- Email: [email protected]
Protein sieving characteristics of sub-20-nm pore size filters at varying ionic strength during nanofiltration of Coagulation Factor IX
Biologicals 41 (2013) 176e183
Contents lists available at SciVerse ScienceDirect
Biologicals journal homepage: www.elsevier.com/locate/biologicals
Protein sieving characteristics of sub-20-nm pore size filters at varying ionic strength during nanofiltration of Coagulation Factor IX Clint J. Winkler*,1, Nuria Jorba 1, Kenneth T. Shitanishi, Steven W. Herring Research and Development Department, Grifols Biologicals Inc., Los Angeles, CA, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 26 June 2012 Received in revised form 20 August 2012 Accepted 16 January 2013
Nanofiltration assures that protein therapeutics are free of adventitious agents such as viruses. Nanofilter pores must allow passage of protein drugs but be small enough to retain viruses. Five nanofilters have been evaluated to identify those that can be used interchangeably to yield a high purity Coagulation Factor IX product. When product preparations prior to nanofiltration were analyzed using electrophoresis, Western blot, liquid chromatography e tandem mass spectrometry and size exclusion HPLC, factor IX, inter e a e trypsin inhibitor and C4b binding protein (C4BP) were observed. C4BP was removed from product by all five nanofilters when nanofiltration was performed at physiological ionic strength. However, at high ionic strength, C4BP was removed by only two nanofilters. HPLC indicated that the Stokes radius of C4BP was larger at low ionic strength than at high ionic strength. The results suggest that C4BP exists in an open conformation at physiological ionic strength and is removed by nanofiltration whereas, at high ionic strength, the protein collapses to an extent that allows passage through some nanofilters. Manufacturers should be aware that protein contaminants in other nanofiltered protein drugs could behave similarly and conditions of nanofiltration must be evaluated to ensure consistent product purity. Ó 2013 The International Alliance for Biological Standardization. Published by Elsevier Ltd. All rights reserved.
Keywords: Nanofiltration C4BP Factor IX
1. Introduction Nanofiltration is a technology that has been used over the last 20 years to enhance the safety of protein biopharmaceuticals [1,2]. Nanofilters have been shown to be capable of removing pathogens such as viruses and the agents associated with transmissible spongiform encephalopathies (TSEs) from protein solutions [3,4]. Removal is accomplished via a sieving mechanism. When used to filter bulk biopharmaceuticals, the pore sizes within the nanofilter allow passage of smaller, pharmaceutically active proteins through the filter while retaining viruses and large proteins [5]. In order to remove the smallest viruses, it is desirable to utilize small-pore nanofilters. However, as the size of the filter pore approaches that of the product protein, the protein less readily passes through the filter, resulting in diminished recovery of
* Corresponding author. Research and Development Department, Grifols Biologicals Inc., 5555 Valley Boulevard, Building 310, Los Angeles, CA 90032, USA. Tel.: þ1 323 227 7232; fax: þ1 323 441 7133. E-mail addresses: [email protected] (C.J. Winkler), nuria.jorba@grifols. com (N. Jorba), [email protected] (K.T. Shitanishi), steven.herring@ grifols.com (S.W. Herring). 1 These authors contributed equally.
product in the filtrate. For this reason, nanofilter manufacturers produce filter products of different mean pore size, ranging from a mean of 10 nm (suitable for recovery of proteins with molecular weights < 70 kDa [6]) to a mean of 50 nm (suitable for recovery of proteins with molecular weights up to 300 kDa) [7]). When the active protein within a product is 100 kDa, the use of nanofilters having mean pore sizes of 20 nm or less is preferred in order to maximize removal of viruses and TSEs. Nanofilters with mean pore sizes in this range, regardless of filter manufacturer, appear to perform similarly with respect to viral clearance [8e10]. However, their performance with respect to protein sieving is not as well understood and should be studied further. Such studies are important because the sieving characteristics of a nanofilter can impact removal of high molecular weight protein contaminants from the product and this will ultimately affect product purity. It is therefore important to know if nanofilters of the same porosity from different manufacturers will perform similarly with respect to protein sieving under a variety of operating conditions. We have studied the protein sieving characteristics of several nanofilters with mean pore sizes of 20 nm or less using intermediate process materials from the manufacture of the human Coagulation Factor IX product, AlphaNineÒ SD. This Coagulation Factor IX product is a human plasma derived product that is used
1045-1056/$36.00 Ó 2013 The International Alliance for Biological Standardization. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biologicals.2013.01.001
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
extensively in the treatment of Hemophilia B patients [11,12]. It comprises almost entirely factor IX (FIX) protein with an apparent molecular weight (MWapp) of 65 kDa [13,14]. Summaries of the AlphaNineÒ SD biochemical characteristics, safety profile and manufacturing process have been published [10]. This Coagulation Factor IX product is currently nanofiltered through a ViresolveÒ 70 (V70) filter (Millipore Corporation, Billerica, MA) which has a narrow pore size distribution with a mean pore size that is <20 nm. In an effort to identify nanofilters that may serve as an alternate to the V70 nanofilter, we evaluated four other nanofilters having mean pore sizes 20 nm. Results suggest that the nanofilters have different protein sieving characteristics and that ionic strength during nanofiltration can affect the molecular structure and sieving characteristics of one accompanying protein, C4b-binding protein (C4BP), in the FIX preparation. 2. Material and methods 2.1. Nanofiltration sample The Coagulation Factor IX bulk material used in these studies was obtained from the Coagulation Factor IX (AlphaNineÒ SD) manufacturing process at a stage just prior to nanofiltration. This material had undergone all purification steps in the manufacturing process and represented the elution pool from the final chromatography column in the process. The elution pool contained sodium citrate and sodium chloride at pH 6.8. The sodium concentration of the elution pools used in this study was 0.58 0.08 molar equivalents (Eq) per L. For some experiments, elution pools with lower sodium concentrations (i.e., 0.12, 0.08 and 0.06 Eq/L) were prepared by diluting the sample at nominal sodium concentration with buffer containing 0.06 Eq/L Na (to reach a concentration of 0.12 Eq/L) or with water (to reach a concentration of <0.1 Eq/L). After salt adjustment, the samples were ultrafiltered in order to reach the approximate protein concentration of the nominal sample. 2.2. Nanofiltration Nanofiltration through V70 cassettes (Millipore Corporation, Billerica, MA, USA) was performed at process scale (1 m2) in tangential flow mode using pumps before and after the nanofilter to control flow. Nanofiltration through PlanovaÒ 15 hollow fiber units (Asahi Kasei Medical Co. Ltd., Glenview, IL, USA) was performed at process (1 m2) and laboratory (0.001 m2) scale in constant flow “dead-end” mode using a pump ahead of the filter to control flow. Process scale and laboratory scale filtrations with the PlanovaÒ 15 were performed under equivalent conditions (volume per filter surface area, protein mass per surface area, flux) and generated equivalent results. All other nanofiltrations were performed at laboratory scale (0.001 m2). In all cases, the membranes were rinsed with water and equilibrated with 0.02 M sodium citrate buffer, pH 6.8, containing sodium chloride at amounts appropriate to adjust the buffer sodium concentration to that of the filtration sample. All steps were performed at 4 C. The Planova BioEX (Asahi Kasei Medical Co. Ltd., Glenview, IL, USA) filtrations were performed in constant flow “dead-end” mode with a pump ahead of the filter. The Viresolve Pro Micro (3.1 cm2) (Millipore Corporation, Billerica, MA, USA) and DV20 (0.001 m2) (Pall Corporation, Covina, CA, USA) filtrations were performed under a constant pressure of 30 psi using a regulated nitrogen stream. In all cases, the filters were used according to the manufacturer’s recommended specifications and pressure was monitored (PressureMAT sensor, PendoTECH, Princeton, NJ, USA). Filters run at constant flow were adjusted to maintain recommended pressure limits if fouling of the membrane occurred.
