Immobilized metal affinity chromatography optimization for poly-histidine tagged proteins

Immobilized metal affinity chromatography (IMAC) is a technique primarily used in research and development laboratories to purify proteins containing engineered histidine tags. Although this type of chromatography is commonly used, it can be problematic as differing combinations of resins and metal chelators can result in highly variable chromatographic performance and product quality results.

Journal of Chromatography A 1629 (2020) 461505 Contents lists available at ScienceDirect Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma Immobilized metal affinity chromatography optimization for poly-histidine tagged proteins Valeria Riguero a,∗, Robert Clifford a, Michael Dawley e, Matthew Dickson d, Benjamin Gastfriend f, Christopher Thompson c, Sheau-Chiann Wang b, Ellen O’Connor a,∗ a Purification Process Sciences, AstraZeneca, One MedImmune Way, Gaithersburg, MD 20878, USA Analytical Biotechnology, AstraZeneca, One MedImmune Way, Gaithersburg, MD 20878, USA Data Science and Modelling, AstraZeneca, One MedImmune Way, Gaithersburg, MD, 20878, USA d Texcell North America, 4991 New Design Road, Suite 100, Frederick, MD, 21703, USA e Quality Engineering and Validation, Genentech, 1000 New Horizons Way, Vacaville, CA, 95688, USA f Department of Chemical and Biological Engineering, University of Wisconsin-Madison, Madison, WI, USA b c a r t i c l e i n f o Article history: Received 15 July 2020 Revised 18 August 2020 Accepted 20 August 2020 Available online 21 August 2020 Keywords: Poly-histidine tag Protein purification IMAC Resin/metal screening Process optimization a b s t r a c t Immobilized metal affinity chromatography (IMAC) is a technique primarily used in research and development laboratories to purify proteins containing engineered histidine tags Although this type of chromatography is commonly used, it can be problematic as differing combinations of resins and metal chelators can result in highly variable chromatographic performance and product quality results To generate a robust IMAC purification process, the binding differences of resin and metal chelator combinations were studied by generating breakthrough curves with a poly-histidine tagged bispecific protein The optimal binding combination was statistically analyzed to determine the impact of chromatographic parameters on the operation Additionally, equilibrium uptake isotherms were created to further elucidate the impact of chromatographic parameters on the binding of protein It was found that for protein expressed in CHO cells, Millipore Sigma’s Fractogel EMD Chelate (M) charged with Zn2+ and GE’s pre-charged Ni Sepharose Excel displayed the highest binding capacities When the protein was expressed in HEK-293, GE’s IMAC Sepharose Fast Flow charged with either Co2+ or Zn2+ bound the greatest amount of protein The study further identified the metal binding capacity of the resin lot, the protein capacity to which the resin is loaded, and the ratio of poly-histidine tag residues on the protein all impacted the chromatographic performance and product quality These findings enabled the development of a robust and scalable process The CHO expressed cell culture product was directly loaded at a high capacity onto variable metal binding affinity Fractogel EMD Chelate (M) A 250 mM imidazole elution condition ensured the product contained monomeric and 6-histidine tagged bispecific proteins The optimized IMAC process conditions determined in this study can be applied to a wide variety of poly-histidine tagged proteins in research and development laboratories as various poly-histidine tagged proteins of differing molecular weights and formats expressed in either HEK-293 or CHO cells were successfully purified © 2020 The Authors Published by Elsevier B.V This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) Introduction Affinity chromatography is a preferred method to capture proteins from cell culture harvest It utilizes a highly selective ligand to bind a protein of interest For example, Fc-containing proteins, such as monoclonal antibodies, are easily captured using Protein ∗ Corresponding authors at: Purification Process Sciences, BioPharmaceutical Development, AstraZeneca, One MedImmune Way, MD 20878, Gaithersburg, USA E-mail addresses: valeria.riguero@astrazeneca.com (V Riguero), ellen.o’connor@astrazeneca.com (E O’Connor) A resin Many other protein formats lack readily available affinity capture modality and, therefore, require tags for affinity capture One such modification, the histidine tag, is a key tool used in biopharmaceutical research and development For this approach, a poly-histidine tag is incorporated on a protein during protein engineering The resulting tagged protein can be captured from the cell culture harvest using immobilized metal affinity chromatography (IMAC) through specific binding of the poly-histidine tag to the metal chelates present in the IMAC media While IMAC is a widely used technology in research and development laboratories, process optimization and manufacturing operational challenges exist which https://doi.org/10.1016/j.chroma.2020.461505 0021-9673/© 2020 The Authors Published by Elsevier B.V This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 can limit its use One challenge is the need for clarification and diafiltration of the cell culture media prior to purification This additional unit operation to enable chromatography increases the overall process time and decreases yield [1] IMAC resins also have relatively low binding capacities compared to other chromatography resins and variability in resin selectivity, both resulting in additional cycles required to complete purifications [2,3] Additionally, the need to strip metal and recharge the resin after each cycle adds processing time These challenges all contribute to IMAC being a lower yielding and slow purification operation Many innovative immobilized metal binding methods are in development to overcome the challenges with the goal of streamlining purification of poly-histidine tagged molecules One current development area focuses on directly applying a cell culture suspension containing the tagged protein of interest to the IMAC resin without the need for clarification Studies have found that chromatography performed with large diameter IMAC agarose beads allowed for the passage of cells while capturing the protein [4] Other studies have used charged and chelated detergent micelles to extract tagged proteins directly from cell culture supernatant [5] Nanoparticle technologies are also being explored to enable direct purification from cell lysate One group has developed nanoparticles with a core functionalized with pentadentate chelate affinity ligand that can be chelated with a variety of metals The charged particles, which can be loaded to a high capacity with cell lysate containing protein, were magnetically separated and protein was eluted at high purity and yield [6,7] Another group examined silica nanospheres containing dual chelating groups, which separated the proteins from cell lysate [8] Another development area focuses on identifying novel chelating ligands Groups have developed 1,4,7-Triazacyclononane (TACN) and 5,5-dithiobis-(2-nitrobenzoic) acid (DTNP) ligands that have been coordinated with Cu2+ or Ni2+ [2,9–13] The development of a new generation of chelating ligands is advantageous for several reasons Compared to currently available ligands, these ligands can increase the resin binding capacity to make the purification more efficient The novel ligands can also increase the selectivity of the resin to elute a product of increased purity The ligands also have high ion stability so that the new generation resins not need to be recharged every cycle Novel chelating ligands are also being applied to non-column separation methods Polyelectrolytes that chelate metal ions have been applied to membrane adsorbers, binding proteins at equal capacity to traditional resin beads, with the added benefit of a 15 purification time [14] Monolith convective interaction adsorbents have also been developed, enabling proteins purified to high purity levels at high binding capacities [3,15,16] The advantage of the monolith technology is the fast flow rate compared to traditional media Liquid-liquid extraction techniques are also being explored A TACN liquid sorbent was developed that can coordinate with Cu2+ , Ni2+ , or Zn2+ to create an ionic liquid sorbent [17] Aptamer technology has also been developed to separate histidine tagged proteins The aptamer complex approach is advantageous because imidazole is not necessary for purification and the purified protein is more pure compared to traditional resin based technology [18] Although new IMAC technologies are being developed to overcome the challenges of currently available purification methods, academic and industrial purification still relies on traditional resinbased processes Therefore, it is useful to understand and better control this widely used technology Studies have contributed to the current knowledge of resin based IMAC In one study, IMAC resins containing different chelators, iminodiacetic acid (IDA) and nitrilotriacetic acid (NTA) were coordinated with nickel and evaluated for purification of 6-histidine tagged cytotoxin associated gene A (CagA) Results were equivalent on all resins regardless of the chelator that was used [19] Another study observed a dif- ference in the purification product resulting from IDA and NTA chelators charged with different metals It was found that Cu2+ charged IDA yielded higher purity 6-histidine tagged viral coat protein product compared to other metal-chelator combinations [20] In a closer examination of IDA ligand mediating purification, four different sorbents; HiTrap Chelating HP, TSK Cheltate-5PW, Poros 20 MC, and monolithic glycidyl methacrylate–ethylene dimethacrylate were studied TSK Chelate-5PW bound the strongest while Poros matrix had a high degree of non-specific binding Agarosebased columns showed high selectivity and specificity [17] In this study, a robust and scalable IMAC process was developed for the purification of a poly-histidine tagged bispecific protein First, a resin and metal chelator screening study was performed to select conditions that would allow for direct protein binding from the cell culture harvest onto a charged resin at a both a high dynamic binding capacity and flow rate In the presence of HEK-293 medium, the optimal resin and metal combination was IMAC Sepharose FF charged with either Zn2+ or Co2+ Optimal combinations in the presence of CHO conditioned medium occurred on Fractogel EMD Chelate (M) charged with Zn2+ and Ni Sepharose Excel Second, a statistical study was performed to understand the impact of operational parameters on IMAC performance and product quality It was found that the metal binding capacity of the resin lot and the extent to which the resin was loaded heavily impacted the chromatographic performance and resulting product quality Finally, equilibrium isotherms were generated to understand the binding behavior of the and 6-histidine tagged materials comprising the poly-histidine tagged bispecific protein Although equivalent maximum adsorption capacities were independently achieved, when combined, 6-histidine tagged bispecific protein competed with and displaced 4-histidine tagged bispecific protein The three studies resulted in the development of a controlled and scalable IMAC purification operation The developed process conditions described herein were successfully applied to other poly-histidine tagged molecules Materials and methods 2.1 Materials and equipment 2.1.1 Materials FreeStyleTM MAX 293 and FreeStyleTM MAX CHO Expression