Biochemical characterization and stability of immune globulin intravenous 10% liquid (Panzyga®).

Panzyga® is a new glycine-formulated immune globulin intravenous 10% liquid for the treatment of patients suffering from immunodeficiencies and autoimmune diseases. Panzyga® is a high purity, native and functional IgG product with an IgG subclass distribution equivalent to normal plasma. The levels of hemagglutinins and accompanying plasma proteins (including IgA and IgM) are low. Potential procoagulant activity is not detectable. Functional activity of the IgG was demonstrated by opsonophagocytosis and receptor binding assays. Dynamic light scattering measurements and fluorescent dye binding were used to characterize the integrity of the IgG molecule. Panzyga® is stable under refrigerated storage for at least two years regarding all assessed physicochemical and functional parameters; it can also be stored at room temperature for at least twelve months within its total shelf-life.

Elevated Endotoxin Levels in Human Intravenous Immunoglobulin Concentrates Caused by (1->3)-{beta}-D-Glucans.

Endotoxins have been measured routinely in the final product and during the production process to produce non-pyrogenic parenterals. Limulus-amoebocyte-lysate-reactive material was found in in-process samples and final product of one of Octapharma's intravenous immunoglobulin (IVIG) preparations. Limulus-amoebocyte-lysate (LAL) is activated by bacterial endotoxins and by (1→3)-ß-D-glucans. The contribution of both compounds on the LAL-related signal was determined by three different approaches: (1) using a test specific for (1→3)-ß-D-glucans, (2) by addition of ß-glucan blocker, and (3) by the use of a recombinant endotoxin assay. It was shown that none of our IVIG concentrates contained elevated endotoxin contents but that the higher LAL reaction could be ascribed to (1→3)-ß-D-glucans extracted from cellulose filter pads. The use of an endotoxin test kit highly sensitive for (1→3)-ß-D-glucans might lead to false-positive results. (1→3)-ß-D-glucans spike solutions did not evoke an increase of temperature in rabbits, suggesting that a pyrogenic reaction is not expected in patients. LAY ABSTRACT: Endotoxins have been measured routinely in the final product and during the production process to produce non-pyrogenic parenterals. Limulus-amoebocyte-lysate-reactive material was found in in-process samples and final product of one of Octapharma's intravenous immunoglobulin (IVIG) preparations. Limulus-amoebocyte-lysate (LAL) is activated by bacterial endotoxins and by (1→3)-ß-D-glucans. The contribution of both compounds on the LAL-related signal was determined by three different approaches: (1) using a test specific for (1→3)-ß-D-glucans, (2) by addition of ß-glucan blocker, and (3) by the use of a recombinant endotoxin assay. It has been shown that none of our IVIG concentrates contained elevated endotoxin contents but that the higher LAL reaction could be ascribed to (1→3)-ß-D-glucans extracted from cellulose filter pads. The use of an endotoxin test kit highly sensitive for (1→3)-ß-D-glucans might lead to false-positive results. (1→3)-ß-D-glucans spike solutions did not evoke an increase of temperature in rabbits suggesting that a pyrogenic reaction is not expected in patients.

Isolation of esterified fatty acids bound to serum albumin purified from human plasma and characterised by MALDI mass spectrometry.

Human serum albumin (HSA) is the most abundant protein in plasma. It is known to transport drugs as well as endogenous ligands, like free fatty acids (FFA). A mass spectrometry based method was applied to analyze the albumin bound lipid ligands. HSA was isolated from a human plasma pool by cold ethanol fractionation and ion exchange chromatography. HSA was defatted using a solvent extraction method to release the copurified lipids bound to the protein. The extracts were then analyzed by matrix-assisted laser desorption ionisation (MALDI) mass spectrometry (MS). Using this method, phospholipids and acylglycerols were detected. The phospholipids were identified to be lyso-phosphatidylcholine (lyso-PC) with distribution of different fatty acids (palmitic, stearic, oleic, and linoleic acids). An abundant species in the HSA lipid extract was found to be a diacylglycerol, composed of two linoleic and/or oleic acid chains. The identified motifs reflect structures that are known to be present in plasma. The binding of lysophospholipids has already been described but it is the first ever-reported evidence of native diacylglycerol ligands bound to HSA. Besides the native ligands from plasma a triacylglycerol was detected that has been added during the albumin preparation steps.