177
2.3. Electrophoresis and western blotting Protein samples before and after nanofiltration were separated by SDS-PAGE on Novex Tris-Glycine Gels (Invitrogen, Carlsbad, CA, USA) and visualized by protein staining or Western blot analysis. Acrylamide gels were run at constant voltage (125 V) in Novex TrisGlycine SDS Running Buffer (Invitrogen, Carlsbad, CA, USA) for approximately 90 min. Total protein was observed by staining either with Colloidal Blue Staining Kit (Invitrogen, Carlsbad, CA, USA) or IRDye Blue Protein Stain (LI-COR, Lincoln, NE, USA) according the manufacturer’s recommended protocol. Gels stained with IRDye Blue Protein Stain were imaged with an Odyssey Imager (LI-COR, Lincoln, NE, USA) at 700 nm. FIX and inter-a-trypsin inhibitor identities were verified by Western blot analysis. For Western blots, proteins were transferred to an ImmobilonPVDF membrane (Millipore Corporation, Billerica, MA, USA) in 1X Tris-Glycine transfer buffer. Membranes were blocked with 1.5% albumin, 1X TBS. The membranes were then incubated with the manufacturer’s recommended concentration of primary antibody (ITI-H1 mouse monoclonal antibody [SC-69788, clone# 40B10], Santa Cruz Biotechnology, Santa Cruz, CA, USA; FIX mouse monoclonal antibody [AHIX-5041], Haematologic Technologies Inc, Essex Junction, VT, USA) in 1X TBS, 0.5% albumin, 0.01% SDS, 0.1% Tween20. The non-specifically bound primary antibody was washed from the membrane with 1X TBS, 0.01% Tween-20. Goat-anti-mouse IRDye 680 (LI-COR, Lincoln, NE, USA) was diluted 1:10,000 in 1X TBS, 0.01% Tween-20 and incubated with the membrane. Non-specifically bound secondary antibody was removed by washing the blot 3 times with 1X TBS, 0.01% Tween-20 and 1 time with 1X TBS. The membranes were imaged with an Odyssey Imager at 700 nm. The identity of C4BP was verified by in-gel Western analysis. For the in-gel Western, the gel was fixed with 5% acetic acid, 50% isopropanol and then rinsed with water. The C4BP mouse monoclonal antibody (1 mg/ml) (05-867, clone# GMA-036, Millipore Corporation, Billerica, MA, USA) was diluted 1:250 in 1X TBS, 5% albumin, 0.1% Tween-20. The gel was incubated for 1 h in the primary antibody before washing away the non-specifically bound antibody with 1X TBS, 0.1% Tween-20. Goat-anti-mouse IRDye 680 was diluted 1:2000 in 1X TBS, 0.5% albumin and 0.1% Tween-20 and incubated with the gel. Non-specifically bound secondary antibody was washed away with 1X TBS, 0.1% Tween-20. The gel was imaged with an Odyssey Imager at 700 nm. 2.4. Liquid chromatography tandem mass spectrometry (LC-MS-MS) The Coagulation Factor IX filtrate from nanofilter A was resolved by non-reducing SDS-PAGE. The resolved FIX protein band and two accompanying bands were excised from the gel. The protein bands were subjected to in-gel trypsin digestion (Sequencing grade Trypsin [V511A], Promega, Madison, WI, USA). The trypsin digested samples were subjected to LC-MS-MS on an LT-QFT Mass Spectrometer (Thermo Scientific, Waltham, MA, USA) at The Pasarow Mass Spectrometry Laboratory, The NPI-Semel Institute, David Geffen School of Medicine, UCLA (Los Angeles, CA, USA). Ions generated from LC-MSMS were given an ion score defined as 10Log(P), where P is the calculated probability that the observed match between the experimental data and the protein database sequence is a random event. A probability-based Mowse Score was generated from the combined ion scores using the Mowse algorithm to determine the most significant protein candidates in each gel band. 2.5. HPLC size exclusion chromatography (HPSEC) The same Coagulation Factor IX bulk material used for the nanofiltration was concentrated to a target A280 of 3.6 units in high
178
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
(0.56 Eq/L Na) or low (0.06 Eq/L Na) buffer. Equivalent amounts (w20 mL) of protein sample were injected on a TSK-GEL G4000SW column (7.5 mm 60 cm) with a TSK-GEL Guard column SWXL (6 mm 4 cm) (TOSOH America Inc, Grove City, OH). The sample was eluted under high salt conditions (0.56 Na Eq/L) (83% e 600 mM NaCl, 20 mM Na Citrate, pH 6.8, 17% e 20 mM Na Citrate, pH 6.8) and low salt conditions (0.06 Na Eq/L) (20 mM Na Citrate, pH 6.8) at a flow rate of 0.5 ml/min. A series of pre-runs were performed to condition the columns to the salt and protein conditions. After conditioning, the results of 5 runs were collected for quantification. Protein elution was monitored by absorbance at 280 nm. One minute time based fractions were collected for analysis of peaks by SDS-PAGE/Western Blot. A two-tailed t-test for unpaired samples assuming equal variances was used for statistical analysis of peak areas. A P < 0.05 was considered statistically significant. 3. Results Four different nanofilters were evaluated for their protein sieving characteristics during nanofiltration of a Coagulation Factor IX (AlphaNineÒ SD) bulk eluate and results were compared to
results obtained with the V70 nanofilter currently used to nanofilter the Coagulation Factor IX bulk eluate. All nanofilters, other than the V70 are identified only by a randomly assigned letter from A to D. All nanofilters had mean pore sizes of approximately 20 nm or less. At the start of nanofiltration, samples were placed in reservoirs. Flow through the nanofilters was controlled with a pump or via pressurized gas (N2). Filtration was allowed to proceed until the entire sample volume was filtered or sample was no longer flowing through the nanofilter (“plugging”). The normalized volume of eluate that could be passed through each nanofilter prior to plugging ranged from 34 L/m2 to 48 L/m2. 3.1. Nanofiltration using the V70 (current) nanofilter Samples before and after nanofiltration through the V70 were analyzed by SDS-PAGE under non-reducing and reducing conditions (Fig. 1A and B, respectively). Under both non-reducing and reducing conditions, preparations before and after V70 nanofiltration contained a major protein band with an apparent molecular weight (MWapp) of 65 kDa which corresponds to the MWapp of FIX. When analyzed under non-reducing conditions (Fig. 1A),
Fig. 1. SDS-PAGE analysis of the Coagulation Factor IX bulk eluate before and after nanofiltration. Non-reduced gel (a) comparing V70 and filter A: lane 1, molecular weight markers; lane 2, FIX bulk; lane 3, FIX bulk post V70 nanofiltration; lane 4, FIX bulk; lane 5, FIX bulk post filtration with nanofilter A. Reduced gel (b) comparing V70 and filter A: lane 1, molecular weight markers; lane 2, FIX bulk; lane 3, FIX bulk post V70 nanofiltration; lane 4, FIX bulk; lane 5, FIX bulk post filtration with nanofilter A. The 70 kDa band observed in reducing conditions is marked with an asterisk. Non-reduced gel (c) comparing filters B,C and D: lane 1, FIX bulk; lane 2, FIX bulk post filtration with nanofilter B; lane 3, FIX bulk; lane 4, FIX bulk post 0.22 mm filtration; lane 5, FIX bulk post 0.22 mm filtration and filtration with nanofilter C; lane 6, FIX bulk; lane 7, FIX bulk post 0.22 mm filtration; lane 8, FIX bulk post 0.22 mm filtration and filtration with nanofilter D.
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
two minor bands were observed in the source material prior to V70 nanofiltration. One of these bands had a MWapp of 210 kDa and the other had a MWapp of >500 kDa (i.e., outside the resolving range of the gel). When the same source material was analyzed under reducing conditions (Fig. 1B), the 210 kDa protein band was observed but not the >500 kDa high molecular weight protein (HMWP) band. Instead, a protein band with MWapp of 70 kDa (marked by asterisk), not seen under non-reducing conditions, was observed, suggesting that the HMWP was an oligomer comprising 70 kDa disulfide linked subunits. Following nanofiltration, analysis under non-reducing conditions showed that the 210 kDa and 65 kDa protein bands were present in the filtered product but not the HMWP (Fig. 1A) indicating that the HMWP (but not the 210 kDa and 65 kDa proteins) were removed by V70 filtration. Under reducing conditions, the 210 kDa and 65 kDa protein bands were present in the filtered product but not the 70 kDa band (Fig. 1B). Since filtration results for the 65 kDa and 210 kDa proteins indicate that a 70 kDa protein should not have been removed by the V70, absence of the 70 kDa protein in the V70 filtered product confirms that this polypeptide was a subunit of the larger HMWP that was removed by the V70. 3.2. Nanofiltration through nanofilters A, B, C, and D Fig. 1 also shows SDS-PAGE results for samples before and after filtration through nanofilter A. When samples were analyzed under non-reducing conditions, the HMWP was present in both the source material and the filtered material (Fig. 1A). Under reducing conditions, the HMWP 70 kDa subunit was also observed in the source material and the filtered material (Fig. 1B). These results indicate that nanofilter A did not remove the HMWP. Fig. 1C, shows SDS-PAGE results from samples before and after filtration through nanofilters B, C and D and the 0.22 mm pre-filter. The SDS-PAGE was performed under non-reducing conditions in order to visualize the HMWP. Pre-filtration did not affect the protein composition of the product. However, filtration through nanofilter B completely removed the HMWP. Filtration through nanofilter D only partly removed the HMWP and nanofiltration through filter C did not remove the HMWP.