systems were purchased from Thermo Fisher (Waltham, MA, USA) Fractogel EMD Chelate (M) was purchased from MilliporeSigma (Billerica, MA, USA) IMAC Sepharose Fast Flow, Ni Sepharose Excel, and Capto Q were purchased from GE Healthcare (Uppsala, Sweden) Profinity IMAC, Profinity IMAC Ni-Charged, and CHT Type II were purchased from Bio-Rad (Hercules, CA, USA) Ni-NTA Superflow was purchased from Qiagen (Hilden, Germany) TOSOH TSK gel 30 0 was purchased from Tosoh (Minato, Japan) The ProPacTM WCX-10 BioLCTM column, NuncTM 96-Well Cap Mats, Plate Sealers and 3.5 kDa MWCO Slide-A-LyzerTM Cassettes were obtained from Thermo Scientific (Waltham, MA, USA) The reverse phase column Intrada, WP-RP, μm (4.6 × 250mm) was purchased from Imtakt, USA (Portland, OR, USA) AcroPrepTM Advance 96-well filter-plates with 0.45 μm polypropylene membranes were purchased from Pall (Port Washington, NY, USA) iCE280 Analyzer and ChromPerfect software were purchased from ProteinSimple (San Jose, CA, USA) All other chemicals including metal salts and imidazole were purchased from Avantor Performance Chemicals (Center Valley, PA, USA) 2.1.2 Equipment For the null cell culture growth, shake flasks were grown using the Multitron incubation shaker from Infors HT (Bottmingen, Switzerland) For bench scale purification experiments, resins were V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 packed in Tricorn columns purchased from GE Healthcare (Uppsala, Sweden) or Vantage columns purchased from MilliporeSigma (Billerica, MA, USA) Purification was controlled by Unicorn 6.0 software on an Äkta Avant purchased from GE Healthcare (Uppsala, Sweden) For scale up chromatography, Fractogel EMD Chelate (M) was packed in a BPG column purchased from GE Healthcare (Uppsala, Sweden) Purification was controlled by the GE Akta Prime (Uppsala, Sweden) For statistical analysis, Jmp version 15 software was purchased from SAS (Cary, NC, USA) The analog rocker was from VWR (Radnor, PA, USA) The SorvallTM LegendTM RT+ Centrifuge and Cimarec+ TM stir plates were purchased from Thermo Scientific (Waltham, MA, USA) Protein quantification was determined using NanoDropTM 20 0 from Thermo Scientific (Wilmington, DE, USA) High performance liquid chromatography (HPLC) systems were purchased from Agilent Technologies (Santa Clara, CA, USA) 2.2 Purification methods 2.2.1 Null cell culture In order to evaluate the cell culture expression systems typically found in research and early development settings, transient transfections for both HEK-293 and CHO cell lines were performed using Thermo Fisher’s FreeStyleTM MAX Expression System The cells were thawed, expanded, and grown per the FreeStyleTM MAX protocols At transfection, the cells were transfected with non-coding plasmid DNA at mg of DNA/L of cells The transfected cells were grown in 5L fed batch shake flasks for 10 days, allowing for typical growth conditions and host cell by-products to be present in the cell culture harvest media The shake flasks were harvested by centrifugation at 40 0× g for 30 The null cell culture media was then 0.2 μm filtered 2.2.2 IMAC resin screening Resin and metal combination screening of the poly-histidine tagged bispecific protein was performed for both purified and null cell media (from Section 2.2.1) spiked materials Tricorn columns of 0.5 cm diameter were packed to a 10 cm bed height, mL column volume (CV), with each resin listed in Table For each binding experiment, resin was rinsed with CVs of water, followed by CVs of 250 mM metal charge solution After charging, the resin was rinsed with CVs of water and then CVs of 500 mM sodium chloride The resin was next equilibrated with CVs of 20 mM tris, 150 mM sodium chloride, pH 7.5 prior to loading purified protein, of 99% monomer, at a mg/mL initial concentration either in the equilibration buffer or in the null cell culture media As the protein was loaded, the flow-through material was tested for protein concentration by RPLC The loaded resin was re-equilibrated for CVs prior to a step elution with 20 mM tris, 150 mM sodium chloride, 500 mM imidazole, pH 7.5 Between cycles, the resin was stripped with 50 mM EDTA, 500 mM sodium chloride All chro- matographic steps were performed at 300 cm/h (1 mL/min), which corresponded to a residence time of The dynamic binding capacity was determined by plotting the protein concentration as a fraction of the inlet feed concentration and normalized to the CV Linear interpolation of the resulting protein breakthrough data was used to find the 10% breakthrough point 2.2.3 Initial process development Two different poly-histidine tagged bispecific protein cell culture lots were purified using two different Fractogel EMD Chelate (M) resin lots Cell culture lot contained 85.0% 4-histidine tag and 12.4% 6-histidine tag, while lot contained 76.3% 4-histidine tag and 20.5% 6-histidine tag Resin lots and respectively had metal binding affinities of 71 and 97 μmol/mL for Cu2+ Both resin lots were packed in 1.1 cm Millipore Vantage L columns to bed heights of 20 cm (19 mL CV) For each experiment, resin was rinsed with CVs of water prior to charging with CVs of 250 mM ZnCl2 The charged resin was then washed with CVs of water prior to CVs of 500 mM sodium chloride The resin was then equilibrated with 20 mM tris, 150 mM sodium chloride, pH 7.5 prior to loading the mg/mL concentration conditioned media to 20 g/L capacity After the load was complete, the protein loaded resin was re-equilibrated for CVs prior to a 20 CV elution gradient to 20 mM tris, 150 mM sodium chloride, 500 mM imidazole, pH 7.5 The resin was regenerated by a CV 50 mM EDTA, 500 mM sodium chloride wash All chromatography steps were performed at 300 cm/h (4.7 mL/min), corresponding to a residence time Elution fractions were collected for analysis of the quantity of protein, the ratio of and 6- histidine tag by cIEF, and the levels of monomer and aggregate by HPSEC 2.2.4 Statistical study Jmp software was used to outline a full factorial experimental design for the statistical study The study tested the impact of factors on IMAC performance: the Cu2+ binding capacity of the Fractogel EMD Chelate resin (from 71 μmol/mL to 97 μmol/mL), the amount of ZnCl2 charged onto the resin (4 CVs of 18 mM to 250 mM), the amount of protein loaded per volume of resin (10 to 30 g/L), and the ratio of to 6-histidine tagged bispecific protein present in the cell culture material (4-histidine 76.3%, 6-histidine 20.5% and 4-histidine 85.0%, 6-histidine 12.4%) Each resin lot was packed in a 1.1 cm diameter Millipore Vantage L column to a bed height of 20 cm (19 mL CV) For each of 18 experiments, including two center points, resin was rinsed with CVs of water followed by charging with CVs of either 18 or 250 mM ZnCl2 The resin was rinsed with CVs of water followed by CVs of 500 mM sodium chloride After equilibration with CVs of 20 mM tris, 150 mM sodium chloride, pH 7.5, the resin was loaded with to a targeted capacity with either cell culture lot (both with titers of mg/mL) After re-equilibration the protein was eluted using a 20 CV gradient to 20 mM tris, 150 mM sodium chloride, 500 mM imi- Table Commercially available IMAC resins IMAC resins tested in the screening study are listed along with their backbone composition and ligand molecule The particle sizes, pore diameters and metal binding capacity of the resins are also shown Resin Backbone Ligand Particle Size (μm) Pore Diameter (nm) Metal Affinity Capacity (μmol/mL) Fractogel EMD Chelate (M) (Merck-Millipore) IMAC Sepharose Fast Flow (GE) Profinity IMAC (Bio-Rad) Ni-NTA Superflow (Qiagen) Ni Sepharose Excel (GE) Profinity IMAC Ni-Charged (Bio-Rad) Cross-linked polymeth-acrylate Agarose UNOsphere Cross-linked agarose Agarose UNOsphere IDA IDA IDA NTA NTA IDA 40–90 45–165 45–90 60–160 90 45–90 80a 30b 130b 23c 30b 130b 60–100 25 12–30 12 54–70 12–30 a b c MilliporeSigma product information Carta, Giorgio and Jungbauer, Alois Protein Chromatography Weinheim: Wiley-VCH, 2010 Kastner, Michael Protein Liquid Chromatography New York: Elsevier, 20 0 4 V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 dazole, pH 7.5 Between cycles, the resin was stripped with 50 mM EDTA, 500 mM sodium chloride All chromatography steps were performed at 300 cm/h (4.7 mL/min) Elution fractions were collected for analysis of the quantity of protein, the ratio of and 6histidine tag by isoelectric chromatography (IEC), and the levels of monomer and aggregate by HPSEC Data was analyzed using Jmp software Least squares fit analyses were used to create models for the prominent elution peaks based on the combination of significant factors 2.2.5 Equilibrium binding experiments Separate monomeric 4-histidine and 6-histidine tagged bispecific protein solutions were formulated at mg/mL in equilibration buffer, 20 mM tris, 150 mM sodium chloride, pH 7.5 containing either 5, 100, or 170 mM imidazole For each imidazole concentration, the two stock solutions were mixed together to create protein solutions with different ratios of the and 6-histidine tagged bispecific proteins Each of the different ratio solutions were then diluted using either equilibration buffer containing imidazole or null CHO cell culture medium (from Section 2.2.1) into protein concentrations ranging from mg/mL to 0.5 mg/mL Three lots of Fractogel EMD Chelate (M) resin with Cu2+ binding capacities of 71, 83, and 94 μmol/mL were obtained Slurries of the three resin lots were prepared in suspension Preparation in suspension proceeded as follows First, resin was rinsed with volumes of water with gentle mixing Centrifugation was performed at 500× g for to remove the rinse Four volumes of 250 mM ZnCl2 were next applied to the resin with gentle mixing Centrifugation was used to remove the charge solution Five CVs of water and 500 mM sodium chloride were then respectively applied, mixed, and removed from the resin by centrifugation Each slurry was then prepared at a 10 % concentration in 20 mM tris, 150 mM sodium chloride, pH 7.5 Each well of a 96-well 0.45 μm filter plate was loaded with 100 μl of charged equilibrated slurry of the desired Cu2+ binding capacity After gentle mixing the equilibration buffer solution was removed from each well via centrifugation of the plate at 500× g for 200 μL of equilibration buffer containing either 5, 100, or 170 mM imidazole was then added to the charged resin After gentle mixing, the new equilibration buffer solution was removed via centrifugation of the plate at 500× g for After the resin was prepared, 200 μL aliquots of histidine tagged bispecific protein solutions varying in protein concentration (load capacity), imidazole concentration, and histidine tag ratio were applied to desired wells The filter plate was then incubated overnight with agitation to ensure protein binding Once incubation was complete, the filter plate was centrifuged at 500× g, and the filtrate of each well was collected 50 μL of the equilibration buffer containing either 5, 100, or 170 mM imidazole was then added to each well and the plate was incubated for hour After incubation, the plate was centrifuged at 500× g and the centrifugal filtrate was collected The two filtrate solutions containing non-bound protein were then pooled together 200 μL elution buffer, 20 mM tris, 150 mM sodium chloride, 500 mM imidazole, pH 7.5 was then added to each well and the plate was incubated for hours After incubation in elution buffer, the plate was centrifuged at 500× g and the centrifugal filtrate (the eluate) was collected in a microplate Each filtrate containing non-bound or eluted materials was analyzed for protein concentration and quantity of and 6-histidine tagged bispecific proteins using IEC 2.2.6 Impact of competition on binding capacity A 0.5 cm diameter tricorn column was packed to a 10 