Purpose: Characterization of the critical quality attribute of higher order structure (HOS) is essential during all stages of the life-cycle of a protein therapeutic, since misfolding or aggregation of mAbs can lead to loss of efficacy or elicit unintended immune responses. High resolution two-dimensional (2D) heteronuclear 1H-13C-methyl correlated nuclear magnetic resonance (NMR) spectroscopy presents an excellent choice for HOS characterization, since even small changes in chemical environment and structure give rise to readily observable changes in corresponding spectra. Indeed, 2D NMR has been demonstrated to be a robust and reproducible method to characterize HOS of biotherapeutics, including monoclonal antibodies (mAbs), with atomic resolution at natural isotopic abundance. However, analysis of such data currently relies primarily on subjective comparison of spectra and qualitative spectral matching; establishment of more objective and quantitative measures is needed.
Methods: Structural isoforms of the NISTmAb reference material were produced by enzymatic remodeling of the Fc attached glycans as well as by preparation in different formulation conditions. Using multivariate approaches, such as Principal Component Analysis (PCA), applied directly to the spectral data matrices, systematic collections of 2D 1H-13C methyl spectra of the NISTmAb isoforms were statistically analyzed to correlate spectra according specific attributes of analyte HOS. Loading vectors, which describe the source of variance in each principal component, were recapitulated as spectral loading plots and used to map the spectral variance to their structural sources.
Results: Results of this study demonstrate that chemometric analysis can be used to identify and differentiate the NISTmAb isoforms from one another with low levels of detection. These spectral differences can be described and quantified using various distance and cluster metrics applied to component space vectors. Loading plots mapped spectral variation from glycan remodeling to specific affected residues in the Fc domain and provisionally elucidated mechanisms for hydrodynamic changes as a function of formulation concentration.
Conclusion: 2D 1H-13C methyl NMR can be used to characterize the HOS of protein therapeutics, including mAbs, with high resolution and sensitivity. New chemometric approaches to analyses of such spectra show promise for the objective and quantitative characterization of HOS, while still providing atomic-level detail about salient features of structure. These methods should help move HOS characterization from the realm of qualitative spectral matching to a more rigorous analysis grounded in statistics.