Polishing Complex Therapeutic Proteins

ABSTRACT

Sartobind STIC (salt-tolerant interaction chromatography, Sartorius Stedim Biotech, Goettingen, Germany), a salt-tolerant anion-exchange membrane adsorber, has demonstrated proof-of-concept in removing residual host-cell impurities from high-salt, packed-bed affinity chromatography eluate. Although the new platform process using Sartobind STIC has fewer unit operations, it produces drug substance with comparable quality attributes to current processes, thus significantly improving productivity and reducing cost of goods. The study presented herein focuses on implementing a novel membrane adsorber for optimized polishing.

ISTOCKPHOTO/THINKSTOCK SOURCE

Packed-bed chromatography is the main workhorse for the downstream processing of therapeutic proteins and monoclonal antibodies (mAbs). Packed-bed columns provide good binding capacity and scalability, combined with excellent resolution. The mass transfer process in packed-bed chromatography comprises several steps, including convection, pore diffusion, and film diffusion. The rate-limiting step of this process is pore diffusion (i.e., the slow diffusion of solutes into the dead-ended pores inside the chromatography media where the majority of the binding sites are located). As a result, residence time is an important parameter for column chromatography and often becomes the limiting factor of how fast the process can be run. High back pressure is another concern when operating packed-bed columns at a high flow rate. Membrane-adsorber (MA) chromatography technology was developed to overcome this mass transfer limitation (1). By coupling functional groups onto a filter-like porous matrix, diffusion-based mass transfer is eliminated because the filter pores are flow-through pores and allow solutes to be transported to the binding sites via convection. With MA chromatography, only film diffusion may limit mass transfer rate and the binding capacity is generally independent of the load flow rate over a wide range (2,3). Importantly, MAs can be manufactured with extremely shallow matrix beds (e.g., bed-heights in the mm range) that have very large cross-sectional area-to-volume ratio. Thus, MAs can be operated at much shorter residence time compared with packed-bed columns (e.g., seconds vs. minutes), thereby reducing the process time and increasing throughput at a large scale (4).

It has been noted for some time that an anion-exchange MA is an attractive alternative to anion-exchange columns when operated in flow-through mode to remove low levels of impurities, such as DNA, host-cell protein (HCP), and virus (5). Flow-through polishing columns are usually sized for speed to achieve desired flow rate and process time, using only a small fraction of the binding capacity available for impurity removal. Because MAs allow for a faster flow rate, a small MA device can replace a bigger column and still provide sufficient binding capacity for impurity clearance (6–8). Today, membrane chromatography has proven to be a robust alternative to Q column chromatography for polishing in flow-through mode, and multiple case studies have demonstrated the popularity of their implementation (9–13). Single-use MAs not only reduce process time, buffer usage, and floor space, but also eliminate the column packing and cleaning validation activities required for packed-bed columns. A detailed cost analysis showed that single-use Q membranes can be cost competitive compared with a reused Q Sepharose fast-flow column in a mAb process when its process capacity is sufficiently high (7). More recent analyses also show that using a disposable MA in flow-through mode provides comparable or a lower cost of goods (CoG) than using a packed-bed column (14, 15). Overall, for a flow-through polishing step, replacing a packed-bed column with a single-use MA can provide cost savings.

As with any other ion-exchange chromatography, conductivity and pH have significant effect on the performance of anion-exchange MA as a flow-through polishing unit operation. One report stated that to ensure sufficient impurity clearance, the ideal range for Sartobind Q flow-through step in a mAb process is 3–4 mS/cm at pH 7.0–7.2 (13). Similarly, another study using Q MA from Millipore found that to achieve > 1 log removal of host cell protein, the load had to be conditioned to pH 8.0 and a conductivity of < 4.0 mS/cm (6). The low salt tolerance of these Q MAs means that a dilution step prior to loading is often required to achieve desired impurity clearance, which increases process complexity. In recent years, efforts were made, both in academia and industry, to develop new types of anion-exchange MAs that have better salt tolerance, which will enable greater process flexibility and potentially lead to wider usage of MA flow-through polishing in the downstream processes. A systematic screening study by Riordan et al. identified three factors that contributed to salt tolerance of anion-exchange MA: ligand net charge, ligand density, and molecular structure of the ligand (16). Interestingly, the study also found that available hydrogens on the amine-binding group improved the salt tolerance of the ligand, indicating that primary amines might have better salt tolerance than quaternary amines.

