Removal of endotoxins from recombinant antithrombin

Endotoxins (also known as lipopolysaccharides or ETs) are a pyrogenic byproduct of the breakdown of gram-negative bacterial cell-walls, such as E. coli. They cause fever and septic shock in humans, and are therefore highly regulated by the FDA and USP. ETs are a common contaminant in recombinant protein solutions, derived from cell cultures which also have ETs. ETs are angiogenic and stimulate the inflammatory response, making them extremely important to remove for research on these cellular functions. The high stability of ETs makes them very difficult to remove and methods must be designed specifically for this purpose. The protein of interest in this study is recombinant antithrombin (AT), which is potentially antiangiogenic and therefore could be used for reatment of diseases caused by excess angiogenesis, like diabetic retinopathy and cancer. This study used two methods of endotoxin removal shown to be effective in cleaning protein solutions: phase extraction with Triton X-114 and affinity column chromatography. Heparin affinity columns were used, due to the strong interaction between heparin and AT. Triton X-114 is also used to wash the affinity column. Results show that both a combination protocol of the phase extraction and affinity chromatography and an affinity column only protocol are effective in removing endotoxins.

REMOVAL OF ENDOTOXINS FROM RECOMBINANT ANTITHROMBIN by Stephanie Ann Lietzke A Senior Thesis Submitted to the Faculty of The University of Utah In Partial Fulfillment of the Requirements for the Honors Degree in Bachelors of Science In Biomedical Engineering Approved: _____________________________ Susan Bock, Ph.D. Thesis Faculty Supervisor ______________________________ Dr. Patrick Tresco, Ph.D Chair, Department of Biomedical Engineering _____________________________ Dr. Kelly Broadhead, Ph.D. Honors Faculty Advisor ______________________________ Sylvia D. Torti, PhD Dean, Honors College April 2016 Copyright © 2016 All Rights Reserved ABSTRACT Endotoxins (also known as lipopolysaccharides or ETs) are a pyrogenic byproduct of the breakdown of gram-negative bacterial cell-walls, such as E. coli. They cause fever and septic shock in humans, and are therefore highly regulated by the FDA and USP. ETs are a common contaminant in recombinant protein solutions, derived from cell cultures which also have ETs. ETs are angiogenic and stimulate the inflammatory response, making them extremely important to remove for research on these cellular functions. The high stability of ETs makes them very difficult to remove and methods must be designed specifically for this purpose. The protein of interest in this study is recombinant antithrombin (AT), which is potentially antiangiogenic and therefore could be used for treatment of diseases caused by excess angiogenesis, like diabetic retinopathy and cancer. This study used two methods of endotoxin removal shown to be effective in cleaning protein solutions: phase extraction with Triton X-114 and affinity column chromatography. Heparin affinity columns were used, due to the strong interaction between heparin and AT. Triton X-114 is also used to wash the affinity column. Results show that both a combination protocol of the phase extraction and affinity chromatography and an affinity column only protocol are effective in removing endotoxins. ii TABLE OF CONTENTS ABSTRACT ii INTRODUCTION 1 METHODS 4 RESULTS 12 DISCUSSION 17 ACKNOWLEDGEMENTS 19 REFERENCES 20 iii 1 INTRODUCTION Diabetic retinopathy is a well-known complication of diabetes, a disease that affected 285 million individuals globally in 2010 [1], and is expected to increase to 439 million by 2030. Excess angiogenesis, the formation of new capillaries on the retina, characterizes retinopathy and leads to vision impairment [2]. In order to treat retinopathy, an antiangiogenic agent is clinically indicated. Current treatments include injection of immune therapies, such as bevacizamab and ranibizumab [2], monoclonal antibodies used to treat metastatic breast cancer and glioblastoma and therefore carry the risk of triggering an immune response [3]. Antithrombin (AT) is an inhibitor protein in the coagulation cascade, which is activated by an ionic bond to heparin and then binds to thrombin at a reactive peptide loop. Certain conformers of latent antithrombin (AT), where the reactive peptide loop is unavailable to thrombin as it is forced inside the alpha sheet of AT (Fig. 1), have been Fig.1: a. Thrombin-reactive AT; b. Latent (Thrombin non-reactive) AT [14]. 