Purification of intravenous immunoglobulin G from human plasma--aspects of yield and virus safety.

Plasma-derived intravenous immunoglobulin (IVIG) preparations have been successfully applied for the prophylactic prevention of infectious diseases in immunodeficient patients. In addition to its replacement therapy of primary and secondary antibody deficiencies, IVIG has found increased use in autoimmune and inflammatory diseases. IVIG has become the major plasma product on the global blood product market. The world wide consumption nearly tripled between 1992 and 2003, from 19.4 to 52.6 tons. Classical manufacturing processes of IVIG, but also new strategies for purification are discussed with respect to practicability and yield. Ethanol fractionation is still the basis for most IVIG processes, although isolation and purification of immunoglobulin G (IgG) by chromatography has gained ground. The efficiency of virus inactivation methods and virus removal techniques in terms of logarithmic reduction factors are analyzed, but also the IgG losses are taken into consideration. Some of these methods also have the ability to separate prions. High pathogen safety and high yields have become the dominant goals of the plasma fractionation industry.

Thermodynamic stability and formation of aggregates of human immunoglobulin G characterised by differential scanning calorimetry and dynamic light scattering.

The final process step of polyclonal human immunoglobulin G is formulation with agents such as sugars, polyols, amino acid and salts. Often the most stable formulations were empirically identified. Physicochemical methods, such as differential scanning calorimetry and dynamic light scattering, provide a deeper insight on the biophysical properties of such a protein solution. The combination of these methods proved to be sensitive enough to detect fine differences in the properties relevant for the development of stable protein solutions. The influence of additives, such as maltose and glycine in combination with water or low concentrations of salts, on human immunoglobulin preparations was analysed. Differential scanning calorimetry illustrated that 0.2 M glycine had better stabilising effects compared to 10% maltose. Dynamic light scattering and differential scanning calorimetry revealed that solutions preventing aggregation were not optimal in terms of thermodynamic stability. Aggregation was minimised with increasing ionic strength, shown by dynamic light scattering, whereas thermodynamic stability for heat sensitive parts of human immunoglobulin G, analysed with differential scanning calorimetry, was decreased.

Effect of oriented or random PEGylation on bioactivity of a factor VIII inhibitor blocking peptide.

Peptides as therapeutic substances are efficient agents in the treatment of several diseases. However, they often have to be chemically modified in order to be suited as therapeutic agent. Conjugation to large carrier molecules is often required. A critical step is the identification of available sites for chemical reaction, without influencing bioactivity. Peptide 238/S1 with the sequence NH(2)-PYWKWQYKYD-COOH previously selected from a combinatorial decapeptide library, has the ability to block inhibitory antibodies against blood clotting factor VIII (FVIII) and therefore, it constitutes a lead for developing a drug to treat patients suffering from development of such antibodies. The aims of this study were (i) to identify sites of the peptide, which are suited for modification without losing bioactivity and (ii) to find out the influence of molecular size of polyethylene glycol (PEG) for bioactivity of the peptide. The contribution of each amino acid residue to biological functionality was investigated by mutational analysis. This method confirmed that the N-terminus is crucial for activity, whereas both lysine residues could be exchanged by other L-amino acids. Using mutational analysis it was possible to identify peptides with higher reactivity compared to the wild type 238/S1. PEGylation experiments demonstrated that conjugation of the peptide to PEG 20,000 resulted in a loss of reactivity, while PEG 5,000 could maintain the bioactivity when conjugated in a site directed manner. The peptide lost its neutralization properties when PEG was coupled to the N-terminus, again indicating that this part of the peptide is important for functionality.

Mapping of FVIII inhibitor epitopes using cellulose-bound synthetic peptide arrays.