Sartobind STIC (salt tolerant interaction chromatography, Sartorius Stedim Biotech, Goettingen, Germany) is a weak anion-exchange MA that is less sensitive to increasing salt concentration than standard Q membranes (17). It carries the polyallylamine ligand that provides high charge density and salt tolerance. The new double-porous membrane replaced the previous generation of membrane with hydrogel, which was shown to shrink and reduce binding site accessibility under high salt conditions (18). Sartobind STIC was shown to provide significantly higher binding capacity and higher LRV of model viruses compared with Sartobind Q in the presence of 150 mM NaCl (16.8 mS/cm) (17). This enhanced salt tolerance allows the MA polishing step to be conducted without load dilution, thus reducing process time and complexity.

This article describes the development work at Bayer to evaluate Sartobind STIC as a platform polishing unit operation for complex protein therapeutics. Specifically, the authors looked at product yield, and HCP and DNA removal by Sartobind STIC from very high salt intermediates, and how to improve the overall purification platform. Data suggest that using a salt tolerant MA enables the number of unit operations in a platform process to be reduced, potentially reducing cost of goods.

MATERIALS AND METHODS

Sartobind Q and Sartobind STIC membrane adsorbers in LP15 (0.41 mL) and nano (1 mL) formats were provided by Sartorius Stedim (Goettingen, Germany). The ligand for Sartobind STIC is a polyallylamine compared to the quaternary amine for Sartobind Q. The LP15 prototype is a 0.41 mL disk format device with three membrane layers in a polysulfone housing. The design of the LP15 device is similar to the commercially available syringe unit Sartobind MA15. Sartobind nano is the commercially available scale-down device with 15 layers and 36.4 cm² total surface area. Scalability of Sartobind nano (1 mL) to process scale capsules has been well demonstrated through the entire range of product up to 1.62 L membrane. Cylindrical format, radial flow distribution, and bed height were kept constant to allow for linear scale up.

The recombinant protein feed materials for this study were obtained from clinical manufacturing at Bayer Berkeley, CA. The immunoaffinity columns used were directly scaled down from clinical manufacturing processes.

As in current processes, all purifications were carried out at 2–8 °C to obtain maximum product stability. Step gradient experiments and breakthrough experiments were carried out on an AKTA Explorer 100 chromatography system from GE HealthCare (Uppsala, Sweden). Simple flow-through runs were carried out using a Watson Marlow 101U peristaltic pump (Falmouth, England). Initial development studies used Sartobind STIC LP15 devices at a flow rate of 3 mL/min (7.3 MV/min). Sartobind STIC nano was used for laboratory scale purifications of two different complex glycoproteins, Bay-A001 and Bay-A002, at a flow rate of approximately 5 mL/min (5 MV/min). The full factorial study on pH and conductivity was designed and data analyzed in JMP (SAS, Cary, NC). All buffers used in the anion exchange MA operation were imidazole based. No multivalent anion, such as phosphate or citrate, was used as it may interfere with protein adsorption because of its strong interaction with AEX ligands.

Bay-A001 and Bay-A002 have theoretical isoelectric points (pI) of 7.4 and 6.4, respectively, based on the amino acid sequences. However, due to the heterogeneity of post-translational modifications, the recombinant proteins exhibit high levels of heterogeneity in pI, which prevented pI determination using isoelectric focusing (IEF) electrophoresis.

All assays were performed by the analytical development and support group at Bayer, Berkeley, CA.

