Nanoscale integration of single cell biologics discovery processes using optofluidic manipulation and monitoring

Abstract The new and rapid advancement in the complexity of biologics drug discovery has been driven by a deeper understanding of biological systems combined with innovative new therapeutic modalities, paving the way to breakthrough therapies for previously intractable diseases. These exciting times in biomedical innovation require the development of novel technologies to facilitate the sophisticated, multifaceted, high‐paced workflows necessary to support modern large molecule drug discovery. A high‐level aspiration is a true integration of “lab‐on‐a‐chip” methods that vastly miniaturize cellulmical experiments could transform the speed, cost, and success of multiple workstreams in biologics development. Several microscale bioprocess technologies have been established that incrementally address these needs, yet each is inflexibly designed for a very specific process thus limiting an integrated holistic application. A more fully integrated nanoscale approach that incorporates manipulation, culture, analytics, and traceable digital record keeping of thousands of single cells in a relevant nanoenvironment would be a transformative technology capable of keeping pace with today's rapid and complex drug discovery demands. The recent advent of optical manipulation of cells using light‐induced electrokinetics with micro‐ and nanoscale cell culture is poised to revolutionize both fundamental and applied biological research. In this review, we summarize the current state of the art for optical manipulation techniques and discuss emerging biological applications of this technology. In particular, we focus on promising prospects for drug discovery workflows, including antibody discovery, bioassay development, antibody engineering, and cell line development, which are enabled by the automation and industrialization of an integrated optoelectronic single‐cell manipulation and culture platform. Continued development of such platforms will be well positioned to overcome many of the challenges currently associated with fragmented, low‐throughput bioprocess workflows in biopharma and life science research.


| INTRODUCTION OF LIGHT-INDUCED ELECTROKINETICS AND IMPACT ON BIOLOGICS DISCOVERY
A new generation of techniques based on forces exerted by a light beam (known as optical manipulations) is enabling interactive biology at the cellular level, thus opening new opportunities in drug discovery. Optical manipulation-which enables highly selective and dynamic processes in micro-and nanoscopic systems-has proven to be a versatile and integrated technology throughout many scientific areas. This technology is based on light-induced electrokinetics that gives rise to designated forces on both solid and fluidic structures.
Since the discovery of the optical gradient and scattering forces in 1970 by Ashkin et al. (1970) a wide variety of optical manipulation methods have been developed including optical tweezers, plasmonbased optical trapping/plasmonic tweezers, and optoelectronic tweezers (OET).
Optical tweezers utilize radiation pressure and gradient force from a single laser beam, focused by a high numerical aperture microscope objective, to trap and manipulate micro-sized particles with forces at piconewton scales and nanometer range distances, (Figure 1a;Ashkin, 1970;1992;Ashkin, Dziedzic, Bjorkholm, & Chu, 1986;Grier, 2003). Through the use of holographic optical tweezers, the ability to manipulate multiple particles in parallel has been enabled and advanced by the Grier and Dufresne labs (Curtis, Koss, & Grier, 2002;Dufresne, Spalding, Dearing, Sheets, & Grier, 2001;Mejean, Schaefer, Millman, Forscher, & Dufresne, 2009;Polin, Ladavac, Lee, Roichman, & Grier, 2005). As a result, optical tweezers have become a primary methodology for a variety of physical, chemical, and biological experiments. In particular, their capability to achieve highly accurate measurements on spatial (sub-nanometer) and temporal (sub-millisecond) regimes has ranked them as one of the forefront single-molecule manipulation techniques with a wide range of applications (Fazal & Block, 2011;Moffitt, Chemla, Smith, & Bustamante, 2008;Neuman & Block, 2004;Neuman & Nagy, 2008;Pang & Gordon, 2011). Figure 1b shows an example of optical tweezers applied to the study of nanomechanical properties of double-stranded DNA (Fazal & Block, 2011). A thorough review of the theory and practice of optical tweezers across multiple size scales and applications has been presented by Polimeno et al. (2018).
