The high speed quantitative high content analysis of cells also makes flow cytometry an attractive technology for drug discovery applications and it is used at many stages of this process. Recently it is also finding a niche in drug screening laboratories sharing bench space with other high throughput technologies. Flow cytometry is becoming an ideal tool, particularly in an environment where primary cell-based assays are increasingly being deployed to monitor drug responses.
low cytometry is a microscopical technique dating back to the 1930s1 and instruments have been commercially available since the 1970s. It was developed to allow high-speed quantitative analysis of cells and other particles. Cells suspended in a liquid are passed through a focused laser beam to generate optical signals, such as light scattering and fluorescence. These signals are typically processed in real time.

Large numbers of particles are analysed in a short period of time (1,000- 50,000/s) to provide statistically robust information about the cell population. Although single cells are the particles of interest, flow cytometry can be used for examining other types of particle, such as chromosomes, synthetic microspheres and other insoluble particulates2-4. Flow cytometry has long been recognised for its multiparameter capability (12 parameters or more is common practice) as well as high speed analysis and sorting capabilities (>30,000 cells per second). In addition, its multiparameter capabilities allow it to distinguish tens of cell subsets in multiplex format.

Plate-based flow cytometry
Multi-parametric analysis enables an extensive investigation of the complex interrelated mechanisms of drug action in cell-based systems. The ability to make high-content measurements has made flow cytometry an important tool used for drug discovery. It is used at every stage of the drug discovery cycle both in Pharma and Biotech companies, including target identification and validation, hit identification, lead and candidate selection and safety studies. In vitro and ex vivo flow cytometry methods are routinely employed in toxicology studies assisting in identifying and characterising off-target effects at the single cell level5. Flow cytometry is used during clinical testing in order to assess both safety (eg anti-drug antibody testing) and pharmacokinetic assessments to monitor the plasma levels of protein and peptide based therapeutics6.

While flow cytometry has found many applications in several stages of drug discovery, its routine widespread use for high throughput drug screening has been somewhat limited. This is unlike the analogous technology – automated high content imaging (HCI), which in terms of drug screening is an established technology common in screening laboratories7. The failure of flow cytometry to keep pace with HCI is largely down to low individual sample throughput on commercially available instruments. Most flow cytometers use a single tube sampling mechanism and are largely used for immune phenotyping of single (patient) clinical samples8. Automated (bar-coded) tube sampling using carousels/racks, with capacities of up to 40 tubes slightly improves the throughput of sample tubes. But the arrival of microwell plate-based sampling flow cytometers are key to usage in more routine applications of drug discovery. This also allows the use of further integrated robotics, peripheral automation and liquid handling automation associated with a typical pharma drug screening laboratory.

The increasing use of phenotypic cell-based assays in drug discovery9 has led to a drive to find ways of increasing flow cytometry throughput. At GSK we worked with Beckman Coulter to automate an FC500 MPL flow cytometer (Figure 1). The flow cytometer was integrated with an automated refrigerated plate stacker and automatic transfer system. This configuration has been successfully utilised for target-based drug discovery where several flow cytometry-based mechanistic and functional assays were used to profile hundreds of compounds in concentration-response profiling. Table 1 lists some flow cytometry assays that we used for drug screening using an automated FC500. These assays were routine and run weekly over long periods of time during lead optimisation campaigns. Such assays would be impossible to prosecute without appropriate facilities, peripheral automation and liquid handling. This included an on-site blood donation unit, allowing access and delivery of fresh human blood and the ability to handle it effectively and safely.

One assay run routinely was used to identify antagonists of actin polymerisation in granulocytes (Figure 2). Inhibition of cell migration is the rationale behind a number of anti-inflammatory drug discovery programmes, with neutrophil chemotaxis assays often used as an assay to monitor cellular drug activity. Actin polymerisation can be used as a surrogate measure of cell chemotaxis10, and monitoring the inhibition of granulocyte actin polymerisation in human whole blood offers considerable time and cost savings over conventional cell migration assays. Additionally, automation is a key factor in the measurement of actin polymerisation. This is because actin polymerisation occurs within seconds following stimulation10 and in this assay was optimal after 45 seconds of stimulation with IL-8. Scheduling software on a Biomek FX automated liquid handler was crucial to ensure reproducibility and concomitant increased throughput.

High Throughput Flow Cytometry (HTFC)
While the use of automated plate-based flow cytometry has been successfully applied to smaller target-based assays, it has not until recently entered high throughput drug screening laboratories, where typically 100,000s of samples are run in low volume microplate-based assays. Flow cytometry, in the past, has been severely limited in throughput rates of individual discrete samples, with sample times of several minutes per well. Typical 96-well microwell plate read times obtained with the Beckman Coulter FC500 automated system (Figure 1) were at least 60 minutes or longer per plate depending on the assay. Some instruments, such as Becton Dickinson’s LSRII or Fortessa with the high throughput sampling attachment (HTS), have a slightly higher sampling rate, but collectively this low plate processing time was a significant limitation and could not meet the size and scale of assays run in drug screening laboratories. Even testing small (1,000s) compoundfocused screening libraries was a challenge.