2 shown to have these antiangiogenic effects [4]. Latent AT is derived from its induced expression in cultures of baculovirus-infected insect cells, which is often contaminated with gram-negative bacteria due to the yeastolate in the media. These contaminants release endotoxins (ETs), lipopolysaccharides found in the cell walls of gram-negative bacteria [5]. As a result, these solutions of recombinant antithrombin are contaminated with endotoxins, making them unsuitable for in vitro, animal, and human studies due to their cellular and systemic effects. ETs are pyrogenic and cause endotoxemia (fever, hypotension, and septic shock) in humans [6]. At the cellular level, ETs can also stimulate angiogenesis [7]. Removing ETs from potential therapeutics is important to provide a consistent baseline for research and for preclinical studies, especially in order to study angiogenesis inhibition, and is strictly controlled for FDA approval of human trial use. ETs bind to basic and hydrophobic domains in proteins [8], allowing them to bind nonspecifically to any proteins with domains containing high amounts of basic and hydrophobic peptides such as histidine and tryptophan, respectively [8], as well as arginine and lysine, which are both basic and hydrophobic. This nonspecific binding makes removing ETs from protein solutions very difficult. The method of extraction must be customized to the individual protein, taking into account the properties of the molecules in solution [9]. A non-ionic detergent, which is amphipathic by nature, can be used to extract ETs via phase separation [10]. The detergent can form micelles around the hydrophobic ETs, pulling them off the proteins. The detergent micelles are formed in the aqueous media beneath the cloud point (when the detergent is soluble in aqueous solutions), and then warmed to 37°C (above the cloud point) and centrifuged to allow the detergent to separate from the 3 aqueous phase, which is then removed from the aqueous sample. ETs can also be removed using affinity columns, when the column is used beneath the cloud point, allowing the micelles to form [9]. A non-ionic detergent is needed, so as not to disturb the binding of AT to heparin, which is a tight, ionic bond [11]. This allows both native AT and recombinant AT in which the heparin binding domain is retained (such as latent forms used in this study) to be purified using a heparin affinity column [12], which can also be exploited for endotoxin removal. I will describe a simple method for removing endotoxin from samples of latent antithrombin expressed from insect cells. Two protocols will be studied: a combination of detergent phase extraction and affinity column extraction, versus only the affinity column extraction. Once ETs are removed, the AT samples without ET can be safely used for both preclinical studies and for human trials for various therapeutic uses. 4 METHODS A. Antithrombin Samples and Experimental Preparations All latent AT samples were expressed at 37°C in baculovirus-infected insect cell cultures (Schnieder 2, Drosophilia melanogaster, Invitrogen Corporation, Carlsbad, California) [13]. AT synthesis was induced using copper sulfate and the cultures were incubated to allow time for sufficient AT production. The cell media was centrifuged and the supernatant was removed. The supernatant was then loaded and purified on a heparin affinity column [11] at 4°C, collecting the eluted fractions [12]. The fractions in which the recombinant AT was found, using SDS-PAGE as described below and verification with factor-Xa inhibition assays (discussed later in this section), were pooled and frozen at -80°C in 1 mg aliquots in 1xPNE. Additional thrombin-active AT from reconstituted 8.86 mg/mL THROMBATE (Talecris Biotheraputics, Triangle Park, North Carolina, 2005) was also used for testing of column protocols and comparison in thrombin and factor Xa-complexing bioactivity assays. Endotoxin-free (non-pyrogenic) materials