Assessment of monoclonal antibody glycosylation: a comparative study using HRMS, NMR, and HILIC-FLD.

Monoclonal antibodies (mAbs) represent the largest class of therapeutic protein drug products. mAb glycosylation produces a heterogeneous, analytically challenging distribution of glycoforms that typically should be adequately characterized because glycosylation-based product quality attributes (PQAs) can impact product quality, immunogenicity, and efficacy. In this study, two products were compared using a panel of analytical methods. Two high-resolution mass spectrometry (HRMS) workflows were used to analyze N-glycans, while nuclear magnetic resonance (NMR) was used to generate monosaccharide fingerprints. These state-of-the-art techniques were compared to conventional analysis using hydrophilic interaction chromatography (HILIC) coupled with fluorescence detection (FLD). The advantages and disadvantages of each method are discussed along with a comparison of the identified glycan distributions. The results demonstrated agreement across all methods for major glycoforms, demonstrating how confidence in glycan characterization is increased by combining orthogonal analytical methodologies. The full panel of methods used represents a diverse toolbox that can be selected from based on the needs for a specific product or analysis.

The Use of Mass Spectrometry in Therapeutic Protein Biologics License Applications: A Retrospective Review Revisited.

Biologic license applications (BLAs) for 93 therapeutic proteins approved between 2016 and 2020 were analyzed for use of mass spectrometry (MS) as a follow up to a previous study that assessed MS use in BLAs from 2000 to 2015. Thirty percent of these BLAs were biosimilars, while only one biosimilar BLA was approved prior to 2016. This analysis evaluated the use of a variety of MS techniques and instrumentation. Results were further interpreted based on the relationship of MS use over time, between drug types, and between new drugs and biosimilars. MS data were included in 93 BLAs examined. The top eight quality attributes most assessed by MS in rank order were amino acid sequence, molecular mass, oxidation, disulfide bonds, deamidation, glycosylation, N-terminal sequence variants, and C-terminal sequence variants. These attributes were the same top attributes seen previously from BLAs approved between 2000 and 2015, and the use of MS to analyze them generally continued to increase across the new time frame. The average number of attributes analyzed by MS per BLA also continued to increase over the extended time frame of 21 years. High-resolution, accurate mass instrumentation such as the Orbitrap and time-of-flight (TOF) usage increased over time for all assessed attributes, while matrix-assisted laser desorption/ionization (MALDI)-TOF/(TOF) usage decreased. From highest to lowest rank, the top 11 attributes were antibody drug conjugate (ADC) characterization (i.e., drug load distribution/drug to antibody ratio (DAR), ADC and linkage site, and synthetic linker), isomerization, folding/higher-order structure (HOS), truncation, host cell proteins (HCPs), sequence variants (amino acid substitutions), succinimidation, glycation, PEGylation, charge variants, and oxidation.

Method validation and new peak detection for the liquid chromatography-mass spectrometry multi-attribute method.

The multi-attribute method (MAM) is a liquid chromatography-mass spectrometry (LC-MS) peptide mapping technique that has been proposed as a replacement for several conventional quality control (QC) methods for therapeutic proteins. In addition to quantification of multiple product quality attributes (PQAs), MAM can also monitor impurities using a new peak detection (NPD) feature. Here, results are provided from method validation and NPD studies of an MAM approach applied to rituximab as a model monoclonal antibody (mAb). Twenty-one rituximab PQAs were monitored, including oxidation, pyroglutamination, deamidation, lysine clipping, and glycosylation. The PQA monitoring aspect of the method was validated according to ICH Guidance. Accuracy, precision, specificity, detection and quantitation limits, linearity, range, and robustness were demonstrated for this MAM approach with minimal issues. All PQAs were successfully validated except for several oxidation sites, which did not pass intermediate precision criteria. The variability found in oxidation measurements was attributed to artificial oxidation during sample preparation and could likely be alleviated through several approaches. The NPD aspect of the method was also evaluated. A spike-in approach was used to assess the limits of detection and quantitation (LOD/LOQ) of the NPD feature of MAM. For NPD, the peak intensity threshold was found to be the most critical parameter for accurate detection of impurities since a low threshold can result in false positives while a high threshold can obscure the detection of true peaks. Overall, the MAM approach presented and validated here has been demonstrated to be suitable for both targeted monitoring of rituximab PQAs and non-targeted detection of new peaks that represent impurities.

Mechanistic insights into F420-dependent glucose-6-phosphate dehydrogenase using isotope effects and substrate inhibition studies.

F420-dependent glucose-6-phosphate dehydrogenase (FGD) is involved in the committed step of the pentose phosphate pathway within mycobacteria, where it catalyzes the reaction between glucose-6-phosphate (G6P) and the F420 cofactor to yield 6-phosphogluconolactone and the reduced cofactor, F420H2. Here, we aim to probe the FGD reaction mechanism using dead-end inhibition experiments, as well as solvent and substrate deuterium isotope effects studies. The dead-end inhibition studies performed using citrate as the inhibitor revealed competitive and uncompetitive inhibition patterns for G6P and F420 respectively, thus suggesting a mechanism of ordered addition of substrates in which the F420 cofactor must first bind to FGD before G6P binding. The solvent deuterium isotope effects studies yielded normal solvent kinetic isotope effects (SKIE) on kcat and kcat/Km for both G6P and F420. The proton inventory data yielded a fractionation factor of 0.37, suggesting that the single proton responsible for the observed SKIE is likely donated by Glu109 and protonates the cofactor at position N1. The steady state substrate deuterium isotope effects studies using G6P and G6P-d1 yielded KIE of 1.1 for both kcat and kcat/Km, while the pre-steady state KIE on kobs was 1.4. Because the hydride transferred to C5 of F420 was the one targeted for isotopic substitution, these KIE values provide further evidence to support our previous findings that hydride transfer is likely not rate-limiting in the FGD reaction.