Epitope mapping using antibodies against factor VIII (FVIII) has been performed using blotting techniques with truncated and/or digested FVIII molecules. Here, we focused on the precise mapping of affinity purified IgG from patients with an immune response against blood clotting FVIII using synthetic peptide arrays on cellulose membranes comprising the entire sequence of FVIII. The aim was to elucidate the epitope profile from different inhibitors and possibly detect new epitopes, which have not been described before. The epitope patterns from five patients showed reactivity with all domains in the FVIII molecule, but were different between various patients. These results included epitopes usually buried within the folded protein. However, in competition assays using FVIII as competitive agent in a mixture with inhibitor IgG, the most immunogenic regions were located in the FVIII light chain. Our results show that the C1 domain was the region with highest immunogenicity in all patients. Here, we demonstrate that the SPOT method is very well suited for the precise location of epitopes in the core of the protein, which usually cannot be detected by other methods.

Combinatorial peptides directed to inhibitory antibodies against human blood clotting factor VIII.

The development of antibodies against blood clotting factor VIII is a major complication affecting 20-30% of hemophilia A patients receiving replacement with FVIII concentrates. This study investigated generating peptides acting as broadly neutralizing agents to block factor VIII antibodies. These peptides were selected from dual positional scanning decapeptide libraries on cellulose membranes. From this library comprising 6.8 x 10(12) peptides we selected 468 peptides for further screening rounds. Finally we identified two decapeptides with the ability to block 8 out of 10 inhibitory antibodies from sera of patients with FVIII inhibitors demonstrated by competition assays. Sequence alignment of the peptides showed similarity with several domains in the FVIII molecule demonstrating the mimotope nature of the selected peptides. Our results show the efficiency of the combinatorial library approach and show the potential of combinatorial peptides to compete out polyclonal inhibitor IgG from a broad range of patients' sera. Combinatorial peptides could be novel and highly effective drug candidates for alternative treatment in patients with factor VIII inhibitors.

Characterization of cysteinylation of pharmaceutical-grade human serum albumin by electrospray ionization mass spectrometry and low-energy collision-induced dissociation tandem mass spectrometry.

Three samples of albumin derived from human plasma (pharmaceutical grade, HSA) obtained from different commercial sources were investigated for their micro-heterogeneities by means of electrospray ionization (ESI) ion trap mass spectrometry (ITMS). The study covered MS analyses of the intact proteins as well as on the tryptic peptide level. The intact protein samples were analyzed without any separation step except for simple desalting. With these samples we observed in the positive ion ESI mass spectra that the multiply charged ion signals of HSA consisted of a number of fully or partly resolved peaks with relative intensities depending on the analyzed sample. The non-modified form of HSA was detected in the three HSA preparations at m/z values of 66448 +/- 3.6, 66450 +/- 0.6 and 66451 +/- 3.2 ([MH]+), respectively. The value calculated from the amino acid sequence was 66439. The second compound present with high intensity (in two cases the base peak in the deconvoluted mass spectrum) is interpreted as a modified HSA, and the molecular mass increase in relation to the unmodified HAS was between 116 and 118 Da (m/z of 66 564, 66 567 and 66 569), suggesting the presence of a covalently bound cysteine residue. A further peak in the deconvoluted ESI spectra was found in all three samples with rather low signal/noise ratio at m/z 66 619, 66 621 and 66 613, respectively, which may correspond to a non-enzymatic glycation described in the literature. The verification of the proposed covalent HSA modifications was subsequently done on the peptide level using high-performance liquid chromatography (HPLC)/ESI-MS and HPLC/ESI-MS/MS including low-energy collision-induced dissociation (CID). Prior to the tryptic digestion, the HSA samples were alkylated without a prior reduction step. Following this procedure we detected peptides of the sequence T21-41 that included the Cys-34 residue in both forms: cysteinylated (m/z 639.15 [M+4H]4+) as well as vinylpyridine-alkylated (m/z 635.69 [M+4H]4+, which means in its previously native free SH form). In the next step on-line LC/ESI low-energy CID MS/MS experiments were performed to verify these two proposed structures. By means of MS/MS analysis of the mentioned ions the described modification (cysteinylation) at the Cys-34 residue could be proven. This abundant modification of HSA in pharmaceutical-grade preparations could be unambiguously identified as cysteinylation at the Cys-34 residue. On the other hand, the proposed non-enzymatic glycation was not detectable on the peptide level in the on-line HPLC/ESI-MS mode, maybe due to the low concentration in the three samples under investigation.