must be used in all experiments. Experiments used certified endotoxin-free or non-pyrogenic pipette tips (Cape Cod Inc, East Falmouth, Massachusetts), microcentrifuge tubes (Fisher Scientific, Hampton, New Hampshire), non-pyrogenic syringes and needles (Becton, Dickinson, and Company, Franklin Lanes, New Jersey), and 96-well Falcon® plates (Corning Inc., Corning, New York). All buffers were made in surgical irrigation water, which is sterile, non-pyrogenic water used in surgical practices and approved by the FDA as pyrogen-, including ET, free (Baxter, Deerfield, Illinois). Buffers were made in pyrogen-free containers, using 5 solidware baked at 250°C to fully deactivate any ET present. ET-free 1xPNE (10mM NaCl, 0.1 mM EDTA, and 100 mM phosphate) was the buffer used as a base for the experiments. B. Cyclic Phase Extraction A schematic of this method of phase separation can be seen in figure 2. Triton X114 (Pierce, Thermo Fisher Scientific, Waltham, Massachusetts), a non-ionic detergent with a cloud point of 22°C [14], was added to AT to 1% by volume at 4°C [9]. The sample was then homogenized at 4°C by vortexing the sample for 10 seconds every 5 minutes. It was then warmed to 37°C and incubated for 10 minutes, followed by centrifuging at 16000xG for 1 minute. This creates a sample with two distinct phase layers. The top phase is the aqueous layer that contains the AT, while the bottom phase is the detergent phase that holds the ET [9]. The top phase was removed to a different Fig. 2: Scheme for endotoxin removal [9]. 6 microcentrifuge tube by pipetting, the recovered fluid volume was measured, and TX114 is again added to 1% by volume. The phase separation process is repeated two to four additional times. C. Affinity Column Chromatography A 1mL heparin affinity column (HiTrap, GE Healthcare, Little Chalfont, UK) was placed in line on an automated column chromatograph (Bio-Rad Laboratories, Hercules, California) with an automated fraction collector (Bio-Rad Laboratories, Hercules, California) and chart recorder (Bio-Rad Laboratories, Hercules, California), which records UV absorbance and conductivity. Additionally, the chart recorder denotes the fractions collected. The chromatography set-up was plumbed using non-pyrogenic IV tubing and buffers made in ET-free water in pyrogen-free bottles. All chromatography was done at 4°C [11], below TX114’s cloud point. Three buffers were used in the affinity chromatography: 1xPNE, 1xPNE + 2.5 M NaCl, and 1xPNE + 0.1% TX114. The sample of AT (pre-purified from the Schneider cell media and stored in 1xPNE) was thawed, an aliquot of 60 μL as removed for ET testing, and the sample was loaded in a method based on volume. The elution was carried out on a 1 mL heparin-sepharose affinity column (GE Healthcare, Little Chalfont, UK). Samples larger than 1 mL were diluted in 1xPNE to 12 mL, and then loaded through the column at 0.5 mL/min, with serial dilutions of 1xPNE when the remaining sample volume was under 2 mL, until a total of 15 mL had been loaded onto the column. Samples smaller than 1 mL were injected directly into the column with a non-pyrogenic syringe and allowed to bind to the column for 10 minutes before beginning the program 7 to clean and elute the sample. The program began with a baseline wash of 1xPNE for 10 mL at 1mL/min, followed by a 5 mL TX114 wash, also at 1mL/min, to remove ET. The TX114 was then washed from the column with 1xPNE for 30 ml at 1mL/min. The elution then occurs using a salt gradient from 0.1 M to 2.5 M NaCl over 25 ml at 1mL, followed by 10 m of 1xPNE + 2.5M NaCl at 1 mL/min to fully dislodge any remaining AT. The program finishes with a variable length re-equilibration of 1xPNE. An alternative method of salt elution was used in THROMBATE elution, where the sample was loaded and washed with TX114 in the same method as above, but the gradient is split into the slow gradient used from 0.1-1 M NaCl over 25 mL to elongate the peaks of latent antithrombin. The salt content is held at 1 M for 5 minutes, and then the remaining AT is Fig. 3: Schematic of affinity column elution protocols. Un-interrupted line and small dashed line are the two methods of salt elution, and the long dash line represents where the TX114 is used on the column. eluted with a sharp gradient from 1-2.5 M NaCl in 5 mL. All of the program steps were run at a rate of 0.5 mL/min. These two protocols can be visualized in figure 3. Fractions of 5 mL were collected for 75 mL, followed by 1.5 mL fractions for 45 mL, and finally, 8 2-5 5mL fractions were collected. The 5 mL fractions were collected in non-sterile tubes and then discarded, and the 1.5 mL fractions were collected in 2 mL pyrogen-free microcentrifuge tubes and stored at -4°C. D. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was used to determine the location of the elution peak of the AT from samples emerging at different times from the affinity chromatography column. The chart readout from the chromatograph was examined for the small peak in the UV absorbance. The fractions surrounding and including the peak were sub-sampled to remove 20 μl, using pyrogenfree pipette tips. Then, 10 μl of 3xSDS+BME (3% sodium dodecyl sulfate, 30% glycerol, 188 mM Tris-HCl, 0.01% bromphenol blue, and 15% beta mercaptoethanol in de-ionized water) is added to each of the 20 μl sub-samples. The sub-samples are then vortexed and boiled for 3 minutes. After centrifuging the tubes, 15 μl of each sub-sample were then loaded into the lanes of a 10% reducing SDS-PAGE 15 lane gel. The gel was run at 45 mA per gel until the dye was 1 cm above the bottom of the gel using a Bio-Rad miniPROTEAN tetra cell apparatus (Bio-Rad Laboratories, Hercules, California) [9]. The gels were then stained in CBB dye (0.25% Coomassie Brilliant Blue, 40% methanol and 10% acetic acid in de-ionized water) for 1 hour and de-stained using 40% methanol and 10% acetic acid (by volume) in de-ionized water for between 1 hour and 1 day. The gels were then photographed individually using a visible light transilluminator from below (Bio-Rad Laboratories, Hercules, California), which sent photographs of the gel to a PC (Microsoft, USA). Using the photograph of the gel and a standard for lane protein 9 content, the amount of protein in each lane was estimated by eyeballing and visual comparison to a gel standard of known concentrations of latent AT, which provided an approximation for the protein amount in the peak through visual comparison. E. Protein Quantification Protein amounts in the peak were also quantified using a Bradford protein assay with bovine serum albumin (BSA, Thermo Fisher Scientific, Waltham, Massachusetts) standards. A standard curve of BSA ranging from 400 μg/ml to 84 μg/ml, using a 67% dilution scheme in triplicate was constructed in the first 3 columns of a 96-well microplate. Up to 9 samples can be assayed on each plate. Each sample was diluted in a two-fold scheme down the column of the microplate. A 20% solution Coomassie Brilliant Blue (Bio-Rad Laboratories, Hercules, California) in 1xPNE was then added to each well and the plate was read at 405 nm using a microplate spectrophotometer (Biotek Instruments, Winooski, Vermont) after 12 minutes of incubation and converted to Excel™ data in Delta Soft (DeltaSoft Inc, USA). The optical absorbances were loaded into ExcelTH (Microsoft, USA), where the triplicate BSA standards were averaged and the standard deviation and standard error of the mean were found to verify that the three standards were similar enough to offer a standard curve with a standard error of the mean of less than 10% and an r2 value > 0.98. Calculations were done using built-in Excel™ functions. The absorbance of the standards was plotted against their known concentrations and linear regression was used to find the concentration of the protein in the samples. 10 F. Endotoxin Quantification Limulus Ameobocyte Lysate (LAL) assays (Pierce, Thermo Fisher Scientific, Waltham, Massachusetts) were used to quantify ET levels. Content was measured using endotoxin units (EU). A standard curve of 2 blanks (ET-free water from Pierce kit), 0.1 EU/ml, 0.25 EU/ml, 0.5 EU/ml, and 1 EU/ml was constructed. This assay is sensitive from 0.1 EU/ml to 1 EU/ml, where the concentration follows a linear slope and therefore linear regression can be used as described above to find the unknown concentration of a sample. A 50 μl aliquot of each sample was placed in each well of a 96-well microplate, and the lysate and chromogenic substrate were added at 37°C as specified by Pierce. The