179
matches for the 210 kDa band were all components of inter-atrypsin inhibitor (ITI) and included the heavy chain isoforms, H1 and H2, and the light chain, bikunin [15]. The highest probability match for the 65 kDa protein was FIX. The molecular weights of all identified proteins were consistent with the protein mobility observed in SDS-PAGE. The LC-MS-MS results were confirmed by in-gel Western and Western blot analyses. As shown in Fig. 2, when the elution pool prior to nanofiltration was analyzed, a mouse monoclonal antibody specific for C4BP recognized the HMWP band and monoclonal antibodies specific for ITI and FIX recognized the 200 kDa and 65 kDa bands, respectively. 3.4. Effect of ionic strength We considered the possibility that the high ionic strength in the elution pool might affect C4BP structure, allowing its passage through the pores of some nanofilters. Therefore, additional nanofiltration studies were performed under conditions of decreasing sodium concentration. Filtration through nanofilter A (which was previously shown not to remove C4BP) was performed at high (0.50 Eq/L), mid (0.12 Eq/L) and low (0.06 Eq/L) levels of sodium and the nanofiltered samples analyzed by SDS-PAGE as shown in Fig. 3A. The protein concentration was kept constant for all sodium conditions. Pre-filtration through a 0.22 mm filter before nanofiltration did not change the protein composition of the source material. At 0.50 Eq/L of sodium, C4BP was not removed by nanofilter A. At 0.12 Eq/L, some, but not all of the C4BP was removed and at 0.06 Eq/L essentially all of the C4BP was removed. Quantification of gel bands and normalization to FIX protein indicated that removal of C4BP was dependent upon the sodium concentration during nanofiltration (Fig. 3B). In contrast, ITI, despite also having a relatively high apparent molecular weight, was not removed from the product under any of the test conditions (Fig. 3A and B). We considered the possibility that nanofilters C and D which, like nanofilter A, did not remove C4BP at high sodium levels, could
3.3. Identification of nanofiltered proteins The HMWP and the two other protein bands observed in the gels (MWapp of 210 kDa and 65 kDa) were identified by LC-MS-MS. Table 1 shows the three highest probability matches for each protein band as determined by LC-MS-MS and Mowse scoring. The highest probability matches for the HMWP were the C4b-binding protein (C4BP) alpha and beta chains. The highest probability
Table 1 LC-MS-MS protein identification. Protein name and rank ( )
Protein GI number
High molecular weight protein (>500 kDa) band 4502503 (1) C4BP [a chain] 4502505 (2) C4BP [b chain] 85544070 (3) C4BP [a chain, N-terminal fragment] 210 kDa protein band (1) ITI [H1, heavy chain] 189054356 (2) ITI [H2, heavy chain] 55958063 (3) ITI [alpha-1-microglobulin, light chain] 579676 65 kDa protein band (1) FIX 182613 (2) FIX [N-terminal fragment] 100017830 (3) Low scoring plasma proteins N/A
Probability based Mowse score 2648 147 106 1927 1894 390 7731 4513 238
Fig. 2. In-gel western of C4BP and western blot detection of inter-a-trypsin inhibitor and Coagulation Factor IX. Lanes 1, 3 and 5, molecular weight markers; lane 2, detection of C4BP with a monoclonal antibody by in-gel western; lane 4, detection of inter-a-trypsin inhibitor with a monoclonal antibody by western blot; lane 6, detection of FIX with a monoclonal antibody by western blot.
180
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
this peak consisted solely of FIX (data not shown). A broad peak eluting from 30 to 35 min under both high and low salt conditions was shown to consist of ITI when analyzed by Western blotting. In the lower portion of Fig. 5, the absorbance scale was expanded for the elution volume from 15 to 30 min in order to better visualize the high molecular weight peaks. In this region, two peaks were observed, an early eluting peak with a retention time of about 20 min and a later eluting peak with a retention time of about 24.5 min. Both peaks have been shown to consist of C4BP by Western blot (results not shown), suggesting two populations for C4BP having different Stokes radii. For the earlier eluting peak, the area under the peak was increased two-fold under low salt conditions when compared to the peak area obtained under high salt conditions (0.21 0.01% versus 0.11 0.01% of total area). In contrast, for the later eluting peak, the area under the peak was decreased almost three-fold under low salt conditions when compared to the peak area obtained under high salt conditions (4.18 0.37% versus 1.32 0.11% of total area). In both cases, the observed change in peak area due to salt concentration was highly significant (P < 0.001). Fig. 5C presents graphically the average peak areas from five runs under each condition for both peaks. Similar shifts in peak areas did not occur for either ITI or FIX under high and low salt conditions. The results indicate that, under high salt conditions, there is a greater percentage of C4BP with a smaller Stokes radius (later retention time) than under low salt conditions. Conversely, under low salt conditions, there is a greater percentage of C4BP with a larger Stokes radius (earlier retention time) than under high salt conditions. 4. Discussion
Fig. 3. C4BP removal at varying sodium concentrations. SDS-PAGE analysis (A) of nanofilter A filtrates was performed in non-reducing conditions. The gel was stained with IRDye 680 (Odyssey, LI-COR system). Reduction of sodium concentration results in a dose dependent increase in the removal of C4BP and decreased amounts of C4BP in the filtrate after nanofiltration. Ratios of C4BP to FIX and ITI to FIX after nanofiltration divided by the ratios before nanofiltration (B). A ratio of 1.0 indicates no removal of C4BP or ITI.
remove the protein at a lower sodium level. Therefore, additional nanofiltration studies were performed with nanofilters C and D together with nanofilter A under conditions of low sodium concentration (0.08 Eq/L). The nanofiltered samples were analyzed by SDS-PAGE under non-reducing conditions as shown in Fig. 4. C4BP was removed during filtration of the Coagulation Factor IX bulk eluate through all nanofilters at low sodium, although to a lesser extent when passed through nanofilter C as indicated by the presence of C4BP in filtrate from nanofilter C. 3.5. HPLC size exclusion chromatography (HPSEC) The effect of ionic strength on C4BP structure was evaluated by monitoring changes in the Stokes radius of the protein via HPSEC. Fig. 5A and B show representative chromatograms of the Coagulation Factor IX bulk eluate prior to nanofiltration analyzed under conditions of high and low sodium, respectively. FIX elutes as the largest peak with a retention time of 38.7 min at high salt and 36.7 min at low salt. SDS-PAGE and Western blot analysis confirmed that
Nanofiltration is a procedure that is included in the manufacture of many plasma derived products to provide additional assurance that these products are free of adventitious agents such as viruses. However, as described here, the nanofiltration step included at the end of the AlphaNineÒ SD manufacturing process can also remove large plasma proteins such as C4BP, small amounts of which can accompany the FIX protein throughout the purification process [16]. Results presented here provide insight into the molecular structure of C4BP and the sieving characteristics of several nanofilters, all of similar pore size (<20 nm), from different manufacturers. C4BP is a large plasma glycoprotein having a molecular weight of 570 kDa [17]. It consists of seven a-chains connected to one b-chain by disulfide interactions [18e21]. The a-chains contain the binding site for C4b while the b-chain contains a high-affinity binding site for the vitamin K-dependent protein S [22,23]. Electron microscopy has shown that the a-chains are rod-like proteins of about 330 Å in length [24]. The shape of C4BP has often been described as octopus-like with the a-chains forming tentacle-like structures surrounding the core b-chain. This open structure would have a diameter of approximately 66 nm and should not pass through nanofilters with pores 20 nm. The observation that C4BP did pass through several such nanofilters at sodium concentrations >0.5 Eq/L but not at concentrations <0.1 Eq/L suggested differences in its structure related to ionic strength. We considered the possibility that the change in C4BP structure was due to the binding of C4b/protein S to C4BP at low salt and the dissociation of these proteins at high salt. However, when C4BP was isolated from Coagulation Factor IX preparations under high and low salt conditions, only a small amount of a protein S-sized material was observed (by SDS-PAGE) and the amount relative to C4BP remained constant regardless of salt concentration (results not shown). Therefore, we focused on the possibility that C4BP undergoes salt concentration-dependent conformational changes and
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
181
Fig. 4. Non-reducing SDS-PAGE analysis of Coagulation Factor IX elution pool following nanofiltration at a sodium concentration of 0.07e0.08 Eq/L. Lane 1, FIX bulk; lane 2, FIX bulk post 0.22 mm filtration; lane 3, FIX bulk post filter A nanofiltration; lane 4, FIX bulk; lane 5, FIX bulk post 0.22 mm filtration; lane 6, FIX bulk post filter C nanofiltration; lane 7, FIX bulk; lane 8, FIX bulk post 0.22 mm filtration; lane 9, FIX bulk post filter D nanofiltration.
