Translatability of In Vitro Stress for Predicting Deamidation and Oxidation Biotransformation on Biotherapeutics.

Biotransformation leading to single residue modifications (e.g., deamidation, oxidation) can contribute to decreased efficacy/potency, poor pharmacokinetics, and/or toxicity/immunogenicity for protein therapeutics. Identifying and characterizing such liabilities in vivo are emerging needs for biologics drug discovery. In vitro stress assays involving PBS for deamidation or AAPH for oxidation are commonly used for predicting liabilities in manufacturing and storage and are sometimes considered a predictive tool for in vivo liabilities. However, reports discussing their in vivo translatability are limited. Herein, we introduce a mass spectrometry workflow that characterizes in vivo oxidation and deamidation in pharmacokinetically relevant compartments for diverse protein therapeutic modalities. The workflow has low bias of <10% in quantitating degradation in the relevant pharmacokinetic concentration range for monkey and rabbit serum/plasma (1-100 µg/mL) and allows for high sequence coverage (∼85%) for discovery/monitoring of amino acid modifications. For oxidation and deamidation, the assay was precise, with percent coefficient of variation of <8% at 1-100 µg/mL and ≤6% method-induced artifacts. A high degree of in vitro and in vivo correlation was observed for deamidation on the six diverse protein therapeutics (seven liability sites) tested. In vivo translatability for oxidation liabilities were not observed for the 11 molecules tested using in vitro AAPH stress. One of the molecules dosed in eyes resulted in a false positive and a false negative prediction for in vivo oxidation following AAPH stress. Finally, peroxide stress was also tested but resulted in limited success (1 out of 4 molecules) in predicting oxidation liabilities.

CD3 Target Affinity Chromatography Mass Spectrometry as a New Tool for Function-Structure Characterization of T-Cell Engaging Bispecific Antibody Proteoforms and Product-Related Variants.

T-cell engaging bispecific antibodies (TCBs) targeting CD3 and tumor-specific antigens are very promising therapeutic modalities. Since CD3 binding is crucial for the potency of TCBs, understanding the functional impact of CD3 antigen-binding fragment modifications is of utmost importance for defining critical quality attributes (CQA). The current CQA assessment strategy requires the integration of structure-based physicochemical separation and functional cell-based potency assays. However, this strategy is tedious, and coexisting proteoforms with potentially different functionalities may not be individually assessed. This increases the degree of ambiguities for defining meaningful CQAs, particularly for complex bispecific antibody formats such as TCBs. Here, we report for the first time a proof-of-concept study to separate and identify critically modified proteoforms of TCBs using functional CD3 target affinity chromatography (AC) coupled with online mass spectrometry (MS). Our method enabled functional distinction of relevant deamidated and glycosylated proteoforms and the simultaneous assessment of product-related variants such as TCB mispairings. For example, CD3 AC-MS allowed us to separate TCB mispairings with increased CD3 binding (i.e., knob-knob homodimers) within the bound fraction. The functional separation of proteoforms was validated using an established workflow for CQA identification based on thoroughly characterized ion-exchange fractions of a 2+1 TCB. In addition, the new method facilitated the criticality assessment of post-translational modifications in stress studies and structural variants in early stage clone selection. CD3 AC-MS has high impact for streamlining the integration of functional and structural characterizations of the large landscape of therapeutic CD3 targeting TCBs from early stage research to late stage characterization.

Discovery of a dual pathway aggregation mechanism for a therapeutic constrained peptide.

