The development of stable protein formulations: a close look at protein aggregation, deamidation, and oxidation.

The biochemical literature has been surveyed to present an overview of the three most common protein degradation pathways: protein aggregation, deamidation, and oxidation. The mechanisms for each of these degradation routes are discussed with particular attention given to the effect of formulation conditions such as pH, ionic strength, temperature, and buffer composition. Strategies to reduce protein degradation are also discussed. These strategies are based on an understanding of the degradation mechanisms and the effect of changes in the storage conditions and formulation components on the degradation. The effects of each of the degradation routes on pharmaceutically relevant properties such as biological activity, metabolic half-life, and immunogenicity are summarized. Predicting a priori the alteration of pharmaceutical properties caused by the three degradation routes is difficult, and must be determined on a case-by-case basis for each protein. The difficulty in predicting the effect of degradation and analyzing the temperature dependence of reaction rates in proteins results in longer development times for protein formulations than for small molecule formulations. Although the use of accelerated stability to predict protein shelf life is difficult, conditions are discussed whereby the Arrhenius equation can be used to shorten formulation development time.

Electron-microscopic and hydrodynamic characterization of recombinant apolipoprotein (a) and its association with LDL.

A recombinant apo(a) containing 17 kringle 4 domains as well as the kringle 5 and protease domains of apo(a) was characterized by hydrodynamic studies and electron microscopy. Recombinant apo(a) is a monomer in solution with a molecular weight of 325,000 by sedimentation equilibrium and 320,000 by sedimentation and diffusion, and it is a highly asymmetric molecule with a frictional ratio of 2.2. In the electron microscope recombinant apo(a) is visualized as a flexible chain of domains approximately 800 A long. Sedimentation velocity studies also demonstrate that when it is mixed with LDL, recombinant apo(a) reversibly forms an Lp(a)-like complex with a 1:1 stoichiometry; moreover, complex formation is inhibited by 6-amino hexanoic acid. Hydrodynamic modeling and electron microscopy suggest that only a small portion of the r-apo(a) molecule interacts with the LDL and the rest of the chain extends into solution. Preliminary studies indicate that recombinant apo(a) also binds mouse LDL.

Influence of calcium ions on the structure and stability of recombinant human deoxyribonuclease I in the aqueous and lyophilized states.

The effect of calcium ions on the structure and stability of recombinant human DNase I (rhDNase) in the aqueous and solid (lyophilized) states was investigated. Fourier transform infrared (FTIR) spectroscopy was used to examine the overall secondary structure, while chemical stability was monitored in terms of deamidation and soluble aggregate formation at 40 degrees C. The exogenous calcium was removed by EGTA. This process can remove all but approximately one calcium ion per protein molecule. Analysis of the FTIR spectra in the amide III region in either the aqueous or lyophilized state demonstrated that removal of exogenous Ca2+ by EGTA-treatment had little effect on the secondary structure (and lyophilization-induced rearrangement thereof). For the aqueous solution, circular dichroism was used as an independent technique and confirmed that there was no large overall change in the secondary or tertiary structure upon the removal of calcium. The primary degradation route for the aqueous protein was deamidation. For the EGTA-treated protein, there was also severe covalent aggregation, e.g., formation of intermolecular disulfides facilitated by the cleavage of Cys173-Cys209. The aggregates exhibited a markedly different secondary structure compared to the native protein. For instance, the beta-sheet band observed at ca. 1620 cm-1 wavenumber in the amide I second derivative spectra was increased. Enzymatic activity was completely lost upon aggregation, consistent with the cleavage of the aforementioned native disulfide. For the protein lyophilized in the presence of Ca2+, there was no increase in deamidated species during solid-state storage; however, some aggregation was observed. For the lyophilized EGTA-treated protein, aggregation was even more pronounced, and there was some loss in enzymatic activity upon reconstitution. Thus, the removal of calcium ions by EGTA-treatment decreased the stability of rhDNase in both the aqueous and solid states even though no large overall calcium-induced structural changes could be observed by the techniques used in this study.

Effect of mannitol crystallization on the stability and aerosol performance of a spray-dried pharmaceutical protein, recombinant humanized anti-IgE monoclonal antibody.

