Adoption of analytical technology usually starts in R&D labs with first-generation instruments operated by skilled and well-trained early adopters. These people work inside the particular institution to solve critical applications for the institution. Eventually, the technology becomes well known. Over time, successive models with simplified operation are adopted in manufacturing environments for process and quality control.
After four decades, this appears to be happening in biotech with mass spectrometry (MS), multiple reaction monitoring (MRM), and liquid chromatography/ mass spectrometry (LC/MS). Significantly, regulators have developed sufficient confidence that MS is accepted for release assays of biotherapeutics. This has opened the door to other purposes, including control of post-translational modification (PTM) and predictive control of reactors.
This meeting review covers the 11th Symposium on the Practical Applications of Mass Spectrometry in the Biotechnology Industry, held September 9–12 in Napa, CA.
MS in product release
Dr. Patrick Bulau of Roche Diagnostics (Penzberg, Germany) lectured that MS has successfully been used for product release at Roche for over a decade. During this time and over several products, there have been no out-of-specification results (OSRs) attributed to problems with MS analytics. The next lecture, by Jun T. Park of the FDA’s Center for Drug Evaluation and Research (CDER, Silver Spring, MD), assured that MS was suitable for inclusion in regulatory submissions and QC, assuming a rigorous validation study was performed. But this is not unique to MS; validation is required for any method or technology.
Dr. Patrick Swann of Biogen Idec (Cambridge, MA) opened the meeting with a report on how MS can enable process improvement. Legacy products can be analyzed by new techniques, such as high resolution and high mass range. Bridging studies can then show that the new process delivers a product that is indistinguishable from the legacy process.
But what if the new process has a different product profile? What if the impurity profile is different? One then must go back to the structure of the active pharmaceutical ingredient (API), relate this to the critical quality attributes (CQAs), and then determine how the new product would affect the quality attributes.
The process starts with general recognition that it should be possible to improve a process by using a new technology, such as time-of-flight/MS (TOF/ MS) to identify input and output materials, including associated impurities.
Initially, TOF/MS was used to build the bridgehead on the current practice side; with these data it was possible to predict the outcome results, including tolerance band of the new TOF/MS-based process control. Next was isotopic labeling with 13C to track process improvements. This produced r2 correlations with CQAs of 0.9998. The bridging study involved about 2000 samples over 12 months. Such studies are essential to show comparability with legacy benchmarks.
Sample banking protocols are used to monitor history, assess stability, and provide reference material for QC of products from subsequent runs. Dr. Sushmita Roy (Capricon Proteomics, Menlo Park, CA) addressed sample storage and processing from the biobank. Every step needs to be evaluated and verified. Dr. Roy started with storage temperature. Storage at –20 °C is often considered to be okay, but she showed that storage at –80 °C can lead to artifacts, since what appears to be a homogeneous solution or intracellular liquid at room temperature can be a mixture of ice and concentrated solution at –20 °C and also at –80 °C. Worse, at –80 °C, the ice dehydrates the sample, including the hydration layer of the analyte. Without the protection provided by the hydration layer, one can envision destabilization of the sample facilitating all sorts of reactions and rearrangements. She also described other parameters such as freeze/thaw cycles and time on the bench.
Predictive analytics for cell culture
Intact cell matrix-assisted ionization (MALDI)-TOF/MS (ICM/MS) is established for identification bacteria in clinical and environmental microbiology. Dr. Sebastian Schwamb (Center for Applied Biomedical Mass Spectrometry, Mannheim, Germany) advocates extension to Chinese hamster ovary (CHO) cell culture. Using ICM/MS, he was able to see signs of cell stress, including apoptosis many hours, even days, before it was evident by other techniques. Plus, he could see changes in the PTM patterns and yield. By getting early warning of pending difficulty, it is possible to respond by adjusting the growth environment, thus extending the run and yield.
Several papers focused on methods to determine the drug-to-antibody ratio (DAR) in antibody drug conjugates (ADCs). Controlling the DAR is important in formulation and stability, plus therapeutic efficacy and safety. The drugs are generally chosen from a cohort of highly cytotoxic compounds that have narrow therapeutic windows.