Exact molecular mass determination of various forms of native and de-N-glycosylated human plasma-derived antithrombin by means of electrospray ionization ion trap mass spectrometry.

Human plasma-derived antithrombin was characterized in both the native and de-N-glycosylated forms (without separation of isoforms) by means of electrospray ionization ion trap mass spectrometry (ESI-ITMS). In order to determine the limits of the instrument set-up, the molecular mass precision and accuracy of the ESI-ITMS analysis was evaluated with the standard protein enolase and some instrumental data acquisition parameters were optimized. Mass precision was determined as a function of the number of averaged mass spectra (= scans) and data acquisition time. For this study, 20 and 50 scans were averaged and the data acquisition time was chosen to be between 0.5 and 5 min. It turned out that data acquisition times longer than approximately 2 min show no significant differences of the standard deviation of the determined molecular mass. Furthermore, the ion trap scan rate was varied at constant acquisition time of 2 min and the number of averaged scans was set to 20. At the scan rate of 13,000 u s(-1) a mass precision of +/-1.8 Da and a mass accuracy of +0.026% were determined. On reducing the scan rate to 5500 u s(-1), better agreement with the theoretical molecular mass was obtained, showing a mass accuracy of +0.012% but with a decrease in the mass precision to +/-3.0 Da. Using the optimized scan rate of 13,000 u s(-1) and a data acquisition time of 2 min, the exact molecular mass was determined of the three forms of antithrombin, namely the alpha-form, the beta-form and the natural mixture (present in human plasma) containing both forms. The protonated molecular masses were found to be 57,854 and 55,664 Da for the affinity chromatography-isolated alpha-and beta-form, respectively. The mass difference of 2190 Da is attributed to the known difference in carbohydrate content at one specific site. The protonated molecular mass of the dominating species of the natural mixture in human plasma was shown to be 57,850 Da, corresponding to the alpha-form, the major component in native plasma. In this mixture the beta-form was also detected, exhibiting a protonated molecular mass of 55,655 Da, but showing a much lower abundance, as expected. To obtain a complete release of the N-glycan residues by means of PNGase F, a denaturation, reduction and alkylation step of the glycoproteins was performed before the enzymatic reaction. After enzymatic removal of all N-glycans, the protonated molecular masses obtained were 49,399, 49,380 and 49,391 Da for the alpha-form, the beta-form and the unseparated natural mixture, respectively. These values are in good agreement (+0.026% for the alpha-form, -0.012% for the beta-form and +0.010% for the unseparated mixture) with the calculated molecular mass based on the SwissProt data. The determined molecular masses after reduction/alkylation and de-N-glycosylation of the alpha-and beta-forms are almost equal, indicating that no major differences exist between the three preparations on the amino acid level.

Characterization of antithrombin III from human plasma by two-dimensional gel electrophoresis and capillary electrophoretic methods.

The isoforms distribution of the glycoprotein antithrombin III (ATIII) derived from human plasma was investigated by means of isoelectric focusing (IEF) in polyacrylamide gels with immobilized pH gradients (IPG) and two-dimensional gel electrophoresis (2-DE) as well as capillary electrophoretic methods. It turned out that the presence of high concentrations of chaotropics (urea, thiourea) and zwitterionic detergents (3-[(3-cholamidepropyl)dimethylammonio]-1-propanesulfonate (CHAPS)) was decisive for attaining good resolution of the protein isoforms. Resolution by IPG-IEF was obtained with excellent reproducibility and pI differences down to 0.01 pH units could be distinguished. ATIII-alpha and ATIII-beta-fractions preseparated by heparin affinity chromatography showed an analogous but shifted spot pattern consisting each of one major and three minor isoforms. The main isoforms of ATIII-alpha and ATIII-beta exhibit pI values of 5.18 and 5.32, respectively, both values determined in the presence of high concentrations of urea. The pI difference of 0.14 pH units correspond to the effect of two sialic acids absent in ATIII-beta. The formation and occurrence of ATIII dimers and trimers turned out to be dependent on the sample preparation. The results obtained by 2-DE were compared with those of capillary zone electrophoresis (CZE) and capillary IEF (CIEF). Quantitative analysis regarding the CZE separated isoforms of plasma derived ATIII yielded a content of about 70% ATIII-alpha main isoform and about 6.6% of ATIII-beta. The pI values of ATIII determined by CIEF with internal calibration were in fair agreement with the pI values of the main isoforms achieved with 2-DE.

