Characterization of site-specific glycation during process development of a human therapeutic monoclonal antibody.

A therapeutic recombinant monoclonal antibody (mAb1) was found to be highly susceptible to glycation during production. Up to 42% glycation was observed in mAb1, which was significantly greater than the glycation observed in 17 other monoclonal antibodies (mAbs). The majority of the glycation was localized to lysine 98 of a unique sequence in the heavy chain complementarity determining region 3. Upon incubation with 5% glucose at 37 °C for 5 days, the level of glycation rose to 80% of the total protein where the majority of the additional glycation was on the lysine 98 residue. These data suggested that the lysine 98 residue was highly susceptible to glycation. However, three other mAbs with a lysine residue in the same position did not show high rates of glycation in the forced glycation assay, suggesting that primary and perhaps secondary structural constraints could contribute to the rate of glycation at that lysine. Interestingly, a portion of the glycation in mAb1 was found to be reversible and upon incubation in phosphate buffer (pH 7) at 37 °C for 5 days, the glycation dropped from starting levels of 42% to 20%. Variation was observed in the total glycation levels between different lots of mAb1. The variability in glycation introduced charge heterogeneity in the form of an acidic peak on cation exchange chromatography and lead to product inconsistency. Mutation of lysine 98 to arginine reduced the starting level of glycation without any impact on potency.

Cold chemical oxidation of proteins.

Various methods of protein footprinting use hydrogen peroxide as an oxidant. Its removal by various solid-phase desalting methods, catalase treatment, or freeze drying after the footprinting is critical to ensure no uncontrolled oxidation. Although catalase treatment removes hydrogen peroxide with little loss of protein or additional protein oxidation, we discovered that freeze drying or freezing of the protein in a peroxide solution does lead to protein oxidation. Interestingly, the oxidation is not a result of freeze or thaw processes but is dependent on the temperature and length of time for incubation. After 2 h, apomyoglobin undergoes almost-complete single oxidation at -80 degrees C and double oxidation at -15 degrees C. Minimal oxidation is observed at 4 and 22 degrees C, compared to oxidation at -80 or -15 degrees C. The concentration of hydrogen peroxide is critical; 75 mM (0.2%) is required to oxidize >50% of the protein at -15 degrees C and 100 mM (0.3%) is required at -80 degrees C. In addition to Met, approximately 5% of the tryptophan and tyrosine residues are oxidized, as well as lower amounts of His and Phe. Oxidation of Val 68 and Val 17 (a buried residue) also occurs, with the oxidation of Val 17 likely occurring by electron transfer from one of two of the oxidized aromatic residues that are in contact with Val 17. Here, we describe the need to remove the hydrogen peroxide prior to cold storage of proteins, and we also report some preliminary results pertaining to the mechanism of cold, solid-state oxidation.

Detection and quantitation of IgG 1 hinge aspartate isomerization: a rapid degradation in stressed stability studies.

In biopharmaceutical process development, it is desirable to identify sites of covalent degradations to ensure product consistency. One characterization method used for therapeutic immunoglobulin gamma (IgG) 1 antibodies is limited LysC proteolysis followed by reversed-phase LC/MS. Limited LysC proteolysis leads to high efficiency cleavage at the C-terminal side of the hinge lysine 222 residue, generating Fab and Fc fragments. In this report, we show that IgG 1 samples incubated under mildly acidic conditions at elevated temperatures were partially resistant to LysC cleavage at the hinge and resulted in a species where one of the Fab arms remained connected to the Fc region (Fab-Fc). The growth of the Fab-Fc species was proportional to the duration and storage temperature of the incubation period and correlated with the amount of isomerization of the aspartic acid residue preceding lysine 222, determined by peptide mapping. The isomerization rates of samples stored for up to one year at 4 degrees C, 6 months at 29 or 37 degrees C, or 3 months at 45 degrees C were determined, and the activation energy for this conversion was calculated to be approximately 33 kJ mol(-1). The apparent isomerization rate constant was only 0.02 week(-1) for samples stored at 4 degrees C, which resulted in a modest increase from 5.1 to 6.0% isoD after twenty four weeks of storage and, hence, is not a significant concern under normal storage conditions typically used for monoclonal antibodies. However, when stored at 29 degrees C, the apparent rate constant of this reaction was found to be 0.06 week(-1) and resulted in an increase from 5.1 to 21.1% isoD after twenty four weeks of storage and is a major degradant in stressed IgG 1 antibodies.