RESULTS

Current platform purification process for Bayer recombinant protein therapeutics

To reduce the cost of development and the time to clinic, the downstream process development team at Bayer recently established a platform purification process for Bayer's complex recombinant protein therapeutics (see Figure 1b). An older generation purification process is also included in Figure 1 for comparison (see Figure 1a). Not only does the current process have fewer unit operations, but it also uses more modern separation technologies, resulting in significantly higher yield and shorter process times, thus reducing the cost of goods.

Two of the unit operations in the current platform process use MA technology. A Q-MA capture step operating in bind-and-elute mode was developed for quick isolation of unstable, low concentration protein products from large volumes of cell-culture harvest generated by perfusion-based bioreactors. The fast flow property of MA allowed us to quickly concentrate and stabilize the products, which will otherwise gradually lose activity in the crude harvest.

Following the MA capture step, an immunoaffinity column was used to provide the majority of purification power of the entire process. Because of the products' sensitivity to non-neutral pH, elution from the immunoaffinity column was achieved with a high concentration of chaotropic salt instead of a pH change. As a result, the eluate was of very high conductivity and a dilution step, in some cases more than 10-fold, was needed before proceeding to the next unit operation. To further remove trace amount of impurities, one or two polishing columns were included after immunoaffinity. The second Q-MA step was a flow-through step designed specifically for removal of residual DNA. To ensure good product recovery, the MA polishing step operated under fairly high salt concentration (> 20 mS/cm at 5 °C), under which no HCP or viral clearance was observed. A viral filtration step provided robust non-enveloped virus clearance and enhanced the pathogen safety profiles of the products. Lastly, a ultrafiltration/diafiltration (UF/DF) step concentrated and formulated the drug substance for frozen storage.

The salt tolerance of Sartobind STIC may provide the opportunity to further improve the process by potentially adding HCP and/or viral clearance capability from high-salt feed streams, reduce the need for dilution, and reduce the number of unit operations.

Sartobind STIC has greater binding strength than Sartobind Q

The first step was evaluating the level of salt tolerance of Sartobind STIC in comparison with Sartobind Q (see Figure 2a). Purified Bay-A001, a Bayer recombinant protein product, was loaded to either Sartobind Q or Sartobind STIC under neutral pH and low salt concentration so that the protein binds to the membranes. A step-wise NaCl gradient wash was conducted to determine the NaCl concentration necessary to elute Bay-A001 from each MA. Some split peaks and conductivity fluctuations were observed, which were probably caused by nonideal flow distribution inside the LP15 membrane holder. Running the same gradient through the bypass line did not produce any conductivity fluctuation (data not shown). This flow distribution issue, however, did not affect the interpretation of the results. The elution of Bay-A001 from Sartobind Q started in the 0.2 M NaCl fraction and continued in the 0.3 and 0.4 M NaCl fractions. This product was inherently heterogeneous and the elution into multiple fractions as shown in the study was consistent with what was seen before. In comparison, no significant elution was observed from Sartobind STIC at NaCl concentrations up to 0.6 M. An elution peak was observed in the last fraction with 1 M NaCl, although the integrated peak area was significantly smaller than the total integrated peak areas from the Sartobind Q chromatogram. Because the same amount of protein was loaded to each MA, the data indicated that not all protein was eluted from Sartobind STIC at 1 M NaCl. Overall, this experiment demonstrated the salt tolerance of Sartobind STIC. It also showed that a high salt concentration was needed to ensure good product recovery for processing proteins, such as Bay-A001, in a flow-through mode using Sartobind STIC.

Figure 1: Schematic comparison of three generations of purification processes for non-mAb recombinant therapeutic proteins at Bayer. a) Older generation purification process. b) Current platform purification process. c) Future platform purification process (i.e., STIC process). A is affinity, FR is blast freeze, IA is immunoaffinity, IE is ion exchange, MA is membrane adsorber, VI is viral inactivation, VF is viral filtration, UF is ultrafiltration, and DF is diafiltration. (ALL FIGURES ARE COURTESY OF THE AUTHORS)

