels in the STIC purified Mab-T were summarized in Figure 1. The residual HCP was <50 ppm at all tested conditions. Higher HCP removal was achieved when the operating conditions moved to the center of pH-NaCl contour plot (see Figure 1a). In most cases, for a given NaCl concentration, with increasing pH, HCP removal efficiency increased to the highest point and then started to decrease. This finding suggests that the optimal pH operating window for Mab-T is at pH 7.0–7.5.

Figure 1

The presence of an optimal operating pH window is consistent with the amine protonation hypothesis reported previously (21, 23). As pH increases from 6.5 to 8.5, amine groups are less protonated. Thus, positive charges on the ligands available to bind impurities decrease (21). Meanwhile, with increasing pH, there is an increase in the net negative charge of host cell proteins, which results in more efficient binding to the positively charged ligands on the membrane adsorber. The presence of an optimal pH operating window is due to the combination of amine protonation on the STIC membrane adsorber and changes in protein surface charges.

Advertisement

In addition, in the pH range of 6.5–8.0, HCP removal was not dramatically affected by NaCl concentration, supporting the salt tolerant nature of the STIC membrane adsorber. Through this quick, full factorial DOE study using 96-well plates, optimal buffer conditions for HCP removal were identified.

We next examined the impact of equilibration buffer conditions on the removal of HMW from the partially purified Mab-T. With the understanding that in most cases, HMW level can be controlled to below 2.0% through pre-polishing steps, the goal of HMW removal in this study was to reduce HMW species from 5.0% in the load to 3.0% in the flowthrough. When operating pH was increased from 6.5 to 8.5, the HMW in the purified Mab-T increased from 2.7% to 4.0% as shown in Figure 1b. A concomitant decrease in the IgG monomer was observed, suggesting that the HMW removal was less efficient as the pH increased. By contrast, HMW removal was not sensitive to NaCl concentration, particularly in the range of 20–120 mM NaCl. These findings further suggest that the process performance of Sartobind STIC is a result of its salt tolerant nature, supporting a wide design space of solution ionic strength or NaCl concentration. In order to reduce the HMW in the final product to 3.0% and HCP to less than 30 ppm, pH 7.0–7.5 and 25–75 mM NaCl were selected for further condition optimization.

Figure 2

The initial buffer conditions developed in the screening experiments were further optimized through 12 additional experimental runs on a STIC 96-well plate via a central composite design (pH: 7.0–7.5, and NaCl concentration: 25–75 mM). The STIC response surfaces of process yield, residual HCP, and HMW level, were defined based on these runs. Again, each well in a STIC 96-well plate represented one unique combination of experimental conditions. As expected, >94% process yield was achieved in all experimental runs. The sweet spot of the equilibration buffer conditions is illustrated as a pH-NaCl contour plot (see Figure 2). When STIC was operated in the window of pH 7.2–7.3, and 30–60 mM NaCl, HCP was reduced to a lower level (< 20 ppm) and HMW to 3.0%.

Figure 3a

Dynamic loading capacity using STIC Nano

Breakthrough curves were used to determine the DLC of antibodies on 1 mL STIC Nano membrane adsorber. When the low-pH- treated protein A eluate of Mab-T was neutralized to the pH value (defined in the previous Condition Screening and Optimization section) and clarified, the conductivity fell into the optimal operating window. This conditioned protein A eluate was then directly applied into the STIC Nano in flowthrough mode. The Mab-T chromatogram is shown in Figure 3a. As expected, the sharp rising shape of the breakthrough curve during the load and sharp decreasing UV trace in the wash step suggest that mass transfer in STIC membrane adsorber is convective flow, and not limited by diffusion as in the case of porous chromatography resins. This finding is consistent with the results of previous works on other membrane adsorbers (15–17, 24). In addition, compared with Q column chromatography, the product pool was not diluted significantly by the wash, as the load volume was 150 MV while the wash volume was only 5 MV (see Figure 3a). Thus, STIC might provide the benefit of a lower dilution factor because of the smaller volume of buffer required in the wash step, which is extremely valuable when there is a limit on tank capacity in manufacturing.

Figure 3b

We examined the HCP breakthrough by collecting different flowthrough fractions and determining the HCP level in each fraction. As shown in Figure 3b, Mab-T dynamic loading capacity was 0.5 g Mab-T/mL-STIC at 10 ppm HCP breakthrough or 0.9 g Mab-T/mL-STIC at 20 ppm breakthrough, which was higher than that achieved from Q column chromatography in a flowthrough mode (data not shown).