Selective solid-phase isolation of methionine-containing peptides and subsequent matrix-assisted laser desorption/ionisation mass spectrometric detection of methionine- and of methionine-sulfoxide-containing peptides.

Methionine residues and the oxidised forms in proteins are becoming more and more important in view of their biological function. In particular, methionine sulfoxide seems to have a regulatory function. This paper presents a fast strategy for simultaneous determination of methionine- and methionine-sulfoxide-containing peptides, involving application of methionine-specific solid-phase reagent chemistry combined with matrix-assisted laser desorption/ionisation mass spectrometry (MALDI-MS). In the first step, methionine-containing peptides are covalently bound as sulfonium salts to glass beads, whereas methionine-sulfoxide-containing peptides and other methionine-free peptides are not bound and are washed out. The wash solution is used for MALDI-MS analysis to determine the molecular masses of these peptides and to perform, if necessary, seamless post-source decay (PSD) fragment ion analysis. Methionine-sulfoxide-containing peptides can be identified due to the characteristic metastable loss of methanesulfenic acid from the protonated molecules. In the second step, the bound peptides are cleaved from the matrix of the beads by addition of 2-mercaptoethanol at pH 8.5-8.8. The resulting peptides, mainly methionine-containing peptides, are analysed in a straightforward manner by MALDI-MS and seamless PSD. The strategy allows the fast identification of methionine- and methionine-sulfoxide-containing peptides even in complex tryptic digests, as demonstrated here for the glycoprotein antithrombin. These results show that sometimes methionine-containing tryptic peptides are not detected due to steric restrictions (e.g. glycosylation near the methionine residue) on the binding reaction, and that, on the other hand, some methionine-free peptides can be quite strongly bound non-covalently to the matrix of the beads. The latter observation indicates the necessity of seamless PSD fragment ion analysis for unambiguous identification. Furthermore, there are indications that oxidation of some methionine residues occurred to a minor extent during the solid-phase isolation steps.

Preparation of vitamin K-dependent proteins, such as clotting factors II, VII, IX and X and clotting inhibitor protein C.

A review is given of preparative methods for the isolation of the vitamin K-dependent clotting factors II, VII, IX, X and clotting inhibitor protein C, all derived from human plasma. Factor II, activated factor VII and activated protein C are also obtained from recombinant animal cells. The methods for their purification are described. The problem of difference in posttranslational modifications between plasma derived and recombinant protein is discussed with regard to therapeutic proteins.

Molecular mass determination of plasma-derived glycoproteins by ultraviolet matrix-assisted laser desorption/ionization time-of-flight mass spectrometry with internal calibration.