theorized that, at the lower ion concentrations, pockets of similarly charged residues on the a-chains prevent them from moving close to each other, resulting in an open structure that cannot penetrate the nanofilter pores. In contrast, at the higher ion concentrations, charges on the a-chains would be neutralized, allowing the a-chains to collapse on one another forming a long, perhaps flexible, bundle of a-chain fibers. This is consistent with studies that have identified clusters of positively charged amino acids on the a-chains of C4BP [25,26]. Since the diameter of the individual a-chain is about 3 nm, the overall diameter of the bundle could be much less than 20 nm and the length of the bundle about half that of the open structure. This hypothesis seemed plausible since Perkins et al. [27] had proposed such a structure for C4BP based on data from solution x-ray scattering experiments performed on C4BP and had calculated a radius of gyration, RG, of 13 nm for C4BP. This RG is less than half the radius of the open C4BP structure and makes it more likely that C4BP would penetrate nanofilter pores with approximate diameters of 20 nm. The light scattering studies were performed at 0.25 Eq/L NaCl and we propose that at higher salt concentrations, such as those present during the nanofiltration of the Coagulation Factor IX bulks (i.e., >0.5 Eq/L), greater charge neutralization occurs, resulting in an even more compact and flexible C4BP structure with increased potential for passage through the nanofilter. This scenario is consistent with results obtained for nanofilters A, C and D in this study, all of which allowed substantial passage of C4BP at high salt concentrations but little, if any, passage at low salt concentrations. The studies performed with nanofilter A showed clearly that passage of C4BP through this nanofilter was salt concentration dependent,
with no C4BP removed at the highest salt concentration (>0.5 Eq/ L), all C4BP removed at the lowest salt concentration (0.06 Eq/L) and an intermediate amount removed at a mid-range concentration (0.12 Eq/L). The results from our HPSEC studies indicated that the Stokes radius of C4BP was smaller at a high salt concentration (0.6 Na Eq/L) than at a low salt concentration (0.06 Na Eq/L) and provides additional evidence that a more compact C4BP structure at the higher salt is responsible for passage of C4BP through some, but not all, of the nanofilters evaluated. Two of the five nanofilters evaluated, the V70 and nanofilter B, did not permit passage of C4BP at any salt condition, suggesting that the mean pore size of these two filters may be smaller than that of the other nanofilters. However, it should be mentioned that any difference in pore size between the two groups of nanofilters is likely small; none of the filters in either group retained ITI, another large (250 kDa) plasma protein, at any salt condition studied. Differences in molecular sieving characteristics between these nanofilters of similar porosity seem to be limited to large proteins whose shape or flexibility is altered under different ionic conditions. It should be emphasized that no differences in virus retention were noted when the virus removal capacity of a nanofilter that did not retain C4BP was compared to nanofilters that did (results not shown). This was true for the smallest virus studied, porcine parvovirus (25 nm), at the highest salt concentration (0.6 Na Eq/L). Our results emphasize the need for manufacturers of biological products to evaluate the effect of a nanofilter on product (protein) composition over the full range of anticipated operating ionic strengths prior to introduction of the nanofilter into the process.
182
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
Fig. 5. HPSEC of high and low salt Coagulation Factor IX bulks. High salt (A) and low salt (B) chromatograms of Coagulation Factor IX bulks are shown. Five HPSEC runs for each condition were performed to confirm reproducibility. One representative chromatogram for each condition is shown. In each chromatogram, the y-axis (mAU280 nm) in the range of 15e30 min is magnified to show the two C4BP peaks. The percentage of total area of each peak as an average of the five HPSEC runs is compared for high and low salt conditions (C).
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
Acknowledgments We thank Zong-Zhi Huang and Evelyn Nario for the excellent technical assistance that they provided for this study. All authors have contributed significantly to the design and performance of experiments, the interpretation of results, or the writing and criticism of the manuscript.
[13]
[14]
[15]
References [16] [1] Burnouf T, Radosevich M. Nanofiltration of plasma-derived biopharmaceutical products. Haemophilia: The Official Journal of the World Federation of Hemophilia 2003;9:24e37. [2] Barin F. Viral safety of biologicals. Annales Pharmaceutiques Francaises 2008; 66:129e39. [3] Diez JM, Caballero S, Belda F, Otegui M, Gajardo R, Jorquera JI. Capacity of the manufacturing process of Flebogamma((R)) DIF, a new human high purity intravenous immunoglobulin, to remove a TSE model-agent. Biologicals: Journal of the International Association of Biological Standardization 2010;38: 670e4. [4] Yunoki M, Tanaka H, Urayama T, Hattori S, Ohtani M, Ohkubo Y, et al. Prion removal by nanofiltration under different experimental conditions. Biologicals: Journal of the International Association of Biological Standardization 2008;36: 27e36. [5] Hoffer L, Schwinn H, Biesert L, Josic D. Improved virus safety and purity of a chromatographically produced factor IX concentrate by nanofiltration. Journal of Chromatography B, Biomedical Applications 1995;669:187e96. [6] DiLeo AJ, Allegrezza Jr AE, Builder SE. High resolution removal of virus from protein solutions using a membrane of unique structure. Biotechnology (N Y) 1992;10:182e8. [7] Chang CJ, Yang CS, Chuang YJ, Khoo HS, Tseng FG. Micro-patternable nanoporous polymer integrated with microstructures for molecular filtration. Nanotechnology 2008;19:365301. [8] Tomokiyo K, Yano H, Imamura M, Nakano Y, Nakagaki T, Ogata Y, et al. Largescale production and properties of human plasma-derived activated factor VII concentrate. Vox Sanguinis 2003;84:54e64. [9] Smith MP, Berrie JR. Selecting a virus reduction filter and optimising performance during process scale-up. Recovery of Biological Products XI, Banff; 2003. [10] Herring S. Grifols’ factor IX concentrates: new findings in biochemical characteristics and safety profiles. Haemophilia: The Official Journal of the World Federation of Hemophilia 2010;16(Suppl. 6):3e8. [11] Goldsmith JC, Kasper CK, Blatt PM, Gomperts ED, Kessler CM, Thompson AR, et al. Coagulation factor IX: successful surgical experience with a purified factor IX concentrate. American Journal of Hematology 1992;40:210e5. [12] Lissitchkov T, Matysiak M, Zavilska K, Laguna P, Gercheva L, Antonov A, et al. A clinical study assessing the pharmacokinetics, efficacy and safety of AlphaNine((R)), a high-purity factor IX concentrate, in patients with severe
[17]
[18] [19]
[20]
[21]
[22]
[23] [24]
[25]
[26]
[27]
183
haemophilia B. Haemophilia: The Official Journal of the World Federation of Hemophilia 2011;17:590e6. Herring SW, Abildgaard C, Shitanishi KT, Harrison J, Gendler S, Heldebrant CM. Human coagulation factor IX: assessment of thrombogenicity in animal models and viral safety. The Journal of Laboratory and Clinical Medicine 1993; 121:394e405. Pock K, Buchacher A, Rizzi A, Josic D. Characterization of clotting factor IX in plasma-derived preparations by electrophoretic techniques. Journal of Chromatography A 2001;921:57e67. Bost F, Diarra-Mehrpour M, Martin JP. Inter-alpha-trypsin inhibitor proteoglycan familyea group of proteins binding and stabilizing the extracellular matrix. European Journal of Biochemistry/FEBS 1998;252:339e46. Clifton J, Huang F, Gaso-Sokac D, Brilliant K, Hixson D, Josic D. Use of proteomics for validation of the isolation process of clotting factor IX from human plasma. Journal of Proteomics 2010;73:678e88. Dahlback B. Purification of human C4b-binding protein and formation of its complex with vitamin K-dependent protein S. The Biochemical Journal 1983; 209:847e56. Blom AM. Structural and functional studies of complement inhibitor C4bbinding protein. Biochemical Society Transactions 2002;30:978e82. Blom AM, Villoutreix BO, Dahlback B. Functions of human complement inhibitor C4b-binding protein in relation to its structure. Archivum immunologiae et therapiae experimentalis 2004;52:83e95. Blom AM, Kask L, Dahlback B. Structural requirements for the complement regulatory activities of C4BP. The Journal of Biological Chemistry 2001;276: 27136e44. Kask L, Hillarp A, Ramesh B, Dahlback B, Blom AM. Structural requirements for the intracellular subunit polymerization of the complement inhibitor C4bbinding protein. Biochemistry 2002;41:9349e57. Webb JH, Villoutreix BO, Dahlback B, Blom AM. Localization of a hydrophobic binding site for anticoagulant protein S on the beta -chain of complement regulator C4b-binding protein. The Journal of Biological Chemistry 2001;276: 4330e7. Hillarp A, Hessing M, Dahlback B. Protein S binding in relation to the subunit composition of human C4b-binding protein. FEBS Letters 1989;259:53e6. Dahlback B, Smith CA, Muller-Eberhard HJ. Visualization of human C4bbinding protein and its complexes with vitamin K-dependent protein S and complement protein C4b. Proceedings of the National Academy of Sciences of the United States of America 1983;80:3461e5. Blom AM, Webb J, Villoutreix BO, Dahlback B. A cluster of positively charged amino acids in the C4BP alpha-chain is crucial for C4b binding and factor I cofactor function. The Journal of Biological Chemistry 1999;274:19237e45. Jenkins HT, Mark L, Ball G, Persson J, Lindahl G, Uhrin D, et al. Human C4bbinding protein, structural basis for interaction with streptococcal M protein, a major bacterial virulence factor. The Journal of Biological Chemistry 2006;281:3690e7. Perkins SJ, Chung LP, Reid KB. Unusual ultrastructure of complementcomponent-C4b-binding protein of human complement by synchrotron X-ray scattering and hydrodynamic analysis. The Biochemical Journal 1986; 233:799e807.