We have examined the stability and aerosol performance of the pharmaceutical protein recombinant humanized anti-IgE monoclonal antibody (rhuMAbE25) spray dried with mannitol. The aerosol performance was measured by the fine particle fraction (FPF), and stability was assessed by the formation of soluble aggregates. When mannitol was added to the spray-dried rhuMAbE25 formulation, its ability to stabilize the protein leveled off above about 20% (w/w, dry basis). The FPF of the spray-dried formulations was stable during storage for rhuMAbE25 containing 10% and 20% mannitol, but the 30% formulation exhibited a dramatic decrease upon storage at both 5 degreesC and 30 degreesC, due to mannitol crystallization. We tested the addition of sodium phosphate to a 60:40 rhuMAbE25:mannitol (w:w) mixture, which otherwise crystallized upon spray drying and yielded a nonrespirable powder. The presence of sodium phosphate was successful in inhibiting mannitol crystallization upon spray drying and dramatically lowering the rate of solid-state aggregation. However, over long-term storage some crystallization was observed even for the phosphate-containing samples, concomitantly with increased particle size and decreased suitability for aerosol delivery. Therefore, the physical state of mannitol (i.e., amorphous or crystalline) plays a role both in maintaining protein stability and providing suitable aerosol performance when used as an excipient for spray-dried powders. Agents which retard mannitol crystallization, e.g., sodium phosphate, may be useful in extending the utility of mannitol as an excipient in spray-dried protein formulations.

Dose-response evaluation of a genetically engineered foot-and-mouth disease virus polypeptide immunogen in cattle.

Four groups of 9 cattle each were vaccinated with 10, 50, 250, or 1,250 micrograms of foot-and-mouth disease (FMD) virus A12 VP1 fusion protein that was produced in Escherichia coli and emulsified in an oil adjuvant. The groups given the 10 and 50 micrograms of antigen were revaccinated at 15 weeks and were challenge exposed at 30 weeks; 5 of 9 and 7 of 9 cattle, respectively, were protected from FMD virus infection. The remaining 2 groups, vaccinated with 250 or 1,250 micrograms of antigen, were revaccinated at 32 weeks and were challenge exposed at 45 weeks; 8 of 9 and 9 of 9 cattle, respectively, were protected. The results indicated that the biosynthetic polypeptide FMD vaccine was effective using vaccination intervals frequently followed with conventional whole-virus vaccines.

pH-dependent polymerization of a human leukocyte interferon produced by recombinant deoxyribonucleic acid technology.

The pH-dependent polymerization of human leukocyte interferon A produced by recombinant DNA technology was investigated by potentiometric hydrogen ion titration and sedimentation velocity analytical ultracentrifugation. The titration curve is completely reversible from pH 6.2 to 2.5. Analysis of the side-chain carboxyl residues by using the Linderstrøm-Lang equation yields a linear plot from pH 5.2 to 3.5. The resulting pKint and W values are 4.66 +/- 0.06 and 0.036 +/- 0.004, respectively. The titration curve from pH approximately 9.6 to 7 is also reversible. However, the reverse titration from pH 7 to 6 results in a time-dependent variation of pH. Sedimentation velocity analysis yields an increase in the corrected sedimentation coefficient, S20,w, from 2.0 to 4.0 S over a pH range of 2.5-9 at 0.38-0.25 mg/mL total protein concentration. The concentration dependence of the sedimentation coefficient at pH 7.0 over the range of 0.05-6.0 mg/mL results in a positive slope which is typical for self-aggregating systems. This aggregation phenomenon is discussed in terms of its ramifications for interferon therapy.

Characterization of aerosols of human recombinant deoxyribonuclease I (rhDNase) generated by jet nebulizers.