Sartobind STIC as a polishing step for high-salt immunoaffinity eluate

The possible implementation of Sartobind STIC into the platform purification process was evaluated. One option was to use Sartobind STIC to replace the current Q MA polishing step, which served solely as a DNA removal step without the requirement for any feed stream adjustment. The drawback, however, was that a salt addition into the feed stream would be needed to ensure good recovery from Sartobind STIC. This added operation contradicted the goal of process improvement. The only step in the process where the salt concentration was high enough for Sartobind STIC flow-through operation was the immunoaffinity eluate. It was therefore decided to use Sartobind STIC to polish the immunoaffinity eluate in flow-through mode to remove residual DNA and HCP. The salt tolerance feature of Sartobind STIC meant that less dilution of the immunoaffinity eluate was needed. The impurity clearance performance by Sartobind STIC would decide whether any other polishing steps were required.

Figure 2: a) Step-wise salt elution of Bay-A001 from Sartobind Q and Sartobind STIC. Brown line is conductivity trace, red line is A280 trace from Sartobind Q, and blue line is A280 trace from Sartobind STIC. Numbers at the bottom are molar concentrations of NaCl in different fractions. b) Step-wise elution of Bay-A001 from Sartobind STIC using increasing percentage of immunoaffinity elution buffer. Brown line is conductivity trace and blue line is A280 trace. Numbers at bottom are percentage points of immunoaffinity elution buffer in different fractions.

An important parameter for the proposed Sartobind STIC operation was the maximum dilution on the immunoaffinity eluate, which should give a salt condition high enough to give good product recovery, while maximizing impurity removal. To determine the dilution target, purified Bay-A001 was loaded to a Sartobind STIC LP15 device at low salt concentration. The membrane was then washed with a 0–50% step-wise gradient of the immunoaffinity elution buffer (see Figure 2b). The chromatogram showed that with 30% immunoaffinity elution buffer, almost all the proteins were eluted. The conductivity of the 30% immuno-affinity elution buffer solution was measured to be 39 mS/cm at 5 °C.

Figure 3: Dynamic binding capacity of host cell impurities by Sartobind STIC. a) Host cell proteins breakthrough curve with host cell protein (HCP) spike-in. b) DNA breakthrough curve with representative immunoaffinity eluate.

The authors then investigated whether Sartobind STIC could clear DNA and HCP under the same buffer conditions. HCP spike-in for HCP clearance evaluation was chosen because the low HCP level in the immunoaffinity eluate and its high salt content prevented the authors from getting reliable assay results. A small aliquot of immunoaffinity load material, which had HCP as the major protein content, was spiked into a diluted immunoaffinity elution buffer at 39 mS/cm at 5 °C. The solution was loaded to a Sartobind STIC LP 15 device and flow-through fractions were collected. Interestingly, the HCP assay showed an immediate 5–7% breakthrough (see Figure 3a), which remained steady at HCP load densities up to 800 µg/mL membrane. Because HCP consisted of a mix of proteins, it was likely that some of the more basic proteins did not bind to Sartobind STIC under the testing conditions, causing immediate breakthrough. Nevertheless, HCP clearance observed in this experiment was significant and possibly sufficient to reduce HCP to a level within the acceptable range for the Bay-A001 drug substance. For the evaluation of DNA clearance, Bay-A001 immunoaffinity eluate was diluted to 39 mS/cm at 5 °C and loaded to a Sartobind STIC LP15 device. Flow-through fractions were collected and the DNA content in load and flow-through fractions was analyzed by quantitative polymerase chain reaction (qPCR). As shown in Figure 3b, no DNA can be detected in the flow-through at DNA load densities up to 45 µg/mL membrane volume. The DNA load density was limited by the availability of feed material and the maximum DNA binding capacity was expected to be much higher. A DNA binding capacity of 24 g/L for Sartobind STIC at 16.7 mS/cm was previously reported (17). Overall, it appeared that by diluting Bay-A001 immunoaffinity eluate to 39 mS/cm at 5 °C, Sartobind STIC in flow-through mode could be used to remove DNA and HCP with good product recovery. Although the maximum impurity load densities achieved in these studies were not high, they still represented good process throughput because the impurity concentrations in immunoaffinity eluate were very low.

