AbstractSimulated moving bed chromatography (SMBC) and its variant multicolumn continuous chroma-

tography (MCC) have the potential to elevate the industrial chromatographic platform by conver-sion of conventional batch processes to more efficient continuous processes. The high productivity, recovery and purity achieved by SMBC on an industrial scale for small molecules hold promise for biomolecule manufacture; however, adoption of continuous chromatography in bioprocessing has been relatively slow. There are good reasons for this, including the complexity of the target mol-ecules and overall process, variability between targets, and sterility and regulatory requirements. In addition, there has been a relative paucity of equipment available to perform SMBC or MCC at the research scale.

The Octave System developed by Semba Biosciences is a line of laboratory instruments capable of performing SMBC/MCC protocols on scales from 10 g/day to 500 g/day. This unique system pro-vides the opportunity to rapidly test materials and develop methods in continuous operating modes that correspond to large-scale operations. Flexibility in programming column and flow configura-tions enable optimization of purity, yield, and adsorbent utilization for various separation chemistries including affinity, ion exchange, hydrophobic interaction, and size exclusion.

For this study we purified a humanized IgG1a monoclonal antibody from CHO culture fluid using the Octave System in a continuous Protein A capture (PAC) process and an AKTA System in a stan-dard batch PAC process. We compared the performance of three commercial Protein A adsorbents in continuous (SMB) and single-column (SC) modes.

Bench top continuous chromatography: an enabling platform for bioprocess development

Robert Mierendorf, Alla Zilberman, Bruce Thalley, and Anthony GrabskiSemba Biosciences, Inc., 505 South Rosa Road, Madison, WI 53719 USA www.sembabio.com

*This work was supported in part by an SBIR grant from the US National Cancer Institute of the NIH.

Figure 3. Representative chromatograms from continuous PAC runs.Peaks represent A280 of elutions from successive columns. Each run was performed for at least 4 complete cycles (32 elutions). Purity analysis was performed on samples taken after 3 cycles.

Materials & Methods

Figure 1. OctaveTM Chromatography SystemCapable of performing SMBC/MCC and other continuous chromatography protocols• Runs up to 8 columns, up to 8 pumps• Proprietary valve block design; 72 two-way valves; low dead volume; non-metallic flow path• Scalable from 12 ml/min to 300 ml/min flow rates; grams to kilograms per run•

8 X 1-ml Protein A Capture columns

Results & Discussion

Protocol

Figure 2. Design of the continuous process

1. Equilibrate 10 CV PBS2. Bind (Feed) 80-100% static capacity at various res times3. Wash 10 CV 25 mM Na phosphate, 1 M NaCl, pH 6.74. Elute 5 CV 100 mM citrate, pH 2.5 or 3.55. Clean 4 CV 100 mM NaOH, 1 M NaCl

Octave inlet and outlet assignments

Column configuration

Octave program

Script Step 1A

Script Step 1B

AnalyticsmAb concentrations in culture fluid and process fractions were determined using a 2.1 x 30 mm • POROS® PA ImmunoDetection® Sensor Cartridge (Life Technologies) on an Agilent 1100 HPLC system.Leached Protein A and host cell protein (HCP) were measured by ELISA using commercial kits (CYGNUS • Technologies). Host cell DNA was measured by qPCR (assays performed by WuXi AppTec). Structural integrity was determined by analytical size exclusion chromatography using a TSK-GEL • G3000SWXL 7.8 mm x 30 cm column (Tosoh Bioscience) on an Agilent 1100 HPLC system.Static binding capacities were determined by loading 1-ml columns at a linear velocity of 57 cm/h (3 min • residence time) with filtered CHO cell culture fluid containing 2.37 mg/ml mAb. Loading continued until the mAb concentration in flow through equaled the concentration in the starting material. The columns were washed, eluted, and mAb in eluted fractions quantified. Results: MabSelect SuReTM (GE Healthcare); 62 mg/ml; Toyopearl® AF-rProtein A (Tosoh Bioscience), 47 mg/ml; POROS MabCaptureTM A (Life Technolo-gies), 33 mg/ml.

MabSelect SuRe Toyopearl AF-rProtein A POROS MabCapture A

Table 1. Results of SC and SMB runs.

Purity. mAb purity was equivalent or higher with the continuous SMB-PAC process vs. SC-PAC with all three resins. HCP and DNA contamination were significantly lower with the SMB-PAC pro-cess on MabSelect SuRe and ToyoPearl AF-rProtein A resins.

Recovery. Compared with the corresponding SC-PAC process, eluted fraction yields were 5-7% lower with SMB-PAC on MabSelect SuRe, 15-20% lower with SMB-PAC on Toyopearl AF-rProtein A, and roughly the same on POROS MabCapture A. Measurements of mAb recovered in flow-through, wash, and elution steps showed that the differences in yield we observed between SMB-PAC and SC-PAC, especially for Toyopearl AF-rProtein A resin, were primarily due to losses during the wash step. To provide consistency for comparisons we performed this work using identical buffer condi-tions for all of the PAC resins. It is likely that recoveries can be improved by optimizing the wash and/or elution conditions for each adsorbent when operated in continuous mode.