To further extend nanoscale optical trapping and overcome the diffraction limitations on spatial confinement associated with optical tweezers, plasmonic optical tweezers (POT) were developed (Juan, Righini, & Quidant, 2011;Miao & Lin, 2007;Reece, 2008;Shoji & Tsuboi, 2014). This approach combines optical tweezers with nanostructured gold substrates resulting in localized surface plasmons ( Figure 1c). The tightly spaced surface plasmons generate a strong electric field enhancement and radiation pressure and enable stable optical trapping of particles at the nanoscale range. POTs facilitate efficient trapping of nanoparticles with laser intensities weaker than conventional optical tweezers (Figure 1d). In addition, the tailored design of the nanostructured substrates allows precise nanoscale control of the motion of nanoparticles.
Conventional and plasmonic optical tweezers have become powerful tools in biology allowing high-resolution experiments on trapped single cells. However, both are limited to a small manipulation area due to large optical intensity requirements generated via high-power lasers (at least 10 4 W/cm 2 ) 11 . In addition, the high optical power density causes cell damage over time and limits the duration of experiments. To overcome these limitations, a novel optical manipulation technique known as OETs was developed. OET enables massively parallelized optical manipulation of single cells via the utilization of optical beams to generate patterned virtual electrodes on a photoconductive material (Chiou, Ohta, & Wu, 2005;Hsu et al., 2010;M. C. Wu, 2011). OET operates within a much lower optical intensity (3 W/cm 2 ) over a much larger addressable area (up to 11 mm 2 as demonstrated on commercially-available OET platforms applied AC bias between the two electrode layers generates a localized electric-field gradient. Specifically, the electric-field gradient results from the many orders of magnitude increase in conductivity of the a-Si:H electrode layer when illuminated. This results in the creation of electron-hole pairs in the electrode layer causing the voltage to drop across the fluidic chamber. In turn, objects such as particles or cells, experience dielectrophoresis (DEP) force in the presence of the lightinduced electric-field gradient. The force F DEP applied to a particle with radius a is proportional to the electric field E applied (Zhang, Nikitina et al., 2018). The force can be positive (attractive) or negative (repulsive) depending on the relative values of the complex permittivity of the media m ϵ ⁎ and particle p ϵ ⁎ (Q. Chen & Yuan, 2019 Grier, 2003). (b) Example of a nucleic acid system studied using optical tweezers, showing nanomechanical properties of filaments subjected to twist: The relative extension of DNA is monitored as trapped DNA is twisted with an optical torque wrench. The coiled DNA undergoes a phase transition from a twisted to a plectonemic form approximately 0.14 supercoiling density (Fazal & Block, 2011). (c) Plasmonic tweezers schematic: Patterned gold nanopillars give rise to localized surface plasmons causing a strong field enhancement under the trapping beam and suppression of the Brownian motion (characteristic of POT) resulting in improved particle confinement (Reece, 2008). (d) Molecular manipulation via plasmonic tweezers: A single bovine serum albumin (BSA) protein is trapped in the gap of a double hole nanostructure (Pang & Gordon, 2011). (e) Schematic of an optoelectronic tweezer device consisting of: a photoconductive layer of hydrogenated amorphous silicon (a-Si:H) on an indium tin oxide (ITO) coated glass substrate (bottom layer), a liquid containing microparticles is sandwiched between the bottom layer and the top ITO-coated glass layer. An AC electrical signal between the top and bottom layers in combination with patterned illumination create a nonuniform electric field that results in particle manipulation via dielectrophoresis (DEP; M. C. Wu, 2011 C. Wu, 2011). The phototransistor has 500x higher conductivity than amorphous silicon when illuminated. This allows long-term culturing of cells and direct observation of the heterogeneity of doubling rates among the cells, bottom of Figure 1f . As shown in Figure 1f, high-resolution patterned illumination results in accurate single-cell encapsulating compartments, which can also define their motion. OETs overcome many limitations associated with other lightbased techniques for micro-particle manipulation, particularly in biological applications. First, OETs require a significantly low optical power (10 −1 W/cm 2 ) in comparison to conventional and plasmonic tweezers (10 4 -10 6 W/cm 2 ), making them noninvasive to achieve cell control and motion without compromising cell viability. Secondly, the low optical power requirements eliminate the necessity of a high numerical aperture objective to tightly focus a high-power laser, which otherwise limits the area of particle control and manipulation.
OETs significantly increase the manipulation area by two orders of magnitude in comparison to optical tweezers via the utilization of a 10x objective and a light-emitting diode (LED) as the illumination source, thus facilitating high-throughput processes (Chiou et al., 2005). Thirdly, the advancement from structural to high-resolution virtual electrodes enables massively parallel and dynamic single-cell manipulation.