Contents lists available at SciVerse ScienceDirect
Biologicals journal homepage: www.elsevier.com/locate/biologicals
Protein sieving characteristics of sub-20-nm pore size filters at varying ionic strength during nanofiltration of Coagulation Factor IX Clint J. Winkler*,1, Nuria Jorba 1, Kenneth T. Shitanishi, Steven W. Herring Research and Development Department, Grifols Biologicals Inc., Los Angeles, CA, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 26 June 2012 Received in revised form 20 August 2012 Accepted 16 January 2013
Nanofiltration assures that protein therapeutics are free of adventitious agents such as viruses. Nanofilter pores must allow passage of protein drugs but be small enough to retain viruses. Five nanofilters have been evaluated to identify those that can be used interchangeably to yield a high purity Coagulation Factor IX product. When product preparations prior to nanofiltration were analyzed using electrophoresis, Western blot, liquid chromatography e tandem mass spectrometry and size exclusion HPLC, factor IX, inter e a e trypsin inhibitor and C4b binding protein (C4BP) were observed. C4BP was removed from product by all five nanofilters when nanofiltration was performed at physiological ionic strength. However, at high ionic strength, C4BP was removed by only two nanofilters. HPLC indicated that the Stokes radius of C4BP was larger at low ionic strength than at high ionic strength. The results suggest that C4BP exists in an open conformation at physiological ionic strength and is removed by nanofiltration whereas, at high ionic strength, the protein collapses to an extent that allows passage through some nanofilters. Manufacturers should be aware that protein contaminants in other nanofiltered protein drugs could behave similarly and conditions of nanofiltration must be evaluated to ensure consistent product purity. Ó 2013 The International Alliance for Biological Standardization. Published by Elsevier Ltd. All rights reserved.
Keywords: Nanofiltration C4BP Factor IX
1. Introduction Nanofiltration is a technology that has been used over the last 20 years to enhance the safety of protein biopharmaceuticals [1,2]. Nanofilters have been shown to be capable of removing pathogens such as viruses and the agents associated with transmissible spongiform encephalopathies (TSEs) from protein solutions [3,4]. Removal is accomplished via a sieving mechanism. When used to filter bulk biopharmaceuticals, the pore sizes within the nanofilter allow passage of smaller, pharmaceutically active proteins through the filter while retaining viruses and large proteins [5]. In order to remove the smallest viruses, it is desirable to utilize small-pore nanofilters. However, as the size of the filter pore approaches that of the product protein, the protein less readily passes through the filter, resulting in diminished recovery of
* Corresponding author. Research and Development Department, Grifols Biologicals Inc., 5555 Valley Boulevard, Building 310, Los Angeles, CA 90032, USA. Tel.: þ1 323 227 7232; fax: þ1 323 441 7133. E-mail addresses: [email protected] (C.J. Winkler), nuria.jorba@grifols. com (N. Jorba), [email protected] (K.T. Shitanishi), steven.herring@ grifols.com (S.W. Herring). 1 These authors contributed equally.
product in the filtrate. For this reason, nanofilter manufacturers produce filter products of different mean pore size, ranging from a mean of 10 nm (suitable for recovery of proteins with molecular weights < 70 kDa [6]) to a mean of 50 nm (suitable for recovery of proteins with molecular weights up to 300 kDa) [7]). When the active protein within a product is 100 kDa, the use of nanofilters having mean pore sizes of 20 nm or less is preferred in order to maximize removal of viruses and TSEs. Nanofilters with mean pore sizes in this range, regardless of filter manufacturer, appear to perform similarly with respect to viral clearance [8e10]. However, their performance with respect to protein sieving is not as well understood and should be studied further. Such studies are important because the sieving characteristics of a nanofilter can impact removal of high molecular weight protein contaminants from the product and this will ultimately affect product purity. It is therefore important to know if nanofilters of the same porosity from different manufacturers will perform similarly with respect to protein sieving under a variety of operating conditions. We have studied the protein sieving characteristics of several nanofilters with mean pore sizes of 20 nm or less using intermediate process materials from the manufacture of the human Coagulation Factor IX product, AlphaNineÒ SD. This Coagulation Factor IX product is a human plasma derived product that is used
1045-1056/$36.00 Ó 2013 The International Alliance for Biological Standardization. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biologicals.2013.01.001
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
extensively in the treatment of Hemophilia B patients [11,12]. It comprises almost entirely factor IX (FIX) protein with an apparent molecular weight (MWapp) of 65 kDa [13,14]. Summaries of the AlphaNineÒ SD biochemical characteristics, safety profile and manufacturing process have been published [10]. This Coagulation Factor IX product is currently nanofiltered through a ViresolveÒ 70 (V70) filter (Millipore Corporation, Billerica, MA) which has a narrow pore size distribution with a mean pore size that is <20 nm. In an effort to identify nanofilters that may serve as an alternate to the V70 nanofilter, we evaluated four other nanofilters having mean pore sizes 20 nm. Results suggest that the nanofilters have different protein sieving characteristics and that ionic strength during nanofiltration can affect the molecular structure and sieving characteristics of one accompanying protein, C4b-binding protein (C4BP), in the FIX preparation. 2. Material and methods 2.1. Nanofiltration sample The Coagulation Factor IX bulk material used in these studies was obtained from the Coagulation Factor IX (AlphaNineÒ SD) manufacturing process at a stage just prior to nanofiltration. This material had undergone all purification steps in the manufacturing process and represented the elution pool from the final chromatography column in the process. The elution pool contained sodium citrate and sodium chloride at pH 6.8. The sodium concentration of the elution pools used in this study was 0.58 0.08 molar equivalents (Eq) per L. For some experiments, elution pools with lower sodium concentrations (i.e., 0.12, 0.08 and 0.06 Eq/L) were prepared by diluting the sample at nominal sodium concentration with buffer containing 0.06 Eq/L Na (to reach a concentration of 0.12 Eq/L) or with water (to reach a concentration of <0.1 Eq/L). After salt adjustment, the samples were ultrafiltered in order to reach the approximate protein concentration of the nominal sample. 2.2. Nanofiltration Nanofiltration through V70 cassettes (Millipore Corporation, Billerica, MA, USA) was performed at process scale (1 m2) in tangential flow mode using pumps before and after the nanofilter to control flow. Nanofiltration through PlanovaÒ 15 hollow fiber units (Asahi Kasei Medical Co. Ltd., Glenview, IL, USA) was performed at process (1 m2) and laboratory (0.001 m2) scale in constant flow “dead-end” mode using a pump ahead of the filter to control flow. Process scale and laboratory scale filtrations with the PlanovaÒ 15 were performed under equivalent conditions (volume per filter surface area, protein mass per surface area, flux) and generated equivalent results. All other nanofiltrations were performed at laboratory scale (0.001 m2). In all cases, the membranes were rinsed with water and equilibrated with 0.02 M sodium citrate buffer, pH 6.8, containing sodium chloride at amounts appropriate to adjust the buffer sodium concentration to that of the filtration sample. All steps were performed at 4 C. The Planova BioEX (Asahi Kasei Medical Co. Ltd., Glenview, IL, USA) filtrations were performed in constant flow “dead-end” mode with a pump ahead of the filter. The Viresolve Pro Micro (3.1 cm2) (Millipore Corporation, Billerica, MA, USA) and DV20 (0.001 m2) (Pall Corporation, Covina, CA, USA) filtrations were performed under a constant pressure of 30 psi using a regulated nitrogen stream. In all cases, the filters were used according to the manufacturer’s recommended specifications and pressure was monitored (PressureMAT sensor, PendoTECH, Princeton, NJ, USA). Filters run at constant flow were adjusted to maintain recommended pressure limits if fouling of the membrane occurred.
