Bioseparation Steps To Purify Endotoxins Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

The objective of this report is to propose a series of bioseparation steps to purify endotoxins introduced by gram-negative bacteria in protein synthesis. Feasibility in industrial scale-up is also a consideration.


Endotoxins have been identified as a disease-causing substance that is a peripheral structure to the membrane of gram-negative bacteria It is commonly associated with the induction of fever, affecting structures and function at cellular and organic level, and causing shock in patients with endotoxin exposure. Bacterial endotoxins have been shown to have strong biological effects even at very low concentrations when introduced into the blood stream.

The production of human proteins including growth hormones and interferons, insulin and monoclonal antibodies is often achieved through the expression of recombinant DNA in gram-negative bacteria. Endotoxins introduced into the process in this manner is often accumulated into high levels, which is not suitable use in parenteral treatments. The allowable endotoxin limits given by the European Pharmacopoeias is set at 5 endotoxin units (EU) per kg of body weight and hour[1]. Hence, it is a necessity to remove endotoxin residuals, especially when the treatment involves the use of larger dosages, such as that of monoclonal antibody preparations.

Endotoxins are classified as lipopolysaccharides that consist of three main parts. The non-polar lipid component, called lipid A, the core oligosaccharide and a heteropolysaccharide (O-antigen). Lipid A, found on the cell interior, is the least varied and toxically active part of the endotoxin. Most strains have a Lipid A part consisting of a -1,6 linked disaccharide of glucosamine, covalently linked to 3-hydroxy-acyl substituents with 12-16 carbon atoms through ester and amide bonds. The core oligosaccharide region that is closer to lipid A and the lipid A itself are partially phosphorylated. [2] This results in endotoxins having a negative charge in common protein solutions. The O-antigen is made up of a chain of repeating oligosaccharide units of around three to eight monosaccharide units each. The make-up of this chain is strain specific. The middle section of the toxin known as the core oligosaccharide consists of the inner KDO-heptose region and the outer hexose region. This section is strain specific. Both the O-antigen and the core oligosaccharide are found on the cell's exterior.

Despite having the same general assembly, the makeup of each of the three parts differs for different endotoxins from different bacteria. For example, in the lipid A part, the phosphate groups may be substituted with arabinose, ethanolamine. Also, in the O-antigen, the single saccharide units may be acetylated, sialylated or glycosylated. Such alterations are in response to varying environmental conditions the bacteria exists in. As such, each endotoxin might have different properties from each other. Thus, because of such chemical and physical heterogeneity, there is no general method for removal of endotoxins from protein solutions.

Materials and Methods


Ultrafitration is a pressure-driven separation technique that is used for the selective removal of low-molecular-weight substances via permeability-based membranes. It also has the ability to concentrate solutions through the removal of water during the separation process. In general, the main separation mechanism is sized-based sieving and the driving force arises from the transmembrane pressure across the membrane [3].

Particularly for endotoxins removal, membranes with a molecular weight cut off (MWCO) of greater than 100 kDa are employed so as to allow the permeation of target proteins (permeate) while retaining the endotoxins (retentate). One of the disadvantages of ultrafiltration is due to its dependency on the solute properties thus rendering separation difficult. In this case, there may be possibility of partial decomposition of endotoxin aggregates which leads to the release of the toxic lipid A. Moreover, smaller units of endotoxins may also be present in the solution due to protein dissociation. These may result in the permeation of lipid A or endotoxin monomers through the membrane with the target proteins, causing ultrafiltration to be ineffective [2].

Based on the study conducted by Li and Luo (1998), it was found that the addition of calcium ions, Ca2+, to the mixture of protein (haemoglobin) and endotoxin allows the reaggregation of endotoxin subunits. An endotoxin reduction from 5 μg ml-1 to less than 6 pg ml-1 can be achieved with the use of 300 kDa nominal MWCO membrane. Hence Ca2+ addition helps to improve the endotoxin removal and separation efficiency of ultrafiltration. However, the limitation of this method is that it is restricted to proteins that are not adversely affected by presence of high concentrations of Ca2+ [4].

Considering all the disadvantages mentioned, it can be concluded that ultrafiltration alone, is not capable of decontaminating the protein solutions through the removal of endotoxins. Despite this fact, it is still involved in the purification steps as it does not introduce foreign components during the process. This is advantageous as product loss can be greatly minimized when no additional step is required to separate the target proteins from the foreign components. Furthermore, with ultrafiltration, both reduction in endotoxin content and concentrating effect can be achieved hence making it easier for downstream separation. This explains why it is often applied together with a series of downstream purification steps before the final target proteins can be obtained [5].