Effects of isoleucine 135 side chain length on the cofactor donor-acceptor distance within F420H2:NADP+ oxidoreductase: A kinetic analysis.

F420H2:NADP+ Oxidoreductase (Fno) catalyzes the reversible reduction of NADP+ to NADPH by transferring a hydride from the reduced F420 cofactor. Here, we have employed binding studies, steady-state and pre steady-state kinetic methods upon wtFno and isoleucine 135 (I135) Fno variants in order to study the effects of side chain length on the donor-acceptor distance between NADP+ and the F420 precursor, FO. The conserved I135 residue of Fno was converted to a valine, alanine and glycine, thereby shortening the side chain length. The steady-state kinetic analysis of wtFno and the variants showed classic Michaelis-Menten kinetics with varying FO concentrations. The data revealed a decreased kcat as side chain length decreased, with varying FO concentrations. The steady-state plots revealed non-Michaelis-Menten kinetic behavior when NADPH was varied. The double reciprocal plot of the varying NADPH concentrations displays a downward concave shape, while the NADPH binding curves gave Hill coefficients of less than 1. These data suggest that negative cooperativity occurs between the two identical monomers. The pre steady-state Abs420 versus time trace revealed biphasic kinetics, with a fast phase (hydride transfer) and a slow phase. The fast phase displayed an increased rate constant as side chain length decreased. The rate constant for the second phase, remained ~2 s-1 for each variant. Our data suggest that I135 plays a key role in sustaining the donor-acceptor distance between the two cofactors, thereby regulating the rate at which the hydride is transferred from FOH2 to NADP+. Therefore, Fno is a dynamic enzyme that regulates NADPH production.

Investigating the Reaction Mechanism of F420-Dependent Glucose-6-phosphate Dehydrogenase from Mycobacterium tuberculosis: Kinetic Analysis of the Wild-Type and Mutant Enzymes.

F420-dependent glucose-6-phosphate dehydrogenase (FGD) catalyzes the conversion of glucose-6-phosphate (G6P) to 6-phosphogluconolactone, using F420 cofactor as the hydride transfer acceptor, within mycobacteria. A previous crystal structure of wild-type FGD led to a proposed mechanism suggesting that the active site residues His40, Trp44, and Glu109 could be involved in catalysis. We have characterized the wild-type FGD and five FGD variants (H40A, W44F, W44Y, W44A, and E109Q) by fluorescence binding assays and steady-state and pre-steady-state kinetic experiments. Compared to wild-type FGD, all the variants had lower binding affinities for F420, thus suggesting that Trp44, His40, and Glu109 aid in F420 binding. While all the variants had decreased catalytic efficiencies, FGD H40A and W44A were the least efficient, having lost ∼1000- and ∼2000-fold activity, respectively. This confirms a crucial catalytic role for His40 in the FGD reaction and suggests that aromaticity at residue 44 aids catalysis. To investigate the proposed roles of Glu109 and His40 in acid-base catalysis, the pH dependence of kinetic parameters has been determined for the E109Q and H40A mutants and compared to those of the wild-type enzyme. The log kcat-pH profile of wild-type FGD and E109Q revealed two ionizable residues in the enzyme-substrate complex, while H40A displayed only one ionization event. The FGD E109Q variant displayed pH-dependent kinetic cooperativity with respect to the F420 cofactor. The multiple-turnover pre-steady-state kinetics were biphasic for wild-type FGD, W44F, W44Y, and E109Q, while the H40A and W44A variants displayed only a single phase because of their reduced catalytic efficiency.

Evidence of Negative Cooperativity and Half-Site Reactivity within an F420-Dependent Enzyme: Kinetic Analysis of F420H2:NADP(+) Oxidoreductase.

Here, we report the very first example of half-site reactivity and negative cooperativity involving an important F420 cofactor-dependent enzyme. F420H2:NADP(+) oxidoreductase (Fno) is an F420 cofactor-dependent enzyme that catalyzes the reversible reduction of NADP(+) through the transfer of a hydride from the reduced F420 cofactor. These catalytic processes are of major significance in numerous biochemical processes. While the steady-state kinetic analysis showed classic Michaelis-Menten kinetics with varying concentrations of the F420 redox moiety, FO, such plots revealed non-Michaelis-Menten kinetic behavior when NADPH was varied. The double reciprocal plot of the varying concentrations of NADPH displays a downward concave shape, suggesting that negative cooperativity occurs between the two identical monomers. The transient state kinetic data show a burst prior to entering steady-state turnover. The burst suggests that product release is rate-limiting, and the amplitude of the burst phase corresponds to production of product in only one of the active sites of the functional dimer. These results suggest either half-site reactivity or an alternate sites model wherein the reduction of the cofactor, FO occurs at one active site at a time followed by reduction at the second active site. Thus, the data imply that Fno may be a functional regulatory enzyme.