Recombinant human deoxyribonuclease I (rhDNase) is a new therapeutic agent developed to improve clearance of purulent sputum from the human airways. It is delivered by inhalation. Four jet nebulizers, T Up-Draft II (Hudson), Customized Respirgard II (Marquest), Acorn II (Marquest), and Airlife Misty (Baxter), were evaluated in vitro for their ability to deliver aerosols of rhDNase. The aerosols were generated from 2.5-mL aqueous solutions of rhDNase, at concentrations of either 1 or 4 mg/mL. In all experiments, the Pulmo-Aide Compressor (De Vilbiss) was used to supply the air to the nebulizers. Between 20 and 28% of the rhDNase dose initially placed in the nebulizers was delivered to the mouthpiece in the respirable range (1-6 microns). Evaluation of the rhDNase following nebulization in all four devices indicated that there was no loss in enzymatic activity and no increase in aggregation. Circular dichroism spectrophotometry indicated there was no change in either the secondary or the tertiary structure in rhDNase following nebulization. These results show that all four nebulizers are essentially equivalent in their ability to deliver respirable doses of rhDNase in an intact, fully active form. Changing the concentration of the solution in the nebulizer from 4 to 1 mg/mL rhDNase leads to a proportional reduction in the respirable dose delivered to the mouthpiece.

Self-association of human and porcine relaxin as assessed by analytical ultracentrifugation and circular dichroism.

The self-association properties of recombinant DNA derived human relaxin, and porcine relaxin isolated from porcine ovaries, have been studied by sedimentation equilibrium analytical ultracentrifugation and circular dichroism (CD). The human relaxin ultracentrifuge data were adequately defined by a monomer-dimer self-association model with an association constant of approximately 6 x 10(5) M-1, whereas porcine relaxin was essentially monomeric in solution. An approximate 5-fold increase in weight fraction of human relaxin monomer elicited by dilution of the protein resulted in no change in the far-UV CD spectrum at 220 nm. In contrast, after the same increase in weight fraction of monomer, the near-UV circular dichroism spectra for human relaxin exhibited a significant decrease in the amplitude for the CD bands near 277 and 284 nm. These CD bands, which may be assigned to the lone tyrosine in human relaxin, are superimposed on a broad envelope that is probably due to the three disulfide chromophores. Although both the human and porcine proteins contain two tryptophan residues, the near-UV CD spectra exhibit only a broad shoulder near 295 nm rather than the strong CD bands often found for tryptophan. Moreover, there is little change in this broad band after dilution of human relaxin to concentrations that resulted in a 4-fold increase in monomer weight fraction. These data suggest that dissociation of the human relaxin dimer to monomer is not accompanied by large overall changes in secondary structure or alteration in the average tryptophan environment, whereas there is a significant change in the tyrosine environment.(ABSTRACT TRUNCATED AT 250 WORDS)

Quadruplex structure of d(G3T4G3) stabilized by K+ or Na+ is an asymmetric hairpin dimer.

The ends of chromosomes contain repeats of guanine-rich sequences that can assume highly compact conformations and are presumed necessary for their biological role in chromosomal stabilization and association. We have investigated the conformational behavior of d(G3T4G3) as a function of the addition of either KCl or NaCl, in the concentration range of 50-200 mM, by using a spectrum of physical techniques and conclude that these salts induce a quadruplex species composed of two strands, each in a hairpin conformation. When salt is added, a large positive signal appears near 290 nm in the CD spectra. UV thermal denaturation curves show a single concentration-dependent transition and provide data for quantitating the thermodynamics of quadruplex formation. In electrophoresis experiments, the quadruplex structure migrates as a single species and more rapidly than the unstructured single strand. NMR spectra in the presence of KCl or NaCl indicate that the structure formed is asymmetric. Equilibrium ultracentrifugation studies confirm that these quadruplexes are composed of two strands of d(G3T4G3). Possible models for this structure are discussed.

Characterization of complex formation by humanized anti-IgE monoclonal antibody and monoclonal human IgE.

The interaction of human IgE with high-affinity IgE Fc receptors on cells of the immune system plays an essential role in the type I hypersensitivity reaction. A proposed therapy is to use an anti-IgE monoclonal antibody to block the binding of IgE to its high-affinity receptor on mast cells and basophils, thus preventing subsequent release of the inflammatory agents after exposure to allergen. We report here the solution characteristics of immune complexes formed by a humanized anti-IgE monoclonal antibody (rhuMAb E25) and IgE using sedimentation analysis and size exclusion chromatography. We demonstrate that the rhuMAb E25 is able to form a variety of complexes with IgE at different molar ratios. The largest complex was identified by sedimentation equilibrium analysis as a heterohexamer with very high stability. The intermediate complex formed when one of the interacting components is in large molar excess appears to have a trimeric structure. The high-affinity interaction of rhuMAb E25 and IgE has also been confirmed. Furthermore, by using hydrodynamic modeling, we show that the largest complex may be represented by a cyclic structure.

