Streamlined Multi-Attribute Assessment of an Array of Clinical-Stage Antibodies: Relationship Between Degradation and Stability

Abstract

Clinical antibodies are an important class of drugs for the treatment of both chronic and acute diseases. Their manufacturability is subject to evaluation to ensure product quality and efficacy. One critical quality attribute is deamidation, a non-enzymatic process that is observed to occur during thermal stress, at low or high pH, or a combination thereof. Deamidation may induce antibody instability and lead to aggregation, which may pose immunogenicity concerns. The introduction of a negative charge via deamidation may impact the desired therapeutic function (i) within the complementarity-determining region, potentially causing loss of efficacy; or (ii) within the fragment crystallizable region, limiting the effector function involving antibody-dependent cellular cytotoxicity. Here we describe a transformative solution that allows for a comparative assessment of deamidation and its impact on stability and aggregation. The innovative streamlined method evaluates the intact protein in its formulation conditions. This breakthrough platform technology is comprised of a quantum cascade laser microscope, a slide cell array that allows for flexibility in the design of experiments, and dedicated software. The enhanced spectral resolution is achieved using two-dimensional correlation, co-distribution, and two-trace two-dimensional correlation spectroscopies that reveal the molecular impact of deamidation. Eight re-engineered immunoglobulin G4 scaffold clinical antibodies under control and forced degradation conditions were evaluated for deamidation and aggregation. We determined the site of deamidation, the overall extent of deamidation, and where applicable, whether the deamidation event led to self-association or aggregation of the clinical antibody and the molecular events that led to the instability. The results were confirmed using orthogonal techniques for four of the samples.

Keywords: 2D-COS; 2T2D; Biotherapeutic; HPLC; asparagine and glutamine deamidation; cation exchange; co-distribution correlation spectroscopy; correlation spectroscopy; extent of deamidation; high-performance liquid chromatography; protein aggregation; protein stability; quantum cascade laser microspectroscopy; two-dimensional correlation spectroscopy; two-trace two-dimensional.

Figure 1.

Streamlined ASN and GLN deamidation comprehensive assessment: (a) QCLM equipped with a heated accessory, (b) 3D representation of a 2D-COS asynchronous plot used to monitor the deamidation process, (c) schematic representation of the slide cell array containing gray wells representing the negative controls or reference standard, green and red wells represent ideal and failed candidates, respectively. The failed candidates exhibit loss of stability, ASN or GLN deamidation, with regions prone to aggregation resulting in increased immunogenicity risk for which the deamidation sites may be localized within the CDRs or the Fc region shown with red spheres on the mAb model associated with the failed candidate. The method involves directly monitoring key signature peaks during the deamidation process: (d) bar graphs that summarize the intensity changes monitored in real time during the deamidation process: (1) ν(C=O) at 1678 cm^–1, (2) δ(NH₂) at 1612 cm^–1 of the ASN side chain and (3 and 4) ν(COO^–) at 1572 and 1567 cm^–1 for the aspartate products, (e) ASN deamidation mechanism, and (f) bar graphs summarizing the real-time intensity changes monitored for the GLN deamidation process: (5) ν(C=O) at 1670 cm^–1, (6) δ(NH₂) at 1581 cm^–1 of the GLN side chain and (7 and 8) ν(COO^–) at 1554 and 1545 cm^–1 for the glutamate products, and (g) the GLN deamidation mechanism. Highlighted in yellow is the backbone within each product.

Figure 2.

Quantum cascade laser (QCL) microspectroscopy spectral overlay for all clinical antibody samples in the array within the spectral region of 1785–1425 cm^–1 and during the temperature range of 28–64 °C with 6 °C temperature intervals. The columns are for the control and forced degraded samples and the rows (a) gemtuzumab, (b) pembrolizumab, (c) evolocumab, (d) briakinumab, (e) atezolizumab, (f) nivolumab, (g) trastuzumab, and (h) ustekinumab. Both the quality and consistency of the spectral data can be observed.

Figure 3.

Breakthrough IgG4 clinical antibody array comparative assessment during the real-time monitoring of ASN and GLN deamidation within the temperature range of 28–64 °C, that at times led to aggregation. The analysis involved the use of 2D correlation and co-distribution spectroscopies for: (a) gemtuzumab, (b) pembrolizumab, (c) evolocumab, (d) briakinumab, (e) atezolizumab, (f) nivolumab, (g) trastuzumab, and (h) ustekinumab. 2D-COS asynchronous plots are especially suited to the determination of the relationship of deamidation, aggregation and stability of the clinical antibodies. Adding further understanding is the co-distribution analysis, which allows the evaluation of the distribution of proteins in solution within the sample and defines if the event of deamidation or the deamidation that led to aggregation is a concern for the majority of the clinical antibodies within the sample analyzed. Evidence of aggregation–deamidation is highlighted with the red arrow and was observed in the forced degraded samples of (a) gemtuzumab and (g) trastuzumab compared to their respective controls.

Figure 4.

Two-dimensional correlation spectroscopy (2D-COS) synchronous plots for all clinical antibodies in the array within the spectral region of 1785–1425 cm^–1 and the temperature range of 28–64 °C. The columns are for the control and forced degraded samples and the rows: (a) gemtuzumab, (b) pembrolizumab, (c) evolocumab, (d) briakinumab, (e) atezolizumab, (f) nivolumab, (g) trastuzumab, and (h) ustekinumab. An auto peak characteristic of aggregation at 1614 cm^–1 (red arrow) is observed for the forced degraded sample assigned to: (a) gemtuzumab, (c) evolocumab, and (g) trastuzumab. Furthermore, self-association events were more subtle auto peak features on the synchronous plots (blue arrow) also observed for the forced degraded samples: (e) atezolizumab and (h) ustekinumab.

Figure 5.

Sequential order of molecular events for all eight clinical antibodies during the temperature perturbation of 28–64 °C in the array: (a) gemtuzumab, (b) pembrolizumab, (c) evolocumab, (d) briakinumab, (e) atezolizumab, (f) nivolumab, (g) trastuzumab, and (h) ustekinumab. In each case, the sequential order of molecular events corresponds to: (top row) control and (bottom row) forced degraded samples.

Figure 6.

Bar graph summarizing the overall ASN and GLN extent of deamidation for the control and forced degraded trastuzumab samples after six months of storage at 2–8 °C. The results presented were for a series of five separate evaluations after the 2–8 °C storage period. Specifically, an average of five separate runs for triplicate samples resulting in a total of n  =  30 determinations for which a statistically significant result was obtained. These results confirm our initial evaluation in Tables II and III, whereby the risk of deamidation for trastuzumab occurs after the sample is subject to a period of forced degradation.