Figure 4

The same screening and optimization procedures using 96-well plates were also applied to three other antibodies, Mab-S, Mab-D and Mab-K. The DLC was determined under the conditions defined by screening and optimization experiments. For Mab-S, an equilibration buffer at pH 7.0 and 9.53 mS/cm, equivalent to 60 mM NaCl, was used in the experiment. Again, different flowthrough fractions were collected and HCP was determined. With increasing load of Mab-S, HCP in the flowthrough remained at a background level up to 2.5 g mAb/mL-STIC (see Figure 4). After that HCP started to increase gradually and reached 10 ppm at 3.0 g mAb/mL-STIC. By contrast, only 0.5 g /mL-STIC DLC was observed for Mab-T. The significant difference in the capacity might be due to the initial HCP level (578 ppm for Mab-T vs. 212 ppm for Mab-S), as well as the initial HMW level (2.0% for Mab-T vs. 0.9% for Mab-S). Similarly, the DLC for Mab-D and Mab-K was 3.5 and 3.7 g-mAb/mL-STIC, respectively (see Table II), under the tailored operating conditions developed on 96-well plates. Mab-D and Mab-K thus showed a reasonably high process capacity on the STIC membrane adsorber.

Table II

Process performance analysis

After the low-pH-treated protein A eluate of Mab-T was neutralized to pH 7.25 and clarified, it was directly applied into 1 mL STIC Nano at a loading capacity of 0.5 g/mL-STIC in flowthrough mode. As expected, 94% of Mab-T was successfully recovered in the flowthrough/wash pool (see Figure 5). The residual HCP was reduced to 10 ppm while aggregates were reduced to 1.02%. Furthermore, levels of residual DNA and leached protein A in the STIC purified mAb-T met the requirement for drug substance.

We also examined whether residual impurities were efficiently removed through the STIC membrane adsorber for Mab-D, Mab-K and Mab-S (see Figure 5). The residual HCP was reduced to less than 10 ppm. The clearance of residual DNA and leached protein A were sufficient to meet product specifications. The STIC thus served as a polishing step for Mab-D, Mab-K and Mab-S with acceptable process performance and product quality.

Figure 5

Salt tolerant nature of STIC membrane adsorber

In traditional ion exchange chromatography, the interaction strength of proteins with resin ligands is controlled by solution pH and NaCl concentration. At a given pH, the interaction decreases dramatically with increasing NaCl concentration. Thus, both product recovery and impurity clearance are functions of pH and NaCl concentration, which were evaluated as the critical operating parameters during the condition screening and optimization in this study. As shown in Figure 1, at a given pH condition, HCP clearance did not vary significantly with NaCl concentration, which highlighted the unique salt tolerant nature of the STIC membrane adsorber. A consistent process performance has been achieved in the tested conditions of pH and NaCl concentration (see Figures 2 and 5), indicating a wide design space for the STIC polishing step.

It should be noted that both Mab-D and Mab-S have solubility issues in the current Q equilibration buffer (conductivity < 5 mS/cm), which posed challenges when the current purification production process and related operating conditions were applied. A low process yield was observed at both pre-Q column TFF and Q column polishing steps. When the solution ionic strength was increased to overcome the solubility issue, the separation efficiency of the Q column chromatography diminished appreciably. In both cases, STIC provided a viable alternative to the Q polishing step, able to remove trace amount of impurities from antibodies at a higher ionic strength buffer condition. This strategy can be applied to other antibodies, exhibiting similar solubility issues as Mab-D and Mab-S.

Transition from 96-well plates to capsules

There are three layers of STIC membrane in 96-well format compared with 15 layers in the Nano capsule, which has the same number as the large-scale STIC devices. An equivalent fold of impurity removal was achieved using Nano capsules compared to 96-well plates with the same process load under identical operating conditions as shown in Figure 6. This might be due to the fact that interactions between antibodies and STIC ligands vary with solution conditions, not with the number of layers of membrane or membrane volume. Parameters collected on the 96-well plate can thus be applied to Nano capsule. However, as reported previously, the antibody loading capacity (or process capacity) is dependent on both solution conditions and membrane volume or number of layers (24). The DLC should be determined from a small STIC membrane capsule such as Nano or Pico, which can be directly applied to larger membrane adsorbers.

Figure 6

Removal of HMW species using STIC membrane adsorber

All four model mAbs were derived from stable CHO cell lines and partially purified using MabSelect SuRe resin under current platform operating conditions. It was reasonable to believe that the host cell protein profiles in these protein A-purified materials were similar. The highest HCP load was, 578 ppm, or 0.578 mg in 1 g of antibody (in the case of Mab-T). In addition, residual DNA and leached protein A only accounted for a small portion of STIC binding capacity (data not shown). The STIC binding sites could not be saturated by residual impurities at a process load of 1 g antibody/mL-STIC. The mechanism behind the lower HCP clearance during Mab-T STIC purification was therefore investigated.