Two-Phase Extraction

Two-phase extraction is found to be one of the effective methods in endotoxins removal through the addition of surfactants. The success of separation and purification is dependent on the difference between the physicochemical environments in the micelle-rich phase and micelle-poor phase. Above the critical micelle concentration (CMC), non-polar interactions between alkyl chains of lipid A and the surfactant tail groups cause endotoxins to be engulfed in a micellar-like structure. Formation of a new phase then occurs when temperature is raised above that of the cloud point which results in aggregation of micelles to droplets with low water content as shown in Figure .


Figure : Schematic diagram of Triton X-114 micellar solution phase separation [6]

With most of the endotoxins present in the surfactant-rich (bottom) phase, centrifugation or further increases in temperature can be applied to separate the two phases. Extraction cycles can be repeated with the surfactant-poor (top) phase several times so as to achieve the desired clearance of endotoxins.

Surfactants of the Triton series such as Triton X-114 and Triton X-100 displays miscibility gap in aqueous solutions. In this case, Triton X-114 with a low cloud point of 22 °C is a better choice in protein purification as compared to Triton X-100. This is due to the fact that Triton X-100, which has a cloud point of 75 °C, may lead to potential denaturation and hence loss of activity of target proteins [6].

It was reported by Adam et al. (1995) that two-phase extraction can perform better separation as compared to the other separation techniques. In fact, a 100-fold endotoxins reduction can be achieved from exopolysaccharide and plasmid DNA with the use of Triton X-114. The former results in a final endotoxins content of 30 EUmg-1 at 50 % loss in bioactivity while the latter ends up with an endotoxins content of 0.1 EU in 6 μg DNA. A comparison study between affinity adsorption and two-phase extraction for endotoxins removal from recombinant proteins cardiac troponin I, myoglobin and creatine kinase isoenzymes has been conducted by Liu et al. (1997). Results showed that phase separation is able to reduce endotoxins content to as high as 99 %, far more effective than affinity adsorption.

Nevertheless, there are some drawbacks with the use of two-phase extraction that must be taken note of. Firstly, 10 to 20 % product loss is inevitable since removal of surfactant form the target protein by adsorption or gel filtration is required. Secondly, it is a time consuming process as multiple cycles will be required to obtain desired endotoxins clearance. Thirdly, only biopolymers that can partition in water phase can adopt this method. Lastly, in view of the sensitivity of proteins towards temperature changes, there may be difficulty in scaling up of the process due to temperature shifts [2].


Besides ultrafiltration and two-phase extraction, adsorption techniques are also used to remove endotoxins from protein solutions. Figure shows the principle of endotoxin adsorption. Anion-exchange and affinity adsorption ligands have a net positive charge. This causes the net negatively charged proteins and endotoxins to adsorb at low ionic strengths. After the adsorbent's capacity is reached, protein recovery approaches 100%. [7] The competition for the binding sites of the adsorbent and repulsion of proteins affects the efficiency of endotoxin removal.

Figure : Mechanism of Endotoxin Adsorption from Protein Solutions [7]

Non-selective adsorption using activated carbon cannot be used to decontaminate protein solutions as the adsorption is irreversible. [7] Hence, the anion-exchange chromatography adsorption method is discussed.

Anion-Exchange Chromatography

The net negative charge of endotoxins due to the presence of phosphate groups from lipid A enables the use of anion-exchange polymeric matrices to remove them. A common matrix used is the Diethylaminoethanol (DEAE)-Sepharose. At usual endotoxin levels found in the feed, which is less than 10ng/ml, around three to four orders of clearance can be achieved by anion-exchange chromatography. [7] However, for acidic protein solutions, anion-exchange matrices are not suitable for the selective removal of endotoxins. They have high adsorbing capacities that will adsorb both the endotoxins and acidic proteins, making them ineffective in the purification process. [2] Therefore, anion-exchange chromatography shows the best results for basic proteins. [7]

To make it possible for anion-exchange chromatography to be applied on acidic protein solutions, the pH of the proteins can be changed to reach or exceed the isoelectric point (pI). [7] This suppresses the adsorption of the proteins to the matrix, hence reducing competition for the binding sites. The requirement for this to be feasible is that the acidic proteins must be soluble and stable enough at the isoelectric point.

