Development of in silico models to predict viscosity and mouse clearance using a comprehensive analytical data set collected on 83 scaffold-consistent monoclonal antibodies.

Biologic drug discovery pipelines are designed to deliver protein therapeutics that have exquisite functional potency and selectivity while also manifesting biophysical characteristics suitable for manufacturing, storage, and convenient administration to patients. The ability to use computational methods to predict biophysical properties from protein sequence, potentially in combination with high throughput assays, could decrease timelines and increase the success rates for therapeutic developability engineering by eliminating lengthy and expensive cycles of recombinant protein production and testing. To support development of high-quality predictive models for antibody developability, we designed a sequence-diverse panel of 83 effector functionless IgG1 antibodies displaying a range of biophysical properties, produced and formulated each protein under standard platform conditions, and collected a comprehensive package of analytical data, including in vitro assays and in vivo mouse pharmacokinetics. We used this robust training data set to build machine learning classifier models that can predict complex protein behavior from these data and features derived from predicted and/or experimental structures. Our models predict with 87% accuracy whether viscosity at 150 mg/mL is above or below a threshold of 15 centipoise (cP) and with 75% accuracy whether the area under the plasma drug concentration-time curve (AUC0-672 h) in normal mouse is above or below a threshold of 3.9 × 106 h x ng/mL.

Differentiation of Aspartic and Isoaspartic Acid Using 193 nm Ultraviolet Photodissociation Mass Spectrometry.

Spontaneous conversion of aspartic acid (Asp) to isoaspartic acid (isoAsp) is a ubiquitous modification that influences the structure and function of proteins. This modification of Asp impacts the stability of biotherapeutics and has been linked to the development of neurodegenerative diseases. We explored the use of 193 nm ultraviolet photodissociation (UVPD) to distinguish Asp and isoAsp in the protonated and deprotonated peptides. The differences in the relative abundances of several fragment ions uniquely generated by UVPD were used to differentiate isomeric peptide standards containing Asp or isoAsp. These fragment ions result from the cleavage of bonds N-terminal to Asp/isoAsp residues in addition to the side-chain losses from Asp/isoAsp or the losses of COOH, CO2, CO, or H2O from y-ions. Fragmentation of Asp-containing tryptic peptides using UVPD resulted in more enhanced w/w + 1/y - 1/x ions, while isoAsp-containing peptides yielded more enhanced y - 18/y - 45/y - 46 ions. UVPD was also used to identify an isomerized peptide from a tryptic digest of a monoclonal antibody. Moreover, UVPD of a protonated nontryptic peptide resulted in more enhanced y ions N- and C-terminal to isoAsp and differences in b/y ion ratios that were used to identify the isoAsp peptide.

Unequivocal Identification of Aspartic Acid and isoAspartic Acid by MALDI-TOF/TOF: From Peptide Standards to a Therapeutic Antibody.

Aspartic acid (Asp) to isoaspartic acid (isoAsp) isomerization in therapeutic monoclonal antibodies (mAbs) and other biotherapeutics is a critical quality attribute (CQA) that requires careful control and monitoring during the drug discovery and production processes. The unwanted formation of isoAsp within biotherapeutics and resultant structural changes in the peptide backbone may negatively impact the efficacy, potency, and safety of the molecule or become immunogenic, especially if the isomerization occurs within the mAb complementarity determining region (CDR). Herein we describe a MALDI-TOF/TOF mass spectrometry method that affords unequivocal identification of the presence and the exact position of the isoAsp residue(s) in peptide standards ranging in size from a tripeptide to a docosapeptide (22 residues). In general, the peptide bond immediately N-terminal to the isoAsp residue is more susceptible to MALDI-TOF/TOF fragmentation than its unmodified counterpart. In some of the peptides evaluated in this study, fragmentation of the peptide bond C-terminal to the isoAsp residue (the aspartate effect) is also enhanced when compared to the control. Relative quantification by MALDI-TOF/TOF of this chemical modification is dependent upon a successful reversed-phase HPLC (rpHPLC) separation of the control and modified peptides. This method has also been validated on a therapeutic mAb that contains a well-documented isoAsp residue in the heavy chain CDR3 after forced degradation. Moreover, we also demonstrate that higher energy C-trap dissociation of only the singly charged species, and not the multiply charged form, of the isoAsp containing peptide, separated by rpHPLC, results in LC-MS/MS fragmentation that is highly consistent to that of MALDI-TOF/TOF.

Native and Denaturing MS Protein Deconvolution for Biopharma: Monoclonal Antibodies and Antibody-Drug Conjugates to Polydisperse Membrane Proteins and Beyond.

Electrospray ionization mass spectrometry (ESI-MS) is a ubiquitously used analytical method applied across multiple departments in biopharma, ranging from early research discovery to process development. Accurate, efficient, and consistent protein MS spectral deconvolution across multiple instrument and detector platforms (time-of-flight, Orbitrap, Fourier-transform ion cyclotron resonance) is essential. When proteins are ionized during the ESI process, a distribution of consecutive multiply charged ions are observed on the m/z scale, either positive [M + nH]n+ or negative [M - nH]n- depending on the ionization polarity. The manual calculation of the neutral molecular weight (MW) of single proteins measured by ESI-MS is simple; however, algorithmic deconvolution is required for more complex protein mixtures to derive accurate MWs. Multiple deconvolution algorithms have evolved over the past two decades, all of which have their advantages and disadvantages, in terms of speed, user-input parameters (or ideally lack thereof), and whether they perform optimally on proteins analyzed under denatured or native-MS and solution conditions. Herein, we describe the utility of a parsimonious deconvolution algorithm (explaining the observed spectra with a minimum number of masses) to process a wide range of highly diverse biopharma relevant and research grade proteins and complexes (PEG-GCSF; an IgG1k; IgG1- and IgG2-biotin covalent conjugates; the membrane protein complex AqpZ; a highly polydisperse empty MSP1D1 nanodisc and the tetradecameric chaperone protein complex GroEL) analyzed under native-MS, denaturing LC-MS, and positive and negative modes of ionization, using multiple instruments and therefore multiple data formats. The implementation of a comb filter and peak sharpening option is also demonstrated to be highly effective for deconvolution of highly polydisperse and enhanced separation of a low level lysine glycation post-translational modification (+162.1 Da), partially processed heavy chain lysine residues (+128.1 Da), and loss of N-acetylglucosamine (GlcNAc; -203.1 Da).

PADLOC: a powerful tool to assign disulfide bond connectivities in peptides and proteins by NMR spectroscopy.

The determination of the disulfide bond connectivity in a peptide or protein represents a significant challenge. It is notoriously difficult to use NMR spectroscopy to assign disulfide connectivities because NMR spectra lack direct evidence for disulfide bonds. These bonds are typically inferred from three-dimensional structure calculations, which can result in ambiguous disulfide assignment. Here, we present a new NMR based methodology, in which the disulfide connectivity is obtained by applying Bayesian rules of inference to the local topology of cysteine residues. We illustrate how this approach successfully predicts the disulfide connectivity in proteins for which crystal structures are available in the protein data bank (PDB). We also demonstrate how this methodology is used with experimental NMR data for peptides with complex disulfide topologies, including hepcidin, Kalata-B1, and µ-Conotoxin KIIIA. In the case of µ-Conotoxin KIIIA, the PADLOC connectivity (1-15,2-9,4-16) differs from previously published results; additional evidence is presented demonstrating unequivocally that this newly proposed connectivity is correct.