Human plasma-derived antithrombin III (AT-III), factor IX (FIX) and vitronectin (VN) were characterized as native glycoproteins and in their de-N-glycosylated form by means of MALDI mass spectrometry. The average molecular masses of the three complex glycoproteins were determined applying internal calibration with high-mass, well-defined protein calibrants. Internal calibration generated for the 47 kDa yeast protein enolase a mass precision in the continuous and delayed extraction mode of +/-0.12 and +/-0.022%, respectively. The achievable mass accuracy for such a high-mass, unmodified protein was in the range of 0.02% in the continuous mode, which turned out to be better than in the delayed extraction mode. Purification of all (glyco) proteins (even the calibration proteins) by means of ZipTip technology and direct elution with a solvent system containing the appropriate MALDI matrix turned out to be a prerequisite to measure the exact molecular masses with an internal calibration. The average molecular masses of the two different forms of AT-III, namely AT-III(alpha) and AT-III(beta), were shown to be 57.26 and 55.04 kDa, respectively. The 2.22 kDa mass difference is attributed to the known difference in carbohydrate content at one specific site (Asn-135). After exhaustive de-N-glycosylation (by means of PNGase F) of the alpha- and beta-form and subsequent MALDI-MS analysis, average molecular masses of 48.96 and 48.97 kDa, respectively, were obtained. These values are in good agreement (-0.15%) with the calculated molecular mass (49.039 kDa) of the protein part based on SwissProt data. The molecular mass of the heavily post-translational modified glycoprotein FIX was found to be 53.75 kDa with a peak width at 10% peak height of 4.5 kDa, because of the presence of many different posttranslational modifications (N- and O-glycosylation at multiple sites, sulfation, phosphorylation, hydroxylation and numerous gamma-carboxyglutamic acids). MALDI-MS molecular mass determination of the native, size-exclusion chromatography-purified, VN sample revealed that the glycoprotein was present as dimer with molecular mass of 117.74 kDa, which could be corroborated by non-reducing SDS-PAGE. After sample treatment with guanidine hydrochloride and mass spectrometric analysis, a single, new main component was detected. The molecular mass turned out to be 59.45 kDa, representing the monomeric form of VN, known as V75. The determined molecular mass value was shown to be on one hand lower than from SDS-PAGE and on the other higher than the calculated amino acid sequence molecular mass (52 277 Da), pointing to the well-known SDS-PAGE bias and to considerable post-translational modifications. Further treatment of the sample with a reducing agent and subsequent MALDI-MS revealed two new components with molecular masses of 49.85 and 9.41 kDa, corresponding to V65 and V10 subunits of VN. PNGase F digest of the V75 and V65 units and MS analysis, exhibiting a molecular mass reduction of 6.37 kDa in both cases, verified the presence of a considerable amount of N-glycans.

Continuous purification of a clotting factor IX concentrate and continuous regeneration by preparative annular chromatography.

Preparative continuous annular chromatography, a method to separate proteins in a truly continuous manner, was investigated in an industrial environment. Plasma-derived clotting factor IX concentrate was used as model protein. Separation of vitronectin, a common impurity in commercial available factor IX concentrates, from factor IX was studied and compared to conventional packed bed chromatography in batch mode. As sorbent, Toyopearl DEAE 650M was used. Regeneration was performed simultaneously with the purification of factor IX in continuous mode. All required parameters applied for preparative annular chromatography such as feed flow-rate and elution flow-rate were first estimated from experiments on conventional batch columns. Then preparative annular chromatography and conventional packed beds were compared regarding enrichment, purity and productivity. Three different process scenarios, the optimal batch process,the preparative annular chromatography process and the batch process equivalent to the preparative annular chromatography process were investigated. The productivity of the optimal batch process was higher than that of the preparative annular chromatography and batch process equivalent to the preparative annular chromatography process. Therefore the throughput could not be increased by the use of the continuous chromatographic system.

Separation of active and inactive forms of human antithrombin by heparin affinity chromatography.

During the manufacturing of an antithrombin preparation, it is necessary to define all steps that may damage or alter the target molecule, and thus decrease the biological activity of the inhibitor in blood coagulation. Pasteurization, commonly used procedure for viral inactivation of plasma derived antithrombin concentrates, was shown to partially alter the conformation of the active native antithrombin to an inactive latent form. To study intensively the different forms of inactive antithrombin that are formed upon heat treatment, human alpha-antithrombin, human beta-antithrombin and an equimolar mixture of the two isoforms were incubated at 60 degrees C for 15 h in the presence of citrate as stabilizing agent. Using two subsequent heparin affinity chromatography steps, three different inactive fractions were separated. By comparison of the heparin binding capacities, isoelectric points and unfolding characteristics of these inactive forms, the alpha-latent and beta-latent antithrombin isoforms could be identified. It was also shown that additional inactive forms such as proteinase cleaved and/or oxidized forms of antithrombin are formed during the heat treatment process. In four commercially available antithrombin preparations, all produced by pasteurization, the amount of inactive protein varied between 0.5% and 9.5%.