Constrained peptides (CPs) have emerged as attractive candidates for drug discovery and development. To fully unlock the therapeutic potential of CPs, it is crucial to understand their physical stability and minimize the formation of aggregates that could induce immune responses. Although amyloid like aggregates have been researched extensively, few studies have focused on aggregates from other peptide scaffolds (e.g., CPs). In this work, a streamlined approach to effectively profile the nature and formation pathway of CP aggregates was demonstrated. Aggregates of various sizes were detected and shown to be amorphous. Though no major changes were found in peptide structure upon aggregation, these aggregates appeared to have mixed natures, consisting of primarily non-covalent aggregates with a low level of covalent species. This co-existence phenomenon was also supported by two kinetic pathways observed in time- and temperature-dependent aggregation studies. Furthermore, a stability study with 8 additional peptide variants exhibited good correlation between aggregation propensity and peptide hydrophobicity. Therefore, a dual aggregation pathway was proposed, with the non-covalent aggregates driven by hydrophobic interactions, whereas the covalent ones formed through disulfide scrambling. Overall, the workflow presented here provides a powerful strategy for comprehensive characterization of peptide aggregates and understanding their mechanisms of formation.

From proof of concept to the routine use of an automated and robust multi-dimensional liquid chromatography mass spectrometry workflow applied for the charge variant characterization of therapeutic antibodies.

The identification and quantification of post-translational modifications (PTMs) is a crucial step required during the development of therapeutic proteins. In particular, the characterization of charge variants separated by cation exchange chromatography (CEX) is a tedious process commonly performed with an off-line manual fraction collection followed by peptide mapping. To improve the efficiency of this time-consuming approach, a generic on-line multi-dimensional LC/MS approach was developed for the characterization of various monoclonal antibody (mAb) isotypes and a bi-specific antibody (BsAb). Fractions collected from 1D CEX analysis were consecutively reduced on a 2D reversed phase liquid chromatography (RPLC) column (polyphenyl), digested within 1-2 min using a 3D immobilized trypsin cartridge, and finally the obtained peptides were separated on another 4D RPLC column (C18), and simultaneously identified with a Q Exactive™ mass spectrometer. 2D RPLC columns and 3D trypsin cartridges from different suppliers were compared, as well as the effects of reducing agents. The effect of 2D and 4D RPLC column temperature, and 2D RPLC column mass load were also systematically studied. Under optimal conditions, the multi-dimensional LC/MS system described in this paper is a robust tool for the on-line digestion of proteins and shows high repeatability. Similar levels of oxidation and deamidation were measured using the off-line and on-line approaches for the same stressed samples. The lower amounts of deamidation and isomerization measured at some asparagine and aspartic acid residues by the on-line approach compared to the manual off-line procedure suggest reduced artifacts using the on-line methodology. The multi-dimensional LC/MS described here enables fast, on-line, automated characterization of therapeutic antibodies without the need for off-line fraction collection and sample pre-treatment (manual approach). The entire workflow can be completed within less than one day, compared to weeks with the manual off-line procedure.

Prediction of methionine oxidation risk in monoclonal antibodies using a machine learning method.

Monoclonal antibodies (mAbs) have become a major class of protein therapeutics that target a spectrum of diseases ranging from cancers to infectious diseases. Similar to any protein molecule, mAbs are susceptible to chemical modifications during the manufacturing process, long-term storage, and in vivo circulation that can impair their potency. One such modification is the oxidation of methionine residues. Chemical modifications that occur in the complementarity-determining regions (CDRs) of mAbs can lead to the abrogation of antigen binding and reduce the drug's potency and efficacy. Thus, it is highly desirable to identify and eliminate any chemically unstable residues in the CDRs during the therapeutic antibody discovery process. To provide increased throughput over experimental methods, we extracted features from the mAbs' sequences, structures, and dynamics, used random forests to identify important features and develop a quantitative and highly predictive in silico methionine oxidation model.

Susceptibility of Antibody CDR Residues to Chemical Modifications Can Be Revealed Prior to Antibody Humanization and Aid in the Lead Selection Process.

