nDSF has been successfully utilized to determine differences in stability between formulations differing only in pH or initial %HWWS. Differences in shape of nDSF samples as well as calculated and predictive methods, such as kD, have been used to predict the long-term stability of 90 mg/mL to 150 mg/mL formulations of the drug product. The results obtained allow the use of this approach to determine relative stability as early as 1 month during long-term stability studies in the development of drug products.

Chemical modifications such as deamidation, isomerization, and oxidation can affect the function of antibody therapeutics and complicate development. Given that experimental assessment is resource-intensive, we present machine learning models, trained on data from over 700 antibody samples, that predict sites where liabilities are likely to occur from sequence input alone. These models can be run throughout the discovery and lead optimization process, allowing for proactive removal of sequence liabilities.

In silico assessment of antibody developability have the potential to speed up antibody development, especially the lead optimization. However, advances in computational tools such as machine learning models are often limited by the lack of suitable training data sets of high quality. Here, I present improvements of common antibody developability assays; e.g., AC-SINS, with the aim of enabling optimal data for modelling. Important parameters like dynamic range, calibration and data processing are discussed.

High-throughput screening techniques for biophysical properties analysis of molecules in early screening studies is warranted due to limited amount of material and large number of molecules. Predictability of early screening results to long-term storage stability is critical as it assists in defining the design space for long-term study. In this study, biophysical properties and 8-week stability of two ADCs in 16 formulations were evaluated, and the predictive capabilities of screening methods were assessed. The study demonstrated that Tagg and B22 is more predictive than conformational stability read-outs (Tm) for long-term storage stability. High-throughput assays also identified poor performing formulations.

Aggregation of recombinant proteins is a major consideration in their developability, safety, and immunogenicity. While detection of aggregates and analysis of their basic properties is routine, understanding the molecular mechanism involved is much more challenging. Structural mass spectrometry techniques such as hydroxyl radical protein footprinting (HRPF) was used to decipher mechanism of aggregate in biotherapeutics development to reduce development timelines

Novel large molecule therapeutic modalities are emerging to deliver sophisticated mechanism of actions to modulate the “undruggable” targets where canonical antibodies fall short. However, little is known about the in vivo stability liabilities (i.e. biotransformation) with these new large molecule modalities. Additionally, the stability findings from in vitro stress conditions may not fully translate in vivo. Here we report a multi-pronged approach using LC-MS and capillary electrophoresis-based methods to characterize and quantify biotransformation liabilities and the in vitro/ex vivo vs. in vivo translatability, including but not limited to linker deconjugation, clipping, and amino acid level modifications.

Therapeutic proteins are susceptible to aggregation throughout their lifetime, with hydrodynamic forces and interfacial stresses being major culprits. We have developed a low-volume extensional flow device (EFD) to understand the kinetic mechanism underpinning flow-induced aggregation. Armed with this knowledge, we can then apply the device as a screening tool for formulations which promote long-term stability under storage conditions.

I will present methodologies developed in my lab for the design and engineering of antibodies. We use machine learning for de novo design and make use of modern technologies to ensure all our designs are developable. We show high yields and other developability criteria of our hit antibodies.

Analytical Quality by Design offers a systematic and robust approach to the development of analytical procedures involving all stages of the product’s lifecycle. The presentation will overview the new draft ICH guidelines on analytical procedure development, validation, and lifecycle management. Special emphasis will be placed on the analytical procedures commonly used in biotherapeutics for DS and DP GMP release and stability.

Therapeutic antibodies are a major class of biopharmaceuticals that can treat a variety of diseases. As an important type of product-related impurities, charge variants and their related heterogeneity arising from post-translational modifications and truncations can affect the stability, efficacy, and safety of the drug product. In this study, we present the development and application of native microfluidic capillary electrophoresis (CE)-mass spectrometry (MS) to monitor the charge variants in therapeutic antibody drug candidates. The ZipChip CE couples front-end charge variant separation with MS for charge variant identifications, and this analysis is applicable for various applications in drug development.

The topic of the presentation is a physico-chemical characterization of vaccine products. Several methods used to measure product attributes at different stages of manufacturing will be discussed. The output can be used for process monitoring, product characterization, and potentially for real-time release.

We present a quantitative single-particle imaging platform that enables simultaneous measurements of the size, mRNA-payload, and dynamic properties of vaccines in cell-like conditions. We investigate the dependence of mRNA-lipid-nanoparticle structure and fusion dynamics on formulation, using commercially available formulations as a starting point. These measurements are made on confined, freely diffusing particles, and during reagent-exchange such as in response to solution pH, in order to emulate intracellular dynamics in a controlled setting. Over the long term and in collaboration with health scientists, we propose to correlate multi-scale data sets including single-particle measurements made in vitro as well as in cells and tissues, with clinical results, to create a throughline of understanding of vaccine effectiveness from the microscopic to clinical scale, to enable and optimize their rational design and engineering.

Light scattering techniques (DLS) for protein size characterization takes an ensemble-based approach to particle sizing. Although being a gold standard, this approach may not accurately distinguish particle sizes in a multimodal sample where the diameter of particles is similar or when particle size distributions are broad. Additionally, concentration of large particles can be overestimated thereby skewing the particle size distribution of the sample. On the other hand, single particle analysis approach involves tunable resistive pulse sensing (TRPS) technology which measures the size of individual particles passing through a nanopore and claims to provide more accurate particle size distribution data within a sample. Particle concentration analysis is also based on single-particle measurements thereby ensuring highly accurate calculations for each size band compared to the ensemble-based approach. The present study performs the comparative analysis of size distribution of vaccine antigens to evaluate the performance of these methods.

Despite the recent groundbreaking advancements in mRNA vaccine development, analytical methods have not evolved at the same rate, leaving a significant need for highly sensitive and rapid purity assessment methods. Here, we propose a dual dynamic staining high-throughput microfluidic electrophoresis analytical method for the detection and characterization of dsRNA contaminants in mRNA vaccines. With an mRNA maximum loading capacity of 13 ng/μL, we can detect dsRNA contaminants as low as 0.1-0.6% of the total concentration.