cm bed height, mL column volume (CV), with Sepharose Fast Flow resin For each experiment, the column was rinsed with CVs of water, followed by CVs of 250 mM Nickel Chloride charge solution After charging, the resin was rinsed with CVs of water, followed by CVs of 500 mM sodium chloride, and equilibrated with 5CVs of 20 mM tris, 150 mM sodium chloride, pH 7.5 For the first experiment, the column was loaded with purified monomeric poly-histidine tagged bispecific protein at a mg/mL concentration in 20 mM tris, 150 mM sodium chloride, pH 7.5 In the second experiment, the column was loaded with equivalent protein spiked to mg/mL concentration in null CHO medium For the third experiment, the column was washed with 100 CV of null CHO medium and then re-equilibrated with CV of 20 mM tris, 150 mM sodium chloride, pH 7.5 The purified monomeric polyhistidine tagged bispecific protein was then loaded onto the column at a mg/mL concentration in the equilibration buffer In the fourth experiment, the purified protein was spiked into null CHO medium and the resulting product was then dialyzed (40 times the product volume) against 20 mM tris, 150 mM sodium chloride, pH 7.5 The dialysis buffer was replaced times with new buffer after the material was dialyzed with mixing for a minimum of hours at 2-8ºC The resulting dialyzed product was then loaded onto the charged Sepharose Fast Flow column at a mg/mL concentration For all experiments, as the protein was loaded, the flow-through material was tested for protein concentration by RPLC The loaded resin was re-equilibrated for CVs prior to a step elution with 20 mM tris, 150 mM sodium chloride, 500 mM imidazole, pH 7.5 Between cycles, the resin was stripped with 50 mM EDTA, 500 mM sodium chloride All chromatographic steps were performed at 300 cm/h (1 mL/min), which corresponded to a residence time of The dynamic binding capacity was determined by plotting the protein concentration as a fraction of the inlet feed concentration and normalized to the CV Linear interpolation of the resulting protein breakthrough data was used to find the 10% breakthrough point 2.2.7 Process scale-up Fractogel EMD Chelate (M) resin was packed in a 20 cm diameter column to a bed height of 20 cm resulting in a 6.3 liter CV The resin was rinsed with CVs of water, prior to application of the CVs of 250 mM ZnCl2 After charging, the resin was rinsed with CVs of water and then CVs of 500 mM sodium chloride The resin was next equilibrated with CVs of 20 mM tris, 150 mM sodium chloride, pH 7.5, prior to loading conditioned media at a g/L initial concentration spiked with mM imidazole to a load of 18 g/L The loaded resin was re-equilibrated with CVs of 20 mM tris, 150 mM sodium chloride, pH 7.5 The column was then washed with CVs of 20 mM tris, 150 mM sodium chloride, 75 mM imidazole, pH 7.5 Product was then eluted with 20 mM tris, 150 mM sodium chloride, 250 mM imidazole, pH 7.5 The resin was stripped with CVs of 20 mM tris, 150 mM sodium chloride, 500 mM imidazole followed by CVs of 50 mM EDTA, 500 mM sodium chloride All chromatographic steps were performed at 300 cm/h (1570 mL/min) This operation was cycled four times to process the 500L reactor product 2.2.8 Application to other poly-histidine tagged proteins The optimized IMAC operation was applied to a variety of polyhistidine tagged molecules Proteins expressed in HEK-293 cells were purified on Sepharose Fast Flow charged with 250 mM CoCl2 5H2 or 250 mM ZnCl2 Proteins expressed in CHO cells were purified using Fractogel EMD Chelate (M) charged with 250 mM ZnCl2 or pre-charged Ni Sepharose Excel For each purification, regardless of scale, a column was packed to a height of 20 cm and operated at 300 cm/hr Each purification included rinsing the column with CVs of water prior to application of CVs of the metal charge solution The charged column was then rinsed with CVs of water followed by CVs of 500 mM sodium chloride Prior to V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 Fig Protein quantitation by RPLC A typical RPLC profile is shown The UV trace of the protein is displayed in blue The red line corresponds to the percentage of ACN/0.1% TFA elution buffer (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) loading the column was equilibrated with CVs of 20 mM tris, 150 mM sodium chloride, pH 7.5 buffer The cell culture harvest containing the poly-histidine tagged protein was then loaded on the charged and equilibrated column at capacities between 10 and 20 g/L After re-equilibration, the protein was eluted with 20 mM tris, 150 mM sodium chloride, 250 mM imidazole Column regeneration proceeded with CVs of 20 mM tris, 150 mM sodium chloride, 500 mM imidazole pH 7.5 and CVs of 50 mM EDTA, 500 mM sodium chloride 2.3 Analytical methods 2.3.1 RPLC titer assay A protein concentration standard curve was generated using pure material by injecting samples onto an Intrada, WP-RP, 3μm column (4.6 × 250 mm) Each sample was eluted with ACN/0.1% TFA at a flow rate of 0.75 mL/min Eluted protein was detected using UV absorbance at 280 nm The protein concentration of each sample was determined using regression analysis from the standard curve of the area under the product peak An example of an RPLC profile is shown in Fig 2.3.2 HPSEC purity assay Each sample was injected onto a TSKgel G30 0SWXL column (7.8 mm × 300 mm, μm) and eluted with 0.1 M disodium phosphate containing 0.1 M sodium sulfate, pH 6.8, at a flow rate of 1.0 mL/min Eluted protein was detected using UV absorbance at 280 nm The results were reported as the area percent of the product monomer peak compared to all other peaks excluding the buffer-related peak observed in the blank run Peaks eluting earlier than the monomer peak were recorded as percent aggregate; high molecular weight (HMW) or dimer Peaks eluting after the monomer peak were recorded as percent fragment A typical HPSEC profile is shown in Fig 2.3.3 Capillary isoelectric focusing (cIEF) assay Samples were concentrated to 2.0 mg/mL with Ultra Pure water using a Microcon 10,0 0 filter Samples were adjusted to 0.25 mg/mL with Ultra Pure water, 1% methylcellulose solution, Pharmalyte pH 3-10, Pharmalyte pH 8-10.5, pI Marker 9.77, and pI Marker 5.85 The samples were loaded onto an iCE280 Analyzer and focused at 1500 V for min, followed by 30 0 V for The resulting electropherograms were analyzed using ChromPerfect software and compared to a reference standard A representative electropherogram showing the migration of the and 6histidine tagged bispecific proteins is shown in Fig 2.3.4 Ion exchange chromatography (IEC) assay To quantify the levels of and 6-histidine tagged bispecific proteins in samples, analytical ion exchange chromatography was performed One milligram of each sample was loaded at mL/min onto a ProPac WCX-10 (4 × 250 mm) analytical high-performance weak cation exchange chromatography column equilibrated with 20 mM phosphate pH The protein was then eluted with an increasing salt gradient of 20 mM phosphate, 100 mM sodium chloride pH buffer Eluted protein was detected using fluorescence detector set to an excitation at 280 nm and an emission of 350 nm The identity of each peak was determined by comparison to a known reference standard containing 4,5, and 6-histidine species The percent peak area for each histidine tagged species was calculated by dividing the peak area by the total peak area An example of the IEC separation of and 6-histidine tagged bispecific proteins is shown in Fig Fig Protein purity by HPSEC A typical HPSEC profile is shown The UV trace of the protein is displayed in blue The high molecular weight species elute first, followed by dimer, and then monomer (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 Fig cIEF profiles of histidine tagged variants A typical isoelectric focusing profile is shown The acidic and basic pI markers are labeled with their corresponding pI values The and 6-histidine tagged bispecific protein peaks are also labeled with their pI values Fig High throughput quantitation of histidine tagged variants by IEC A typical analytical ion exchange chromatography profile is shown The fluorescence trace of the protein is displayed in blue and the elution gradient is displayed in red The 4-histidine tagged bispecific protein elutes first followed by the 6-histidine tagged bispecific protein (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) Results and discussion 3.1 Protein description In this study, we sought to understand and optimize IMAC to enable a controlled large-scale purification of a poly-histidine tagged bispecific protein The 52 kDa protein, shown in Fig 5, consisted of two binding domains The antigen binding domain was engineered to bind to an oncology target, while the T cell binding region was designed to interact with the glycoprotein surface complex To enable IMAC purification, a poly-histidine tag was engineered into the terminus of the T cell binding domain Prior to large-scale purification of the molecule, several studies were performed and are described within 3.2 IMAC resin screening of purified protein A screening study was performed to select IMAC materials that optimized the dynamic binding capacity of the poly-histidine tagged bispecific protein The study consisted of generating breakthrough curves using resin and metal chelator combinations of the commercially available materials shown in Tables and As can be seen in Table 1, studied resins included those having a variety of backbones and both IDA and NTA ligands The particle sizes composing each resin are also shown and ranged between 40 to 165 μm which enabled the superficial velocity of 300 cm/h without the generation of pressure challenges at bench and large Fig Diagram of the poly-histidine tagged bispecific protein The protein consists of two single chain variable fragment domains (scFv) The antigen binding and T cell binding domains are shown for the 52 kDa protein The 4-6 residues of the terminal histidine tag are also depicted V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 Fig Dynamic binding capacities of purified protein The binding data of purified bispecific protein on commercially available resin/chelator combinations are shown 6A displays the protein’s breakthrough profiles when all resins were charged with Ni2+ 6B shows the impact of the metal chelator on the breakthrough profiles 6C includes the dynamic binding capacities at 10% breakthrough for various resin and metal combination scales The pore sizes and metal affinity capacities are also detailed Table lists the studied metal chelators and the corresponding salts used to generate the charged buffers The breakthrough curves of purified bispecific protein on resins charged with Ni2+ are shown in Fig 6A The dynamic binding capacities for each resin differ The Profinity resins exhibited the steepest binding profiles and the least capacities at 10% breakthrough In contrast, shallow curves were observed using the GE resins which also achieved greater than 50 mg/mL capacities at 10% breakthrough The dynamic binding capacities of the resins were next determined when charged with various metals As shown in Fig 6B, regardless of the metal used, the resin binding order is consistent with that of Fig 6A with IMAC Sepharose FF > Fractogel EMD Chelate (M) > Profinity IMAC The GE resin, IMAC Sepharose FF, displayed the highest capacity at 10% breakthrough for each Table Metal chelators The metals and corresponding salts used for charging the commercially available resins for the screening study are shown Metal Salt Nickel Zinc Copper (II) Cobalt Iron (II) Magnesium Manganese NiCl2 6H2 O ZnCl2 CuSO4 5H2 O CoCl2 5H2 O FeCl2 4H2 O MgCl2 6H2 O MnCl2 4H2 O charged metal Charging of this resin with Ni2+ , Co2+ , Zn2+ , or Cu2+ , each resulted in higher capacities