Removal of lipopolysaccharide from protein solution using nanostructured porous supports bearing lipid membranes
© Wakita; licensee Springer. 2013
Received: 17 July 2013
Accepted: 15 October 2013
Published: 5 November 2013
Polymeric lipid membranes of N-octadecylchitosan, which consists of 70 mol% of 2-(octadecylamino)-2-deoxy-d-glucopyranose, 17 mol% of 2-amino-2-deoxy-d-glucopyranose, and 13 mol% of 2-acetamido-2-deoxy-d-glucopyranose, were covalently immobilized to carboxylated porous supports composed of chitosan and used for the adsorption of pyrogenic lipopolysaccharide. When human serum albumin solution, including 5 mg mL-1 of albumin and 5.6 ng mL-1 of lipopolysaccharide, was passed through a column packed with the resulting porous supports bearing lipid membranes assembled in nanoscale, lipopolysaccharide was removed to as low as a detection limit of 0.020 ng mL-1 with a quantitative recovery of protein. On the other hand, in the case of directly N-octadecylated porous supports having cationic and hydrophobic ligands which are not assembled as lipid membranes, lipopolysaccharide could not be removed to the detection limit and protein recovery was lower than the porous supports bearing lipid membranes. The difference above as well as difference from conventional adsorbents suggested that the selectivity was attributable to an interaction between the cationic lipid membranes of N-octadecylchitosan and lipopolysaccharide as well as protein. The porous supports bearing lipid membranes were stable in 0.5 M NaOH and 0.1 M HCl at ambient temperature. Considering the confirmed excellent selectivity and chemical stability, their practical use as separation media in the pharmaceutical manufacturing can be expected.
KeywordsLipid membranes Lipopolysaccharide Chitosan
Lipid membranes have been receiving considerable attention as models for various chemical investigations. Densely packed alkyl chains with hydrophilic head groups can have both a hydrophobic interaction and a hydrophilic one with guest substrates. This specific environment provided by lipid membranes is essential for molecular recognition in biological membranes and has been studied in the areas of chemical sensing and separation [1, 2].
Proteins or other biologically active substances, especially those produced by recombinant microorganisms, are usually contaminated with lipopolysaccharide (LPS) . LPS, which originates from an outer membrane of Gram-negative bacteria, consists of a polysaccharide and a terminal lipid A moiety. Lipid A is composed of a diglucosamine that is highly substituted with amide- and ester-linked long-chain fatty acids and negatively charged with phosphate groups. For pharmaceutical uses of those active substances, LPS has to be removed to not higher than 0.1 ng mL-1 because of its strong pyrogenicity . Intensive studies have been done on the removal of LPS from protein solutions . For example, LPS was selectively removed by ion-exchange chromatography using DEAE-Sepharose CL-6B  and affinity chromatography using adsorbents bearing histidine , polymyxin B , and polycation . However, the removal of LPS is suggested to be extremely difficult when LPS is associated with protein to be purified  and has been an issue in pharmaceutical technology and science.
Materials and general methods
Chitosan was purchased from Dai-ichi Kogyo Seiyaku Co., Ltd. (Kyoto, Japan). The degree of deacetylation was determined as 87 mol% by colloidal titration. Intrinsic viscosity was 1.42 dL g-1 (0.2 M CH3COOH/0.1 M CH3COONa, 30°C) which corresponded to 2.67 × 104 of molecular weight relative to poly(ethylene glycol). A cross-linked porous chitosan having a particle size of 45 to 420 μm and an average pore diameter of 2 μm, a product of Kurita Water Industries, Ltd. (Tokyo, Japan), was used as obtained. 1-Bromooctadecane, succinic anhydride (Kishida Chemical Co., Ltd., Osaka, Japan), 1-ethyl-3-(3-(dimethylamino)propyl)-carbodiimide hydrochloride (water-soluble carbodiimide (WSC), Dojindo Laboratories, Kumamoto, Japan), N-hydroxysuccinimide (HOSu), and d-(+)-glucosamine hydrochloride (Tokyo Kasei Kogyo Co., Ltd., Tokyo, Japan) were used as obtained. A volume of adsorbent was determined after standing for 24 h in a measuring cylinder with water. Infrared (IR) spectra were recorded using IR-810 (Jasco Co., Ltd., Tokyo, Japan). Total organic carbon content analysis and differential scanning calorimetry (DSC) were carried out using TOC-5000A (Shimadzu MFG., Kyoto, Japan) and DSC220C (Seiko Instruments Inc., Tokyo, Japan), respectively. Scanning electron micrographs (SEM) and transmission electron micrographs (TEM) were taken at JEOL DATUM (Tokyo, Japan) using JSM-6400 F (JEOL) and JEM-1200EX (JEOL), respectively. DEAE-Sepharose CL-6B and Pyrosep (histidine-immobilized agarose, Sigma-Aldrich, Tokyo, Japan) were obtained from manufacturers. HSA (20% w/v) and LPS (Escherichia coli serotype O127:B8) were products of Nihon Pharmaceutical Co., Ltd. (Tokyo, Japan) and Difco Laboratories (Detroit, MI, USA), respectively, and used as obtained. Toxicolor (Seikagaku Corporation, Tokyo, Japan), which is a chromogenic Limulus amebocyte lysate test, was used as an assay method for LPS. Samples containing LPS were diluted with Tris–HCl buffer (pH 8.0) to lower than 0.085 ng mL-1 of LPS and assayed by the method recommended by the manufacturer. The detection limit of LPS in this test was as low as 0.020 ng mL-1, which corresponded to 0.06 endotoxin unit. HSA concentration was measured by UV at 236 nm to avoid interference of a stabilizer N-acetyltryptophan showing adsorption at 280 nm.