| BIOLOGICAL APPLICATIONS OF OETs
The utility of OET for particle and cell manipulations has led to a wide variety of advanced OET-based devices and biological applications.
A proof-of-concept experiment demonstrated the unified platform's ability to form a high-density array of droplets and was applied in the biomedical application of viral detection (Figure 2c; Pei et al., 2015).
Another OET-based device integrated with microfluidic channels and chambers has been developed for high-throughput and high-selectivity electroporation of individual cells . A schematic of the device is shown in Figure 2d: lithographically defined channels were integrated with the OET device enabling light-induced electroporation, maintenance of viable cell cultures, and perfusion of different soluble reagents. Electroporation was performed on HeLa cells using the membrane impermeant dye, propidium iodide (PI; Valley et al., 2009).
Initially, low OET bias (0.2 kV/cm) was used to position individual cells in specified locations, followed by application of high electroporation bias (1.5 kV/cm) to selected cells resulting in the intracellular delivery of the PI dye ( Figure 2e). In addition to electroporation, single cell lysis using OET has also been demonstrated Witte et al., 2014). Lastly, a device based on a novel concept, Self-Locking Optoelectronic Tweezers (SLOT), has shown promising results in scaling up single-cell manipulation across a significantly larger area (Y. Yang et al., 2016). The device schematic, Figure 2f, shows a prototype array of ring-shaped phototransistors that control particle/ cell trapping. A unique feature of the SLOT platform is the Al 2 O 3 (an insulating dielectric layer) coating the surface to partially drop the voltage, enabling the single-cell self-locking function in high conductivity media. A device consisting of 250,000 phototransistor traps over a 1-cm 2 area has been shown to enable simultaneous trapping of over 100,000 polystyrene beads. Furthermore, the device enables the selective release of trapped particle via a scanning light beam as shown in Figure 2g, resulting in the formation of four letters standing for UCLA (Y. Yang et al., 2016). Another recent improvement called patterned optoelectronic tweezers enables more flexible particle manipulation by exposing the electrode layer in specific patterns that will hold particles after the light source is removed (Zhang, Shakiba et al., 2018).
Additional methods to retain particles after light removal include the use of microwells to observe the interactions between cancer cells and immune cells (Ke et al., 2017). The schematic diagram of the four-leafclover-shaped chip, Figure  The advanced OET-based devices described above offer a wide variety of potential applications, including single-cell studies involving multiplexed environmental stimuli, high-throughput, and high-resolution genetic transfection, study of cell-to-cell signaling, tissue engineering, in vitro fertilization, immunotherapy, and beyond.
However, these platforms have been limited to proof-of-concept experiments at the bench level.

| INDUSTRIALIZATION OF PARALLELIZED SINGLE-CELL MANIPULATIONS
The broad potential applications of OET have led to the development of an industrial platform, Beacon®, which incorporates opto-electropositioning (OEP) technology, a variation on the OET technology, into an automated standalone system shown in Figure 3a. The system can be broken up into three key components: hardware, software, and consumables (chips and reagents). The hardware itself is approximately the size of a large refrigerator and requires power (120 V) and gas lines (CO 2 , house air, and optional O 2 ) to operate, as well as an optional Ethernet port. An optical module consisting of an epifluorescent microscope with LED illumination and a custom digital mirror display for creating patterned illumination is situated above a motorized stage in the "sample bay" (Figure 3a,b). This three-axis motorized stage holds four individual "nests" for housing up to four separate "chips." Each nest has independently controlled temperature (15-40°C), using liquid-cooled thermoelectric devices, as well as independent fluidic lines to control fluid exchange. Each nest has a dedicated syringe pump, located in a separate "reagent bay." Each chip also has a "needle" that is independently actuated in the z-direction for both input and output of media or reagents. A fifth syringe pump, also located in the reagent bay, is connected to a fifth input/output needle and can be used as a liquid handling robot. All of the input/output needles are situated above two "well plate incubators"  F I G U R E 2 OET Bench-scale applications | (a) Schematic of serial particle concentration: A light pattern is swept across the device concentrating beads at one end of the droplet via OET, followed by droplet splitting via OEW. Reiteration of this process results in an exponential increase in particle concentration (Valley et al., 2011).   (Wrammert et al., 2008;X. Wu et al., 2010). Mettler-Izquierdo et al. As the Beacon technology matures, it will undoubtedly offer unique opportunities to transform antibody discovery workflows for both reagent as well as therapeutic applications.