Mechanism of tobacco mosaic virus assembly: Incorporation of 4S and 20S protein at pH 7.0 and 20 degrees C.

The mechanism of assembly of tobacco mosaic virus (TMV) has been investigated at pH 7.0 and 20 degrees C by analytical ultracentrifugation. Under these conditions the overall rates of interconversion of 4S and 20S TMV coat protein are sufficiently slow to make possible measurements of the concentrations of remaining 4S and 20S TMV coat protein after addition of homologous RNA to solutions containing, initially, various mass ratios of 20S protein to 4S protein. It has been possible to measure, by schlieren boundary analysis, the relative rates of incorporation of 4S and 20S TMV protein into the growing nucleoprotein rod over the range of initial 20S:4S protein mass ratios from 93:7 to 18:82. The results show that the amount of incorporation of 20S TMV protein depends on the initial 20S:4S mass ratio between approximately 100% and 60% 20S protein but that reconstitution can proceed with approximately 100% 20S TMV protein to form full virus-size rods. However, when the initial protein solutions have less than 60% 20S protein, approximately 80% of the reconstituted nucleoprotein is preferentially formed from 4S coat protein. The remaining approximately 20% appears to require preformed 20S coat protein. These results suggest that a larger region of RNA than previously estimated is involved in the rate-limiting nucleation step in assembly and may explain previously conflicting results concerning the elongation phase of assembly when starting with partially assembled rods.

Evidence for the involvement of histidine A(12) in the aggregation and precipitation of human relaxin induced by metal-catalyzed oxidation.

The metal-catalyzed oxidation (ascorbate/cupric chloride/oxygen) of recombinant human relaxin (rhRlx, type II) was shown by Li et al. [Li, S., Nguyen, T. H., Schöneich, C., and Borchardt, R. T. (1995) Biochemistry 34, 5762-5772] to result in the chemical modification of His A(12), Met B(4), and Met B(25). Considering the fact that His A(12) exists in an extended loop that joins two alpha-helices in this protein, we hypothesized that oxidation of this specific amino acid leads to alterations in the secondary and tertiary structures of the protein, resulting in the pH-dependent aggregation/precipitation phenomena observed in our earlier studies (i.e., at pH >6.0 most of the degradants of rhRlx are insoluble). Evidence obtained in the current study that supports this hypothesis includes the following: (i) oxidation of rhRlx with hydrogen peroxide (H(2)O(2)), which leads only to modification of Met B(4) and Met B(25), does not result in the pH-dependent aggregation/precipitation of the protein; and (ii) metal-catalyzed oxidation of porcine relaxin (pRlx), which does not contain His at position A(12), leads to chemical degradation of the protein [e.g., Met A(2) is oxidized] but produces only slight pH-dependent aggregation/precipitation of the protein. In addition, experimental evidence is provided to show that the physical instability of rhRlx observed at pH >6.0 does not appear to be related to the pH-dependent solubility of a common protein degradant. Instead, it appears that several oxidation products of His A(12) are produced in a pH-dependent manner and that these oxidation products produce different effects on the physical stability of the protein. Evidence in support of this conclusion includes the observation that the soluble degradants of rhRlx showed reduced levels of His, reduced levels of the T(2)-T(7) tryptic fragment that contained His A(12), and the presence of 2-oxo-His. Similarly, the precipitated degradants of rhRlx showed reduced levels of His but no 2-oxo-His. In addition, the soluble degradants, which contain 2-oxo-His, appear to exist as monomers having an average molecular weight similar to that of rhRlx. These results suggest that the metal-catalyzed oxidation of His A(12) leads to other, as yet unidentified oxidation products of His A(12) that affect the secondary/tertiary structure of the protein more significantly than does 2-oxo-His and ultimately lead to the physical instability of the protein observed at higher pH values.