Preparation of porous supports bearing lipid membranes
Direct alkylation of porous supports
After being washed by dimethylacetamide, the cross-linked porous chitosan (20.0 mL, including 4.6 mmol of NH2) was reacted with mild stirring using 1-bromooctadecane (4.56 g, 13.7 mmol) in the presence of Na2CO3 (1.45 g, 13.7 mmol) at 70°C for 69 h in dimethylacetamide under nitrogen atmosphere. Particles were recovered by filtration and washed with water, warm ethyl acetate, ethanol, and water successively. Changes in the appearance and porous structure were not observed by SEM. In the IR spectra (KBr pellet), νCH of CH2 2,925 and 2,850 cm-1 was observed. Ion-exchange capacity was 2.3 meq g-1 dry particles. Using this value, the colloidal equivalent of chitosan (5.0 meq g-1, pH 4.0) used for the preparation of the cross-linked porous chitosan and the ion-exchange capacity of the cross-linked porous chitosan (3.8 meq g-1) which was cross-linked by 1,6-diisocyanatohexane, GlcN of the cross-linked porous chitosan, and GlcNC18 of the resulting directly alkylated porous supports were calculated as 69 and 47 mol%, respectively. Therefore, about half of the monosaccharide units of the support particles are deemed to be octadecylated.
Column-wise adsorption of LPS from protein solutions
Purified water and buffer solutions were sterilized using an autoclave at 115°C to 121°C for 15 min. Glass wares were also sterilized using the autoclave at 250°C for 2 h. LPS aqueous solution, which was prepared by vortex mixing and dilution with water, was added to HSA preparation. The solution was filtered with a filter disk having 0.2-μm-diameter pores and diluted with a buffer solution to a desired concentration for column-wise experiments. Phosphate buffer was used for experiments at pH 7.0 and 8.0, and acetic acid buffer was used for those at pH 4.3 and 5.3. Buffer solutions were prepared by adding NaOH solution of a predetermined concentration to phosphoric acid or acetic acid to obtain the desired ionic strength (μ).
Column-wise adsorption was done at 20°C. Adsorbents were suspended in water and fed into a glass column (8 mm i.d. × 100 mm length) using a LC-6A pump (Shimadzu Corp., Kyoto, Japan). The length of the gel bed was between 930 and 980 mm. The resulting column was washed with 0.5 M NaOH, at 10 mL h-1 for 3 h, and left overnight filled with 0.5 M NaOH to decompose LPS in the column (depyrogenation). After washing with water for 3 h, 0.1 M acetic acid was passed through for 1.5 h to convert amino groups of N-octadecylchitosan immobilized on the supports to their ammonium forms. After the buffer solution was passed through for 6.5 h, HSA solution was passed through at 5 mL h-1 for 15 to 16 h. The eluted solution was collected immediately as ten fractions of 7.5 mL each. Fractions of 2, 4, 6, 8, and 10 were analyzed for the concentrations of LPS and HSA.
Porous supports bearing lipid membranes of N-octadecylchitosan were immersed in 0.5 M NaOH or 0.1 M HCl at ambient temperature overnight and then washed with water and subjected to IR spectroscopic analysis. The supports (20 mL) were packed into a column (XK16/20, Pharmacia LKB Biotechnology, Uppsala, Sweden; 16 mm i.d. × 100 mm length). The packed column was filled with 0.5 M NaOH and allowed to stand overnight at 20°C. After washing with 200 mL of water, 50 mL of water was circulated in the column for 24 h at a flow rate of 1 BV h-1. The water was recovered and subjected to an analysis of total organic carbon content .