177
2.3. Electrophoresis and western blotting Protein samples before and after nanofiltration were separated by SDS-PAGE on Novex Tris-Glycine Gels (Invitrogen, Carlsbad, CA, USA) and visualized by protein staining or Western blot analysis. Acrylamide gels were run at constant voltage (125 V) in Novex TrisGlycine SDS Running Buffer (Invitrogen, Carlsbad, CA, USA) for approximately 90 min. Total protein was observed by staining either with Colloidal Blue Staining Kit (Invitrogen, Carlsbad, CA, USA) or IRDye Blue Protein Stain (LI-COR, Lincoln, NE, USA) according the manufacturer’s recommended protocol. Gels stained with IRDye Blue Protein Stain were imaged with an Odyssey Imager (LI-COR, Lincoln, NE, USA) at 700 nm. FIX and inter-a-trypsin inhibitor identities were verified by Western blot analysis. For Western blots, proteins were transferred to an ImmobilonPVDF membrane (Millipore Corporation, Billerica, MA, USA) in 1X Tris-Glycine transfer buffer. Membranes were blocked with 1.5% albumin, 1X TBS. The membranes were then incubated with the manufacturer’s recommended concentration of primary antibody (ITI-H1 mouse monoclonal antibody [SC-69788, clone# 40B10], Santa Cruz Biotechnology, Santa Cruz, CA, USA; FIX mouse monoclonal antibody [AHIX-5041], Haematologic Technologies Inc, Essex Junction, VT, USA) in 1X TBS, 0.5% albumin, 0.01% SDS, 0.1% Tween20. The non-specifically bound primary antibody was washed from the membrane with 1X TBS, 0.01% Tween-20. Goat-anti-mouse IRDye 680 (LI-COR, Lincoln, NE, USA) was diluted 1:10,000 in 1X TBS, 0.01% Tween-20 and incubated with the membrane. Non-specifically bound secondary antibody was removed by washing the blot 3 times with 1X TBS, 0.01% Tween-20 and 1 time with 1X TBS. The membranes were imaged with an Odyssey Imager at 700 nm. The identity of C4BP was verified by in-gel Western analysis. For the in-gel Western, the gel was fixed with 5% acetic acid, 50% isopropanol and then rinsed with water. The C4BP mouse monoclonal antibody (1 mg/ml) (05-867, clone# GMA-036, Millipore Corporation, Billerica, MA, USA) was diluted 1:250 in 1X TBS, 5% albumin, 0.1% Tween-20. The gel was incubated for 1 h in the primary antibody before washing away the non-specifically bound antibody with 1X TBS, 0.1% Tween-20. Goat-anti-mouse IRDye 680 was diluted 1:2000 in 1X TBS, 0.5% albumin and 0.1% Tween-20 and incubated with the gel. Non-specifically bound secondary antibody was washed away with 1X TBS, 0.1% Tween-20. The gel was imaged with an Odyssey Imager at 700 nm. 2.4. Liquid chromatography tandem mass spectrometry (LC-MS-MS) The Coagulation Factor IX filtrate from nanofilter A was resolved by non-reducing SDS-PAGE. The resolved FIX protein band and two accompanying bands were excised from the gel. The protein bands were subjected to in-gel trypsin digestion (Sequencing grade Trypsin [V511A], Promega, Madison, WI, USA). The trypsin digested samples were subjected to LC-MS-MS on an LT-QFT Mass Spectrometer (Thermo Scientific, Waltham, MA, USA) at The Pasarow Mass Spectrometry Laboratory, The NPI-Semel Institute, David Geffen School of Medicine, UCLA (Los Angeles, CA, USA). Ions generated from LC-MSMS were given an ion score defined as 10Log(P), where P is the calculated probability that the observed match between the experimental data and the protein database sequence is a random event. A probability-based Mowse Score was generated from the combined ion scores using the Mowse algorithm to determine the most significant protein candidates in each gel band. 2.5. HPLC size exclusion chromatography (HPSEC) The same Coagulation Factor IX bulk material used for the nanofiltration was concentrated to a target A280 of 3.6 units in high
178
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
(0.56 Eq/L Na) or low (0.06 Eq/L Na) buffer. Equivalent amounts (w20 mL) of protein sample were injected on a TSK-GEL G4000SW column (7.5 mm 60 cm) with a TSK-GEL Guard column SWXL (6 mm 4 cm) (TOSOH America Inc, Grove City, OH). The sample was eluted under high salt conditions (0.56 Na Eq/L) (83% e 600 mM NaCl, 20 mM Na Citrate, pH 6.8, 17% e 20 mM Na Citrate, pH 6.8) and low salt conditions (0.06 Na Eq/L) (20 mM Na Citrate, pH 6.8) at a flow rate of 0.5 ml/min. A series of pre-runs were performed to condition the columns to the salt and protein conditions. After conditioning, the results of 5 runs were collected for quantification. Protein elution was monitored by absorbance at 280 nm. One minute time based fractions were collected for analysis of peaks by SDS-PAGE/Western Blot. A two-tailed t-test for unpaired samples assuming equal variances was used for statistical analysis of peak areas. A P < 0.05 was considered statistically significant. 3. Results Four different nanofilters were evaluated for their protein sieving characteristics during nanofiltration of a Coagulation Factor IX (AlphaNineÒ SD) bulk eluate and results were compared to
results obtained with the V70 nanofilter currently used to nanofilter the Coagulation Factor IX bulk eluate. All nanofilters, other than the V70 are identified only by a randomly assigned letter from A to D. All nanofilters had mean pore sizes of approximately 20 nm or less. At the start of nanofiltration, samples were placed in reservoirs. Flow through the nanofilters was controlled with a pump or via pressurized gas (N2). Filtration was allowed to proceed until the entire sample volume was filtered or sample was no longer flowing through the nanofilter (“plugging”). The normalized volume of eluate that could be passed through each nanofilter prior to plugging ranged from 34 L/m2 to 48 L/m2. 3.1. Nanofiltration using the V70 (current) nanofilter Samples before and after nanofiltration through the V70 were analyzed by SDS-PAGE under non-reducing and reducing conditions (Fig. 1A and B, respectively). Under both non-reducing and reducing conditions, preparations before and after V70 nanofiltration contained a major protein band with an apparent molecular weight (MWapp) of 65 kDa which corresponds to the MWapp of FIX. When analyzed under non-reducing conditions (Fig. 1A),
Fig. 1. SDS-PAGE analysis of the Coagulation Factor IX bulk eluate before and after nanofiltration. Non-reduced gel (a) comparing V70 and filter A: lane 1, molecular weight markers; lane 2, FIX bulk; lane 3, FIX bulk post V70 nanofiltration; lane 4, FIX bulk; lane 5, FIX bulk post filtration with nanofilter A. Reduced gel (b) comparing V70 and filter A: lane 1, molecular weight markers; lane 2, FIX bulk; lane 3, FIX bulk post V70 nanofiltration; lane 4, FIX bulk; lane 5, FIX bulk post filtration with nanofilter A. The 70 kDa band observed in reducing conditions is marked with an asterisk. Non-reduced gel (c) comparing filters B,C and D: lane 1, FIX bulk; lane 2, FIX bulk post filtration with nanofilter B; lane 3, FIX bulk; lane 4, FIX bulk post 0.22 mm filtration; lane 5, FIX bulk post 0.22 mm filtration and filtration with nanofilter C; lane 6, FIX bulk; lane 7, FIX bulk post 0.22 mm filtration; lane 8, FIX bulk post 0.22 mm filtration and filtration with nanofilter D.
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
two minor bands were observed in the source material prior to V70 nanofiltration. One of these bands had a MWapp of 210 kDa and the other had a MWapp of >500 kDa (i.e., outside the resolving range of the gel). When the same source material was analyzed under reducing conditions (Fig. 1B), the 210 kDa protein band was observed but not the >500 kDa high molecular weight protein (HMWP) band. Instead, a protein band with MWapp of 70 kDa (marked by asterisk), not seen under non-reducing conditions, was observed, suggesting that the HMWP was an oligomer comprising 70 kDa disulfide linked subunits. Following nanofiltration, analysis under non-reducing conditions showed that the 210 kDa and 65 kDa protein bands were present in the filtered product but not the HMWP (Fig. 1A) indicating that the HMWP (but not the 210 kDa and 65 kDa proteins) were removed by V70 filtration. Under reducing conditions, the 210 kDa and 65 kDa protein bands were present in the filtered product but not the 70 kDa band (Fig. 1B). Since filtration results for the 65 kDa and 210 kDa proteins indicate that a 70 kDa protein should not have been removed by the V70, absence of the 70 kDa protein in the V70 filtered product confirms that this polypeptide was a subunit of the larger HMWP that was removed by the V70. 3.2. Nanofiltration through nanofilters A, B, C, and D Fig. 1 also shows SDS-PAGE results for samples before and after filtration through nanofilter A. When samples were analyzed under non-reducing conditions, the HMWP was present in both the source material and the filtered material (Fig. 1A). Under reducing conditions, the HMWP 70 kDa subunit was also observed in the source material and the filtered material (Fig. 1B). These results indicate that nanofilter A did not remove the HMWP. Fig. 1C, shows SDS-PAGE results from samples before and after filtration through nanofilters B, C and D and the 0.22 mm pre-filter. The SDS-PAGE was performed under non-reducing conditions in order to visualize the HMWP. Pre-filtration did not affect the protein composition of the product. However, filtration through nanofilter B completely removed the HMWP. Filtration through nanofilter D only partly removed the HMWP and nanofiltration through filter C did not remove the HMWP.