microplate was then read at 405 nm. The two blank samples are averaged, and this value is subtracted from all other samples, to account for background activity. The ET content was determined using linear regression as described above. If the content of sample was calculated to be higher than the standard curve, or if the absorbance was saturated, the sample was diluted 5 to 20 times and assayed again. In order to verify the activity of the AT collected from the column, a factor Xa (fXa)-complexing assays were used. By testing the fractions from the affinity column, we can confirm where the latent AT is, as latent AT does not react with thrombin or fXa, because the reactive peptide loop is inaccessible. In fXa-complexing assays, the chromogenic substrate S2765 (Chromogenix S-2765, Diapharma), which turns from clear to yellow in the presence of fXa, was used and the assay was carried out on a non-sterile, medium binding 96-well plate (Costar) [10]. A standard curve of 1108 ng/mL THROMBATE (Talecris Biotheraputics) was contrasted in triplicate. Subsamples of each fraction were diluted to 25% of their original volume, so that the salt content was 11 under 0.25 M to prevent inhibition of the assay. The fractions and the standards were diluted in 2-fold down the columns of the microplate and a double volume of 6mM fXa, 60 nM unfractionated heparin in PP (1xPNE + 0.1% PEG 6000). The microplate was incubated for 15 minutes, and the fXa-AT binding was halted and chromogenically assayed using 100μL S2765 (Diapharma) and 5 mM polybrene in PP. The plate was immediately read at 405nm (Biotek Instruments, Winooski, Vermont). As with the Bradford protein assay and LAL, a standard curve was made and linear regression was used to find the concentration of active AT in samples assayed in Excel™. Inhibition activity is also compared to the blank samples of 20 μl PP to verify a higher level of inhibition in active forms as well as verify a lack of inhibition in latent samples. This process was used to verify, in tandem with the SDS-PAGE, exactly which fractions collected from the affinity chromatography contained active AT (high inhibition) and which contained the latent AT (low to no inhibition) of interest. 12 RESULTS A. TX114 Phase Separation with Affinity Column Wash The TX114 phase separations used either 3 or 5 cycles. The number of cycles used did not alter the percent of aqueous volume recovery (Table 1). The phase separation took approximately 1 hour per cycle. The endotoxin content as determined from the LAL assay was reduced by 99.5% (Fig. 5), from 6.05 EU/mL in a sample of Table 1: Volume reduction of mutant AT variants 35 and 66, by number of cycles. Note that the total volume reduction is approximately 57% in both cases. Table 2: numerical and reduction in ET values, as illustrated in fig 6. 13 Fig. 4: Decreases in ET concentration by method of removal. From left to right: initial ET content, phase extraction only, phase extraction + affinity column, and affinity column only. For m66, decrease in from initial to phase extraction, and for m35, data is from reduction from phase + affinity column and affinity column only. m66 AT to 0.03 EU/mL (Fig. 4) using only the phase extraction with 3 cycles, a similar finding to Aida [8]. While this is outside the standard curve, the value is still below 0.1 EU/ml, a sufficient reduction (98.4%). After the TX114 extractions, the recovered volume for both 3 and 5 cycles was approximately 57%. After detergent phase extraction, the AT was eluted with NaCl on the heparin affinity column, using the protocol illustrated in fig 3. A sample chart recording, showing where the peaks eluted and how the protocol translates to actual elution is seen in figure 5. The UV absorbance at 280 nm is seen in blue, while the solution conductivity is in red. The first peak in the UV trace is AT that did not bind to heparin (a). The second peak is TX114 (b), and the third peak is the latent AT eluting (c). 