Since HMW species may, through multiple-site attachment, have greater avidity to the AEX resin or membrane adsorber than the monomers, AEX in a flowthrough mode was used for HMW removal as previously reported (12, 17). Practically, the removal efficiency through AEX in a flowthrough mode varies with different antibodies. For some antibodies, HMW can be reduced to a very low level while in other cases HMW removal is not efficient, and in some extreme cases, HMW reduction is not observed at all. Thus HMW removal is challenging and should be evaluated for each case. Additionally, HMW removal with AEX resins may be limited by steric hindrance (25), indicating a potential issue with respect to loading capacity. In the case of membrane chromatography, the mechanism of mass transfer is convective flow. Therefore, the HMW binding capacity on STIC is expected to be much higher.

Figure 7 shows HMW removal from partially purified Mab-T with 1.49% HMW in the load, assessed by SE–HPLC. The HMW in flowthrough was 0.99% with a load of 2.0 g Mab-T/mL-STIC. The amount of HMW bound to STIC in this experiment was 10 mg. However, as expected, HCP removal efficiency decreased slightly as the residual HCP in the flowthrough was 40 ppm. In this case, the HMW species might have stronger interactions with STIC than HCPs. The saturation of binding sites on the membrane adsorber by the HMW species prevented further removal of trace impurities. However, for the three other antibodies tested, either the HMW level in the load was low or only minimal HMW removal was observed, and a much higher process capacity was achieved (based on HCP breakthrough). Thus, caution should be taken if HMW species at an elevated level (>5%) are applied to the membrane chromatography. A competitive binding analysis of HMW species and other trace impurities should be performed. If STIC offers the same or higher clearance of HMW compared to other impurities, process capacity might be compromised. Depending on the scale of purification production, different strategies can be used to mitigate the issue. Membrane chromatography in flowthrough mode with different mechanisms such as hydrophobic interaction can be incorporated. In addition, multiple cycles of STIC operation can be used to provide enough manufacturing capacity.

Figure 7

Prediction on large-scale purification production

Figure 8 presents a mAb purification production scenario using STIC as an alternative polishing step at large scale. In this theoretical case, the starting materials are proteins partially purified using a protein A column from 11,000-L HCCFs at a titer of 5 g/L. The antibody load for STIC is 50 kg. Four cycles of 5-L STIC membrane adsorber operation can provide enough production capacity for Mab-D, Mab-K and Mab-S. Unfortunately, application of STIC to Mab-T in large scale is predicted to be challenging due to its lower process capacity or higher residual HCP level. Incorporation of a HMW mitigation step in the process would be required before being applied to STIC.

Figure 8

The major characteristics of STIC membrane adsorbers are further compared with traditional AEX columns in Table III. A smaller membrane adsorber device can provide required production capacity, and reduce the plant footprint. As a single-use system, the STIC membrane adsorber avoids issues experienced in the packing, unpacking, cleaning, and storage of traditional chromatographic columns. Significant amounts of consumables (e.g., water for injection, buffers, cleaning solutions) are saved and, more importantly, less related labor is required when membrane chromatography is used. In contrast to the major development effort that AEX chromatography requires, process development for STIC membrane adsorber is simple and efficient as demonstrated in the previous sections. Furthermore, the integrity of the membrane adsorber can be assessed using a pre- and post-use filter integrity test protocol, which is straightforward compared to the HETP test used in traditional chromatographic columns. Lastly, because of its unique hydrodynamic characteristics, membrane adsorbers can operate at a much shorter residence time or higher operating flow rate than columns, thus reducing overall processing time and costs. Therefore economic benefits can be achieved using membrane adsorbers for manufacturing of antibodies as described previously (22).

Table III

In summary, STIC provides an alternative to the current AEX polishing step. The two-column production platform can be shortened by removing the pre-AEX TFF or dilution step. It is extremely valuable for antibodies which have solubility issues at low ionic strength conditions. In addition, fast screening and optimization followed by process capacity determination in this article suggests an extremely short development timeline. More importantly STIC can be incorporated into our current platform in a "plug and play" development approach.

CONCLUSION

The application of the STIC membrane adsorber enables an alternative polishing platform for monoclonal antibodies. An operating window of STIC in a flowthrough mode has been established through the use of high throughput screening and optimization on 96-well plates in a relatively short time frame. Acceptable product recovery and efficient clearance of host cell proteins, leached protein A, DNA, and high molecular weight species have been demonstrated on STIC Nano using four model proteins. The possible implication of HMW removal through STIC has been addressed. This simple and efficient polishing step can be integrated into current mAb production platforms.

