Removing endotoxin from biopharmaceutical solutions

October 1, 2009

Pharmaceutical Technology Europe

Pharmaceutical Technology Europe, Pharmaceutical Technology Europe-10-01-2009, Volume 21, Issue 10

Endotoxin removal from a finished product is a major challenge for biopharmaceutical manufacturers; particularly as all endotoxin removal methods have operational limitations and may result in loss of protein.

Recombinant biopharmaceuticals are manufactured using complex biological systems, such as bacteria, yeast, baculoviruses or mammalian cells. However, regulating consistent yields, monitoring contamination and validating product efficacy during the manufacturing process are areas of concern and represent major challenges for biopharmaceutical manufacturers.

High batch rejection rates of biopharmaceuticals, which are common during production, can usually be attributed to contamination by biological agents such as microorganisms or byproducts produced by these microorganisms, including toxins or pyrogens. Endotoxin contamination of biopharmaceuticals is a serious issue and may infringe on workforce safety. Apart from adherence to FDA regulations and strict international quality standards, many US pharmaceutical companies must also deal with the Occupational Health and Safety Administration, which ensures that contamination control is much more than just a housekeeping activity. In Europe, this task is taken up by the European Agency for Safety and Health at Work.

flashfilm/Getty Images

Bacterial endotoxin is another term used for lipopolysaccharides (LPS), complexes that are located in the outer cell membrane of Gram-negative bacteria and blue-green algae.1 Gram-negative bacteria are widely used in biopharm manufacturing to produce recombinant DNA products, such as therapeutic proteins. LPS subunits are complex amphiphilic molecules with a molecular weight (MW) of approximately 10–20 kDa2,3 and vary widely in chemical composition both between and among bacterial species. LPS complexes tend to aggregate and form large structures that have an average MW > 10 kDa.

LPS is a potential cause of pyrogenic reactions in parenteral drug products because these complexes can act as a strong immunostimulant that activates the complement system by the alternative (properdin) pathway4 upon entry into the human blood circulation. This can cause death of the individual who is administered with the drug.5,6

A pyrogenic reaction can be caused by only a small amount of endotoxin — approximately 0.1 ng/kg of body weight. The standard reporting unit for endotoxin data is one endotoxin unit (EU) — the equivalent of 0.1 ng. A typical gram-negative bacterium contains 10–15 g of LPS, which means that at least 105 bacterial cells are required to contribute 0.1 ng of endotoxin.

The chemical structure of endotoxin

LPS consists of three components or regions: Lipid A, an R-polysaccharide and an O-polysaccharide (Figure 1).

Figure 1: Lipopolysaccharide (LPS) structure and location in the bacterial membrane.

Somatic (O) antigen or O-polysaccharide is attached to the core polysaccharide and consists of repeating oligosaccharide subunits made up of three to five sugars. The O-polysaccharide maintains the hydrophilic domain of the LPS molecule and also contains the major antigenic determinant (antibody-combining site) of the gram-negative cell wall.

Toxicity of endotoxin has been found to be associated with Lipid A, whilst immunogenicity is associated with the polysaccharide components.

Endotoxin in the biopharm industry

During the last decade, more than 20 different monoclonal antibodies (mAbs) have been approved for therapeutic use by the FDA.7 Numerous new mAbs and recombinant proteins are in clinical trials or under development, many of which are expected to be approved in the coming years. Advances in understanding protein function are leading to the development of smaller truncated protein therapeutics with functional active sites that may not require post translational modifications. This has led to the re-emergence of bacterial expression systems (E.coli, etc.) as the preferred expression system for manufacturing proteins that do not need to undergo post-translational modifications for their activity. However, endotoxin issues associated with bacterial expression systems remain the toughest operational challenge.

Endotoxin is notoriously resistant to destruction by heat, desiccation, pH extremes and various chemical treatments, but many techniques are now available that remove endotoxins from solutions by exploiting certain characteristics of the molecule:

  • Endotoxin molecules tend to form micelles or vesicles in aqueous solution and can be removed by filtration.2,8

  • Their hydrophobic nature allows separation by a two-phase extraction9 or by hydrophobic interaction chromatography.10

  • Their negative charge can be used for adsorption on anion exchangers and subsequent removal.11

Several affinity ligands also showed good specificity for endotoxin binding, such as polymyxin B,12 histidine,13 dimethylamine ligands,14 deoxycholic acid,15 polycationic ligands,16 poly-ε-lysine and poly-L-lysine.17