at 10% breakthrough than those of the Ni2+ pre-charged resins Differences were also seen in the binding capacity for all resins depending on the metal used for charging Resins charged with Mg2+ , Mn2+ , and Fe2+ , showed lower binding capacity compared to resins charged with Ni2+ , Co2+ , Zn2+ , and Cu2+ , and suggested that these metals were not as effective at coordinating with the chelating ligand and histidine tagged bispecific protein An examination of the breakthrough profiles on IMAC Sepharose FF in Fig 6C revealed that charging with Co2+ , Zn2+ , and Cu2+ resulted in preferable binding, as breakthrough curves generated with the other metals showed steep slopes or lower capacities at 10% breakthrough Overall, the breakthrough curves of the purified poly-histidine tagged bispecific protein showed that IMAC Sepharose FF resin charged with Cu2+ provided the most favorable binding profile and highest dynamic binding capacity It is interesting to note that regardless of the metal chelator, the dynamic binding capacities at 10% breakthrough were consistently ranked IMAC Sepharose FF > Fractogel EMD Chelate (M) > Profinity IMAC Several factors govern dynamic binding capacity; residence time, particle size, and pore diffusivity [21] As described in section 3.1, the molecule used in the study was a 52 kDa protein with a calculated hydrodynamic radius of 3.22 nm [22] A closer examination of the resin specifications in Table showed that the pore sizes of the resins tested were theoretically adequate to allow diffusion of the bispecific protein into the resin pores Diffusion was likely as the protein was at least times smaller than the pore sizes of the Qiagen and GE resins, the resins with the V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 Fig Dynamic binding capacities of protein spiked in null cell media The binding capacities of the protein spiked in null cell media is shown 7A and 7B respectively display the dynamic binding capacities at 10% breakthrough for bispecific protein purified on a variety of resin and metal combinations from HEK-293 and CHO media smallest pores Therefore, for each resin, the protein could access all the available pore volume for coordination Additionally, as Table shows, there were differences in the metal binding capacities of the resins as specified by their manufacturers Fractogel EMD Chelate (M) possessed the greatest metal capacity, 60–100 μmol metal/mL The GE resins, Ni Sepharose Excel and IMAC Sepharose FF, could be respectively charged to 54 –70 and 25 μmol metal/mL Ni-NTA Superflow and Profinity IMAC possessed the least metal binding capacity of 12–30 μmol/mL The data in Fig showed that the highest dynamic binding capacities were observed on the IMAC Sepharose FF resin that was composed of particles with small pores and a moderate level of metal binding capacity Ni-NTA resin was composed of particles with similar pore size and decreased metal binding capacity and demonstrated reduced protein dynamic binding capacity The pore size of Ni Sepharose Excel resin was also comparable to IMAC Sepharose FF and Ni-NTA However, the metal binding capacity of Ni Sepharose Excel exceeded IMAC Sepharose FF, but also demonstrated reduced protein dynamic binding capacity It is likely that that protein binding created steric hindrance, effectively blocking additional proteins from accessing the additional coordination sites A level of optimal ligand density existed, between 25 to 54 μmol metal/mL, the metal binding capacities of IMAC Sepharose FF and Ni Sepharose Excel resins 3.3 IMAC resin screening of protein in CHO and HEK-293 media IMAC is typically used as a capture chromatography operation Therefore, experiments were performed to identify the impact of the cell culture expression material on dynamic binding capacity To mimic the capture step, breakthrough profiles were generated for the poly-histidine tagged bispecific protein in null media The dynamic binding capacities at 10% breakthrough for the protein respectively spiked into HEK-293 and CHO null media are shown in Fig 7A and B As expected, in both media the binding capacities were decreased for all resin and metal combinations compared to those previously generated with pure protein A closer examination of the dynamic binding capacity data shows that compared to the purified protein values, the HEK-293 medium had a greater impact on binding than the CHO medium For example, both Ni Sepharose Excel and IMAC Sepharose FF charged with Ni2+ , Co2+ , Zn2+ , and Cu2+ had greater dynamic binding capacities in CHO than HEK-293 null cell medium Greater capacities in CHO medium were also observed for Fractogel EMD Chelate (M) when charged with Zn2+ and Cu2+ Furthermore, this data demonstrated the dynamic binding capacity results for the purified protein did not extrapolate when in the presence of media, as the best overall binding capacities in media, exceeding 15 g/L, were on Fractogel EMD Chelate (M) charged with Zn2+ and Cu2+ or pre-charged Ni Sepharose Excel The dynamic binding capacities in the presence of null conditioned media were decreased compared to those of the purified protein as the protein competed with cell expression products or media components capable of coordination to the charged metal The data indicated that there were components in HEK-293 null conditioned medium that more effectively competed with the bispecific protein for the charged metal than components in the CHO conditioned medium, as the capacities in CHO medium were generally greater It is also interesting to note that unlike the case for the pure protein, the highest capacities observed in CHO medium were on resins containing the highest metal binding capacities, Fractogel EMD Chelate (M) charged with Zn2+ and Ni Sepharose Excel The availability of the increased metal binding sites likely increased the total quantity of both bound media components and protein The high levels of binding to these resins was not observed in the HEK-293 medium The differing components in the HEK-293 medium may have had increased affinity for Zn2+ and Ni2+ These experiments were impactful for purification development of the poly-histidine tagged bispecific protein because they showed that IMAC resin and metal combinations existed that achieved high dynamic binding capacities at 10% breakthrough without requiring diafiltration of the cell culture harvest media The ideal combinations included IMAC Sepharose FF charged with Co2+ , Zn2+ , or Cu2+ when purifying a protein from HEK-293 medium and either pre-charged Ni Sepharose Excel or Zn2+ charged Fractogel EMD Chelate (M) when purifying a protein from CHO medium 3.4 Initial process development The screening study results were applied to the development of the IMAC capture operation for the poly-histidine tagged bispecific protein The expression system for the cell culture process used CHO cells, therefore Ni Sepharose Excel and Zn2+ charged Fractogel EMD Chelate (M) were considered as capture columns as the protein could be directly loaded onto the resin at high capacity without the need for diafiltration Fractogel EMD Chelate (M) was selected for the process as its integration decreased purification costs Process development of the IMAC capture operation was initiated by performing gradient elutions to aide in the identification V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 Fig Variable IMAC gradient elution profiles Variable IMAC elution profiles are shown for two different lots of Fractogel EMD Chelate (M) resin and two different cell culture lots The elution profiles of cell culture lot 1/resin lot 1, cell culture lot 2/resin lot 1, and cell culture lot 1/resin lot are respectively shown in 8A, 8B, and 8C The blue traces indicate A280 levels, while the red trace displays the progress of the elution gradient In each chromatogram, all three elution peaks are labeled along with their corresponding imidazole elution concentrations 8D, 8E, and 8F show the milligram levels of protein in terms of percent monomer purity and ratio of histidine tag in the load, peak 1, peak 2, peak 3, and combined peaks (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) of step elution conditions As described in Section 2.2.3, two lots of Fractogel EMD Chelate (M) resin differing in Cu2+ binding abilities, 71 and 94 μmol/mL, were used to purify the poly-histidine tagged bispecific protein expressed in two cell culture lots Prior to purification of both lots, the ratio of 4-histidine tagged and 6histidine tagged bispecific proteins was measured Cell culture lot contained 85.0% 4-histidine tagged and 12.4% 6-histidine tagged bispecific proteins, while lot contained 76.3% 4-histidine tagged and 20.5% 6-histidine tagged bispecific proteins Three experiments of differing harvest and resin lot combinations were performed in which 300 milligrams of harvested cell culture product was loaded to a 20 g/L capacity on the IMAC resin prior to a 20 CV imidazole gradient elution Although the same procedures were used to perform the purifications, inconsistent chromatography profiles were observed during the elution gradients Examples of the protein elution profiles are shown in Fig All chromatographic profiles contained three prominent peaks, however the peaks had different peak areas and eluted at different points of the imidazole gradient For example, as shown in Fig 8A and B, protein from two different cell culture lots were purified using the same 71 μmol/mL lot of IMAC resin The first elution peak is more prominent in Fig 8A (cell culture lot 1) whereas the second peak is more prominent in Fig 8B (cell culture lot 2) Cell culture lot was also purified on the 94 μmol/mL IMAC resin lot Three elution peaks were also ob- served, however they eluted with increased imidazole concentrations The findings of these experiments show that variation in cell culture and resin lot properties impact protein elution behavior Due to the observed chromatographic inconsistencies, additional characterizations of the elution peak materials were performed cIEF, HPSEC, and protein concentration analyses were performed for the three prominent peaks of each elution shown in Fig 8A, B, and C As shown in Fig 8D, E, and F, each elution peak within a gradient was not homogeneous in terms of polyhistidine tag length and monomer purity cIEF data revealed that across the three purifications, peak contained predominately 4-histidine tagged bispecific protein, peak contained predominately 6-histidine tagged bispecific protein, and peak contained a mixture of poly-histidine tagged bispecific proteins, with the 6histidine tagged variant in greater proportion For example, of the 255 mg of 4-histidine tagged bispecific protein loaded in Fig 8A, 217 mg of the material eluted in peak 1, whereas mg eluted in both peaks and In the same experiment 37 mg of 6-histidine tagged bispecific protein was loaded onto the resin No significant amount of 6-histidine tagged material was found in peak 1, whereas 19 and 15 mg were each respectively found in peaks and The same trends held true for the chromatography experiments shown in Fig 8B and C Additionally, a decrease in monomer was observed across the gradients as peaks and contained 10 V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 Table Statistical study parameters Parameters were identified that impacted IMAC performance and product quality The parameter ranges were studied in a range finding study as shown The products of each purification were analyzed for quality by the listed assays Parameters Parameter ranges Ratio of 4-histidine:6-histidine tagged bispecific protein in cell culture lot 76.3 : 20.5 85.0 : 12.4 Copper binding ability of Fractogel EMD Chelate (M) (μmol/mL) Resin Charging (mM Zinc) Load Capacity (g/L) 71 18 10 97 250 30 