Table I: Process parameters of Sartobind STIC unit operation in a laboratory-scale STIC process for Bay-A001.

With its DNA and HCP clearance capability from immunoaffinity eluate, Sartobind STIC has the potential to replace polishing columns and the Q MA polishing step in the platform process. For the new platform process, which was tentatively called STIC process, there are only three chromatography unit operations: a Q MA capturing step, an immunoaffinity column, and a Sartobind STIC flow-through polishing step (see Figure 1c). To determine whether STIC process could produce Bay-A001 drug substance comparable to the current platform process, three laboratory-scale purification trains using STIC process were performed. Viral filtration was not performed in the laboratory-scale runs. Based on previous experience, viral filtration can reduce HCP and aggregates in Bay-A001 by approximately two-fold. The process parameters for the laboratory-scale Sartobind STIC step are listed in Table I. Table II compares the performance of laboratory-scale STIC process with that of the current platform process at manufacturing scale. Sartobind STIC has an average step yield of 93%. The combined yield of the three polishing steps that Sartobind STIC replaced is 92%. Thus, no difference in overall yield was expected between the two processes.

Table II: Critical quality attributes of drug substances (DS) from Bay-A001 laboratory-scale STIC process are comparable to those from current process at manufacturing scale. N.D. = not detected; NA = not available. SEC–HPLC is size-exclusion chromatography–high-performance liquid chromatography, SDS-PAGE is sodium dodecyl sulfate polyacrylamide gel electrophoresis.

Critical quality attributes of the drug substances from the two processes were also compared. HCP levels from STIC process are low and comparable to those from the current process, even without the additional clearance from viral filtration. Two STIC process runs had DS DNA levels below detection, as is the case with the current process. The other STIC process run had DS DNA at 2.8 pg/dose, significantly lower than the specification for this product and regulatory guidelines. Aggregates were about two-fold higher in DS from STIC process compared with the current process. However, aggregate levels were expected to be comparable if viral filtration is included in the STIC process based on our experience that the viral filter provides about two-fold reduction in aggregates. Degradation product levels tested slightly higher in STIC run 2 and 3, but were well within acceptable range and should not be a quality concern. Purities as determined by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) Coomassie blue staining were comparable between the two processes. Specific activity was somewhat lower with STIC process, but still within the acceptable range. Data analysis showed that the lower specific activities were correlated to lower step yields at the laboratory scale UF/DF and not directly caused by STIC, which could be an equipment-specific issue. More extensive comparability studies at a larger purification scale following this proof-of-concept will show whether this difference is consistently obtained. The immunoaffinity column does not have significant ligand leaching based on previous experience. Thus, the IgG clearance capability of Sartobind STIC was not characterized. For both processes, the levels of mouse IgG in drug substance were below the assay's limit of detection. Figure 4 shows the SDS-PAGE analysis, including Coomassie blue staining, silver staining, and Western blotting, of STIC process drug substance in comparison with Bay-A001 reference standard. No unknown band or significant change in band pattern was observed in STIC process drug substance. Overall, the STIC process reduced the number of unit operation by two compared with the current platform process, without sacrificing yield or product quality.

Figure 4: Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis of Bay-A001 drug substances generated from laboratory-scale STIC process. a) Commassie blue staining. From left to right: Lane 1: Bay-A001 reference standard, Lane 2: lab scale run 1, Lane 3: lab scale run 2, Lane 4: lab scale run 2, Lane 5: lab scale run 3, Lane 6: lab scale run 3, Lane 7: Bay-A001 reference standard. b) Silver staining, left lane: Bay-A001 reference standard, right lane: laboratory-scale run 1,. c) Western blot, left lane: Bay-A001 reference standard, right lane: laboratory-scale run 1.