A critical part of the clinical development path for a therapeutic antibody involves evaluating the physical and chemical stability of candidate molecules throughout the manufacturing process. In particular, the risks of chemical liabilities that can impact antigen binding, such as deamidation, oxidation, and isomerization in the antibody CDR sequences, need to be controlled through formulation development or eliminated by replacing the amino acid motif displaying the chemical instability. Commonly, the antibody CDR sequence contains multiple sequence motifs (potential hotspots) for chemical instability. However, only a subset of these motifs results in actual chemical modification, and thus, experimental assessment of the extent of instability is necessary to identify positions for potential sequence engineering. Ideally, this information should be available prior to antibody humanization at the stage of parental rodent antibody identification. Early knowledge of liabilities allows for ranking of clones or the mitigation of liabilities by concurrent engineering with the antibody humanization process instead of time-consuming sequential activities. However, concurrent engineering of chemical liabilities and humanization requires translatability of the chemical modifications from the rodent parental antibody to the humanized. We experimentally compared the stability of all sequence motifs by mass spectrometric peptide mapping between the rodent parental antibody and the final humanized antibody and observed a linear correlation. These results have enabled a streamlined developability assessment process for therapeutic antibodies from lead discovery to clinical development.

High-throughput screening of antibody variants for chemical stability: identification of deamidation-resistant mutants.

Developability assessment of therapeutic antibody candidates assists drug discovery by enabling early identification of undesirable instabilities. Rapid chemical stability screening of antibody variants can accelerate the identification of potential solutions. We describe here the development of a high-throughput assay to characterize asparagine deamidation. We applied the assay to identify a mutation that unexpectedly stabilizes a critical asparagine. Ninety antibody variants were incubated under thermal stress in order to induce deamidation and screened for both affinity and total binding capacity. Surprisingly, a mutation five residues downstream from the unstable asparagine greatly reduced deamidation. Detailed assessment by LC-MS analysis confirmed the predicted improvement. This work describes both a high-throughput method for antibody stability screening during the early stages of antibody discovery and highlights the value of broad searches of antibody sequence space.

The Use of a 2,2'-Azobis (2-Amidinopropane) Dihydrochloride Stress Model as an Indicator of Oxidation Susceptibility for Monoclonal Antibodies.

Protein oxidation is a major pathway for degradation of biologic drug products. Past literature reports have suggested that 2,2-azobis (2-amidinopropane) dihydrochloride (AAPH), a free radical generator that produces alkoxyl and alkyl peroxyl radicals, is a useful model reagent stress for assessing the oxidative susceptibility of proteins. Here, we expand the applications of the AAPH model by pairing it with a rapid peptide map method to enable site-specific studies of oxidative susceptibility of monoclonal antibodies and their derivatives for comparison between formats, the evaluation of formulation components, and comparisons across the stress models. Comparing the free radical-induced oxidation model by AAPH with a light-induced oxidation model suggests that light-sensitive residues represent a subset of AAPH-sensitive residues and therefore AAPH can be used as a preliminary screen to highlight molecules that need further assessment by light models. In sum, these studies demonstrate that AAPH stress can be used in multiple ways to evaluate labile residues and oxidation sensitivity as it pertains to developability and manufacturability.

Protein engineering to increase the potential of a therapeutic antibody Fab for long-acting delivery to the eye.

To date, ocular antibody therapies for the treatment of retinal diseases rely on injection of the drug into the vitreous chamber of the eye. Given the burden for patients undergoing this procedure, less frequent dosing through the use of long-acting delivery (LAD) technologies is highly desirable. These technologies usually require a highly concentrated formulation and the antibody must be stable against extended exposure to physiological conditions. Here we have increased the potential of a therapeutic antibody antigen-binding fragment (Fab) for LAD by using protein engineering to enhance the chemical and physical stability of the molecule. Structure-guided amino acid substitutions in a negatively charged complementarity determining region (CDR-L1) of an anti-factor D (AFD) Fab resulted in increased chemical stability and solubility. A variant of AFD (AFD.v8), which combines light chain substitutions (VL-D28S:D30E:D31S) with a substitution (VH-D61E) to stabilize a heavy chain isomerization site, retained complement factor D binding and inhibition potency and has properties suitable for LAD. This variant was amenable to high protein concentration (>250 mg/mL), low ionic strength formulation suitable for intravitreal injection. AFD.v8 had acceptable pharmacokinetic (PK) properties upon intravitreal injection in rabbits, and improved stability under both formulation and physiological conditions. Simulations of expected human PK behavior indicated greater exposure with a 25-mg dose enabled by the increased solubility of AFD.v8.