HPSEC Purity Histidine Tagged Ratio Outputs for Each Peak Protein Concentration mostly monomer, while peak contained mostly aggregate For example, analysis of the chromatography performed in Fig 8A shows that of the 228 mg of monomer loaded onto the resin, 199 mg of monomer (87.3% yield) was recovered in elution peaks of and The monomer purity of peaks and were respectively 84.2% (15.8% dimer) and 71.3% (28.7% dimer) Peak was enriched high molecular weight aggregate, with an aggregate level of 84.6% The same monomer purity level trends held true across the three chromatography experiments The findings of these experiments show that 6-histidine tagged and aggregated protein are likely to interact with more charged sites than 4-histidine tagged or monomeric species as increased levels of imidazole are required to elute these proteins The same chromatographic trends were also observed when the Fractogel EMD Chelate (M) resin was charged with other metals, including Cu2+ , Co2+ , and Ni2+ (data not shown) as the poly-histidine tagged bispecific protein interacted with the charged ligand in a similar manner The trends are likely to be similar on other resins, however the separation would be impacted by the metal binding capacity of the resin These findings demonstrate that variations in the metal binding capacity of an IMAC resin, cell culture product quality, both monomeric purity and quantity of differing histidine tagged levels, complicate development of IMAC capture operations 3.5 Statistical analysis of IMAC operation The results of the three chromatography experiments performed during the initial purification development showed the Cu2+ binding capacity of the IMAC resin as well as variation of the histidine tag ratio resulted in inconsistent chromatographic performance and product quality A statistical study was employed to both determine if additional parameters affect the purification performance, and to understand the impact of possible ranges of the parameters The operating parameters studied are highlighted in Table The poly-histidine bispecific protein materials used in this study were the same as those used in the initial chromatography work Cell culture lot contained 85.0% 4-histidine tagged and 12.4% 6-histidine tagged bispecific protein, while lot contained 76.3% 4-histidine tagged and 20.5% 6-histidine tagged bispecific proteins The resin used in the study also matched that used in the initial development work and spanned the range of Cu2+ binding capacities produced by the resin manufacturer, 71 and 97 μmol/mL In addition to the histidine tag ratio and the resin lot, the extent of resin metal charging and the protein load capacity were also explored Resin metal charging solutions ranged from 18 mM to 250 mM as recommended by manufacturers’ protocols Protein loading ranged from 10 to 30 g/L to observe the impact on product quality as the resin is pushed in excess of 10% DBC Each chromatographic experiment in the statistical study was performed as described in Section 2.2.4 Materials from each of the three characteristic gradient peaks were separately analyzed for protein concentration, HPSEC purity, and histidine tag ratio Required Imidazole Concentration The study identified the Cu2+ binding capacity of the resin and the amount of protein loaded on the column were the two most important factors that influenced the performance and product quality The contour plots displaying the impact of two parameters is shown in Fig As shown in Fig 9A and D, when protein was loaded to 10 g/L onto 97 μmol/mL resin peaks and eluted with higher concentrations of imidazole compared to when loaded to 30 g/L on 71 μmol/mL resin The earlier strong binding condition respectively required 145 mM and 230 mM of imidazole to elute peaks and The later weak binding condition respectively required 110 and 180 mM of imidazole to elute peaks and The Cu2+ binding capacity and the amount of protein loaded on the column also influenced the ratios of and 6-histidine tagged bispecific proteins in elution peaks and As shown in Fig 9B and E, when protein was loaded to 10 g/L onto 97 μmol/mL resin, the histidine tagged ratio within each peak is more homogeneous For example, under the strong binding condition, the peak elution was greater than 99% 4-histidine tagged and peak contained material greater than 87% 6-histidine tagged This is in contrast to weak binding conditions in which 98% of the peak product was 4-histidine tagged and 72% was 6-histidine tagged Peaks and were also analyzed in the statistical study for monomeric content As shown in Fig 9C and F, the monomer content in the elution products was influenced by both the Cu2+ binding capacity of the resin and the amount of protein loaded onto the column The monomer content of peak material was as high as 88% when protein was loaded to 10 g/L on the 97 μmol/mL resin The purity decreased to 76% under the weak binding condition of 30 g/L binding on 71 μmol/mL resin The same trends were observed for peak material in which the monomer purity ranged from 74% to 66% The trends observed in the study are logical Fully charged 97 μmol/mL resin had the greatest possible number of coordination sites Loading that lot of charged resin to the low 10 g/L capacity, allowed for the maximum interactions between the polyhistidine tagged bispecific protein with the resin Elution from this state required the maximum amount of imidazole and resulted in relatively homogeneous histidine tagged material with a high monomer percentage Fully charged 71 μmol/mL resin loaded to 30 g/L had less coordination sites available It is possible that these weak binding conditions created an environment in which the 6histidine tagged bispecific protein competed with the 4-histidine tagged material and the HMW aggregate competed with the 6histidine tagged bispecific protein The elution products therefore contained increased aggregate and an increased heterogeneity in terms of and 6-histidine tagged bispecific proteins The results of the statistical study showed that IMAC performance and product quality were influenced by process parameters For example, to ease the action of performing the purification, any lot of Cu2+ binding capacity resin could be obtained from the manufacturer and fully charged with four CVs of 18 mM to 250 mM metal chelator The charged resin could be loaded to between 10 V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 11 Fig Statistical study of IMAC gradient elution Contour plots containing data from the statistical study are shown The impact of Fractogel EMD Chelate (M) resin metal charge capacity and protein loading on the imidazole required for elution of peaks and is shown in 9A and 9D The impact of the same two parameters is shown on the ratio of histidine tag in products of peaks and is shown in 9B and 9E The monomer purities of the same materials are shown in 9C and 9F to 30 g/L prior to elution with 250 mM imidazole As shown in this study, the 250 mM imidazole elution allowed for the elution of both and 6-histidine tagged bispecific proteins that were enriched for monomer 3.6 Modeling equilibrium isotherms The statistical study revealed differences in the binding and elution patterns of the poly-histidine tagged bispecific protein which resulted from the Cu2+ binding capacity of the resin and the resin protein loading The observations in the statistical study were based on two cell culture lots One lot was composed of 85.0% 4-histidine tagged and 12.4% 6-histidine tagged bispecific protein, while the other lot contained 76.3% 4-histidine tagged and 20.5% 6histidine tagged bispecific proteins Prior to finalizing the IMAC purification operation for the protein, it was desired to further study the impact of a wider range of both and 6-histidine tagged bispecific protein ratios Confirmation of the binding behavior of pure monomeric 4-histidine and 6-histidine tagged bispecific proteins as well as a complete range of their ratios allowed for a thorough understanding of the operation so that any cell culture lot, regardless of the histidine tagged ratio would be purified as expected Single-component equilibria binding experiments were first performed with 100% 4-histidine tagged and 100% 6-histidine tagged bispecific protein on 83 μmol/ml Cu2+ binding capacity resin as described in Section 2.2.5 Adsorption isotherms were created for each protein solution in three mobile phases as determined based on the elution strength required to elute the two variants in the statistical range finding experiments; mM imidazole to prevent non-specific binding, 100 mM imidazole to elute 4-histidine tagged bispecific protein, and 170 mM imidazole to elute 6-histidine tagged bispecific protein The single-component isotherms in Fig 10A and B show both 6-histidine and 4-histidine tagged bispecific proteins exhibit typical Langmuir isotherms The maximum adsorption capacities, qm , in the different mobile phases are shown in Table At mM imidazole, the qm values for both proteins were equivalent at 39.9 mg/mL As imidazole increased in the mobile phase to 100 mM, the adsorption equilibrium constant (kL ) was decreased overall at 26.2 mg/mL The adsorption equilibrium constant for 6-histidine tagged material was greater than the 4-histidine tagged material in both phases Adsorption isotherms were also created to study the impact of a range of 6-histidine tagged and 4-histidine tagged ratios on protein binding as the poly-histidine tagged bispecific protein harvest material contained an uncontrolled mixture of 4-histidine and 6-histidine tagged bispecific proteins Shown in Fig 10C and D, equivalent qm values were observed for protein solutions composed of either 100% or 6-histidine tagged materials, with kL6 greater than kL4 , respectively 7.99 and 3.28 mL/mg As increasing percentages of the opposing molecule were introduced, lower qm values were observed for the 4-histidine tagged material compared to the 6-histidine tagged material For example, the qm of 4-histidine tagged bispecific protein loaded in a mixture of 25% 4histidine tagged and 75% 6-histidine tagged bispecific proteins approached mg/mL, whereas that of the 6-histidine tagged material loaded in a 25% 6-histidine tagged and 75% 4-histidine tagged mixture approached 12 mg/mL The results of the single-component and the multi-component changing histidine tagged ratio experiments were used to generate a competitive Langmuir model that describes the competition of the 6-histidine and 4-histidine tagged bispecific proteins The model is outlined in Fig 10E and the equation is shown below qei represents milligrams of 4-histidine or 6-histidine tagged bispecific protein (qe4 or qe6 ) adsorbed per milliliter of resin cei is equal to the concentration (mg/mL) of soluble 4-histidine (ce4 ) and 6-histidine tagged bispecific protein (ce6 ) qm represents the maximum protein adsorbed per resin volume (mg/mL) kLi repreTable Single component adsorption isotherm results Isotherms were created for each protein solution in three mobile phases listed on 83 μmol/mL copper binding capacity resin The maximum adsorption capacities, qm , in the different mobile phases are shown Both the 4-histidine tagged and 6-histidine tagged bispecific proteins had equivalent qm values Imidazole (mM) 100 170 qm (mg/mL) 39.95 26.18 0.65 12 V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 Fig 10 Equilibrium isotherms of and 6-histidine tagged bispecific proteins Isotherms were generated for purified and 6-histidine tagged bispecific proteins in 5, 100, and 170 mM imidazole mobile phases on 83 μmol/mL Cu2+ binding affinity Fractogel EMD Chelate (M) resin and are shown in 10A and 10B The protein solutions were also mixed into different ratios to examine the competition of the poly-histidine tagged bispecific