14 The fractions surrounding peak c were run using SDS-PAGE, as described above. Out of these lanes on the gel, the lane, and corresponding fraction, with the highest amount of protein was recorded and chosen for the LAL assay (Fig. 6). Fig. 5: Conductivity (red) and UV absorbance (blue) trace from a heparin affinity column of m35. a: initial loss of non heparin-binding AT, b: TX114 wash on the heparin affinity column, c: elution of the AT with during the salt gradient. Fig. 6: SDS-PAGE on eluted fractions from an affinity column of m66 that was preextracted with TX114, with highest protein content fraction circled. 15 Based on the LAL assays of the fractions of columns 8 and 9 (m35) with the highest amount of AT, a 99.9% reduction in ET content was achieved through the combined protocol. The SDS-PAGE visual estimate also gave an estimate of protein amount that was eluted from the content, allowing for a rough estimate of percent yield. Based on these yields, we can estimate the efficacy in terms of percent of remaining mutant AT or native AT, depending on the protocol, and which methods resulted in the highest yield of AT after the protocol is complete. The combination method, with both TX114 extraction and affinity chromatography gave an average percent yield of 47.3% (Table 3). Table 3: percent yield from columns. 16 B. TX114 Wash on Heparin Affinity Column Only When only the affinity column was used to remove ET from AT in the absence of detergent, overall ET levels were still decreased by 96.3% (Fig. 5), for a total endotoxin reduction from 20.45 EU/mL in the sample of m35 AT to 0.75 EU/mL (Fig. 6). C. Activity assays S2765 assays of the latent AT samples showed an average absorbance change of 17.7 AU/min (n=7, standard deviation of 6). Based on the absorbance change of the blank sample (PP, which has no inhibition of fXa), 19.3±4 is within the range of the standards (n=12, 19.3±0.85). It is also more than 4 standard deviations away from the average absorbance rate of the active AT (2.6±0.6, n=3). Assuming the S2765 values follow a standard distribution, 99.8% of the absorbance rates of active AT would fall in the range of 0.8-4.4. Since the absorbance rate of the latent AT is outside of this range, the AT samples of interest are indeed latent (inactive to fXa). 17 DISCUSSION The aim of this study was to develop a simple method of removing endotoxin (ET) from antithrombin (AT). The study shows that the total ET was reduced effectively in AT samples by both the combination of phase extraction with TX114 plus affinity column elution and by affinity column only. While the affinity column only reduces the ET content by 96.3% rather than the 99.9% seen in the combined phase separation and affinity column protocol (Fig. 5), the affinity column alone still reduced the content to less than 1 EU/mL. The ET reduction in both protocols is consistent with the Lui paper, which showed 98-99% reduction in ET content with TX114 phase separation and a 95-99% reduction with affinity chromatography [8]. Overall, this study shows that both phase extraction and affinity chromatography with a TX114 wash are viable options for ET removal from AT. This equates to only 0.9% more ET in the affinity column only method over the joint method. The affinity column method takes approximately 1 hour and 40 minutes, while the joint protocol takes 4 hours and 40 minutes for a 3cycle extraction and 6 hours and 40 minutes for a 5-cycle extraction. Based on the ET reduction in the affinity column only method, as well this protocol being 2-4 hours shorter makes the affinity column-only protocol the most effective method and therefore will be the one used for future AT samples. Additionally, the affinity chromatography can be scaled onto larger heparin affinity columns to allow for larger amounts of concentration. The affinity column also is automated, requiring only the initial programming and loading of the sample. The automation and ability to scale the process makes the affinity column only protocol much more appealing to pharmaceutical companies. 