Automated Affinity Capture and On-Tip Digestion to Accurately Quantitate in Vivo Deamidation of Therapeutic Antibodies.

Deamidation of therapeutic antibodies may result in decreased drug activity and undesirable changes in pharmacokinetics and immunogenicity. Therefore, it is necessary to monitor the deamidation levels [during storage] and after in vivo administration. Because of the complexity of in vivo samples, immuno-affinity capture is widely used for specific enrichment of the target antibody prior to LC-MS. However, the conventional use of bead-based methods requires large sample volumes and extensive processing steps. Furthermore, with automation difficulties and extended sample preparation time, bead-based approaches may increase artificial deamidation. To overcome these challenges, we developed an automated platform to perform tip-based affinity capture of antibodies from complex matrixes with rapid digestion and peptide elution into 96-well microtiter plates followed by LC-MS analysis. Detailed analyses showed that the new method presents high repeatability and reproducibility with both intra and inter assay CVs < 8%. Using the automated platform, we successfully quantified the levels of deamidation of a humanized monoclonal antibody in cynomolgus monkeys over a time period of 12 weeks after administration. Moreover, we found that deamidation kinetics between in vivo samples and samples stressed in vitro at neutral pH were consistent, suggesting that the in vitro stress test may be used as a method to predict the liability to deamidation of therapeutic antibodies in vivo.

Chemical structure and properties of interstrand cross-links formed by reaction of guanine residues with abasic sites in duplex DNA.

A new type of interstrand cross-link resulting from the reaction of a DNA abasic site with a guanine residue on the opposing strand of the double helix was recently identified, but the chemical connectivity of the cross-link was not rigorously established. The work described here was designed to characterize the chemical structure and properties of dG-AP cross-links generated in duplex DNA. The approach involved characterization of the nucleoside cross-link "remnant" released by enzymatic digestion of DNA duplexes containing the dG-AP cross-link. We first carried out a chemical synthesis and complete spectroscopic structure determination of the putative cross-link remnant 9b composed of a 2-deoxyribose adduct attached to the exocyclic N(2)-amino group of dG. A reduced analogue of the cross-link remnant was also prepared (11b). Liquid chromatography-tandem mass spectrometric (LC-MS/MS) analysis revealed that the retention times and mass spectral properties of synthetic standards 9b and 11b matched those of the authentic cross-link remnants released by enzymatic digestion of duplexes containing the native and reduced dG-AP cross-link, respectively. These results establish the chemical connectivity of the dG-AP cross-link released from duplex DNA and provide a foundation for detection of this lesion in biological samples. The dG-AP cross-link in duplex DNA was remarkably stable, decomposing with a half-life of 22 days at pH 7 and 23 °C. The intrinsic chemical stability of the dG-AP cross-link suggests that this lesion in duplex DNA may have the power to block DNA-processing enzymes involved in transcription and replication.

Replication across regioisomeric ethylated thymidine lesions by purified DNA polymerases.

Causal links exist between smoking cigarettes and cancer development. Some genotoxic agents in cigarette smoke are capable of alkylating nucleobases in DNA, and higher levels of ethylated DNA lesions were observed in smokers than in nonsmokers. In this study, we examined comprehensively how the regioisomeric O(2)-, N3-, and O(4)-ethylthymidine (O(2)-, N3-, and O(4)-EtdT, respectively) perturb DNA replication mediated by purified human DNA polymerases (hPols) η, κ, and ι, yeast DNA polymerase ζ (yPol ζ), and the exonuclease-free Klenow fragment (Kf(-)) of Escherichia coli DNA polymerase I. Our results showed that hPol η and Kf(-) could bypass all three lesions and generate full-length replication products, whereas hPol ι stalled after inserting a single nucleotide opposite the lesions. Bypass conducted by hPol κ and yPol ζ differed markedly among the three lesions. Consistent with its known ability to efficiently bypass the minor groove N(2)-substituted 2'-deoxyguanosine lesions, hPol κ was able to bypass O(2)-EtdT, though it experienced great difficulty in bypassing N3-EtdT and O(4)-EtdT. yPol ζ was only modestly blocked by O(4)-EtdT, but the polymerase was strongly hindered by O(2)-EtdT and N3-EtdT. LC-MS/MS analysis of the replication products revealed that DNA synthesis opposite O(4)-EtdT was highly error-prone, with dGMP being preferentially inserted, while the presence of O(2)-EtdT and N3-EtdT in template DNA directed substantial frequencies of misincorporation of dGMP and, for hPol ι and Kf(-), dTMP. Thus, our results suggested that O(2)-EtdT and N3-EtdT may also contribute to the AT → TA and AT → GC mutations observed in cells and tissues of animals exposed to ethylating agents.