proteins The resulting isotherms are shown in 10C and 10D Isotherms are shown with experimental data points The displacement of 4-histidine tagged bispecific protein by 6-histidine tagged bispecific protein is shown in 10E sents the Langmuir equilibrium coefficients for 4-histidine and 6histidine tagged bispecific protein, respectively kL4 and kL6 The denominator is the sum of the k and c terms for the and 6-histidine tagged bispecific proteins qei = 1+ qm kLiCei n j=1 kL j ce j The model predicts the displacement of 4-histidine tagged protein with increasing amounts of 6-histidine tagged protein As shown in Fig 10, the model is consistent with the data points generated from the single and multi-component isotherm experiments The kL4 and kL6 values in conjunction with qm values from both sets of experiments indicated that 4-histidine and 6-histidine tagged materials competed for binding In the and 100 mM imidazole mobile phases, the 6-histidine tagged bispecific protein bound more favorably than 4-histidine tagged bispecific protein, achieving higher adsorbed protein concentrations at lower solution protein concentrations These finding showed that increased protein loading of material containing both 4-histidine and 6-histidine tagged bispecific proteins can impact the histidine tag ratio in the elution products These observations are consistent with the sta- tistical study data which showed displacements of the 4-histidine and 6-histidine proteins at high loading capacities The model was implemented to generate equilibrium isotherms to examine the influence of resins with different Cu2+ binding capacities Fig 11 displays the 4-histidine or 6-histidine tagged bispecific protein single-component isotherms along with overlaying experimental points generated in mobile phases of 5, 100, and 170 mM imidazole on each of three resins with Cu2+ binding capacities of 71, 83, and 94 μmol/mL As Table details, in the mM imidazole environment which is most favorable for poly-histidine tagged bispecific protein binding, both the and 4-histidine tagged material achieved similar qm values, 38, 40, and 39 mg/mL, irrespective of the binding ability of the resin As the imidazole concentration was increased to 100 mM in the mobile phase, the qm values respectively decreased to 25, 26, and 29 mg/mL on the 71, 83, and 94 μmol/mL Cu2+ binding capacity resins In the 170 mM imidazole mobile phase, the respective qm values were 0.2, 0.7, and 3.6 mg/mL for the same three resins For each mobile phase, the lowest qm values were seen for the 71 μmol/ml Cu2+ binding capacity resin Greater qm values were observed as the resin Cu2+ binding capacities increased Table also includes kL4 and kL6 values from the set of experiments The kL6 values were greater than V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 13 Fig 11 Equilibrium isotherms on resins of different Cu2+ binding abilities Equilibria isotherms were generated for and 6-histidine tagged bispecific proteins on three Fractogel EMD Chelate (M) resin lots in mM, 100 mM, and 170 mM imidazole mobile phases respectively shown in 11A and 11B, 11C and 11D, and 11E and 11F Grey, blue, and orange respectively represent 94, 83, and 71 μmol/mL Cu2+ binding capacity resins Experimental data points are shown (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) the kL4 values under all mobile phase and resin conditions evaluated, consistent with the previous single component and multicomponent experimental observation used to construct the model Specifically, kL4 values were two to three times less than kL6 values under conditions favorable for protein binding (5 mM imidazole) and are more than ten times under protein eluting conditions (100 mM imidazole) This observation was consistent with the experimental column chromatography data that shows 4-histidine tagged bispecific protein was more readily eluted from the resin than 6histidine tagged bispecific protein It is interesting to note that the kL4 appeared similar across the three resin lots tested, while kL6 appeared to increase with increased Cu2+ binding capacity under low imidazole (high protein binding) conditions This observation is consistent with the trend seen in the Fig statistical design experiments where the highest protein resolution was seen with the 94 μmol/mL resin Additionally, experiments were performed to study the impact of null cell conditioned media on the binding equilibria as IMAC is typically used as a capture chromatography operation Binding experiments were performed in the mM imidazole mobile phase on 14 V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 Table Influence of resin on poly-histidine tagged bispecific protein binding Single component isotherms were created for 4-histidine tagged and 6-histidine tagged bispecific proteins in three different mobile phases: 5, 100, and 170 mM imidazole on three resins of different copper binding ability: 71, 83, 94 μmol/mL Corresponding qm and k values are shown for both proteins Mobile phase: mM Imidazole Resin Cu2+ binding capacity (μmol/mL) 71 83 94 qm (mg/mL) kL6 (mL/mg) kL4 (mL/mg) 38.08 7.23 3.19 39.95 7.99 3.28 39.16 10.19 3.31 71 25.49 0.89 0.04 83 26.18 1.02 0.08 94 29.46 1.00 0.06 71 0.20 36.15 1.20 83 0.66 11.61 1.88 94 3.56 1.52 0.20 Mobile Phase: 100 mM Imidazole Resin Cu2+ Binding Capacity (μmol/mL) qm (mg/mL) kL6 (mL/mg) kL4 (mL/mg) Mobile Phase: 170 mM Imidazole Resin Cu2+ Binding Capacity (μmol/mL) qm (mg/mL) kL6 (mL/mg) kL4 (mL/mg) Table Impact of conditioned media on protein binding Single-component isotherms were generated by performing binding experiments in the mM imidazole mobile phase on all three copper binding capacity resins both in the presence of and without null conditioned medium The resulting qm and k values are shown Resin: 71 μmol/mL Cu2+ binding capacity qm (mg/mL) kL6 (mL/mg) kL4 (mL/mg) Purified protein CM Protein 38.08 7.23 3.19 32.90 6.46 2.24 Purified Protein 39.95 7.99 3.28 CM Protein 35.87 7.13 2.31 Purified Protein 39.16 10.19 3.31 CM Protein 36.40 10.20 3.02 Resin: 83 μmol/mL Cu2+ Binding Capacity qm (mg/mL) kL6 (mL/mg) kL4 (mL/mg) Resin: 94 μmol/mL Cu2+ Binding Capacity qm (mg/mL) kL6 (mL/mg) kL4 (mL/mg) all three Cu2+ binding capacity resins both in the presence of and without null cell conditioned medium Isotherms generated using the model as well as overlaying experimental points are shown in Fig 12 The qm , kL4, and kL6 values generated from the isotherms are shown in Table The qm, kL4, and kL6 values were all decreased in conditioned medium conditions when compared to the pure protein conditions, consistent with the screening study trends in Fig The decreased binding in the presence of medium was most different on the lower affinity Cu2+ binding resin The differences between the kL4 , kL6, and qm values decreased for the purified proteins compared to the proteins in medium as the Cu2+ binding capacity increased It is likely that a component in the medium or byproduct of cell expression, such as host cell proteins, stripped the metal charge or competed with the histidine tagged bispecific proteins for binding to the charged ligands As the 71 μmol/mL Cu2+ affinity resin contained less binding sites than 83 and 94 μmol/mL Cu2+ affinity resin, the greatest difference between the purified protein and medium spiked protein was observed Overall, the observation that the isotherms produced in this study are consistent with the Langmuir model is significant Previous studies by different groups created isotherms that found the Langmuir model, in which one protein binds to one metal ion, was not adequate For example, while creating isotherms for cytochrome c binding to TSK-chelate resin, isotherms were consistent with the Langmuir model at low levels of copper coordinated on the resin, as immobilized copper decreased the possibility for multiple site interactions As increasing metal was coordinated on the resin, Langmuir-Freundlich isotherms were observed, indicating simultaneous coordination to more than one ion [23] The same group later found that a single histidine residue was capable of coordinating to multiple charged sites [24] Another group conducted a study with a diverse set of proteins including lysozyme, ovalbumin, and pig albumin They found that proteins were capable of coordinating to 13 ions on the resin [25] Yet another group showed extensive Langmuir-Freundlich isotherms in the study of lysozyme, ovalbumin, BSA, conalbumin, and wheat germ agglutinin [26] The qm and kL values were impacted by ionic strength and pH [27] A fourth group extensively studied IMAC binding and developed their own metal affinity interaction model (MAIC) to account for multiple coordination bonds between the protein and the charged sites, low capacities on the IMAC adsorbents, and the concentration of mobile phase modifiers The MAIC model explained the binding behavior of RNAse A, myoglobin, lysozyme, conalbumin, and ovalbumin [28,29] These studies are consistent The proteins used in these studies contained natural histidine residues that exist across the proteins The studied molecules did not contain engineered poly-histidine tags Therefore, multiple surface exposed histidine residues present across a protein can interact with several ions on the highly charged resin The resulting isotherms would therefore exhibit non-Langmuir characteristics These studies showed that the qm and kL values of these curves are impacted by the concentration of imidazole in the mobile phases and the placement and number of histidine residues across the molecules The bispecific protein used in this study contained an engineered poly-histidine tag Other potential interacting residues were removed Therefore, only one site on the protein could interact with the highly charged resin containing the maximum level of accessible ions The generation of this one to one interaction explains the Langmuir consistent isotherms of this study Although the and 6-histidine tagged bispecific proteins each had one region capable of interacting, the 6-histidine tagged bispecific protein bound with a greater kL than the 4-histidine tagged bispecific protein The shift of the isotherms according to total charge and concentration of imidazole in the mobile phase is consistent with those reported in the discussed literature 3.7 Impact of competition on binding capacity The binding isotherm data from Section 3.6 was compared to the resin screening data from Sections 3.2 and 3.3 to better understand binding behavior for the poly-histidine tagged bispecific protein on Fractogel EMD Chelate (M) charged with zinc Comparison of the qm values calculated from the isotherm experiments to the maximum protein binding capacities observed during the screening experiments revealed different binding values The difference in maximum binding values between the purified protein and the protein in the presence of null cell media was greater in the screening experiments (Figs and 7) versus the isotherm binding experiments (Fig 12) In the screening study, the maximum binding capacity of the purified poly-histidine tagged bispecific protein was approximately 30 mg/mL greater than when in the presence of null cell medium In comparison, the difference in maximum binding capacity between the purified protein and that in the presence of null cell medium was significantly less, mg/mL, for the isotherm binding experiments An examination of the experimental conditions revealed that the differences in both the maximum binding values as well as V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 15 Fig 12 Equilibrium isotherms for proteins incubated in null CHO cell medium Equilibria isotherms were generated for and 6-histidine tagged bispecific proteins on three Fractogel EMD Chelate (M) lots of 71, 83, and 94 μmol/mL