Subunit Analysis Zeroes In on PTMs

Post-translational modifications (PTMs) occur naturally in mammalian cells, including mammalian expression systems that produce monoclonal antibodies (mAbs) and therapeutic proteins. PTMs gained attention during the 1980s and 1990s, as biotechnology was industrializing, when patients receiving early-generation mAb treatments developed serious immune responses to the drugs. Human anti-mouse antibody reactions, which severely dose-limited early protein therapies, were in large part blamed on non-human glycosylation patterns. Fast forward to 2021, and PTMs—most prominently glycosylation but also amidation, oxidation, phosphorylation, rearrangement, etc. —are now considered critical quality attributes (CQAs).

Biosimilars, for example, are compared with originator molecules (and deemed ready for approval) based on the identities, branching, and locations of sugar residues added after cells construct the amino acid backbone. PTMs, especially glycosylation, serve as CQAs during process development, to guide developers toward optimizing unit operations, and post-approval to monitor consistency during production and from batch to batch.

Characterizing a protein for glycosylation involves sample preparation followed by analysis, primarily through liquid chromatography (HPLC) and variants, mass spectrometry (MS) and variants, with supportive methods (immunoassay, electrophoresis, cell sorting, etc.) used where necessary.

Because all analytic methods are limited in scope, sensitivity, and utility, and glycosylation itself is so chemically and isomerically diverse, improvements are desirable in either the methods themselves or how they work together.

Ion mobility for subtle differences

There are two main avenues, outside of entirely new methods, to improve workflows: Working on individual instruments or how they are used, or making the pieces work better together. One could, for example, use a HILIC column instead of reverse phase, supercritical HPLC vs. standard UHPLC, or a softer MS ionization method. The other approach involves synergy—making separation and detection work better together.

Agilent Technologies has published an application note describing an ion mobility MS technique that, in addition to normal mass/charge (m/z) determinations, further separates molecules of identical or nearly identical m/z by their shape. This could be useful in quantifying peptides incorporating the same number of PTMs but at different locations, or chemically distinct entities whose nominal molecular weights are within one mass unit.

The new technique achieves this by measuring single-field collisional cross section (CCS), which combined with "4D feature extraction" has shown, for example, that phosphopeptides are more compact than non-phosphorylated molecules with similar m/z values. CCS values also differ when phosphorylation sites differ in number or location.

Ion mobility-MS treats mass spectral data as a large, four-dimensional array of retention time, drift time, m/z, and abundance values. "The output of 4D feature detection is to group, by isotopic cluster, peptide peaks that share those common values," says Rebecca Glaskin, LC/MS applications scientist at Agilent Technologies. "Collision cross section describes a molecule's size and shape, and provides insights into folding as well. This allows for IMS to separate gas-phase peptide ions on the basis of small conformational differences that cannot be determined by mass spectrometry alone."

Synergy and harmony

The second strategy, synergy, was the impetus for the recent collaboration between Waters, an instrument company, and reagents purveyor Genovis, to develop complete biopharmaceutical characterization workflows based on Genovis SmartEnzymes™ and Waters™ Andrew+ pipetting robot. The project covers sample preparation, a major bottleneck in glycosylation analysis in established LC-MS platforms.

Traditional peptide mapping takes time and resources. Bottom-up approaches involve proteolysis, typically with trypsin, and forcing these digestions to proceed as completely as possible. One issue with bottom-up is that many digestion sites are only poorly accessible to proteases, which means these reactions take longer. "It takes time to incubate a sample, and to reduce and alkylate cysteine residues ahead of time," says Matthew Lauber, Ph.D., Sr. Consulting Scientist at Waters. "Five days is not unusually long to prep, analyze, and interpret peptide mapping experiments from ten antibody samples."

By contrast to bottom-up, subunit analysis provides similar information and takes as few as five hours, but this method only works when proteins cleave predictably. Since SmartEnzymes (there are several) each cleave at one specific site, cleavage fragments are much larger than peptides obtained from a bottom-up approach, but small enough that most modern mass spectrometers can generate high-quality mass spectra from them.

"Subunit digestion occurs very quickly and specifically. There’s also no need for sample cleanup ahead of analysis," Lauber adds. "When digestion is complete analysts can go directly to the mass spectrometer to monitor for glycosylation, sequence identity, oxidations, deamidations, and other PTMs."

The collaboration also addresses consistency, which for many labs is the prime directive for automation. "There's a great need in quality control and in routine assays for the hands-off approach that reduces human error and improves reproducibility and traceability," Lauber says. The new workflow assures that "the last sample is prepared in exactly the same way as the first. The Andrew + robot can’t do everything a big liquid-handling platform does, but it can easily prepare a 96-well plate of samples in short order."

"When trypsin and other less-specific enzymes cleave antibodies they leave ragged ends and heterogeneous products, which are observable across the chromatogram and/or the mass spectrum," says Andreas Nägeli, Ph.D., Sr. Application Scientist at Genovis. Over-digestion also contributes to lysate heterogeneity. SmartEnzymes, Nägeli says, do not over-digest. "They only do one thing—cleave at specific sites—but they do it really, really well."

Standard glycosylation analytic workflows have their individual operational and instrument-related limitations. Top-down mAb analysis is rapid because it does not involve sample preparation, but resolution and insights are limited. Bottom-up or tryptic workflows provide adequate resolution but involve a lot of sample prep. Subunit-level analysis appears to strike a balance by requiring limited sample prep but while still answering such critical questions such as: What is my glycosylation profile? How many oxidation sites are present? Are there deamidations? Do I have the right amino acid sequence?