Affinity Adsorption

Affinity adsorption works on the principle of structural recognition and specific interaction between adsorbents and endotoxins. Affinity techniques used to remove endotoxins have a higher recovery of protein compared to other techniques. [7] The mechanism of endotoxin adsorption is the same as that for anion-exchange ligands. As endotoxins are negatively charged and hydrophobic, affinity sorbents can tap on the electrostatic and/or hydrophobic interaction between sorbent and endotoxin to selectively remove it from protein solutions. The types of affinity adsorption include the employment of polymyxin B-immobilized sepharose, histamine and histidine-immobilized sepharose, polycationic ligands, polymeric matrices and immunoaffinity ligands which will be discussed in this section.

Polymyxin B-Immobilized Sepharose

Polymyxin B is an antibiotic that is basic with positive charges. Polymyxin B destroys gram-negative bacteria by disrupting the cell walls when inserted. It is a group-selective ligand which can recognize a range of endotoxins. [7] It can be immobilised on CNBr-activated Sepharose to form columns for affinity adsorption of endotoxins. These columns showed clearance factors of more than 105 from heavily contaminated culture filtrates having concentrations of 10µg/ml. [2]

Despite having advantages such as high removal of endotoxins, there are some disadvantages when polymyxin-Sepharose columns are used. The ionic interactions between the cationic region of polymyxin B and negatively charged proteins at low ionic strengths cause protein losses when they are passed through the column. This accounts for the low DNA recovery (50%) in spite of 200 to 10,000-fold reduction of endotoxins. [7] This method is also not suitable for intravenous (IV) injection solutions due to its high cost [2] and the neurotoxicity and nephrotoxicity of polymyxin B when in solutions. [7]

Histamine and Histidine-Immobilized Sepharose

Histamine and histidine, when immobilized to Sepharose matrix, showed good results of separating ribonucleic acids from endotoxins, which is usually very hard to remove. [2] Besides histamine and histidine, bases such as adenine and cytosine can also be used as adsorbents for endotoxins. Among the different bases, histamine has the highest affinity for endotoxins. [2]

Histamine and histidine have similar efficiencies of removing endotoxins with polymyxin B, with clearance factors ranging from 5 to 200, depending on the conditions and concentration of the protein sample. They are able to decontaminate protein solutions consisting of lysozyme, insulin and myoglobin. [7] Even though histamine has a higher affinity for endotoxins, histidine is a safer choice for IV injection solutions due to histamine's biological activity. [2]

Polycationic Ligands

Polymers with cationic functions such as the hydrophilic polyethyleneimines (PEI) have been found effective as ligands in forming secondary bonds at surface with endotoxins via van der Waals interaction and hydrogen bonding. An efficacy similar to that of polymyxin B was obtainable if PEI is immobilized on cellulose beads, but immobilization on cellulose fibres proved to be more effective. This method has been experimented on -globulin, myoglobin and Cytochrome C solutions and shown a high endotoxin clearance of more than 98% with a high protein recovery (> 98%). [8]

It was also reported by Friedrich and Dagmar that polycationic ligands are most effective to be employed for the removal of endotoxins from net-negatively charged proteins with concentrations less than 1 mg/ml and net-positively-charged proteins such as human lgG, ribonucleases and lysozyme. When immobilized on nylon microfiltration membranes coated with dextran 40,000, results obtained from the experiment has confirmed a high endotoxin clearance of 8000 EU/ml by a single pass through the adsorber membrane, which highlighted the possibility of deploying such technique for rapid processing of large volumes. [9]

Use of ligands other than PEI was also plausible for the effective removal of endotoxins. Examples of such ligands are: poly-L-lysine (PLL) and poly-L-histidine (PLH). Among these ligands, PLH had demonstrated to have greater efficacy. However, it is also relatively expensive and unstable under alkaline conditions. [8]

Polymeric Matrices

For efficient adsorption of endotoxins, poly(-methyl-L-glutamate) beads which are spherical and porous were aminated to provide high endotoxin-binding capacity. These beads were also positively charged to ensure affinity binding and selective removal of endotoxins from the protein solution. This method was claimed to have lower dependence on ionic strength, which enabled it to be effective for up to a 0.4M salt concentration, and high selectivity towards bovine serum albumin (BSA). The diffusion of proteins into pores was also avoided by adjusting reaction conditions to achieve small pore size beads with small pore sizes. As such, high recoveries of net-negatively charged proteins and strong adsorption of endotoxins are achieved. At pH 7, removal efficiencies of 96-99% were obtained at protein recoveries of above 99%. [8]

Polymeric matrix synthesized from immobilization of L-serine ligand (PVDF-Ser) on polyvinylidene fluoride matrix has also proved to be twice as effective as PVDF carrier cartridge in a recent research study conducted by Gao and co-workers. [10] However, this method has yet to prove applicable on human. Furthermore, these cation-containing polymeric matrices have relatively less chemically stable ester bonds which, during synthesis, can only be replaced partially in the presence of amide groups. The remaining ester bonds tend to hydrolyse under harsh environmental conditions, thereby altering the structure of the matrix and reduce the binding capacity.