Robert Birdsall (Waters, Milford, MA) used 2D-UPLC/MS (ACQUITY M-Class and hydrophobic interaction chromatography [HIC] in the first dimension), which elutes by increasing DAR number for drugs bonded at cysteine cross-linking sites. Reversed-phase LC (RPLC)/MS is used in the second dimension to resolve the positional isomers. For example, three positional isomers are expected for the homolog containing two drug groups. With a DAR of four, there are four isomers. The isomer locations are confirmed from the MS spectrum provided by a Waters Xevo G2 QTOF MS.
A similar study using lysine coupling would be much more complicated since the number of possible sites is typically much larger and they are not paired as they are with disulfide/cysteine structure of antibodies. For example, maytansine is a cytotoxin that inhibits the assembly of microtubules by binding to tubulin. Even more potent analogs of maytansine have been synthesized and conjugated to antibodies via N-hydroxysuccinimide linkers to lysine to yield a potent API against HER-2 positive breast cancer.
Dr. Lintto Wang (Immunogen, Waltham, MA) focused on the distribution of maytansinoid conjugates distributed among about 90 lysine sites in an ADC. Both the DAR and site occupancy were relevant. They tried several approaches, but finally settled on a composite of peptide maps, top-down MS of the intact protein, and middle-down LC/MS of the dissected protein to monitor the distribution among 26 accessible sites. The middle-down assay used site-specific cleavage to reduce the size to be more compatible with MS.
ADCs are often not stable, according to Elsa Wagner-Rousse of the Center of Immunology in Strasbourg, France. She reported that MS has been used for years to monitor the stability of antibody-based drugs. In general, compounds with high DAR are more prone to aggregation. Plus, IgG4s are susceptible to chain mixing, becoming significant in a half-day or so.
The majority of proteins are post-translationally modified with conjugation with one or several glycans. Prof. Carlito Lebrilla of the University of California at Davis reported that the glycol part of the glyconjugates is composed of up to 20 monosaccharides. The arrangement of the particular glycan is complex with isomers, including order, linkage, and branching. The number of glycans and the subtle differences complicate a quantitative understanding of protein heterogeneity. Yet glyconjugates are important in human development, cancer, and autoimmune diseases.
The analytical challenge for glyconjugates is even more complicated than lysine coupling since more amino acids are potentially involved. O-linked glycans bond through the oxygen to serine and threonine, plus the N-linked bond through the amide of asparagine the protein.
The analytics involve site- or linkage-specific cleavage enzymes. Isomers, including positional, are often resolved by the ion mobility stage in the mass analyzer train.
The majority of investigational new drugs (INDs) are antibody based, with lots of creativity exhibited in their design. Often the antibody is sliced and diced with parts rearranged to meet particular needs. Genovis (Lund, Sweden) presented a poster describing how the FabULOUS™ (SpeB) protease cleaves the FAB sections of IgG1 antibodies at the upper hinge region. More recently, Genovis introduced the Fabricator® (IdeS) that cleaves the hinge from the lower part, giving pure Fc. This was demonstrated for the drug trastuzumab, where selective digestion facilitated characterization of the structure with electrospray ionization (ESI)/MS. Genovis has also developed and markets other specific enzymes for antibody engineering and characterization. IgGZERO™ specifically removes Fc-glycans. GlycINATOR™ removes Fc-glycans, including high-mannose structures.
Steric exclusion chromatography (SEC) and reversed-phase chromatography are often used together to follow reactions such as PTM, aggregation, and degradation. Usually the main peaks are obvious, but what about the origin of the “foothills” that often flank the main peaks? Which ones correlate with others seen in chromatograms of the other mode? MS might help, but the even high-resolution spectrometers have trouble with small flanking peaks. Paul Salinas and colleagues (Shire, Lexington, MA) studied six samples with different impurity levels of the native protein and protease digest track peaks using the Pairwise Statistical Analysis Tool, which is part of the JMP software package (JMP, Cary, NC). Some of the peaks were easily identified as arising from cysteine dioxidation. Others were also correlated as common between the LC modes.