179
matches for the 210 kDa band were all components of inter-atrypsin inhibitor (ITI) and included the heavy chain isoforms, H1 and H2, and the light chain, bikunin [15]. The highest probability match for the 65 kDa protein was FIX. The molecular weights of all identified proteins were consistent with the protein mobility observed in SDS-PAGE. The LC-MS-MS results were confirmed by in-gel Western and Western blot analyses. As shown in Fig. 2, when the elution pool prior to nanofiltration was analyzed, a mouse monoclonal antibody specific for C4BP recognized the HMWP band and monoclonal antibodies specific for ITI and FIX recognized the 200 kDa and 65 kDa bands, respectively. 3.4. Effect of ionic strength We considered the possibility that the high ionic strength in the elution pool might affect C4BP structure, allowing its passage through the pores of some nanofilters. Therefore, additional nanofiltration studies were performed under conditions of decreasing sodium concentration. Filtration through nanofilter A (which was previously shown not to remove C4BP) was performed at high (0.50 Eq/L), mid (0.12 Eq/L) and low (0.06 Eq/L) levels of sodium and the nanofiltered samples analyzed by SDS-PAGE as shown in Fig. 3A. The protein concentration was kept constant for all sodium conditions. Pre-filtration through a 0.22 mm filter before nanofiltration did not change the protein composition of the source material. At 0.50 Eq/L of sodium, C4BP was not removed by nanofilter A. At 0.12 Eq/L, some, but not all of the C4BP was removed and at 0.06 Eq/L essentially all of the C4BP was removed. Quantification of gel bands and normalization to FIX protein indicated that removal of C4BP was dependent upon the sodium concentration during nanofiltration (Fig. 3B). In contrast, ITI, despite also having a relatively high apparent molecular weight, was not removed from the product under any of the test conditions (Fig. 3A and B). We considered the possibility that nanofilters C and D which, like nanofilter A, did not remove C4BP at high sodium levels, could
3.3. Identification of nanofiltered proteins The HMWP and the two other protein bands observed in the gels (MWapp of 210 kDa and 65 kDa) were identified by LC-MS-MS. Table 1 shows the three highest probability matches for each protein band as determined by LC-MS-MS and Mowse scoring. The highest probability matches for the HMWP were the C4b-binding protein (C4BP) alpha and beta chains. The highest probability
Table 1 LC-MS-MS protein identification. Protein name and rank ( )
Protein GI number
High molecular weight protein (>500 kDa) band 4502503 (1) C4BP [a chain] 4502505 (2) C4BP [b chain] 85544070 (3) C4BP [a chain, N-terminal fragment] 210 kDa protein band (1) ITI [H1, heavy chain] 189054356 (2) ITI [H2, heavy chain] 55958063 (3) ITI [alpha-1-microglobulin, light chain] 579676 65 kDa protein band (1) FIX 182613 (2) FIX [N-terminal fragment] 100017830 (3) Low scoring plasma proteins N/A
Probability based Mowse score 2648 147 106 1927 1894 390 7731 4513 238
Fig. 2. In-gel western of C4BP and western blot detection of inter-a-trypsin inhibitor and Coagulation Factor IX. Lanes 1, 3 and 5, molecular weight markers; lane 2, detection of C4BP with a monoclonal antibody by in-gel western; lane 4, detection of inter-a-trypsin inhibitor with a monoclonal antibody by western blot; lane 6, detection of FIX with a monoclonal antibody by western blot.
180
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
this peak consisted solely of FIX (data not shown). A broad peak eluting from 30 to 35 min under both high and low salt conditions was shown to consist of ITI when analyzed by Western blotting. In the lower portion of Fig. 5, the absorbance scale was expanded for the elution volume from 15 to 30 min in order to better visualize the high molecular weight peaks. In this region, two peaks were observed, an early eluting peak with a retention time of about 20 min and a later eluting peak with a retention time of about 24.5 min. Both peaks have been shown to consist of C4BP by Western blot (results not shown), suggesting two populations for C4BP having different Stokes radii. For the earlier eluting peak, the area under the peak was increased two-fold under low salt conditions when compared to the peak area obtained under high salt conditions (0.21 0.01% versus 0.11 0.01% of total area). In contrast, for the later eluting peak, the area under the peak was decreased almost three-fold under low salt conditions when compared to the peak area obtained under high salt conditions (4.18 0.37% versus 1.32 0.11% of total area). In both cases, the observed change in peak area due to salt concentration was highly significant (P < 0.001). Fig. 5C presents graphically the average peak areas from five runs under each condition for both peaks. Similar shifts in peak areas did not occur for either ITI or FIX under high and low salt conditions. The results indicate that, under high salt conditions, there is a greater percentage of C4BP with a smaller Stokes radius (later retention time) than under low salt conditions. Conversely, under low salt conditions, there is a greater percentage of C4BP with a larger Stokes radius (earlier retention time) than under high salt conditions. 4. Discussion
Fig. 3. C4BP removal at varying sodium concentrations. SDS-PAGE analysis (A) of nanofilter A filtrates was performed in non-reducing conditions. The gel was stained with IRDye 680 (Odyssey, LI-COR system). Reduction of sodium concentration results in a dose dependent increase in the removal of C4BP and decreased amounts of C4BP in the filtrate after nanofiltration. Ratios of C4BP to FIX and ITI to FIX after nanofiltration divided by the ratios before nanofiltration (B). A ratio of 1.0 indicates no removal of C4BP or ITI.
remove the protein at a lower sodium level. Therefore, additional nanofiltration studies were performed with nanofilters C and D together with nanofilter A under conditions of low sodium concentration (0.08 Eq/L). The nanofiltered samples were analyzed by SDS-PAGE under non-reducing conditions as shown in Fig. 4. C4BP was removed during filtration of the Coagulation Factor IX bulk eluate through all nanofilters at low sodium, although to a lesser extent when passed through nanofilter C as indicated by the presence of C4BP in filtrate from nanofilter C. 3.5. HPLC size exclusion chromatography (HPSEC) The effect of ionic strength on C4BP structure was evaluated by monitoring changes in the Stokes radius of the protein via HPSEC. Fig. 5A and B show representative chromatograms of the Coagulation Factor IX bulk eluate prior to nanofiltration analyzed under conditions of high and low sodium, respectively. FIX elutes as the largest peak with a retention time of 38.7 min at high salt and 36.7 min at low salt. SDS-PAGE and Western blot analysis confirmed that
Nanofiltration is a procedure that is included in the manufacture of many plasma derived products to provide additional assurance that these products are free of adventitious agents such as viruses. However, as described here, the nanofiltration step included at the end of the AlphaNineÒ SD manufacturing process can also remove large plasma proteins such as C4BP, small amounts of which can accompany the FIX protein throughout the purification process [16]. Results presented here provide insight into the molecular structure of C4BP and the sieving characteristics of several nanofilters, all of similar pore size (<20 nm), from different manufacturers. C4BP is a large plasma glycoprotein having a molecular weight of 570 kDa [17]. It consists of seven a-chains connected to one b-chain by disulfide interactions [18e21]. The a-chains contain the binding site for C4b while the b-chain contains a high-affinity binding site for the vitamin K-dependent protein S [22,23]. Electron microscopy has shown that the a-chains are rod-like proteins of about 330 Å in length [24]. The shape of C4BP has often been described as octopus-like with the a-chains forming tentacle-like structures surrounding the core b-chain. This open structure would have a diameter of approximately 66 nm and should not pass through nanofilters with pores 20 nm. The observation that C4BP did pass through several such nanofilters at sodium concentrations >0.5 Eq/L but not at concentrations <0.1 Eq/L suggested differences in its structure related to ionic strength. We considered the possibility that the change in C4BP structure was due to the binding of C4b/protein S to C4BP at low salt and the dissociation of these proteins at high salt. However, when C4BP was isolated from Coagulation Factor IX preparations under high and low salt conditions, only a small amount of a protein S-sized material was observed (by SDS-PAGE) and the amount relative to C4BP remained constant regardless of salt concentration (results not shown). Therefore, we focused on the possibility that C4BP undergoes salt concentration-dependent conformational changes and
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
181
Fig. 4. Non-reducing SDS-PAGE analysis of Coagulation Factor IX elution pool following nanofiltration at a sodium concentration of 0.07e0.08 Eq/L. Lane 1, FIX bulk; lane 2, FIX bulk post 0.22 mm filtration; lane 3, FIX bulk post filter A nanofiltration; lane 4, FIX bulk; lane 5, FIX bulk post 0.22 mm filtration; lane 6, FIX bulk post filter C nanofiltration; lane 7, FIX bulk; lane 8, FIX bulk post 0.22 mm filtration; lane 9, FIX bulk post filter D nanofiltration.