In-vitro replication studies on O(2)-methylthymidine and O(4)-methylthymidine.

O(2)- and O(4)-methylthymidine (O(2)-MdT and O(4)-MdT) can be induced in tissues of laboratory animals exposed with N-methyl-N-nitrosourea, a known carcinogen. These two O-methylated DNA adducts have been shown to be poorly repaired and may contribute to the mutations arising from exposure to DNA methylating agents. Here, in vitro replication studies with duplex DNA substrates containing site-specifically incorporated O(2)-MdT and O(4)-MdT showed that both lesions blocked DNA synthesis mediated by three different DNA polymerases, including the exonuclease-free Klenow fragment of Escherichia coli DNA polymerase I (Kf(-)), human DNA polymerase κ (pol κ), and Saccharomyces cerevisiae DNA polymerase η (pol η). Results from steady-state kinetic measurements and LC-MS/MS analysis of primer extension products revealed that Kf(-) and pol η preferentially incorporated the correct nucleotide (dAMP) opposite O(2)-MdT, while O(4)-MdT primarily directed dGMP misincorporation. While steady-state kinetic experiments showed that pol κ-mediated nucleotide insertion opposite O(2)-MdT and O(4)-MdT is highly promiscuous, LC-MS/MS analysis of primer extension products demonstrated that pol κ favorably incorporated the incorrect dGMP opposite both lesions. Our results underscored the limitation of the steady-state kinetic assay in determining how DNA lesions compromise DNA replication in vitro. In addition, the results from our study revealed that, if left unrepaired, O-methylated thymidine lesions may constitute important sources of nucleobase substitutions emanating from exposure to alkylating agents.

The roles of DNA polymerases κ and ι in the error-free bypass of N2-carboxyalkyl-2'-deoxyguanosine lesions in mammalian cells.

To counteract the deleterious effects of DNA damage, cells are equipped with specialized polymerases to bypass DNA lesions. Previous biochemical studies revealed that DinB family DNA polymerases, including Escherichia coli DNA polymerase IV and human DNA polymerase κ, efficiently incorporate the correct nucleotide opposite some N(2)-modified 2'-deoxyguanosine derivatives. Herein, we used shuttle vector technology and demonstrated that deficiency in Polk or Poli in mouse embryonic fibroblast (MEF) cells resulted in elevated frequencies of G→T and G→A mutations at N(2)-(1-carboxyethyl)-2'-deoxyguanosine (N(2)-CEdG) and N(2)-carboxymethyl-2'-deoxyguanosine (N(2)-CMdG) sites. Steady-state kinetic measurements revealed that human DNA polymerase ι preferentially inserts the correct nucleotide, dCMP, opposite N(2)-CEdG lesions. In contrast, no mutation was found after the N(2)-CEdG- and N(2)-CMdG-bearing plasmids were replicated in POLH-deficient human cells or Rev3-deficient MEF cells. Together, our results revealed that, in mammalian cells, both polymerases κ and ι are necessary for the error-free bypass of N(2)-CEdG and N(2)-CMdG. However, in the absence of polymerase κ or ι, other translesion synthesis polymerase(s) could incorporate nucleotide(s) opposite these lesions but would do so inaccurately.