Interaction of human IgE with soluble forms of IgE high affinity receptors.

PURPOSE: Interaction of human IgE with its high affinity receptor (Fc epsilon RI) on mast cells and basophils is an important step for initiating IgE mediated immune responses. To characterize the IgE and Fc epsilon RI interaction, we investigated this interaction in terms of stoichiometry and binding affinity in solution. The binding of IgE and IgE Fc epsilon RI alpha chain, the extracellular portion of IgE high affinity receptor (sFc epsilon RI alpha) was compared with the binding of IgE and IgE immunoadhesin (Fc epsilon RI alpha-IgG). METHODS: The interaction was characterized by analytical ultracentrifugation, size exclusion chromatography, light scattering and ELISA. RESULTS: We show that the sFc epsilon RI alpha is only able to bind to one IgE, while the immunoadhesin can bind to two IgE. The interaction between IgE and Fc epsilon RI is very strong. Both forms of soluble receptors have similar intrinsic binding affinity with IgE. CONCLUSIONS: Both soluble receptors (Fc epsilon RI alpha-IgG and sFc epsilon RI alpha) can block the binding of IgE to its high affinity receptors on cell surface. The Fc epsilon RI alpha-IgG is a better IgE binding protein than sFc epsilon RI alpha at physiological relevant conditions. A humanized anti-IgE monoclonal antibody, rhuMAb E25 that also can block the binding of IgE to its high affinity receptors appears to bind to IgE at slightly different regions or in a different manner as the soluble forms of IgE receptors.

Electrostatic effects in myoglobin. Application of the modified Tanford-Kirkwood theory to myoglobins from horse, California grey whale, harbor seal, and California sea lion.

The modified Tanford-Kirkwood electrostatic theory (Shire et al., 1974a) was applied to ferrimyoglobins from the following animal species: sperm whale (Physeter catodon), horse, California grey whale (Eschrichtius gibbosus), harbor seal (Phoca vitulina), and California sea lion (Zalophus californianus). Computations were made of the overall hydrogen ion titration curves of the proteins, and of pH and ionic strength variations of ionization equilibria for individual groups in the protein, with particular reference to the hemic acid ionization of the iron bound water molecule. Coordinates and static solvent accessibility were estimated in terms of the sperm whale myoglobin structure. Where possible, theoretical results and experimental data are compared. Some comparative features of charge and ionization properties among the various myoglobins are presented.

Purification and immunogenicity of fusion VP1 protein of foot and mouth disease virus.

A procedure has been developed to purify foot and mouth disease virus (FMDV) VP1 surface antigens from recombinant Escherichia coli. The VP1 antigens are expressed as fusion proteins derived from the E. coli Trp operon and VP1 surface protein of FMDV. The procedure is capable of recovering greater than 96% of the desired product at a purity of greater than 96%. The resulting antigens induce significant levels of virus-neutralizing antibody in guinea pigs and cattle as determined by a mouse protection assay [Skinner, H.H. (1952) Proc. Int. Vet. Congr., 15th 1, 195]. E. coli contaminants have a deleterious effect on ion-exchange chromatography as well as immunogenicity of the expressed fusion VP1 antigens. The method presented removes significant E. coli contaminants, yielding fusion VP1 proteins which are immunogenically potent. In particular, virus neutralization titers at 100-micrograms dosage of the fusion VP1 proteins of the O1 and A24 serotypes are similar to that induced by the natural VP1 proteins isolated from FMD virions.

Kinetics and mechanism of tobacco mosaic virus assembly: direct measurement of relative rates of incorporation of 4S and 20S protein.

The mechanism of assembly of tobacco mosaic virus has been investigated under conditions in which the rates of incorporation of the 4S and 20S proteins can each be directly measured by analytical centfrifugation. Under these conditions, pH 6.5, 6.5 degrees C, 0.10 M ionic strength potassium orthophosphate, the protein can be made to exist as a metastable 20S aggregate that is necessary for efficient reconstitution. The overall assembly process consists of an initiation (nucleation) reaction that requires two to three 20S disk aggregates per RNA molecule and is followed by an elongation (growth) reaction. In the elongation phase of assembly the 4S protein is incorporated 50 to 70 times faster than the 20S disk, calculated on the basis of a steady-state kinetic analysis. Therefore, under these conditions, in which the rate of assembly is about 0.06 of that at pH 7, 20 degrees C, 0.10 M ionic strength orthophosphate, the 4S protein preferentially participates in the elongation phase. At this slow reconstitution rate intermediate assembly states (about 70-168 S) can be observed. The kinetics of both protein incorporation and nucleoprotein formation suggest that the elongation process is composed of at least two different, possibly sequential, rate-limiting reactions.