| OPTIMIZATION OF LARGE MOLECULES
The development of large molecule therapeutics with optimal potency, selectivity, and biophysical properties often requires the F I G U R E 4 Primary ASC based antibody discovery | (a) Workflow overview: Immunization schedules designed to enhance the relative frequency of ASCs in specific compartments. Specific organ compartments were harvested, and ASCs enriched using magnetic bead negative selection followed by a multicolor FACS sorting strategy. Iterative multiplex screens were performed on the OEP platform. Selected hits were exported, the VH and VL sequences recovered via multiplex PCR, and the IgGs recombinantly expressed for further characterization. generation and screening of many engineered antibody variants (Chiu & Gilliland, 2016). Depending on the properties desired in the final molecule, these engineering designs may target affinity or selectivity modulation (Kiyoshi et al., 2014;Sellmann et al., 2016;Tiller et al., 2017), reduction in immunogenicity (Presta, 2006), better pharmacokinetics (Haraya, Tachibana, & Igawa, 2019;Hotzel et al., 2012), removal of chemically labile sites (Chelius, Rehder, & Bondarenko, 2005;DiCara et al., 2018;Haberger et al., 2014), and/ or improvement in other biophysical properties related to manufacturing (Jarasch et al., 2015;Seeliger et al., 2015;Xu et al., 2013).
Standard methods of screening panels of molecules that have been engineered for multi-parameter optimization require cloning, expression, and purification of a sufficient quantity of protein to run multiple assays. This workflow is labor-intensive and limits the number of designs that can be tested in a reasonable amount of time (Barnard, Hougland, & Rajendra, 2015;Bos et al., 2014;Estes et al., 2015;Schmitz et al., 2019;Winters, Chu, & Walker, 2015;X. Yang et al., 2013;Yoo, Provchy, Park, Schulz, & Walker, 2014).
To address these limitations, we explored the development of a

| DEVELOPMENT OF A QUANTITATIVE SINGLE-CELL ASSAY
To further expand the utility of the Beacon platform, we explored a variety of library formats, with yeast display as a primary focus due to its wide versatility to perform protein and peptide engineering (Cherf & Cochran, 2015). Yeast display has been used to discover de novo binders (McMahon et al., 2018;Wang et al., 2016), affinity mature existing binders (Tiller et al., 2017), engineer in pH-sensitivity (Schroter et al., 2015), improve protein thermal stability (Jones, Tsai, & Cochran, 2011), and enhance enzyme kinetics (I. Chen, Dorr, & Liu, 2011). Our work used yeast-displayed peptides isolated from an affinity maturation campaign against a receptor ECD to develop a binding affinity assay on the platform. The precise cell manipulation on the Beacon via OEP facilitated the development of a single-cell assay even for small-sized cells such as yeast (diameter 5 microns). The binding affinity assay is schematically summarized in Figure 6a.  Homozygously edited cell lines should not express the marker and therefore a fluorescent assay using anti-CXCR4 antibody was used to identify nonfluorescent clones for selection. Selected clones were exported with a process that splits each clone into two separate exports, one for sequencing and one for proliferation. Expected edits were confirmed by genomic DNA sequencing in some of the selected clones, demonstrating the ability to quickly identify edited clones of interest while simultaneously keeping a live culture ready for scaleup (Mocciaro et al., 2018).

| DEVELOPMENT OF CELL LINES FOR PRODUCTION
The generation of a highly productive manufacturing cell line is a key and important step in the development process for large molecule therapeutics (Estes & Melville, 2014;Fischer, Handrick, & Otte, 2015). To select a cell line that will give high productivity, long-term stability, and consistent product and process consistency, subcloning and extensive screening is required on hundreds of clonal cell lines, measuring growth, productivity and product quality in multiple screening assays (Wurm, 2004). In addition, regulatory agencies require that clinical trials be initiated with material from clonally derived cell lines and banks, so documentation that the cell line is derived from a single cell is required to provide that assurance of clonality (Tharmalingam et al., 2018).
