With the incredibly high demand for new antibody-based biotherapeutics, and, with the need for the rapid production of biologics to treat emerging diseases, it is now more important than ever to speed up biologics discovery and manufacture by streamlining processes and timelines. This optimization will boost speed-to-clinic and get new therapeutics to patients faster. For antibody drug discovery, this means delivering high-affinity, high-potency antibody candidates with better developability profiles through much faster timelines. To achieve this, the future of antibody discovery lies in emerging technologies, automation and artificial intelligence (AI). Together, these three areas are driving improvements in the antibody discovery process.

This article explores the multiplicative forces of emerging technologies, automation and AI for faster antibody discovery, with a focus on leveraging microfluidic platforms to identify target-specific variants in minimal time to accelerate speed-to-clinic.

Challenges in the antibody discovery process
Antibody discovery typically starts by either creating hybridomas by fusing a B-cell and a myeloma cell, or by utilizing technologies such as phage display. All these processes require several iterative screening runs to interrogate antibody libraries and identify leads. The leads are then characterized, optimized and taken to pre-clinical development.

Despite technological and methodological advances, there are several limitations still to be overcome. First, hybridoma generation is time-consuming, costly and inefficient, as this approach may result in missing the rare antigen-binding antibody with the desired characteristics. During hybridoma generation, many B-cells die-off in culture and do not undergo successful fusion and, as a consequence, the key antibody-secreting cells may be lost. Second, traditional approaches only interrogate a small subset of the cell repertoire. In the hybridoma scenario, a large part of the B-cell repertoire is lost during hybridoma generation, so only a fraction of the total population is screened, and in turn, the most potent candidate molecules may not be identified. Considering the population of target cells can be as low as 0.001% of the original ~40 million cells harvested, this is a significant challenge to address. The development of fast, accurate and flexible technologies that enable the deep interrogation of whole B-cell repertoires and hybridoma populations while preserving cell viability is therefore paramount to ensure that key target-specific antibody-secreting cells are not missed.

High–throughput screening (HTS) remains the most successful lead generation strategy; when applied to the development of therapeutic antibodies, it reduces the time required to identify and isolate the rare antigen-specific, antibody-secreting cells from a population of millions. This is typically done by screening the purified B-cells directly with flow cytometry, bypassing both hybridoma fusion and phage display. Flow cytometry has the advantage of being very high-throughput, and antibodies secreted by B-cells can potentially be screened using cold capture, a technique used to prevent the full secretion of antibodies by trapping them at the cell surface. However, this is a representation rather than direct measurement of the antibody secretion profile by a single cell. Additionally, flow cytometry can be harsh on the cells, especially primary cells, and may alter cell function and viability. Alternative screening methods include ELISA and Elispot; however, these techniques often need to be executed manually and, consequently, it becomes too costly and time-consuming to analyze large populations.

After multiple rounds of screening and selection, the positive cells must be sub-cloned into monoclonal populations (lead panels) by employing semi-automated methods like cell-in-well imagers and cell sorting; this multi-step approach adds even more complexity and hands on-time, slowing down the process even more.¹

Achieving improved process efficiency with microfluidics
The rise of microfluidics and, in particular, droplet microfluidics, has further advanced the field of HTS. Microfluidic systems conduct complex multi-step assays with high reliability, cost-efficiency and throughput in a picoliter volume water-in-oil emulsion droplet (picodroplet) format. Picodroplets bring the advantages of miniaturization and automation, thus helping to improve laboratory productivity by facilitating rapid, high-throughput research to interrogate bigger repertoires and find more functional properties.²

Automated single cell analysis platforms, such as Sphere Fluidics’ Cyto-Mine,³ enable the screening of millions of primary B-cells for secreted immunoglobulin, identifying and sorting the antigen-specific candidates, and then gently dispensing them into microtiter plates with visual proof of cell number through high-quality imaging. Encapsulation in picodroplets ensures that the concentration of the molecules secreted by the cell increases extremely quickly, and protects single cells from shear stress, thereby overcoming some of the major barriers of flow cytometry and fluorescence-activated cell sorting. Through facilitating the screening of isolated B-cells directly in these systems, no B-cell immortalization is required, which drastically cuts down the process cycle timelines and minimizes the possibility of losing cells in culture. Each step in the platform, from cell-encapsulation to incubation, to fluorescence detection of antibody secretion and dispensing, is typically controlled through a user-friendly computer interface, allowing for the continuous operation of the platform with minimal human intervention.

Further integration across the lab
To further advance antibody discovery research and accelerate speed-to-clinic, pharmaceutical companies are exploiting the synergy between different technologies. Platforms can be further enhanced through the integration of different automation technologies and software solutions on the same platform, such as the combination of fluidics technology with robotics systems for microplate handling, liquid handling and plate readers/scanners.

The developments in robotic integrations and AI-powered software bring many advantages. First, fully-integrated workflows reduce human intervention to minimize errors and enhance overall platform stability. Second, they eliminate bottlenecks and reduce the number of instruments needed to support scalability, so that multiple experiments can run at scale in a cost-efficient way. Third, they bring further efficiency by providing improved monitoring, data collection and analysis throughout the experimentation time. This, subsequently, requires more sophisticated data handling procedures and analysis solutions to quickly transform the acquired data into meaningful and biologically relevant insights.

The future of antibody discovery
Computational tools and AI are advancing the field of antibody discovery by enabling researchers to design and develop new antibodies using bioinformatic tools and by generating new data on the biophysical properties of specific antibody molecules. Machine learning algorithms will allow in silico prediction of the properties of certain antibodies and de novo antibody design. No wet work is involved until the molecule to test has been identified so, as the field develops, researchers can harness de novo drug design knowledge to effectively shorten timelines and reduce the experimental effort and cost of antibody discovery, as well as potentially lessen the burden of animal testing in research.⁴

De novo design is a state-of-the-art approach that involves using computational methods to specify the target and desired properties for an antibody. Algorithms detect meaningful patterns in databases to discover specific and selective binders computationally, to then make and test in prospective studies. Everything that is learned, successes and failures alike, is fed back into the algorithm to iteratively develop better processes.

Further advancements in the fields of biology, computation, AI and automation will continue to drive productivity in antibody discovery. As traditional multi-step antibody discovery approaches are streamlined with flexible HTS platforms and interfaced with robotic integrations and AI-powered computational tools, companies will increase efficiency and productivity, and, in the process, better serve patients and the wider community. 

References

1. Zhang, H., Wilson, I. A., and Lerner, R. A. (2012). Selection of antibodies that regulate phenotype from intracellular combinatorial antibody libraries. Proceedings of the National Academy of Sciences, 109(39), 15728-15733. doi:10.1073/pnas.1214275109.
2. Matuła, K., Rivello, F., and Huck, W. T. (2020). Droplet Microfluidics: Single‐Cell Analysis Using Droplet Microfluidics (Adv. Biosys. 1/2020). Advanced Biosystems, 4(1), 2070012. doi:10.1002/adbi.202070012.
3. Sphere Fluidics, Products, Cyto-Mine® [online] Available at: https://spherefluidics.com/products/cyto-mine/?v=79cba1185463 (accessed April 2020).
4. Schneider, G., and; Clark, D. E. (2019). Automated De Novo Drug Design: Are We Nearly There Yet? Angewandte Chemie, 131(32), 10906-10917. doi:10.1002/ange.201814681.