Endotoxin and protein interaction

Endotoxins have the tendency to bind and form complexes with some proteins, which are usually target compounds of interest in recombinant biopharmaceuticals. Table 1 depicts the proteins that endotoxin can interact with. Many basic proteins (pH > 7) have been reported to interact with endotoxins.18,19 Electrostatic interactions are usually the driving force for such interaction; however, some studies suggested that hydrophobic interactions are the major driving force and that only limited ionic binding is involved.20 Neutral proteins (pH ~ 7), such as haemoglobin, and even acidic proteins (pH < 7), such as serum albumin, have been known to interact with endotoxin. The mechanisms of such interactions are not quite clear, but can be assumed to be a hydrophobic interaction between the protein and endotoxin. In the case of serum albumin, however, fatty acid binding domains may also be involved.

Table 1: Proteins known to interact with endotoxin.

Dissociation of endotoxin from protein solutions can be achieved by detergent treatment, but an additional problem in the purification process is the removal of the surfactant.21 Significant product loss and low product yield can also result from the separation steps employed to remove endotoxins.22

Regulatory guidelines on endotoxins

The USP chapter "Sterilization and Sterility Assurance of Compendial Articles" recognizes the value of validated processes that eliminate endotoxin. This chapter states that an acceptable depyrogenation process should demonstrate at least a three-log (103) difference between the recoverable input endotoxin ≥ 1000 EU and any residual endotoxin present after processing. The maximum allowable endotoxin limit for biological preparations is laid down by regulatory bodies across the world; for example, according to FDA guidelines, the maximum endotoxin limit (EU/mL) for streptokinase, insulin and meningococcal polysaccharide vaccine is 0.02 EU/100 mL, 2.5 EU/mL and 200 EU/mL, respectively.

Managing endotoxin

Therapeutic proteins expressed in E.coli have the intrinsic risk of high endotoxin levels. Sometimes, endotoxins released from the E.coli cell wall bind to the protein of interest and make the downstream process more challenging, but contamination can also occur at any stage of a production process. Therefore, it is important to monitor the endotoxin levels at different production stages and to take the necessary corrective and preventive action to minimize the risk of endotoxin contamination or carry over.

The author says...

There are two ways to manage endotoxin within the prescribed levels: size-based and charged-based separation.

Minimum cGMP for preparing drug products for administration to humans or animals is laid down in the FDA's 21 CFR 211 document. This document provides guidelines regarding facility design, organization and personnel, equipment, production process, packaging and labelling control, control of components of drug product container closure systems, report generation, quality control checks and laboratory controls. By following these guidelines, the entry of endotoxin into the biotechnology manufacturing process from the surroundings can be reduced.

Endotoxin removal

Endotoxin removal methods can be classified into two main categories based upon the mechanism of removal. These are:

  • based on size exclusion of endotoxin (ultrafiltration [UF])

  • based on electrostatic/hydrophobic interactions (affinity and ion-exchange chromatography, charged membrane/depth filtration).


Endotoxins may exist in monomeric form (10–20 kDa) or may aggregate to form vesicles that have an MW of 300–1000 kDa. Successful removal of endotoxin during the manufacture of water for injection (WFI) has been demonstrated using 10 kDa UF membranes. However, UF processing of protein solutions has some major disadvantages, such as low filtration rates, a high consumption of processing time in the case of viscous products, and the high cost of the UF system and modules. Also, UF membranes cannot be used to remove endotoxin from the protein solutions unless the size of the protein molecule is at least 25 times smaller than the endotoxin molecule. Considering the limitations of molecular weight/size of the product, tangential flow filtration can only be implemented for endotoxin removal from, for example, small peptides and APIs (Table 2).

Table 2: Endotoxin removal by ultrafiltration.

Filtration (charge interaction)

Depth filtration refers to the use of a porous medium that can retain articles from the protein solution throughout its matrix rather than just on its surface.23 Depth filters have a high solid-handling capacity per unit area of filter used and can process larger volumes of feed-streams loaded with a high amounts of colloids, cell debris, nucleic acids or even endotoxin.