18 There are possible repercussions to using the affinity chromatography technique. One is the small yield, ranging from 12-50%. This makes culture-derived latent AT a less desirable to industry, as it reduces profit margins. This also holds up research by producing smaller quantities with more reagents, and therefore more cost. Another possibility is the deactivation of the latent AT due to exposure to the TX114. However, chromogenic studies of the TX114-cleaned active AT confirm inhibition of factor Xa (a sign that it is indeed still the active form), and the same studies reveal little to no inhibition of factor Xa by latent AT. Another possibility with the use of TX114 is the possibility of remaining detergent affecting assays. This could be counteracted by cleaning with gel filtration, either on a chromatographic column filled with gel absorbance or Bio-beads, a type of gel that absorbs the detergent, as was used in the Aida paper [9]. With the development of an AT-specific endotoxin removal technique, latent antithrombins with specific mutations can be grown in culture and then purified for use as pharmaceuticals. Previously, latent AT has been shown to inhibit angiogenesis [4], but has not been developed due to contamination with ET from cell culture. This technique could be used to develop novel medications to treat diabetic retinopathy by preventing angiogenesis. 19 ACKNOWLEDGEMENTS S.A.L. extends great thanks to her PI, Dr. Susan Bock, for her guidance in conducting this research, and to her mentors Ms. Heather Palmer and Dr. Dave Grainger for their knowledge and guidance in writing this thesis. Additional thanks are given to the Ambati Lab at the Moran Eye Institute, the rest of the Bock Laboratory, the Undergraduate Research Opportunities Program, the University of Utah Honors College, the University of Utah Department of Bioengineering, and the Merrill Engineering Fellowship. 20 REFERENCES [1] J.E. Shaw, R.A. Sicree, P.Z.Zimmet, “Global estimates of the prevalence of diabetes for 2010 and 2030,” Diabetes Research and Clinical Practice. ], vol. 87, pp. 4-14, 2010. [2] P. S. Sliva, MD, J. D, Cavallerano, OD, PhD, J. K. Sun, MD, MPH, L. M. Aiello, MD, L. Paul Aiello, MD, PhD, “Effect of systemic medications on onset and progression of diabetic retinopathy,”. Nature Review of Endocrinology,. vol. 6, no. 9, pp. 494-508, 2010. [3] T. T. Hansel, H. Kropshofer, T. Singer, J. A. Mitchell, A. T.J. George, “The safety and side effects of monoclonal antibodies,” Nature Reviews Drug Discovery, vol. 9, 2010.. [4] H. Larsson, T. Sjoblom, J. Dixelius, A. Ostman, K. Ylinenjarvi, I. Bjork, and L. Claesson-Welsh, “Antiangiogenic Effects of Latent Antithrombin through Perturbed Cell-Matrix Interactions and Apoptosis of Endothelial Cells,” Journal of Cancer Research,. . vol. 60, pp. 6723-6729, 2000. [5] C. R. H. Raetz, “Biochemistry of Endotoxins,” Annual Review of Biochemistry, vol. 59, pp.129-170, 1990. S. J. H. van Deventer, H. R. Büller, J. W. ten Cate, L. A. Arden, C. E.Hack, and A. Sturk , “Experimental Endotoxemia in Humans: Analysis of Cytokine Release and Coagulation, Fibrinolytic, and Complement Pathway,” American Society of Hematology, vol. 26, no. 12, pp. 2520-2526, 1990. [6] I. Mattsby-Baltzer, A. Jakobsson, J. Sӧrbo, and K. Norrby, “Endotoxin is angiogenic,” International Journal of Experimental Pathology, vol. 75, no. 3, pp. 191-196. 21 [7] Y. Zhu, B. Ho, J. Ling Ding, “Sequence and structural diversity in endotoxinbinding dodecapeptides,” Biochimica et Biophysica Act – Biomembranes. .vol. 1611, no. 1-2, pp. 234-242. [8] S. Liu, R. Tobias, S. McClure, G. Styba, Q. Shi, and G. Jackowski, “Removal of Endotoxin from Recombinant Protein Preparations,” Journal of Clinical Biochemistry,, vol. 30, no. 6, pp. 455-463, 1997. [9] Y. Aida and M. J Pabst, “Removal of Endotoxin for Protein Solutions by Phase Separation using Triton X114,” Journal of Immunological Methods,, vol. 132, pp. 191-195, 1990. [10] S. T. Olsen, I. Bjork, R. Sheffert, P. A. Craig, J. D. Shore, J. Choay, “Role of the Antithrombin-binding Pentasaccharide in Heparin Acceleration of AntithrombinProteinase Reactions,” Journal of Biochemistry,, vol. 267, no 18, pp. 12528-12538, 1992. [11] R. D Rosenberg and P. S Damus, “The Purification and Mechanism of Action of Human Antithrombin-Heparin Cofactor,” Journal of Biological Chemistry, vol. 248, no. 18, pp. 6490-6505, 1978. [12] R. G. C. dela Cruz, M. A. Jairapuri, S. C. Bock, “Disruption of a Tight Cluster Surrounding Tyrosine 131 in the Native Conformation of Antithrombin III Activates it for Factor Xa Inhibition,” Journal of Biological Chemistry, vol. 281, no. 42, pp. 31668-31676, 2006. [13] Sigma-Aldrich, Product Information. Triton X-114. [14] M Yamasaki et al. Nature 000, 1-4 (2008) doi:10.1038/nature07394