Other Techniques

Isoelectric focussing is carried out by utilizing electrolyser with multi-compartment which is fitted with membranes that have predetermined isoelectric points (pI) to fractionate proteins. Similar technique can also be carried out under continuous circulation of myoglobin solution between membranes at controlled pH of 6.98 to 8.04 and in 1 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES). It was found that at pH 5.1 and over a duration of 3 hours, 99.9% of the endotoxin content was removed. [2]

Colloidal zirconia can also be used as an affinity adsorber since the phosphate and phosphate esters possess high affinity for endotoxins to bind strongly to. However, such adsorbent proved to have decreased efficiency in the presence of BSA. Another form of sorbent makes use of cross-linked chitosan beads. As chitosan becomes positively charged in an acidic environment, it can be employed as a cationic ligand. Quaternised chitosan improves efficacy for processing under alkaline conditions. [2]

To take advantage of hydrophobic interactions for separation of endotoxin under high salt concentrations (> 3 M), histidine-immobilised sorbents may be utilized. Yet, this technique may impose high recovery costs of proteins that still remain in the high salt concentration solution. [2]

Removal of endotoxins from purified recombinant proteins is also proved efficient by Dudley and co-workers using reverse phase high-performance liquid chromatography (HPLC). It was claimed that this technique may be the best strategy used in the final purification of endotoxin-free recombinant proteins, especially if the downstream applications involve endothelial cell or angiogenesis-type assays. [11]


The proposed purification strategy summary for the removal of endotoxins from protein solution in gram-negative bacteria is shown in Figure : Purification Strategy.

Figure : Purification Strategy

The proposed strategy comprises of a combination of purification techniques previously mentioned in Section 3. Before the purification steps can proceed, the cells of gram-negative bacteria have to be first isolated and this is achieved by centrifugation of the fermentation broth. Due the density difference of the components in the broth, two distinct layers will be produced, namely the supernatant and the sediment. The supernatant will comprise of the less dense materials such as the soluble extracellular proteins and nutrients, while the sediment will comprise of the bacteria cells which are much denser. The bacteria cells can now be separated from the broth and are ready for purification.

To purify the intracellular recombinant proteins produced in the bacteria cells, the cells have to be lysed to release the protein contents. There are various methods for cell lysis including enzymatic lysis, sonication, bead-beating, detergent based lysis and high-shear mechanical methods. However, with considerations for the feasibility of large scale purification in terms of cost and large volume operation, high-shear mechanical cell lysis is considered. Rotor stator homogenizer is selected in this purification strategy as disruption of microorganisms can be achieved efficiently with the aid of glass beads. Moreover, it has a huge batch capacity of approximately 19,000 liters and on-line capacity of 68,000 liters/hour. [12] Rotor stator homogenizers also generates very small amount of heat and hence, minimal cooling is required.

After the bacteria cells are lysed, the cell debris and insoluble components have to be separated to further purify the protein samples. Cell lysate can be clarified by means of centrifugation. Centrifugation separates the soluble proteins and endotoxins from the insoluble cell debris. The clarified cell lysate will then be available for the rest of the purification steps.

Ultrafiltration is then used to treat the cell lysate as described in Section 3.1. The efficiency of endotoxin removal using ultrafiltration is highly dependent on the relative size of endotoxin to the target protein in the sample. If endotoxins are larger in size as compared to the target protein, ultrafiltration is capable of reducing the endotoxin levels in the protein sample as they are mostly trapped in filter membrane, with the target protein found in the permeate. However, it should be noted with caution that endotoxins would decompose under conditions of high temperature and high acidity. [2] This allows the lipid A component of endotoxins, which is highly toxic, to permeate through the membrane and hence, contaminating the protein samples. On the other hand, if endotoxins are smaller than the target protein, they will end up in the permeate together with the target protein, making the purification step inefficient. Hence, the pore size of the membrane used for ultrafiltration is crucial in determining the efficiency of this purification step. In addition, additives such as calcium ions (Ca2+) are added to improve endotoxin removal. Ca2+ promotes the aggregation of endotoxin molecules and thus allows more endotoxin molecules to be trapped in the filter membrane, purifying the protein sample. [2] As a result, a membrane with nominal weight cut-off of 300kDa is used to allow permeation of the target protein and at the same time, trapping the large endotoxin aggregates.