The effect of sucrose hydrolysis on the stability of protein therapeutics during accelerated formulation studies.

Stability studies of protein therapeutics are often accelerated by storing potential formulations at elevated temperatures where the rates of various chemical and physical degradation pathways are increased. An often overlooked caveat of using these studies is the potential degradation of the formulation components themselves. In this report, we show that the monoclonal antibody MAB001 aggregated at a faster rate when formulated with sucrose compared to samples that contained sorbitol or no excipient during accelerated stability studies following an initial lag phase where the rates of aggregate formation were similar in all formulations. The duration of the lag phase was both pH and temperature dependent and a significant increase of protein glycation was noticed during this time. These observations indicate that the enhanced rate of antibody aggregation in sucrose containing formulations is likely due to protein glycation following sucrose hydrolysis under accelerated conditions. This hypothesis was confirmed by demonstrating that antibody directly glycated with glucose aggregated at a faster rate than nonglycated antibody stored in the identical formulation. These findings question the utility of using accelerated stability data for predicting protein stability in sucrose containing formulations stored at 2-8 degrees C, where no glycation or change in aggregation rate was observed.

Structural features responsible for the biological stability of Histoplasma's virulence factor CBP.

The virulence factor CBP is the most abundant protein secreted by Histoplasma capsulatum, a pathogenic fungus that causes histoplasmosis. Although the biochemical function and pathogenic mechanism of CBP are unknown, quantitative Ca (2+) binding measurements indicate that CBP has a strong affinity for calcium ( K D = 6.45 +/- 0.4 nM). However, no change in structure was observed upon binding of calcium, prompting a more thorough investigation of the molecular properties of CBP with respect to self-association, secondary structure, and stability. Over a wide range of pH values and salt concentrations, CBP exists predominantly as a stable, noncovalent homodimer in both its calcium-free and -bound states. Solution-state NMR and circular dichroism (CD) measurements indicated that the protein is largely alpha-helical, and its secondary structure content changes little over the range of pH values encountered physiologically. ESI-MS revealed that the six cysteine residues of CBP are involved in three intramolecular disulfide bonds that help maintain a highly protease resistant structure. Thermally and chemically induced denaturation studies indicated that unfolding of disulfide-intact CBP is reversible and provided quantitative measurements of protein stability. This disulfide-linked, protease resistant, homodimeric alpha-helical structure of CBP is likely to be advantageous for a virulence factor that must survive the harsh environment within the phagolysosomes of host macrophages.

Reversed-phase liquid chromatography of immunoglobulin G molecules and their fragments with the diphenyl column.

A diphenyl column was able to resolve two closely related monoclonal IgG2 molecules, while a C8 column failed to separate these IgGs under identical chromatographic conditions. The diphenyl column also showed a better separation of a mixture of two light and two heavy chains than the C8 column. The influence of amino acid side chains from protein sequences in binding to the diphenyl and C8 stationary phases was studied by using a set of synthetic peptides with the sequence GXXLLLKK, where X represents substitution with all of the 20 amino acids. Peptides containing aromatic amino acids showed a greater binding on the diphenyl column than on the C8 column. This increase in retention was attributed to pi-pi interactions between the aromatic amino acid side chains and the diphenyl ligand. Based on the retention of peptides on the diphenyl column, new retention coefficients were assigned for the separation of proteins. A good correlation was observed between the sum of retention coefficients (SigmaRc) for IgGs and their retention time on the diphenyl column. On-column hydrogen-deuterium exchange showed that the diphenyl column had a larger surface of interaction with protein than the C8 column. pi-pi interactions and the large contact surface resulted in improved resolution of IgGs and their fragments on the diphenyl column.

Laser flash photolysis of hydrogen peroxide to oxidize protein solvent-accessible residues on the microsecond timescale.