Productivity. Productivities were in the range expected at the mAb concentration (2.37 g/L) used in our experiments. Figure 4 shows a comparison of productivity predicted for SMB and SC meth-ods at increasing antibody titers, assuming 100% yield. Calculations are based on the process pro-tocols and binding capacities used and obtained in this study. The plateaus indicated for MabSelect SuRe and Toyopearl resins reflect the points at which maximum linear flow rates recommended by the manufacturer are reached.

Pro

duct

ivity

(g/L

-res

in/h

)

mAb titer (g/L)

MabSelect SuRe Toyopearl AF-rProtein A POROS MabCapture A

Titer in these experiments

Figure 4. Predicted PAC productivities of SC and SMB processes with increasing mAb titer.

Table 2. Observed vs. predicted PAC productivities.

Table 2 shows the productivity data obtained from the SC-PAC vs. SMB-PAC performed with loading columns to 80% of their measured static binding capacities. Productivities of the SMB vs. SC process with MabSelect SuRe and POROS MabCapture A were within 3% of the predicted val-ues, whereas SMB productivity was 21% less than that predicted for the Toyopearl AF-rProtein A resin. We believe that this shortfall in productivity was most likely due to the significant difference in yield between SC and SMB processes for this resin.

Conclusions & Future DirectionsThis study has shown that the mAb produced from a continuous PAC process performed on the

Octave platform was equivalent or superior in quality to that produced by the analogous batch pro-cess. Productivity and recovery will be optimized in future studies using higher titer mAb prepara-tions.

As mAb titer increases, productivity is predicted to dramatically increase with SMB vs. SC pro-cesses (see Fig. 4). This is because unlike SC, SMB is primarily a volume-driven process. As titers increase, switch times decrease and flow rates increase. To take full advantage of the potential productivity gains afforded by SMB-PAC for processing high titer mAb, per column residence times must be decreased while maintaining efficient capture. To effectively increase residence time in SMB-PAC, multiple columns can be allocated to the capture zone. In this study we employed 3 col-umns in the capture zone and per column residence times of 0.4-0.5 min for MabSelect SuRe and Toyopearl AF-rProtein A resins. Negligible losses in binding capacity occurred at these residence times for either resin, which is in agreement with results reported by Merck Serono (J. Sutter, 11th Annual Biological Production Forum 2012). The POROS MabCapture A resin utilizes a macropo-rous backbone fundamentally distinct from the other two more conventional resins, and has the advantage of retaining its binding capacity at very high flow rates (9-10-fold higher than the other resins). In this study we operated the SC-PAC and SMB-PAC processes with the POROS resin at residence times of 0.25 and 0.17 min, respectively, with no apparent difference in capture efficiency. Further experiments are needed to determine the practical limits of residence time on the productiv-ity of SMB-PAC.

In addition to the 5-step protocol employed in these studies, the Octave System has been con-figured to run other PAC protocols including as many as 9 steps using 7 pumps (data not shown). Future studies will also investigate alternative adsorbents, column configurations, and protocols.

MabSelect SuRe Toyopearl AF-rProtein A POROS MabCapture ACharacteristic SC SMB SMB SMB SC SMB SMB SMB SC SMBProductivity (g/L-resin/h) 32.4 32.7 39.8 38.7 31.4 25.4 37.0 33.6 89.3 87.4DNA (pg/mg protein) 58.6 42.2 25.6 57.4 523.7 70.8 127.4 100 44.4 65.9HCP (ppm) 150.5 141.4 102.5 90.8 140.7 50.4 88.4 83.1 111.8 117.5Protein A (ppm) 1.8 2.3 1.7 1.6 7.7 1.9 16.5 3.7 10.6 9.8Monomer (%) 98.1 98.6 98.4 98.4 98.5 99.3 99.3 99.4 99.2 96.3Total Recovery (%) 98 99 95 100 108 91 90 97 105 93

Elute (%) 93 88 88 86 94 75 79 74 85 83FT (%) 0 0 0 7 8 5 2 13 14 1Wash (%) 5 11 7 7 6 11 9 10 6 9

Res. Time (min/column) 2.0 0.5 0.4 0.4 2.0 0.5 0.4 0.4 0.25 0.17Flow Rate (cm/h) 86 343 429 429 86 343 429 429 706 1,020mAb Loaded (mg/ml resin) 49 48 50 62 40 42 39 47 25 33

Resin Process Res. Time (min/col)

Predicted prod.

Observed prod.

Efficiency (Obs vs. Pred)

Predicted SMB:SC

Observed SMB:SC

MabSelect SuReSC 2 34.6 32.4 93.6%

1.26 1.23SMB 0.4 43.6 39.8 91.3%

Toyopearl AF-rProtein ASC 2 31.4 31.4 100.0%

1.39 1.18SMB 0.4 43.6 37 84.9%

POROS MabCapture ASC 0.25 104 89.3 85.9%

1.00 0.98SMB 0.17 104 87.4 84.0%