This method has a high rate of error and requires multiple rounds of cloning to satisfy clonality assurance required by regulatory agencies. established. This approach also only measures proteins levels at a single point in time, but protein synthesis, modification, and secretion are a highly dynamic process that is well known to change during the cell cycle (Alber & Suter, 2019). As a consequence of low recovery F I G U R E 7 Comparison of a microtiter plate based cloning workflow vs. a nanofluidic chip subcloning workflow. A depiction of the steps involved in performing a clonal isolation and expansion workflow using two approaches. Differences are highlighted in boxes for FACS-based workflow(solid) and platform workflow (dotted). Standard subcloning operation: A heterogeneous population is isolated and deposited into microtiter plates using FACS-based cell sorting, followed with high-quality, high-throughput whole well imaging to verify a single cell in a well. After growth and repeated imaging, colonies are picked and consolidated using automation liquid handlers. Top clones are then screened in a bioreactor to select the final clone. In contrast, the platform workflow enables single cell isolation, growth assessment, and high-throughput screen on the chip (dotted box) in the nanofluidic workflow, and only those clones that meet the desired criteria are exported and expanded for further evaluation (Le et al., 2018) [Color figure can be viewed at wileyonlinelibrary.com] and poor-quality screening, a typical cloning process will use 50 to 100 or more plates to isolate and screen hundreds to thousands of clones of single cell origin to find a suitable manufacturing line.  (Le et al., 2018). In a head to head comparison with a FACS-enabled microtiter plate-based workflow, we were able to generation of comparable clonal cell lines with reduced resources, summarized in Figure 7. Over all recoveries of the clonal CHO cell lines were shown to be higher than other methods, in part due to the improved conditions nanoscale culture offers. A rich data set is generated to support clonality, tracking, and population understandings to enable early decisions and identification of highly performing cell lines. Single-cell data obtained from cell lines provide insights into defining population characteristics of the production cell lines that are not currently possible with the traditional FACS-based methods, offering the potential to make improved cell line choices and better predict downstream success (Le et al., 2018).

| CONCLUSION
Innovation in basic science has delivered some outstanding advancements to biomedical research over the last 40 years including transgenic mice, PCR, genetic sequencing, gene editing, and many others. Interestingly, most of these innovations were realized using relatively simplistic cell and liquid handling technologies invented in the 1950s: pipettes and well plates. In many cases, the pace of applied science and technology development has been outpaced by basic scientific progress. The traditional approach taken by those in the field has been to utilize straight-forward technologies at the microscale level. These microscale approaches reach practical limits very quickly, as the complexity of rooms filled with single-function equipment and automated robots still struggle with recent advances in research methodology, particularly on individual cells at the nanoscale fluidic regime. The promise of "lab on a chip" technology has been slow to mature towards industrial applications, but the promise remains the same: miniaturization of basic cellular manipulations should lead to faster and more efficient discovery, requiring less reagent and effort due to enhanced sensitivities. The platform technology discussed here, Beacon, overcomes such limitations through the capability to maintain physiologically-relevant culture environments of thousands of segregated cells while performing numerous types of very sensitive assays all under reproducible computer control, otherwise known as "digital cell biology". Importantly, this ability to flexibly string together multiple processes and tests (manipulate, grow, assay, interrogate, select) on a single system allows for full biological workflows to be performed with significantly minimized resources.
Digital cell biology on the platform creates the opportunity to transition many of today's cell-based well plate assays to a unique new format, unlocking a step function increase in workflow scale and speed. Complex and time-consuming assays, like the measurement of growth and IgG secretion rate on thousands of clones, have proven to be straight-forward to transfer onto the platform's nanofluidic environment, making them routine to run on a day to day basis. We envision many more complex assays being reduced to routine software-controlled workflows, such as on-chip functional assays Combining the power of machine learning and AI with digital cell biology offers additional power to unlock deep insights into mechanistic biology that has thus far proven elusive at the macro scale. This technology can be applied to many areas within biopharma including antibody discovery, assay development, antibody engineering, and cell line development. With the ability to select the right cells, the technology can also be applied in the future to cell therapy manufacturing as well as workflows within synthetic biology, and diagnostics. We believe digital cell biology platforms will prove to be transformative to multiple industries that require a dramatic step forward in speed and scale to realize the next wave of breakthroughs.