Results and discussion
Porous supports bearing lipid membranes
Column-wise adsorption of LPS from protein solution by porous supports bearing lipid membranes
Column-wise adsorption of LPS and HSA using the porous supports bearing lipid membranes
Concentration (ng mL-1)
Concentration (ng mL-1)
Column-wise adsorption of LPS and HSA using various adsorbents
Concentration (ng mL-1)
Concentration (ng mL-1)
Porous supports bearing lipid membranes
Pyrosep; histidine-immobilized agarose
Directly alkylated porous chitosan
As shown in Table 1, in the case of the porous supports bearing lipid membranes, LPS was removed to lower than 0.020 ng mL-1 at pH 5.3, 7.0, and 8.0 and to 0.039 ng mL-1 at pH 4.3 with a quantitative recovery of protein. In the case of DEAE-Sepharose CL-6B and histidine-immobilized agarose (Table 2), concentrations of LPS in the recovered solution were higher than those in the porous supports bearing lipid membranes. Since the removal of LPS to lower than the detection limit is usually required for pharmaceutical applications, the above removal ability of the porous supports bearing lipid membranes can be an advantage in practical use.
Mechanism of the selective adsorption of LPS
For the argument of adsorption mechanism, the electric charge of LPS and protein, aggregation behavior of LPS, and interaction between LPS and protein should be reviewed. Since lipid A is partially phosphorylated, LPS exhibits a net negative charge at all pH ranges applied. On the other hand, since pI of albumin is 4.9, it exhibits a net positive charge at pH 4.3 and a net negative charge at pH 5.3, 7.0, and 8.0. About the aggregation of LPS and the interaction between LPS and proteins, it is well known that LPS forms various molecular aggregates in aqueous solutions  and interacts with various proteins to form molecular complexes . From the amphiphilic structure of LPS and the effect of nonionic detergent on the dissociation of LPS , the aggregation between LPS is likely caused by hydrophobic interaction between LPS molecules. Considering our dynamic light scattering study showing that LPS interacts with bovine serum albumin , it seems that LPS interacts with HSA in applied conditions.
Specific interaction for LPS of the porous supports bearing lipid membranes seems to be useful in a forthcoming issue. Antibody drugs with superior efficacy have been developed intensively in the last few decades. However, since they are produced by mammalian cells such as Chinese hamster ovary, their cost is very expensive and needs to be reduced. For this issue, intensive researches have been continued for production of antibody drugs using microorganisms such as E. coli. Although protein A chromatography is useful to recover antibody drugs from preparations, further chromatography is necessary to obtain the required purity for their clinical use. It is likely that production of antibody drugs by E. coli leads to a further requirement of selective removal of LPS. Considering the selectivity for LPS of the porous supports bearing lipid membranes, their application for purification of antibody drugs is interesting.
Chemical stability of porous supports bearing lipid membranes
The elution property of separation medium is a key issue in liquid purification for the pharmaceutical industry. Elution from porous supports bearing lipid membranes of N-octadecylchitosan was evaluated by measuring the total organic carbon content in an eluent from a column packed with 20 mL of supports. The total organic carbon in the recovered water described in the experimental section was 400 μg L-1. Even assuming that all of the organic carbons are due to eluted N-octadecylchitosan, the eluted N-octadecylchitosan (C, 65.7%) is calculated as 3.0 × 10-5 g. This amount is 0.038% of the N-octadecylchitosan immobilized on the 20-mL supports. This stability for alkali is attributable to the stability of the amide linkage used for the immobilization of N-octadecylchitosan as well as the stability of the support material. Any substantial change was not observed in the IR spectra of the porous supports bearing lipid membranes after immersion in 0.5 M NaOH or 0.1 M HCl overnight at ambient temperature. This chemical stability, especially to 0.5 M NaOH immersion which is used as a standard depyrogenation procedure, is robust enough for a practical application in the pharmaceutical industry.
Porous supports bearing cationic lipid membranes of N-octadecylchitosan assembled in nanoscale adsorb LPS selectively from HSA solution at pH 4.3 to 8.0 with the ionic strength of 0.05 to 0.1. LPS was removed to as low as a detection limit of 0.020 ng mL-1 by a column-wise adsorption with a quantitative recovery of HSA. Since LPS includes a terminal diglucosamine which is negatively charged and highly substituted with long-chain fatty acids, LPS is adsorbed by both an ionic interaction and a hydrophobic one. In addition, the low pKa of the chitosan-based material as well as the rigid gel phase of lipid membranes leads to a relatively weak interaction between HSA and results in the selective adsorption of LPS. For a practical use in separation, especially such as liquid purification for the pharmaceutical industry, a lower elution property of the separation medium is a key issue. Considering the excellent selectivity and the chemical stability of the supports bearing cationic lipid membranes of N-octadecylchitosan, their practical use as separation media in pharmaceutical manufacturing can be expected.