Positively charged depth filters (i.e., Millistak+ filters; Millipore, MA, USA) or membrane filters (i.e., Charged Durapore filters; Millipore, MA, USA) have been used to remove endotoxins from water, saline and sugar solutions.24,25 Charged Durapore has been shown to exhibit a > 5 log reduction value (LRV) when challenged with 106 pg/mL of purified E.coli endotoxin (Type 055:B5 LPS). Endotoxin retention and subsequent removal is dependent on the net positive charge on the depth filter; the net charge on filter surfaces is strongly influenced by pH, dissolved solids and the presence of soluble organics.26 In comparison, with positively charged filters, little retention of endotoxin occurred when negatively charged depth and membrane filters were used. Removal efficiency also decreased in the presence of 5% newborn calf serum — probably because serum proteins compete with the endotoxin for adsorptive binding sites on the charged filters — and at pH levels > 8.5, which can be attributed to a decrease in the net charge on the endotoxin molecule.

In the presence of competing negative ions, however, such as in buffer filtration, a charged membrane cannot be used because it loses its adsorptive properties. Depth filters tend to have comparatively higher hold-up volume and release more extractables than membrane filters. Post-use air blow down at low pressure or buffer flush is a standard practice for product recovery from depth filters. However, buffer flush cause dilution of the filtrate. Because of high extractables, implementation of a depth filter towards the end of the manufacturing process is not favoured.


Chromatography-based endotoxin removal techniques require strong selectivity to achieve low residual endotoxin levels without affecting protein recovery. There are two chromatographic methods for the removal of endotoxins: binding the endotoxins to positively charged surfaces and allowing protein solutions to flow through; or binding the proteins to negatively charged surfaces and allowing endotoxins to flow through.

Ion-exchange chromatography is the most popular method for protein purification and is based on the principle that proteins have different charges at different pH values. The solid adsorbents are charged, positive or negative, and will adsorb charged proteins based upon the charges. According to the difference of the interaction forces between the protein and adsorbent, proteins are bound with different strengths to the adsorbent (Figure 2). Thus, upon the use of a suitable buffer, the protein can be displaced easily from the adsorbent and washed out at different velocities; the less the interaction between the adsorbent and the proteins, the faster they will be washed out. Proteins can then be separated according to the sequence of their elution.

Figure 2: Modes of Interaction of endotoxin with chromatographic media matrices.

Endotoxins are negatively charged under conditions commonly encountered during protein purification and, therefore, can be removed using anion-exchange chromatography. If endotoxin binding can be achieved under conditions at which the protein of interest carries a net positive charge (i.e., at a pH below its isoelectric point), then the protein will be repelled from the positively charged matrix and flow through with the mobile phase where it can be collected.

The use of chromatography resins based on quarternary ammonium-based chemistry is well known for endotoxin removal. Membrane adsorbers based on quarternary ammonium and, relatively recently, polyallylamine (ChromaSorb; Millipore, MA, USA) are relatively recent additions to the chromatography-based endotoxin removal techniques. The ChromaSorb membrane adsorber has been validated to remove at least > 3 LRV of endotoxin spiked in buffer (Figure 3). In addition, its binding strength can significantly reduce the risk of changes within a process resulting in earlier breakthrough of impurities.

Figure 3: Endotoxin removal by ChromaSorb in buffer at increasing salt concentrations.

Membrane adsorbers offer many advantages compared with packed bed resin chromatography. One of the principal benefits is that the transport phenomena are convection-driven. Binding sites in membrane adsorbers are exposed to the molecules within short diffusion distances, whereas 90–95% of bead ligands will be reached only by large diffusion distances and pore diffusion. Hence, the open membrane structure combines the advantages of excellent flow characteristics, and thus productivity, with the high selectivity of classical chromatography. Membrane adsorbers have been found to be a useful tool in endotoxin removal, viral vaccine production and DNA purification for gene therapeutic agent production at process scale.20,27 In addition, because membrane adsorbers are single-use devices, they do not require the large capital investment usually associated with ion-exchange resins and columns.

In bead-based chromatography, most of the adsorption sites are internal to the bead and the rate of mass transfer is controlled by pore diffusion. Hence flow rate becomes a limitation. Chromatography has the inherent limitation of costly and time consuming column packing, cleaning validation and requirement of high floor space and buffer usage.

Though membrane adsorber offers a great advantage compared with bead-based chromatography, universal adoption of this technology has been slow because it is limited by the lower binding capacity than that of bead-based chromatography, though the high flux advantages provided by membrane adsorbers would lead to higher productivity.


Efficient removal of endotoxins from biopharm solutions is challenging and a lot of research is being focused on this area. Despite the development of chromatography and membrane adsorbers in recent years, more research is needed to ensure the removal of endotoxin is a cost-effective process.