theorized that, at the lower ion concentrations, pockets of similarly charged residues on the a-chains prevent them from moving close to each other, resulting in an open structure that cannot penetrate the nanofilter pores. In contrast, at the higher ion concentrations, charges on the a-chains would be neutralized, allowing the a-chains to collapse on one another forming a long, perhaps flexible, bundle of a-chain fibers. This is consistent with studies that have identified clusters of positively charged amino acids on the a-chains of C4BP [25,26]. Since the diameter of the individual a-chain is about 3 nm, the overall diameter of the bundle could be much less than 20 nm and the length of the bundle about half that of the open structure. This hypothesis seemed plausible since Perkins et al. [27] had proposed such a structure for C4BP based on data from solution x-ray scattering experiments performed on C4BP and had calculated a radius of gyration, RG, of 13 nm for C4BP. This RG is less than half the radius of the open C4BP structure and makes it more likely that C4BP would penetrate nanofilter pores with approximate diameters of 20 nm. The light scattering studies were performed at 0.25 Eq/L NaCl and we propose that at higher salt concentrations, such as those present during the nanofiltration of the Coagulation Factor IX bulks (i.e., >0.5 Eq/L), greater charge neutralization occurs, resulting in an even more compact and flexible C4BP structure with increased potential for passage through the nanofilter. This scenario is consistent with results obtained for nanofilters A, C and D in this study, all of which allowed substantial passage of C4BP at high salt concentrations but little, if any, passage at low salt concentrations. The studies performed with nanofilter A showed clearly that passage of C4BP through this nanofilter was salt concentration dependent,
with no C4BP removed at the highest salt concentration (>0.5 Eq/ L), all C4BP removed at the lowest salt concentration (0.06 Eq/L) and an intermediate amount removed at a mid-range concentration (0.12 Eq/L). The results from our HPSEC studies indicated that the Stokes radius of C4BP was smaller at a high salt concentration (0.6 Na Eq/L) than at a low salt concentration (0.06 Na Eq/L) and provides additional evidence that a more compact C4BP structure at the higher salt is responsible for passage of C4BP through some, but not all, of the nanofilters evaluated. Two of the five nanofilters evaluated, the V70 and nanofilter B, did not permit passage of C4BP at any salt condition, suggesting that the mean pore size of these two filters may be smaller than that of the other nanofilters. However, it should be mentioned that any difference in pore size between the two groups of nanofilters is likely small; none of the filters in either group retained ITI, another large (250 kDa) plasma protein, at any salt condition studied. Differences in molecular sieving characteristics between these nanofilters of similar porosity seem to be limited to large proteins whose shape or flexibility is altered under different ionic conditions. It should be emphasized that no differences in virus retention were noted when the virus removal capacity of a nanofilter that did not retain C4BP was compared to nanofilters that did (results not shown). This was true for the smallest virus studied, porcine parvovirus (25 nm), at the highest salt concentration (0.6 Na Eq/L). Our results emphasize the need for manufacturers of biological products to evaluate the effect of a nanofilter on product (protein) composition over the full range of anticipated operating ionic strengths prior to introduction of the nanofilter into the process.
182
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
Fig. 5. HPSEC of high and low salt Coagulation Factor IX bulks. High salt (A) and low salt (B) chromatograms of Coagulation Factor IX bulks are shown. Five HPSEC runs for each condition were performed to confirm reproducibility. One representative chromatogram for each condition is shown. In each chromatogram, the y-axis (mAU280 nm) in the range of 15e30 min is magnified to show the two C4BP peaks. The percentage of total area of each peak as an average of the five HPSEC runs is compared for high and low salt conditions (C).
C.J. Winkler et al. / Biologicals 41 (2013) 176e183
Acknowledgments We thank Zong-Zhi Huang and Evelyn Nario for the excellent technical assistance that they provided for this study. All authors have contributed significantly to the design and performance of experiments, the interpretation of results, or the writing and criticism of the manuscript.
[13]
[14]
[15]
References [16] [1] Burnouf T, Radosevich M. Nanofiltration of plasma-derived biopharmaceutical products. Haemophilia: The Official Journal of the World Federation of Hemophilia 2003;9:24e37. [2] Barin F. Viral safety of biologicals. Annales Pharmaceutiques Francaises 2008; 66:129e39. [3] Diez JM, Caballero S, Belda F, Otegui M, Gajardo R, Jorquera JI. Capacity of the manufacturing process of Flebogamma((R)) DIF, a new human high purity intravenous immunoglobulin, to remove a TSE model-agent. Biologicals: Journal of the International Association of Biological Standardization 2010;38: 670e4. [4] Yunoki M, Tanaka H, Urayama T, Hattori S, Ohtani M, Ohkubo Y, et al. Prion removal by nanofiltration under different experimental conditions. Biologicals: Journal of the International Association of Biological Standardization 2008;36: 27e36. [5] Hoffer L, Schwinn H, Biesert L, Josic D. Improved virus safety and purity of a chromatographically produced factor IX concentrate by nanofiltration. Journal of Chromatography B, Biomedical Applications 1995;669:187e96. [6] DiLeo AJ, Allegrezza Jr AE, Builder SE. High resolution removal of virus from protein solutions using a membrane of unique structure. Biotechnology (N Y) 1992;10:182e8. [7] Chang CJ, Yang CS, Chuang YJ, Khoo HS, Tseng FG. Micro-patternable nanoporous polymer integrated with microstructures for molecular filtration. Nanotechnology 2008;19:365301. [8] Tomokiyo K, Yano H, Imamura M, Nakano Y, Nakagaki T, Ogata Y, et al. Largescale production and properties of human plasma-derived activated factor VII concentrate. Vox Sanguinis 2003;84:54e64. [9] Smith MP, Berrie JR. Selecting a virus reduction filter and optimising performance during process scale-up. Recovery of Biological Products XI, Banff; 2003. [10] Herring S. Grifols’ factor IX concentrates: new findings in biochemical characteristics and safety profiles. Haemophilia: The Official Journal of the World Federation of Hemophilia 2010;16(Suppl. 6):3e8. [11] Goldsmith JC, Kasper CK, Blatt PM, Gomperts ED, Kessler CM, Thompson AR, et al. Coagulation factor IX: successful surgical experience with a purified factor IX concentrate. American Journal of Hematology 1992;40:210e5. [12] Lissitchkov T, Matysiak M, Zavilska K, Laguna P, Gercheva L, Antonov A, et al. A clinical study assessing the pharmacokinetics, efficacy and safety of AlphaNine((R)), a high-purity factor IX concentrate, in patients with severe
[17]
[18] [19]
[20]
[21]
[22]
[23] [24]
[25]
[26]
[27]
183
haemophilia B. Haemophilia: The Official Journal of the World Federation of Hemophilia 2011;17:590e6. Herring SW, Abildgaard C, Shitanishi KT, Harrison J, Gendler S, Heldebrant CM. Human coagulation factor IX: assessment of thrombogenicity in animal models and viral safety. The Journal of Laboratory and Clinical Medicine 1993; 121:394e405. Pock K, Buchacher A, Rizzi A, Josic D. Characterization of clotting factor IX in plasma-derived preparations by electrophoretic techniques. Journal of Chromatography A 2001;921:57e67. Bost F, Diarra-Mehrpour M, Martin JP. Inter-alpha-trypsin inhibitor proteoglycan familyea group of proteins binding and stabilizing the extracellular matrix. European Journal of Biochemistry/FEBS 1998;252:339e46. Clifton J, Huang F, Gaso-Sokac D, Brilliant K, Hixson D, Josic D. Use of proteomics for validation of the isolation process of clotting factor IX from human plasma. Journal of Proteomics 2010;73:678e88. Dahlback B. Purification of human C4b-binding protein and formation of its complex with vitamin K-dependent protein S. The Biochemical Journal 1983; 209:847e56. Blom AM. Structural and functional studies of complement inhibitor C4bbinding protein. Biochemical Society Transactions 2002;30:978e82. Blom AM, Villoutreix BO, Dahlback B. Functions of human complement inhibitor C4b-binding protein in relation to its structure. Archivum immunologiae et therapiae experimentalis 2004;52:83e95. Blom AM, Kask L, Dahlback B. Structural requirements for the complement regulatory activities of C4BP. The Journal of Biological Chemistry 2001;276: 27136e44. Kask L, Hillarp A, Ramesh B, Dahlback B, Blom AM. Structural requirements for the intracellular subunit polymerization of the complement inhibitor C4bbinding protein. Biochemistry 2002;41:9349e57. Webb JH, Villoutreix BO, Dahlback B, Blom AM. Localization of a hydrophobic binding site for anticoagulant protein S on the beta -chain of complement regulator C4b-binding protein. The Journal of Biological Chemistry 2001;276: 4330e7. Hillarp A, Hessing M, Dahlback B. Protein S binding in relation to the subunit composition of human C4b-binding protein. FEBS Letters 1989;259:53e6. Dahlback B, Smith CA, Muller-Eberhard HJ. Visualization of human C4bbinding protein and its complexes with vitamin K-dependent protein S and complement protein C4b. Proceedings of the National Academy of Sciences of the United States of America 1983;80:3461e5. Blom AM, Webb J, Villoutreix BO, Dahlback B. A cluster of positively charged amino acids in the C4BP alpha-chain is crucial for C4b binding and factor I cofactor function. The Journal of Biological Chemistry 1999;274:19237e45. Jenkins HT, Mark L, Ball G, Persson J, Lindahl G, Uhrin D, et al. Human C4bbinding protein, structural basis for interaction with streptococcal M protein, a major bacterial virulence factor. The Journal of Biological Chemistry 2006;281:3690e7. Perkins SJ, Chung LP, Reid KB. Unusual ultrastructure of complementcomponent-C4b-binding protein of human complement by synchrotron X-ray scattering and hydrodynamic analysis. The Biochemical Journal 1986; 233:799e807.