Glycoengineering of antibodies
Glycosylation of the Fc part of IgG strongly influences cytotoxic efficacy, according to Dr. Markus Haberger of Roche Diagnostics. Since the need is great, Roche developed in house an automated, high-throughput MS method for glycopeptide analysis following photolytic digestion. Peptides are fractionated by hydrophilic interaction solid phase extraction and analyzed by ESI/MS in positive ion mode. An ion mobility spectrometry (IMS) drift tube resolves isomers. A specialized data processing program provides a report of the relative amounts of various glycopeptides. This analyzer supports process development of glycol engineered IgGs.
Host cell proteins
As discussed in several lectures and posters, host cell proteins (HCPs) are copurified with the product antibody and affect the impurity profile of the finished product. For IgGs, most processes started with expression in CHO cells followed by purification by protein A affinity chromatography. After this stage, HCP contamination was about 2000 ppm. Most of these proteins are complexed to the antibody, and hence are difficult to remove. Amgen (Thousand Oaks, CA) decided to explore reengineering the antibody to reduce complexation, as well as exploring use of mild chaotropes to disrupt the HCP-mAb complex without denaturing the mAb. This is a work in progress.
Capillary electrophoresis MS
Twenty-five years ago, capillary electrophoresis (CE) was a hot topic, with several vendors. Fifteen years ago, CE was the essential technology for sequencing the human genome. Today, interest in CE is low, despite its exceptional separation power, especially when combined with MS. Prof. Andras Guttman of the University of Debrecen in Hungary showed how CE/ESI/MS can be used to characterize trastuzumab by monitoring the protein digests. CE/MS provides 100% sequence coverage, enabling the unambiguous assay of methionine oxidation asparagine–deamidation, and C-terminal lysine heterogeneity, etc., with only 100 fmol of the antibody.
Ion mobility mass analysis extends other modes
Ions traversing a gas-filled drift tube experience drag due to frictional forces. The amount of drag is a function of the molecular profile. Thus, IMS can separate isomers that differ in their mobility.
In a plenary lecture, Dr. Richard Smith of the Pacific Northwest National Laboratory (Richland, WA) showed that, under appropriate conditions, ions can be trapped, stored, and recalled. To improve workflow in MS, we’ve seen examples in IMS and ion traps that have subsecond events, but Dr. Smith showed ion traps that are stable for hours, even days. He calls the technology “SLIM” for Structures for Lossless Ion Manipulation. SLIM starts with ion funnels that focus ions into traps. These are created using printed circuit board technology1 to create traps to store and manipulate ions for MS analysis. The storage traps are planar structures with a 4–6-mm-thick open space for as many as 107 ions. Vacuum gas is nitrogen at 4 Torr or less.
Summary and credits
It would be impossible to overstate the positive impact of MS and LC/MS on our society, particularly healthcare. Today, our pharmaceutical leaders and regulators can talk with confidence about the detailed structure of complex large molecules. ADC and DAR measurements including occupancy are particularly useful metrics. As these are assimilated, scientists will certainly develop even more insightful therapeutics, which should improve efficacy and safety.
This meeting series typically attracts about 160 scientists to a congenial, focused 2.5-day program. Posters are up for the entire meeting, which provides an effective forum for discussion and networking. This is sharp contrast to the ASMS meeting, which is five times larger, much more intense, and addresses the entire range of MS technology and applications.
Dr. Alain Balland of Amgen and Dr. Steve Cohen of Northeastern University in Boston, MA, deserve special credit for organizing a strong and timely technical program focused on biopharm. Ms. Renee Olsen led the CASSS support team that provided the creature comforts that are essential for successful scientific meetings. The 12th edition in this meeting series will be held September 22–25 at the New York Marriott at the Brooklyn Bridge. Please monitor CASSS.org for details.
Robert L. Stevenson, Ph.D., is Editor, American Laboratory/Labcompare; e-mail: [email protected].