All of the methods devised for endotoxin removal possess operational limitations and result in protein loss when the endotoxin level in the therapeutic solution is high, thus, increasing operational costs. Endotoxin removal from pharmaceutical and biotechnology solutions requires strict adherence to cGMP guidelines, as well as the selection, optimization and validation of the endotoxin removal method.


The authors are grateful to Millipore India Pvt. Ltd and Biomab Biopharmaceuticals India Pvt. Ltd for their support in the preparation and review of the manuscript.

Valencio Salema is Process Development Specialist at Millipore India Pvt. Ltd, Bangalore (India). Tel. +91 80 3928 2817 Fax +91 80 2839 6345

Lalit Saxena is a scientist at Biomab Pharmaceuticals (India) Private Limited, Goa (India).

Priyabrata Pattnaik is Technical Manager, Biomanufacturing Sciences Network, at Millipore Corp. Millipore Singapore Pte Ltd, (Singapore).

Microbiological methods

For more about microbiological methods, why not take a look at Pharmaceutical Technology Europe's interview with Martin Cockcroft, Operations Manager at Tepnel Pharmaceutical services? Cockcroft discusses the challenges of harmonizing microbiological methods between the US, European and Japanese pharmacopoeias.


1. L. Anderson and F. Unger, ACS Symposium Series, No. 231, American Chemical Society (1983).

2. K.J. Sweadner, M. Forte and L.L. Nelsen, Appl. Environ. Microbiol., 34(4), 382–385 (1977).

3. T.T. Evans-Strikfaden and K.H. Oshima, J. Pharm. Sci. Technol., 50 (1996).

4. Rietschel et al., FASEB, 8(2), 217–225 (1994).

5. K. Hou and R. Zaniewski, Biotechnol. Appl. Biochem., 12, 315–324 (1990).

6. D.C. Morrison and J.L. Ryan, Annu. Rev. Med., 38, 417–432 (1987).

7. U. Ritzen et al., J. Chrom. B., 856(1–2), 343–347 (2007).

8. J.C. Cradock et al., J. Pharm. Pharmacol., 30(3), 198–199 (1978).

9. Y. Aida and M.J. Pabst, J. Immunol. Methods, 132(2), 191–195 (1990).

10. M.J. Wilson et al., J. Biotechnol., 88(1), 67–75 (2001).

11. Anspach et al., J. Chrom., A711, 81–92 (1997).

12. A.C. Issekutz, J. Immunol. Methods, 61(3), 275–281 (1983).

13. S. Minobe et al., Biotechnol. Appl. Biochem., 10, 143–153 (1988).

14. Z. Yuan et al., Biomaterials, 26(15), 2741–2746 (2005).

15. D. Petsch, E. Rantze and F.B. Anspach, J. Mol. Recognit., 11, 222–230 (1998).

16. S. Mitzner et al., Artif. Organs, 17(9), 775–781 (1993).

17. F.B. Anspach and O. Hilbeck, J. Chromatogr. A., 711, 81–92 (1995).

18. E. Elass-Rochard et al., Biochem. J., 312, 839–845 (1995).

19. D. Berger and H.G. Berger, Drug Res., 38(6), 817–820 (1988).

20. N. Ohno and D.C. Morrison, J. Biol. Chem., 264, 4434–4441 (1989).

21. P. Reichelt, C. Schwarz and M. Donzeau, Protein Exp. Purif., 46(2), 483–488 (2006).

22. D. Petsch and F.B. Anspach, J. Biotechnol., 76(2), 97–119 (2000).

23. J.V. Fiore, W.P. Olson and S.L. Holst, "Depth Filtration," in J.M. Curling, Ed, Methods of Plasma Protein Fractionation (Academic Press, New York, New York, USA, 1980) pp 239–268.

24. C.P. Gerba and K. Hou, Appl. Environ. Microbiol., 50(6), 1375–1377 (1985).

25. K. Hou and R. Zaniewski, J. Parenter. Sci. Technol., 44(4), 204–209 (1990).

26. K. Hou et al., Appl. Environ. Microbiol., 40(5), 892–896 (1980).

27. D.C. Morrison et al., Prog. Clin. Biol. Res., 388, 3–15 (1994).

28. K. Poelstra et al., Lab. Invest., 76, 319–327 (1997).

29. S.A. David, P. Balaram and V.I. Mathan, J. Endotoxin Res., 2(2), 99–106 (1995).