TRAP, the trp RNA-binding attenuation protein of Bacillus subtilis, is a multisubunit complex that appears to recognize G/UAG repeats in the trpEDCFBA and trpG transcripts.

A filter binding assay was developed to study interactions between purified TRAP, the trp RNA-binding attenuation protein of Bacillus subtilis, and trp specific transcripts. TRAP formed stable complexes with trpEDCFBA leader RNA; binding was L-tryptophan-dependent and was complete within 60 s. TRAP binds to a segment of the trp leader transcript that includes part of an RNA antiterminator structure. Binding to this segment allows formation of an RNA terminator structure, thereby promoting transcription termination. Using several trpEDCFBA leader deletion transcripts, we identified several closely spaced trinucleotide repeats (seven GAG and four UAG repeats) in the trp leader transcript that appeared to be required for TRAP binding. We also showed that TRAP binds to a segment of the trpG transcript that includes the trpG ribosome binding site; the nucleotide sequence of this segment contains several appropriately spaced trinucleotide repeats (seven GAG, one UAG, and one AAG). TRAP binding to the trpG transcript would block translation initiation. RNA footprint analysis confirmed interaction between TRAP and the trinucleotide repeats in the various transcripts. TRAP, in the presence or absence of L-tryptophan, appears to consist of 11 or 12 identical 8-kDa subunits. Our findings suggest that each tryptophan-activated TRAP subunit can bind one G/UAG repeat in a target transcript. Multiple protein-RNA interactions are required for stable association.

Investigation of protein-surfactant interactions by analytical ultracentrifugation and electron paramagnetic resonance: the use of recombinant human tissue factor as an example.

PURPOSE: The purpose of this work is to utilize electron paramagnetic resonance (EPR) spectroscopy in conjunction with analytical ultracentrifugation (AUC) to investigate the binding of surfactants to proteins with a transmembrance domain. As an example these methods have been used to study the interaction of a nonionic surfactant, C12E8, to recombinant human tissue factor (rhTF) in liquid formulations. The complementary nature of the two techniques aids in data interpretation when there is ambiguity using a single technique. In addition to binding stoichiometries, the possibility of identifying the interacting domains by using two forms of rhTF is explored. METHODS: Two recombinant, truncated forms of human tissue factor were formulated in the absence of phospholipids. Neither of the recombinant proteins, produced in E. coli, contains the cytoplasmic domain. Recombinant human tissue factor 243 (rhTF 243) consists of 243 amino acids and includes the transmembrane sequences. Recombinant human tissue factor 220 (rhTF 220), however, contains only the first 221 amino acids of the human tissue factor, lacking those of the transmembrane region. EPR and AUC were used to investigate the interactions between these two forms of rhTF and polyoxyethylene 8 lauryl ether, C12E8. RESULTS: Binding of C12E8 to rhTF 243 is detected by both EPR spectroscopy and AUC. Although a unique binding stoichiometry was not determined, EPR spectroscopy greatly narrowed the range of possible solutions suggested by the AUC data. Neither technique revealed an interaction between rhTF 220 and C12E8. CONCLUSIONS: The complementary nature of EPR spectroscopy and AUC make the combination of the two techniques useful in data interpretation when studying the interactions between rhTF and C12E8. By utilizing these techniques in this study, the binding stoichiometry of rhTF 243 to C12E8 ranges from 1.2:1 to 1.3:0.6 based on an aggregation number of 120. This binding is consistent with previously reported activity data that showed an increase in clotting rate when rhTF 243 is in the presence of C12E8 micelles. From the rhTF 220 data, it can further be concluded that the transmembrane domain of rhTF is necessary for interactions with C12E8.