ACKNOWLEDGMENTS

The authors would like to thank Greg Liebisch of the Cell Expression/Scale-up team for providing harvested cell culture fluid used in this study. The authors also thank David Zhou, Dr. Yujing Yang and Dr. Uwe Gottschalk at Sartorius Stedim Biotech for providing STIC 96-well plates and for helpful discussion.

Yun (Kenneth) Kang* is a principal scientist and head of the Purification Team in BioProcess Sciences; Stanley Ng is a principal research associate in BioProcess Sciences; Julia Lee is a senior research associate in BioProcess Sciences; Josaih Adaelu is a principal research associate in Process Development; Bo Qi is a director in Process Development; Kris Persaud is a director in BioProcess Sciences; Dale Ludwig is a vice president in BioProcess Sciences; and Paul Balderes is a director in BioProcess Sciences, all at ImClone Systems, a wholly-owned subsidiary of Eli Lilly and Company, New York, NY. *To whom correspondence should be addressed, yun.kang@imclone.com.

PEER REVIEWED

Article submitted: Oct. 4, 2011.

Article accepted: Oct. 28 2011.

REFERENCES

1. D. K. Follman and R. L. Fahrner, J. Chromatogr. A 1024 (1,2), 79–85 (2004).

2. A. Groenberg, et al., BioProcess Intl. 5 (1), 48–56 (2007).

3. B.D. Kelley, Biotechnol. Prog. 23 (5), 995–1008 (2007).

4. A.A.Shukla et al., J Chromatogr. B 848 (1), 28–39 (2007).

5. B.D. Kelley, G. Blank, and A. Lee, "Downstream Processing of Monoclonal Antibodies: Current Practices and Future Opportunities," in Process Scale Purification of Antibodies, U. Gottschalk, Ed. (John Wiley & Sons, Inc., Hoboken, NJ, 1st ed., 2009), pp. 1–23.

6. L. Giovannoni, M. Ventani, and U. Gottschalk, BioPharm Intl. 21 (12), 48–52 (2008).

7. J. Glynn et al., supplement to BioPharm Intl. 23 (3), s4–s10 (2010).

8. S. Ghose et al., "Integrated Polishing steps for Monoclonal Antibody Purification," in Process Scale Purification of Antibodies, U. Gottschalk, Ed. (John Wiley & Sons, Inc., Hoboken, NJ, 1st ed., 2009), pp. 145–167.

9. Y. Li, et al., "Development of a Platform Process for the Purification of Therapeutic Monoclonal Antibodies," in Process Scale Purification of Antibodies, U. Gottschalk, Ed. (John Wiley & Sons, Inc., Hoboken, NJ, 1st ed., 2009), pp. 187–201.

10. H. Yang, et al., presentation at 241st ACS National Meeting & Exposition (Anaheim, CA, 2011).

11. R.L. Fahrner, et al., Biotechnol. Genet. Eng. Rev. 18, 301–327 (2001).

12. B.D. Kelley, et al., Biotechnol. Bioeng. 101(3), 553–566 (2008).

13. J. H. Knox and H. M. Pyper, J. Chromatogr. A 361 (1), 1–30 (1986).

14. D.S.M. Strauss, et al., Biotechnol. Bioeng. 104 (2), 371–380 (2009).

15. J.X. Zhou and T. Tressel, supplement to BioProcess Intl . 3 (9), 32–37 (2005).

16. J.X. Zhou, and T. Tressel, Biotechnol. Prog. 22 (2), 341–349 (2006).

17. J.X. Zhou, T. Tressel, and S. Guhan, supplement to BioPharm Intl. 20 (2), s26–s35 (2007).

18. C. Boi, J. Chromatogr. B. 848 (1), 19-27 (2007).

19. R. Faber, Y. Yang, and U. Gottschalk, supplement to BioPharm Intl. 22 (10), s11–s14 (2009).

20. W.T. Riordan, et al., Biotechnol. Prog. 25 (6), 1695–1702.

21. A. Mehta et al., presentation at 241st ACS National Meeting & Exposition (Anaheim, CA, 2011).

22. N. Fraud, et al., BioPharm Intl . 23 (8), 44–52. (2010).

23. S.R. Holmes-Farley et al., J.M.S.–Pure Appl. Chem. A36 (7–8), 1085–1091 (1999).

24. H. L. Knudsen et al., J. Chromatogr. A 907 (1–2), 145–154 (2001).

25. Y. Yao and A.M. Lenhoff, J. Chromatogr. A 1126 (1–2), 107–119 (2006).