Cu2+ binding capacities Respective isotherms are shown in 12A and 12B, 12C and 12D, and 12E and 12F Grey and orange respectively represent isotherms generated when the proteins were bound in the presence of null media and without null media Experimental data points are shown (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) the magnitude of the differences were attributed to the presence and interactions of imidazole and null cell media The resin screening experiments were performed in the absence of spiked imidazole In contrast, the isotherm binding experiments mimicked the developed process and were performed in the presence of imidazole Imidazole was present in the developed process to prevent the non-specific binding of media components or undesired cellular expression products to allow for the elution of a product with increased purity Without imidazole in the screening experiments, the purified protein bound to the charged resin, achieving the 48 mg/mL binding capacity In the presence of null cell media and absence of imidazole, the maximum binding value dramatically decreased to 21 mg/mL The decrease was attributed to the competition of the protein with media components or host cell proteins for binding to the resin The presence of imidazole in the isotherm binding experiments prevented the high level of binding of the pu- 16 V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 Fig 13 Impact of null cell medium on binding capacity Dynamic binding capacity profiles were generated for the poly-histidine tagged bispecific protein on Ni2+ charged Sepharose Fast Flow Orange triangles and red squares respectively represent the pure protein and the pure protein in CHO null cell medium The green diamond represents binding of the pure protein to resin washed with CHO null cell medium The blue circle shows the pure protein dialyzed from CHO null cell medium to tris buffered saline (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) rified poly-histidine tagged bispecific protein that was observed in the screening experiments as it may have competed with the protein for binding In this case, a qm value of 39 mg/mL was determined In the presence of both null cell medium and imidazole, the qm value was found to range from 32 to 36 mg/mL In addition to null cell medium and imidazole competing with the protein for binding, it is probable that imidazole interacted with null cell components The interaction of imidazole and null cell medium components may have allowed for increased protein to bind when compared to the null cell spiked resin screening condition A DBC study was performed to study the impact of null cell medium components on poly-histidine tagged bispecific protein binding to IMAC resins The results of the binding study on Ni2+ charged Sepharose Fast Flow resin are shown in Fig 13 DBC profiles, performed with equilibration buffers that did not contain imidazole, were generated for the purified molecule, the molecule in the presence of CHO null cell medium, and the molecule buffer exchanged from the medium into buffer A DBC curve was also created for the purified poly-histidine tagged bispecific protein applied to the charged resin that was washed with null cell culture medium prior to loading The purified poly-histidine tagged bispecific protein achieved a 10% DBC of 57 mg/mL as neither imidazole nor null cell medium were present As expected, a dramatic decrease in binding, to a 10% DBC of 12 mg/mL, was seen when the protein was loaded in the presence of null cell medium as components competed with the poly-histidine bispecific protein for binding Interestingly, this DBC profile matches that of the purified molecule applied to the null cell medium washed column, indicating that medium components interacted with a significant amount of binding sites, effectively blocking the binding of the poly-histidine tagged bispecific protein Multiple media components are likely responsible for the interaction as the 10% DBC value for the null cell spiked molecule dialyzed into buffer is an intermediate level The buffer exchange removed small molecules, less than 3.5 kDa, and retained larger host cell protein molecules Several possibilities can account for the small molecules resulting in decreased binding values in the presence of null cell media First, the metal could have been stripped from the charged ligand by EDTA which is frequently found in cell culture media The scavenging of metal would eliminate binding sites and result in a de- creased binding ability Second, it is possible that the medium contained components that competed with the histidine tag for the charged ligand These medium components included amino acids that exist in zwitterionic form with a deprotonated carboxyl group such as; L-Cystine, L-Glycine, L-glutamine, L-Isoleucine, L-Leucine, L-Lysine, L-Methionine, L-Phenylalanine, L-Serine, L-Threonine, LTryptophan, L-Tyrosine, and L-Valine Media can also contain free L-Histidine in its anionic state In addition to amino acids, salts were abundant in the medium The negative ions comprising the salts included chloride, nitrate, sulfate, bicarbonate, and phosphate groups that could have coordinated with the positively charged ligand and competed with the histidine tag The coordination of free amino acids or negative ions to the charged ligand would decrease the available binding sites for the histidine tagged bispecific protein Finally, it is also important to note that the cell culture medium also contained positively charged metal ions including zinc, cadmium, selenium, copper, barium, tin, and silver These metals could have coordinated with the unbound histidine tagged bispecific protein, blocking the protein from being able to interact with the charged resin Any or the combination of all three events could potentially decrease the binding of histidine tagged bispecific proteins in cell culture media to IMAC resins 3.8 Scale-up chromatography The enhanced understanding of IMAC performance gained from the three studies described in this paper were applied to a gram scale purification of the poly-histidine bispecific protein The goal of the purification was to use IMAC to capture the protein directly from the CHO cell culture harvest that contained, as shown in Fig 14, mostly 6-histidine tagged material and 70% monomer The chromatography operation was to be high yielding and include a step elution that contained both 4-histidine and 6-histidine tagged monomeric proteins Fractogel EMD Chelate (M) was selected for the scale-up purification because, as determined in the described studies, it was able to bind CHO expressed poly-histidine tagged bispecific protein directly from the harvested cell culture product in a cost effective manner One lot of this resin, with 83 μmol/mL Cu2+ binding capacity, was obtained and packed in a 20 cm diameter column to V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 17 Fig 14 Scale-up IMAC purification The bispecific protein was purifed using Fractogel EMD Chelate (M) at the gram scale 14A shows the chromatographic profile of the purification The A280, pH, conductivity, and pressure measurments are displayed 14B shows the mass values for the load and product in terms of histidine tag length and monomer purity Table Application of IMAC protocol to other molecules Various histidine tagged molecules expressed in either CHO or HEK-293 cells were purified on either Fractogel EMD Chelate (M) or IMAC Sepharose Fast Flow resin The yields and purities of the elution products for each purification are listed Histidine tagged protein Molecular weight (KDa) Expression cell line Yield (%) Monomer purity (%) Fab #1 Fab #2 Fab #3 Fab #4 Fab #5 Fab #6 Fab #7 Fab #8 Hormone Fusion #1 Fusion #2 Exotoxin Receptor Receptor Bispecific T Cell Engager #1 Bispecific T Cell Engager #2 Bispecific T Cell Engager #2 50 50 50 50 50 50 50 50 120 71 39 136 150 52 52 52 HEK-293 CHO CHO CHO CHO CHO CHO CHO HEK-293 CHO HEK-293 CHO CHO HEK-293 HEK-293 HEK-293 HEK-293 72 79 85 70 75 72 68 73 60 78 80 75 80 82 85 84 78 95 82 76 81 91 82 79 83 86 75 78 89 74 73 92 88 74 a bed height of 20 cm The column dimensions were consistent with the geometry of the statistical study, resulting in a 6.5 liter CV Based on the size of the column and the quantity of protein expressed, the load capacity was expected to be 18 g/L per cycle In order to ensure the step elution would elute the totality of the 4-histidine and 6-histidine tagged bispecific monomeric proteins at the same ratio as the starting product, data from the statistical study and the equilibrium binding experiments were analyzed The scale-up input parameters of 18 g/L load capacity and 83 μmol/mL Cu2+ binding resin capacity were overlaid on Fig 9A and D The overlay showed that the and 6-histidine tagged bispecific proteins would respectively elute with 129 mM and 211 mM imidazole, while the aggregate would elute with 299 mM imidazole The data also indicated that displacement of the 4-histidine and 6-histidine proteins would not occur as the scale-up parameters overlaid in the homogenous design space for and 6-histidine tags on Fig 9B and E Additionally, the scale-up load material contained a higher ratio of 6-histidine tagged bispecific protein compared to those used in the screening and statistical studies However, the equilibrium binding study in Fig 11 demonstrated that a load of 100% 6histidine tagged bispecific protein showed binding across the range of Cu2+ binding capacity resins in the mM mobile phase In the presence of the 170 mM mobile phase, only minor binding was observed to the 97 μmol/mL resin This indicated that imidazole concentrations greater than 170 mM would elute the pure 6-histidine tagged bispecific protein on the 83 μmol/L Cu2+ binding capacity resin The imidazole concentration required for preventing the binding of the pure 6-histidine tagged material in the binding study was consistent also with the imidazole levels necessary to elute peak in the statistical study Together, the statistical and equilibrium binding studies indicated that 250 mM imidazole step elution would elute both the and 6-histidine tagged bispecific proteins while leaving the high molecular weight aggregate to be stripped by the 500 mM imidazole buffer The scale-up purification also implemented the imidazole spiking of the load material and a wash prior to the elution These steps were included to decrease non-specific binding and allow for 18 V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 separation of host cell protein from the eluted protein The mM imidazole spike was supported by the equilibrium binding data generated in the mM imidazole mobile phase The 75 mM imidazole wash was supported by the statistical study in Fig 9A that showed 110 mM imidazole was the minimal concentration that eluted protein from the weakest binding conditions The optimized IMAC process parameters were included in the purification of harvested cell culture material, as described in section 2.2.7, with successful results Fig 14 shows a representative scale-up chromatogram as well product quality data The product quality of: 28.6% 4-histidine, 67.2% 6-histidine, mg/mL, and 74.4% monomer met the requirements for scale-up processing The IMAC product intermediate was successfully further polished to meet the final product targets 3.9 Application to other protein classes The IMAC conditions developed in this study apply to the purification of molecules of various molecular formats The IMAC procedure described in Section 2.2.7 was applied to a diverse poly-histidine tag containing molecules Table lists the diverse molecules studied, molecular weights, cell lines used for expression, and resulting monomer purities and yields All proteins were directly purified from either CHO or