The stability of peptide and protein biotherapeutics directly impacts their pharmacokinetic profile, efficacy, and safety. It is therefore essential to characterize the stability of a given bio-therapeutic including both in-vivo and in-vitro catabolism, thereby providing a fundamental understanding of potential metabolic soft-spots. To facilitate an automated accurate mass LC-MS workflow for the detection and confirmation of peptide biotransformations and their profiling across a time course, SCIEX has introduced several modifications to the existing MetabolitePilot software platform.  In this research grade version of MetabolitePilot, the LC-MS peak finding strategy has been expanded for peptide and protein-based drugs by (i) supporting higher charge states, (ii) generating putative catabolic products by hydrolytic cleavage of the amide backbone (including cleavage of disulfide bonds), (iii) considering the isotopic distribution of the catabolic product and (iv) interpreting MS/MS data using conventional peptide fragmentation patterns. 

In this blog I provide an overview of the key features of the research grade version of MetabolitePilot and share some initial data from the determination of Exenatide catabolites formed in-vitro, as well as survey other reports regarding the implementation of this software for tackling the challenges associated with peptide/protein catabolism.

Exenatide is a 4.2 kDa glucagon-like peptide-1 agonist (GLP-agonist) with amino acid sequence HGEGTFTSDLSKQMEEEAVRLFIEWLKNGGPSSGAPPPS-NH₂, and is used for the treatment of type 2 diabetes. In the current investigation, in vitro catabolites of Exenatide were determined following incubation in rat whole blood using SWATH^® high-resolution accurate mass data acquired on a SCIEX TripleTOF 5600⁺ mass spectrometer.  In SWATH mode (Sequential Window Acquisition of All Theoretical MS), fixed 50 Da wide MS/MS windows were applied at 45eV collision energy for precursor masses ranging from 100 Da to 1250 Da.

The post-acquisition processing workflow in research grade MetabolitePilot initially involves defining potential bio-transformations.  In addition to the provided default modifications in the Biologics set, custom biotransformations can be entered, and for Exenatide included deamidation (N/Q), phosphorylation (S/T/H), sulphate conjugation (S/T/W), acetylation (K), formylation (G), oxidative deamination to alcohol (K), desaturation (K/Q/E), methylation (G), demethylation (A/T) and demethylation + oxidation (A/T).  The MetabolitePilot processing algorithm always considers hydrolytic cleavages, and for Exenatide generated 702 theoretical catabolites.

The next stage in the MetabolitePilot workflow defines the Peak Finding Strategy, which leverages the TOF-MS data derived from the first experiment in the SWATH acquisition.  Within the Processing Parameters dialog, the peptide sequence of Exenatide is entered along with an MS/MS reference spectrum from which two diagnostic product ions (y₃-ion at m/z 299.1717 and y₄-ion at m/z 396.2245) were also incorporated into the peak finding algorithm (Figure 1).

With the Biotransformation and Processing Parameters established, incubated samples at t₀, t₃₀, t₆₀, t₁₂₀ and t₂₄₀ minutes were interrogated against control samples.  Proposed catabolites are presented in the Results workspace and were considered only if the measured parent mass was within 10 ppm of theoretical and the response was 3-fold greater than that observed in control samples. This same mass accuracy was applied to the assignment of product ions measured against theoretical b- and y-ion masses for proposed catabolite sequences (Figure 2).  Results from each time-point were compiled in the Correlation workspace, and potential catabolites plotted (Figure 3).  In the case of Exenatide, only one catabolite demonstrated increased response with incubation time and coincided to the N-terminal HG clipping biotransformation product Exenatide(3-39), whose chromatographic profile and MS/MS spectrum are compared to Exenatide in Figure 4.Exenatide(4-39), (5-39), and (7-39) catabolites were also detected, however only at incubation times  ≥ t₁₂₀, and therefore could not be fully Correlated within the timeframe of the experiment (Figure 3). Regardless, each minor putative catabolite generated the diagnostic y3– and y4– ions of the Exenatide reference spectrum, and each parent mass was measured within 5 ppm of theoretical. Of particular note, Exenatide(3-39) and (4-39) were chromatographically unresolved, and their [M+5H]⁺⁵ precursor masses were simultaneously transmitted through the same SWATH window (i.e. m/z 749 – 800), thereby generating a mixed MS/MS spectrum.  While a formidable challenge to deconvolute this complex scenario in applications such as PeakView, the advanced algorithm used in MetabolitePilot successfully re-constructed the MS/MS spectrum derived from each of Exenatide(3-39) and (4-39), thereby aligning the TOF-MS XICs with confirmatory b– and y- fragmentions, as outlined in Figure 5.

Recently, the research grade version of MetabolitePilo was combined with accurate mass measurements from a TripleTOF^® 6600 system to understand the multi-chain complexity of insulin catabolism.¹ Data was obtained using both SWATH and Information Dependent Acquisitions (IDA).  More than 30,000 potential catabolites were predicted, and when using an enhanced peak finding approach with filtering of singly-charged material, unexpected catabolites could also be identified at the TOF-MS level of the data.  Searching by characteristic product ions in the TOF-MS/MS experiment further contributed toward the unexpected catabolite pool.  Finally, the list of isobaric catabolites for a given metabolite peak could be reduced when considering the characteristic fragment ions containing the C-terminal residue.

The expanded capabilities of MetabolitePilot have also been applied to the investigation of the in-vitro metabolism of two model therapeutic peptides Bivalirudin and Peptide A².  Bivalirudin was found to be relatively stable in plasma with methylation as the most abundant minor catabolite, while Peptide A underwent proteolytic hydrolysis to form numerous catabolites through sequential C- and N-terminal residue losses.  The SCIEX TripleTOF 6600 was used in IDA mode with dynamic background subtraction and focusing only on multiply charged precursor ions.

Conclusion
The research grade version of MetabolitePilot software, when combined with accurate mass measurements from a TripleTOF platform operated in SWATH or IDA mode, represents an automated platform for catabolism studies, supporting both linear peptides and more complex multi-chain peptide- and protein-based drugs. Also, the versatility and ease of use of the software facilitates a rapid interrogation of results such that appropriate modifications may be made to improve the stability of the biotherapeutic. The advanced capabilities of the MetabolitePilot algorithm were exemplified in properly assigning co-eluting catabolites simultaneously transmitted through the same SWATH window, a daunting feat with any other software platform.

References

1. Patel J, Cancilla M, Yu X, Popescu AD, Payne G, Duchoslav E, Ramagiri S.  Emerging Bioanalytical Tool to Characterize Drug and Drug-Products of Biotherapeutics. ASMS 2016
2. Duchoslav E, Woodward M, Patel J, Ramagiri S.  Software Aided Workflow for Streamlining the Investigation of Catabolic Degradation Products during Peptide Therapeutics Compound Optimization.  ASMS 2015