pH and conductivity are critical parameters for Sartobind STIC operation

To evaluate how pH and conductivity variations may affect Sartobind STIC performance, a full factorial screening DOE study was designed in JMP with a pH range of 6.8–7.4, and a conductivity range of 36–42 mS/cm at 5 °C. Aliquots from Bay-A001 immunoaffinity eluate were adjusted to pH and conductivity targets, as listed in Table III. Each aliquot was then loaded to a Sartobind STIC LP15 device. Step yield, DNA, and HCP clearances from each run were also listed in Table III. A contour plot showing trends of yield and HCP clearance in response to pH and conductivity changes was generated in JMP (see Figure 5). As expected from any anion-exchange flow-through operation, increasing pH decreased product yield but increased HCP clearance, while increasing conductivity increased product yield but decreased HCP clearance. Robust DNA clearance was observed because DNA levels in STIC FT from all four runs were either at or below the limit of detection. Careful control of pH and conductivity is, therefore, crucial for ensuring robust performance of Sartobind STIC.

Table III: Full factorial design of experiments (DOE) screening study shows the effect of pH and conductivity on yield and host-cell protein (HCP) clearance. N.D. = not detected.

The performance of Sartobind STIC with Bay-A002

To demonstrate that Sartobind STIC can be a platform unit operation, performance in processing another Bayer recombinant protein product, Bay-A002, was tested. As outlined in the platform purification process (see Figure 1b), Bay-A002 is also captured from cell-culture harvest using a large-scale Q MA in bind-and-elute mode, followed by purification using an immunoaffinity column. An ion-exchange column and a Q MA flow-through step were used to further polish the product. The process was tested using Sartobind STIC in flow-through mode to polish the high salt immunoaffinity eluate, which was diluted to a conductivity of 39 mS/cm at 5 °C for STIC loading. A laboratory-scale purification run showed that Sartobind STIC gave an excellent yield of 96%. It reduced HCP to 0.1 µg/dose, a six-fold reduction, and reduced DNA to below the limit of detection. These HCP and DNA levels were comparable to those in the MA flow-through in the current process, indicating that Sartobind STIC can replace both the ion-exchange column and the Q MA steps (see Figure 6). Because anion-exchange is a versatile purification technique, the authors believe Sartobind STIC can be easily adapted to processing various proteins by finding the optimal pH and conductivity settings for each protein, thus making it a true platform technology.

Figure 5: Contour plot of pH and conductivity on Sartobind STIC yield and host-cell protein (HCP) clearance. Red lines and numbers represent predicted yield, and green lines and numbers represent predicted HCP fold clearance. The data used to generate this contour plot are listed in Table III. DNA levels in Sartobind STIC FT are either below or close to the limit of detection of 2.5 pg/mL, so no contour plot was generated for DNA clearance.

DISCUSSION

In this study, proof-of-concept for using Sartobind STIC as a platform-polishing unit operation was achieved. When operated in flow-through mode, Sartobind STIC is capable of removing HCP and DNA from high-salt feed streams with good product recovery. A new purification process incorporating Sartobind STIC has fewer unit operations than the current platform process, but produces drug substance with similar yield and comparable quality attributes. Sartobind STIC allows a further streamlined, future manufacturing platform for complex recombinant proteins. This new platform will have only three chromatography unit operations: a reusable Q MA capturing step, an affinity column providing the majority of purification, and a single-use Sartobind STIC polishing step. This one-column-two-MA platform process is well suited for purifying low titer, unstable, complex proteins, which are an important part of Bayer's biologics pipeline. With fewer unit operations and a single-use polishing step, the new platform process is expected to reduce process time, increase productivity, and reduce the cost of goods.

Figure 6: Sartobind STIC was tested to replace two unit operations in Bay-A002 purification process. The step yield is 96%. The host cell protein (HCP) in STIC FT is 0.1 µg/dose, a 6-fold reduction from STIC load and comparable to the HCP level in current process. The DNA level in STIC FT is below the limit of detection.