Footprinting of proteins by hydroxyl radicals generated on the millisecond to minute timescales to probe protein surfaces suffers from the uncertainty that radical reactions cause the protein to unfold, exposing residues that are protected in the native protein. To circumvent this possibility, we developed a method using a 248 nm KrF excimer laser to cleave hydrogen peroxide at low concentrations (15 mM, 0.04%), affording hydroxyl radicals that modify the protein in less than a microsecond. In the presence of a scavenger (20 mM glutamine), the radical lifetimes decrease to approximately 1 microsecond, yet the reaction timescales are sufficient to provide significant oxidation of the protein. These times are arguably faster than super-secondary protein structure can unfold as a result of the modification. The radical formation step takes place in a nanoliter flow cell so that only one laser pulse irradiates each bolus of sample. The oxidation sites are located using standard analytical proteomics, requiring less than a nanomole of protein. We tested the method with apomyoglobin and observed modifications in accord with solvent accessibility data obtained from the crystal structure of holomyoglobin. Additionally, the results indicate that the F-helix is conformationally flexible in apomyoglobin, in accord with NMR results. We also find that the binding pocket is resistant to modifications, indicating that the protein pocket closes in the absence of the heme group-conclusions that cannot be drawn from current structural methods. When developed further, this method may enable the determination of protein-ligand interfaces, affinity constants, folding pathways, and regions of conformational flexibility.

ZZ domain of CBP: an unusual zinc finger fold in a protein interaction module.

CREB-binding protein (CBP) is a large, multi-domain protein that provides a multitude of binding sites for transcriptional coactivators. The site of interaction of the tumor suppressor p53 and the oncoprotein E1A with CBP/p300 has been identified with the third cysteine-histidine-rich (CH3) domain, which incorporates two zinc-binding motifs, ZZ and TAZ2. We show that these two domains fold independently and do not interact in solution. Our experiments demonstrate conclusively that the interaction of p53 and E1A with the CH3 domain resides exclusively in the TAZ2 domain, with no contribution from the ZZ domain. We report also the three-dimensional solution structure of the ZZ domain of murine CBP. The 52 residue ZZ domain contains two twisted antiparallel beta-sheets and a short alpha-helix, and binds two zinc ions. The identity of the zinc coordinating ligands was resolved unambiguously using NMR spectroscopy of the ZZ domain substituted with (113)Cd. One zinc ion is coordinated tetrahedrally via two CXXC motifs to four cysteine side-chains, and the second zinc ion is coordinated tetrahedrally by a third CXXC motif, together with an unusual HXH motif coordinating via the N(epsilon2) atom of His40 and the N(delta1) atom of His-42. The first zinc cluster of the ZZ domain is strictly conserved, whereas the second zinc cluster shows variability in the position of the two histidine residues, reflecting the wide variety of molecules that incorporate ZZ domains. The structure of the ZZ domain shows that it belongs to the family of cross-brace zinc finger motifs that include the PHD, RING, and FYVE domains; however, its biological function is unclear. Mapping of the positions of conserved residues onto the calculated structures reveals a face containing exposed aromatic and hydrophobic side-chains, while the opposite face contains a series of conserved charged or hydrophilic groups. These homologies suggest that the ZZ domain is involved in ligand binding or molecular scaffolding, with specificity provided by the variability of the sequence that contains the helix in the murine CPB ZZ domain structure.

Characterization of the binding site for inhibitors of the HPV11 E1-E2 protein interaction on the E2 transactivation domain by photoaffinity labeling and mass spectrometry.

An indandione-containing class of inhibitors abrogates DNA replication of human papillomavirus (HPV) types 6 and 11 by binding reversibly to the transactivation domain (TAD) of the viral E2 protein and inhibiting its interaction with the viral E1 helicase. To locate the binding site of this class of protein-protein interaction inhibitors, a benzophenone derivative was used to generate an irreversibly labeled E2-TAD polypeptide. The single site of covalent modification of the E2-TAD was identified by proteolytic digestions using trypsin, LysC, and V8 proteases and characterization of the resulting peptides by LC-MS procedures. Through this methodology, the benzophenone attachment point was located at the terminal methyl of residue Met101. Evidence further pinpointed the site of photoaffinity attachment to the terminal carbon atom, which is significant in providing a definitive example of the ability to locate photoinduced cross-linking to a polypeptide with atomic resolution using solely mass spectrometric detection. The location of the inhibitor binding site vis-à-vis the Glu39 and Glu100 residues sensitive to mutation for HPV 11 E2-TAD is discussed in relation to the crystal structure of the E2-TAD from the related HPV type 16.