The author thanks Mr. Tsuneyasu Adachi and Mr. Jun-ichi Ida (Kurita Water Industries) for the valuable technical assistance.
- Kim Y-R, Jung S, Ryu H, Yoo Y-E, Kim SM, Jeon T-J: Synthetic biomimetic membranes and their sensor applications. Sensors 2012, 12: 9530–9550. 10.3390/s120709530View Article
- Stibius K, Bäckström S, Hélix-Nielsen C: Passive transport across biomimetic membranes. In Biomimetic Membranes for Sensor and Separation Applications. Edited by: Hélix-Nielsen C. New York: Springer; 2012:137–155.
- Westphal O, Lüderitz O: Chemical research on lipopolysaccharides of Gram-negative bacteria. Angew Chem 1954, 66: 407–417. 10.1002/ange.19540661312View Article
- Westphal O, Lüderitz O, Galanos C, Mayer H, Riestschel ET: The story of bacterial endotoxin. In Advances in Immunopharmacology 3. Edited by: Chedid L, Hadden JW, Speafiro F. New York: Pergamon; 1986:13–34.View Article
- Magalhäst PO, Lopes AM, Mazzola PG, Rangel-Yagui C, Penna TCV, Pesspa A Jr: Methods of endotoxin removal from biological preparations: a review. J Pharm Pharmaceut Sci 2007, 10: 338–404.
- Shibatani T, Kakimoto T, Chibata I: Purification of high molecular weight urokinase from human urine and comparative study of two active forms of urokinase. Thromb Haemostasis 1983, 49: 91–95.
- Matsumae H, Minobe S, Kindan K, Watanabe T, Sato T, Tosa T: Specific removal of endotoxin from protein solutions by immobilized histidine. Biotechnol Appl Biochem 1990, 12: 129–140.
- Issekutz AC: Removal of Gram-negative endotoxin from solutions by affinity chromatography. J Immunol Methods 1983, 61: 275–281. 10.1016/0022-1759(83)90221-1View Article
- Sakata M, Inoue T, Todokoro M, Kunitake M: Limulus amebocyte lysate assay for endotoxins by an adsorption method with polycation-immobilized cellulose beads. Anal Sci 2010, 26: 291–296. 10.2116/analsci.26.291View Article
- Wakita M, Hashimoto M: Covalent immobilization of polymeric bilayer membranes to porous supports. Langmuir 1995, 11: 4013–4018. 10.1021/la00010a063View Article
- Wakita M, Adachi T, Ida J, Hashimoto M: Selective adsorption of lipopolysaccharide from protein solutions by porous supports bearing cationic lipid membranes. Bull Chem Soc Jpn 1996, 69: 1017–1021. 10.1246/bcsj.69.1017View Article
- Wakita M, Hashimoto M: Bilayer vesicle formation of N -octadecylchitosan. Jpn J Polymer Sci Technol 1995, 52: 589–593.
- Shands JW Jr, Graham JA, Nath K: The morphologic structure of isolated bacterial lipopolysaccharide. J Mol Biol 1967, 25: 15–21. 10.1016/0022-2836(67)90275-6View Article
- Aida Y, Pabst M: Removal of endotoxin from protein solutions by phase separation using triton X-114. J Immunol Methods 1990, 132: 191–195. 10.1016/0022-1759(90)90029-UView Article
- Wakita M, Hashimoto M: Selective adsorption of lipopolysaccharide in protein solution by polyion-complexed lipid membrane. Influence of the membrane rigidity on the adsorption selectivity. Langmuir 1995, 11: 607–611. 10.1021/la00002a041View Article
- Ihara H: New organic phase for biomimetic HPLC enhanced molecular-shape selectivity through molecular orientation. Chromatograhy 2000, 21: 179–185.
- Simmons LC, Reilly D, Klimowski L, Raju TS, Meng G, Sims P, Hong K, Shields RL, Damico LA, Rantacore P, Yansura DG: Expression of full-length immunoglobulins in Escherichia coli : rapid and efficient production of aglycosylated antibodies. J Immunol Methods 2002, 263: 133–147. 10.1016/S0022-1759(02)00036-4View Article
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.