Q MA in flow-through mode also provides viral clearance in mAb processes (19). The conductivity for the process, however, has to be low (3–4 mS/cm) to prevent viral particles from breaking through the membrane. Low conductivity is often feasible with mAbs because many mAbs have high pI and will not bind to an anion-exchanger, even at low conductivity. XMuLV and PPV clearance by Sartobind STIC was tested under the process conditions described in this paper. No significant clearance was observed at 30–39 mS/cm at 5 °C. Because of the strong binding of the products to Sartobind STIC, the conductivity cannot be lowered further to achieve viral clearance without severely affecting product yield. The lack of viral clearance, however, does not disqualify Sartobind STIC as a platform polishing step. For example, no viral clearance claim was made on any of the polishing steps in the Bay-A001 process; hence, replacing those polishing steps with Sartobind STIC as shown in Figure 1 has no effect on the viral clearance claims of the process. With proteins that bind less strongly to anion-exchangers, such as mAbs, Sartobind STIC could provide significant viral clearance at conductivity settings higher than what would be required for a Q MA step. This could be a significant advantage that may eliminate the need for a dilution step.

The authors will continue to evaluate Sartobind STIC for processing new protein therapeutics, including mAbs, in Bayer's pipeline and will also seek opportunities to test the new platform process at the pilot scale to demonstrate its scalability. It could also be beneficial to evaluate other salt tolerant anion exchange MAs, such as ChromaSorb from EMD Millipore, for the same application. These new-generation membrane adsorbers show how new ligand chemistry and new matrix structure can lead to improved separation performance, which is not achievable with older generation chromatography media. This type of technological innovation allows continuous improvement of platform manufacturing processes, as demonstrated in this study.

ACKNOWLEDGEMENTS

The authors thank the Bayer Berkeley GBD Analytical Development group for performing all the assays.

Min Lin*, PhD, is a staff development scientist, Ashley Hesslein, PhD, is a senior staff development scientist, and Jens H. Vogel, PhD, is director of isolation and purification, all at Global Biological Development, Bayer HealthCare, Berkeley, CA; Nathalie Frau, PhD, is a senior scientist, R&D Process Technologies, at Sartorius Stedim North America, Bohemia, NY; and Rene Faber is vice-president, R&D Process Technologies, at Sartorius Stedim Biotech GmbH, Goettingen, Germany.

*To whom correspondence should be addressed, min.lin@bayer.com.

REFERENCES

1. S. Brandt et al., Nat. Biotechnol. 6, 779–782 (1988).

2. J. Thommes and M.R. Kula, Biotechnol. Progress 11, 357–367 (1995).

3. R. Ghosh, J. Chromatogr. A 952, 13–27 (2002).

4. J.H. Vogel et al., Biotechnol. Bioeng. 109 (12) 3049–3058 (2012).

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

6. M. Phillips et al., J. Chromatogr. A 1078, 74–82 (2005).

7. J.X. Zhou and T. Tressel, Biotechnol. Progress 22, 341–349 (2006).

8. N. Fraud, Bioprocessing J. 7 (2), 26–29 (2008).

9. A. Arunakumari, J. Wang, and G. Ferreira, "New Standards in Virus and Contaminant Removal" supplement to BioPharm Int. 20, s6–s10 (2007).

10. A. Clutterbuck, J. Kenworthy, and J. Liddell J, "New Standards in Virus and Contaminant Removal" supplement to BioPharm Int. 20, s11–s14 (2007).

11. D. Farb, presentation at Downstream Technology Forum (King of Prussia, Pennsylvania, 2006).

12. M. Kuczewski et al., Biotechnol. Bioeng. 105 (2), 296–305 (2010).

13. J.X. Zhou et al., J. Chromatogr. A 1134, 66–73 (2006).

14. N. Fraud et al., BioPharm Int. 23 (8), 44–52 (2010).

15. J.A.C. Lim et al., BioProcess Int. 5, 60–64 (2007).

16. W. Riordan et al., Biotechnol. Bioeng. 103 (5), 920–929 (2009)

17. R. Faber, Y. Yang, and U. Gottschalk,, Supplement to BioPharm Int. 22, 11-14 (2009).

18. I. Tatarova et al., J. Chromatogr. A 1216, 941–947 (2009).

19. J.X. Zhou et al., Biotechnol. Bioeng. 100 (3), 488–496 (2008).