HEK-293 cell harvests Fractogel EMD Chelate (M) charged with 250 mM ZnCl2 or Ni Sepharose Excel was used to capture products expressed in CHO cell culture IMAC Sepharose FF charged with either 250 mM ZnCl2 or CoCl2 5H2 was used to capture proteins expressed in HEK293 cell culture The ease of operation, protein yields, and product purities were consistent with the screening findings and showed a broad application of the developed procedure Polishing of the IMAC products only required one additional aggregate removal operation to deliver material of acceptable purity was loaded The results of the study showed that regardless of the resin lot and protein loading, an elution buffer containing 250 mM imidazole was adequate to elute the desired monomeric and 6histidine species with separation from the high molecular weight aggregate A third set of experiments assessed the binding of 4-histidine and 6-histidine tagged bispecific proteins As cell culture lots contained uncontrolled ratios of 4-histidine tagged and 6-histidine tagged bispecific proteins, equilibria isotherms were generated for the individual species as well as combined at different ratios Isotherms were also generated for different resin lots and in the presence of null expression medium It was observed that although the 4-histidine and 6-histidine tagged bispecific proteins exhibited different kL values, they obtained equivalent qm values However, when combined, 6-histidine tagged bispecific protein competed with and displaced 4-histidine tagged bispecific protein The presence of imidazole and null cell culture medium were also found to decrease the qm values as they competed with the histidine-tagged protein for binding The combination of the three sets of experiments enabled a thorough understanding of the IMAC process for the histidine tagged bispecific protein The results of the resin-metal chelator choice and performance experiments in conjunction with the binding experiments were utilized in the scale-up purification of the poly-histidine tagged bispecific protein as well as proteins of a variety of other molecular classes; Fab, hormone, receptor, and exotoxin For each class of poly-histidine tagged protein, the developed process allowed for the elimination of pre-load conditioning diafiltration The molecules also achieved high protein loading on the IMAC resin All purifications were high-yielding with acceptable product quality to enable successful downstream polishing Academic and industrial scientists can apply these findings and suggested process to quickly enable their purifications both at bench and large scale Conclusions Funding IMAC purification process performance is known to be variable, resulting in unpredictable product quality This study contributes to the understanding of factors impacting its variability and how to control those factors This understanding is critical to providing confidence prior to using IMAC in both bench and large-scale research and development settings The first study in this work revolved around choosing an appropriate resin-metal chelator combination for the purification of a poly-histidine tagged bispecific protein It was found that screening results from the purified protein could not be extrapolated to the protein present in cell culture harvest media, as the greatest dynamic binding capacity for the purified protein was achieved on IMAC Sepharose FF charged with Cu2+ In the presence of HEK-293 medium, the optimal resin and metal combination was IMAC Sepharose FF charged with either Zn2+ or Co2+ The highest binding capacity in CHO conditioned medium occurred on Fractogel EMD Chelate (M) charged with Zn2+ and Ni Sepharose Excel The binding differences were attributed to media components or host cell line by-products that competed for binding and the extent of resin metal binding capacity These identified combinations allowed for the direct capture of expressed protein from media at high loading capacities A second study examined interacting factors that impacted the purification performance of the chosen resin-metal chelator combination, Fractogel EMD Chelate (M) charged with Zn2+ This statistical study showed that the amount of imidazole needed for elution of both the and 6- histidine tagged bispecific proteins as well as the product yield, purity, and histidine-tagged protein ratio were influenced by two main factors Those factors included the metal binding capacity of the resin lot and the extent to which the resin This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper CRediT authorship contribution statement Valeria Riguero: Conceptualization, Methodology, Investigation, Writing - original draft, Resources, Supervision Robert Clifford: Methodology, Investigation, Writing - original draft Michael Dawley: Methodology, Investigation Matthew Dickson: Methodology, Writing - original draft Benjamin Gastfriend: Methodology, Software, Investigation Christopher Thompson: Data curation, Formal analysis Sheau-Chiann Wang: Methodology, Resources Ellen O’Connor: Conceptualization, Methodology, Formal analysis, Investigation, Writing - original draft, Resources, Supervision Acknowledgements We would like to thank Christopher Afdahl, Jenny Feng, Coral Fulton, Jeffrey Gill, Gerard Lacourciere, and Eugene Sun for their initial observations Thanks also to Marcia Carlson for her encouragement to initiate the project We would also like to thank Alan Hunter and William Wang for their careful reviews of the manuscript V Riguero, R Clifford and M Dawley et al / Journal of Chromatography A 1629 (2020) 461505 References [1] C Kittinger, A Barnes, A Hunter, L Machiesky, S Phipps, A Shannon, R Stadelman, S Wilson, E O’Connor, A high yielding IFNAR1 ECD mammalian expression process for use in autoimmune disease drug development, Protein Expres Purif 167 (2020) 1–12 [2] M Flickinger, Downstream industrial chromatography, Recovery and Purification, Wiley, Hoboken, New Jersey, 2013 [3] M Peterka, M Jarc, M Banjac, V Frankovic, K Bencina, M Merhar, V Gaberc– Porekar, V Menart, A Strancar, A Podgornik, Characterisation of metal-chelate methacrylate monoliths, J Chromatogr A 1109 (2006) 80–85 [4] A Kinna, B Tolner, E Rota, N Titchener-Hooker, D Nesbeth, K Chester, IMAC capture of recombinant protein from unclarified mammalian cell feed streams, Biotechnol Bioeng 113 (2016) 130–140 [5] S Wang, S Bai, X Dong, Y Sun, Partitioning behavior of enhanced green fluorescent protein in nickel-chelated affinity-based aqueous two-phase micellar system and its purification from cell lysate, Sep Purif Technol 133 (2014) 149–154 [6] P Fraga Garcia, M Freiherr von Roman, S Reinlein, M Wolf, S Berensmeier, Impact of nanoparticle aggregation on protein recovery through a pentadentate chelate ligand on magnetic carriers, Appl Mater Interfaces 16 (2014) 13607–13616 [7] P Fraga Garcia, M Brammen, M Wolf, S Reinlein, M Freiherr von Roman, S Berensmeier, High-gradient magnetic separation for technical scale protein recovery using low cost magnetic nanoparticles, Sep Purif Technol 150 (2015) 29–36 [8] Y Yin, G Wei, X Zou, Y Zhao, Functionalized hollow silica nanospheres for His-tagged protein purification, Sens Actuators B 209 (2015) 701–705 [9] N Bortone, M Fidaleo, Immobilization of the recombinant (His)6 -tagged l-arabinose isomerase from Thermotoga maritima on epoxy and cupper-chelate epoxy supports, Food Bioprod Process 95 (2015) 155–162 [10] Y Li, X Xiao, R Zhao, K Yu, C Hou, J Liang, Study of the fabrication and characterization of porous Ni using polystyrene sphere template, Synth React Inorg Metal 46 (2016) 286–290 [11] Y Chou, C Ko, L Chen, C Yen, J Shaw, Purification and immobilization of the recombinant Brassica oleracea chlorophyllase (BoCLH1) on DIAION R CR11 as potential biocatalyst for the production of chlorophyllide and phytol, Molecules 20 (2015) 3744–3757 [12] C Zhang, D Fredericks, D Longford, E Campi, T Sawford, M Hearn, Changed loading conditions and lysate composition improve the purity of tagged recombinant proteins with tacn-based IMAC adsorbents, Syst Synth Biol Nano Med 10 (2015) 480–489 [13] J Mooney, D Ferdericks, C Zhang, T Christensen, C Jespergaard, C Schiodt, M Hearn, Purification of a recombinant human growth hormone by an integrated IMAC procedure, Protein Expr Purif 94 (2014) 85–94 19 [14] W Ning, S Wijeratne, J Dong, M Bruening, Immobilization of carboxymethylated polyethylenimine–metal-ion complexes in porous membranes to selectively capture his-tagged protein, Appl Mater Interfaces (2015) 2575–2584 [15] R Prasanna, M Vijayalakshmi, Characterization of metal chelate methacrylate monolithic disk for purification of polyclonal and monoclonal immunoglobulin G, J Chromatogr A 1217 (2010) 3660–3667 [16] M Cheeks, N Kamal, A Sorrell, D Darling, F Farzaneh, N Slater, Immobilized metal affinity chromatography of histidine-tagged lentiviral vectors using monolithic adsorbents, J Chromatogr A 1216 (2009) 2705–2711 [17] D Ren, N Penner, B slentz, F Regnier, Histidine-rich peptide selection and quantification in targeted proteomics, J Proteome Res (2004) 37–45 [18] F Bartnicki, E Kowalska, K Pels, W Strzalka, Imidazole-free purification of His3 -tagged recombinant proteins using ssDNA aptamer-based affinity chromatography, J Chormatogr A 1418 (2015) 130–139 [19] C Karakus, M Uslu, D Yazici, B Salih, Evaluation of immobilized metal affinity chromatography kits for the purification of histidine-tagged recombinant CagA protein, J Chromatogr B 1021 (2016) 182–187 [20] W Choe, R Clemmitt, H Chase, A Middelberg, Comparison of histidine-tag capture chemistries for purification following chemical extraction, J Chromatogr A 953 (2002) 111–121 [21] W Kast, (in German), Adsorption aus der Gasphase: Ingenieurwissenschaftliche Grundlagen und technische Verfahren, VCH, Weinheim, 1988 ISBN 3527267190, OCLC 35459784 [22] Hydrodynamic radius calculator, Fluidic Analytics 2020 https://www.fluidic com/resources/Toolkit/hydrodynamic- radius- Converter/ [23] R Todd, R Johnson, F Arnold, Multiple-site binding interactions in metal-affinity chromatography: I Equilibrium binding of engineered histidine-containing cytochromes C, J Chromatogr A 662 (1994) 13–26 [24] R Johnson, R Todd, F Arnold, Multipoint binding in metal-affinity chromatography II Effect of pH and imidazole on chromatographic retention of engineered histidine-containing cytochromes C, J Chromatogr A 725 (1996) 225–235 [25] M Belew, T Yip, L Andersson, J Porath, Interaction of proteins with immobilized Cu2+ : Quantitation of adsorption capacity, adsorption isotherms and equilibrium constants by frontal analysis, J Chromatogr A 403 (1987) 197–206 [26] S Sharma, G Agarwal, Interactions of proteins with immobilized metal ions: a comparative analysis using various isotherm models, Anal Biochem 288 (2001) 126–140 [27] S Sharma, G Agarwal, Interactions of proteins with immobilized metal ions, J Colloid Interface Sci 243 (2001) 61–72 [28] S Vunnum, S Gallant, Y Kim, S Cramer, Immobilized metal affinity chromatography: Modeling of nonlinear multicomponent equilibrium, Chem Eng Sci 50 (1995) 1785–1803 [29] S Vunnum, S Cramer, IMAC: nonlinear elution chromatography of proteins, Biotechnol Bioeng 54 (1997) 373–390  ... was developed for the purification of a poly-histidine tagged bispecific protein First, a resin and metal chelator screening study was performed to select conditions that would allow for direct protein... at 10% breakthrough for each Table Metal chelators The metals and corresponding salts used for charging the commercially available resins for the screening study are shown Metal Salt Nickel Zinc... contained 85.0% 4-histidine tagged and 12.4% 6-histidine tagged bispecific proteins, while lot contained 76.3% 4-histidine tagged and 20.5% 6-histidine tagged bispecific proteins Three experiments