To obtain even higher purity of the target protein, two-phase extraction with Triton X-114 has to be employed. The use of Triton X-114 is supported with its convenient cloud point of 25°C. [5] Although the Triton X-114 is only slightly denser than water, the density difference between the two phases can be widened by lowering water phase density which was accomplished by ultrafiltration in the previous step. A substantial density difference is essential in determining the efficiency of endotoxin removal. Above the cloud point, Triton X-114 exists as a different phase with water and extracts endotoxin, leaving the target protein in the aqueous phase. Still, a number of such extraction cycles have to be performed to attain higher percentage of endotoxin removal. This strategy uses 3 cycles of extraction to achieve 99% removal of endotoxins. [5]

The last step of the purification strategy is the process of chromatography. The phosphate group present on the lipid A tail of endotoxins gives them a net negative charge. Although anion-exchange chromatography can be employed to adsorb endotoxins on the stationary phase surface which is positively charged, this separation is only effective when the target protein does not have an overall charge. If the target protein is also negatively charged, it may also be co-absorbed onto the stationary phase, leading to a loss of target protein and hence a low yield. Conversely, a positively charged target protein would form complexes with endotoxins, thus minimizing the endotoxin removal efficiency. Therefore, affinity chromatography is employed to provide better separation efficiency. Endotoxins have both anionic and hydrophobic properties. Hence, the ligand used must be both cationic and hydrophobic to ensure strong adsorption of endotoxins to the chromatography matrix. Poly(ε-lysine) immobilized on cellulose is used for this purification step as it is commonly used for endotoxin removal in protein solutions. Moreover, it is capable of reducing endotoxin level to 0.1 EU/ml. [7]

The purification strategy proposed has certain limitations with respect to large scale operation. Firstly, under a large scale production, flow rate will be high. This leads to serious problem of membrane fouling, which has to be changed frequently, and such solution is not economically-wise.

While ultrafiltration has proven to be efficient in the separation of endotoxins, it is worthwhile to note that such efficiency is only possible in a protein-free solution. In the purification of endotoxins from the target protein, the strong agitation forces of the ultrafiltration in a large scale production will result in damage of target protein, rendering the whole effort futile [13].

An important step in the purification strategy involves a two-phase extraction with Triton X-114. This step has the potential of removing up to 99% of endotoxins depending on the number of cycles performed [12]. While this step is simple, and cost-effective in large scale production, the disadvantage of it is the requirement for a high speed centrifugation step. Moreover, the optimum conditions are also unknown to date. This puts the cap on the scale-up possibility of such a separation procedure.

After the two-phase extraction process, it is possible that some detergent remains in the protein solution. As such, an additional separation step is most likely to occur. According to Aida Y. it is most likely that this process will lead to a 10-20% product loss [12]. In terms of large scale production, this could lead to high purity, but low yield of target protein. This translates to expensive protein purification steps.

A recommendation that is possible to overcome these problems is the use of the expanded bed adsorption (EBA) method. With the use of the EBA, crude sample can be applied to it, such that the steps of capture, clarification and concentration can be done in a single step. This allows large scale production without the need to up scale the centrifugation step. The EBA involves the introduction of feed from the bottom, the expansion of the bed such that it fluidize and the adsorption of the endotoxin on the beads through affinity or ion-exchange. The effluent will be the purified target protein exiting through the column. However, the optimum conditions have to be further investigated before it can be implemented in reality.


Given the benefits and downsides of the three methods of endotoxin purifications proceeding cell lysis and clarification; namely, ultrafiltration, two-phase extraction and adsorption discussed in the text, each of the three methods are employed in a sequence of protein purification steps to achieve a much higher purification factor than if were to be done alone. (Figure )

However, as discussed, the upscaling of the purification strategy is to be costly due in part to the lowered protein yield, inevitably lost in the two-phase extraction method [2]. Furthermore, both ultrafiltration and adsorption methods are limited in scale, and application. As discussed, adsorption methods which include anion-exchange chromatography and affinity adsorption must be tailored to the properties of the desired proteins. In summary, while the upscaling of the said strategy is likely to be possibly, the high cost of purification is likely to raise the cost of the final protein product, and hence, pass the cost to end consumers. Therefore, the application of the purification steps is likely to be limited to treatment that necessitates the high purity of the final product, such as monoclonal antibody preparations [2].

The use of an expanded bed adsorption (EBA) method could possibly overcome these limitations, by incorporating the purification steps into a single step. However, the purification factor of this method is not yet optimized for large scale operations [13]. Further research could involve the development of new adsorbents with improved protein-binding capacity. In combination with the use of highly specific and stable ligands, it will allow a higher product yield with the initial recovery of products even in large scale feedstock volumes used. [14]