The main features of conventional flow cytometry that hamper their ability to handle samples rapidly is a combination of data processing and the mechanics concerned with running one sample to the next. There are significant delays relating to saving individual sample associated files, and in the cell suspension sampling mechanism, that often involves tube priming and flushing11. Two successive generations of high throughput flow cytometry sample handling technology have evolved to address these issues. Both were developed at the University of New Mexico11. The first of these is Plug flow cytometry. A flow injection analysis approach is used in which individual sample suspensions are sequentially inserted as plugs of precisely defined volumes into a flowing fluid, which delivers them into the flow cytometer12. The second generation technology, now commercialised by Intellicyt is the HyperCyt. This uses a peristaltic pump in combination with an auto-sampler (Figure 3). In contrast to the typical sample handling mechanism, the HyperCyt approach is to continuously deliver the entire sample, eg 96 wells, with each sample (well) separated by air bubbles (Figure 3). The data from all the samples in the plate are acquired and stored in a single file. A high resolution time parameter is also recorded during data acquisition. Temporal gaps in cell detection are created in the data stream by the passage of the air bubbles, allowing the individual cell suspensions to be easily distinguished and separately evaluated when measured in conjunction with the time parameter. Cell suspension volumes as low as 1 to 2ul can be sampled depending on the cell concentration. The HyperCyt is compatible with a number of different flow cytometers and the system can be automated with plate loaders. We took the approach of integrating the system with a Beckman Coulter Biomek NX. We found that this was more suitable than a conventional robotic arm as it allowed us to perform just-in-time additions to plates as well as providing extra lab liquid handling capacity (Figure 4).

The HyperCyt can sample both 96 and 384 well plates in less than four and 12 minutes respectively. The plate sampling time is highly dependent on the cell concentration. Plate read times as low as four minutes can be obtained when the cell number is not limiting, for example, when using immortalised cells. However, sampling times are higher when analysing rarer sub populations of cells from more complex heterogeneous populations, such as those found in PBMCs.

One example where we have deployed the HyperCyt is for the routine analysis of GRO stimulated CD11b upregulation in neutrophil populations in whole human blood (Figure 5). CD11b is expressed on the surface of many leucocytes including granulocytes. Functionally it regulates leukocyte adhesion and migration to mediate the inflammatory response13 and its upregulation in granulocytes can be used as a surrogate marker for chemotaxis, similar to actin polymerisation. The assay only requires 10ul of whole blood per well in a 96-well plate, reducing blood usage and avoiding costly neutrophil purification procedures. Plates are stacked on the Biomek deck and automatically transferred on to the HyperCyt. Typical plate read time is less than 10 minutes using a HyperCyt, whereas at least 60 minutes is required using the FC500 system. The multiparametric nature of flow cytometry allows the detection of CD11b directly on neutrophils using a combination of light scattering and CD16 surface maker labelling. Other invaluable information is also captured, such as cell number and morphology giving an indication of potential compound toxicity.

We have also been exploring the applications of HTFC for antibody screening. Traditionally, antibody screening for cell surface target antigens often consists of several sequential steps. Each step involves different tests for binding and specificity followed by cell-based assays. Screening hybridoma supernatants for specific antibodies that bind cell-based antigen is a critical component of monoclonal antibody generation and often a bottleneck in the process. Binding of monoclonal antibodies to whole cells expressing target protein is preferred over solid phase methods such as standard ELISAbased technologies. It is highly advantageous if cell-based assays can be performed at the primary screening stage in the monoclonal antibody selection process. This is because in cell-based assays ligands remain in their natural confirmation. The FMAT (Fluorometric Microvolume Assay Technology) is recognised as a gold standard platform for primary therapeutic antibody screening using whole cells expressing the target protein of interest14. Since thousands of clones are routinely screened, flow cytometry was not viewed as a practical option for primary high throughput hybridoma screening. However, flow cytometry offers several advantages over the FMAT such as sensitivity and its ability to measure several parameters simultaneously. The simple use of flow cytometry light scattering allows the discrimination of dead cells from test populations and thus reduces the numbers of false positives, a constant issue we identify with the FMAT. Previously throughput has limited flow cytometry use to secondary confirmation screens. Using the HyperCyt HTFC screening for therapeutic antibodies is now possible and at the same time offers all of the full armoury associated with flow cytometry providing a more robust and sensitive screening platform.

During hybridoma screening it is usual to test specific binding of secreted antibodies from each hybridoma supernatant to a positive human target cell (eg, a transient cell line expressing the target protein of interest), a negative expressing cell (expressing a similar related protein) and one or two orthologue species expression the protein of interest. This would normally result in screening hybridoma supernatants several times in separate assays to test for binding. Using flow cytometry all the different cell types can be mixed together and run as one, thus increasing the throughput of the assay, making significant time savings and saving costly reagents. This is accomplished by using a technique known as fluorescent cell bar coding15, where different populations of cells can be identified using a different fluorescent signature. For example, each cell population can be labelled with a different concentration of the fluorogenic substrate calcein AM. The cells are then mixed and aliquoted into microtitre plate wells. Test antibody supernatants plus controls are added to the mixture and then analysed using HTFC. Each population can be differentiated based on their fluorescent intensity, and the corresponding antibody binding to each cell type can be quantified simultaneously (Figure 6).

Discussion and future perspectives
Traditional flow cytometry has an important role throughout the drug development cycle and is represented at each phase in some capacity. Often flow cytometry is the preferred technology to use for drug screening when running phenotypic assays, but throughput limitations have led to the development of less desirable recombinant assays. The introduction of plate-based sampling on flow cytometers began to realise the potential of flow cytometry as a drug screening platform. The introduction of faster sampling technologies, such as the HyperCyt and the use of peripheral automation and liquid handling has transformed flow cytometry as a drug screening tool. HTFC is still at ‘the lower end’ of the capacity scales in a typical drug screening laboratory where one to two million compounds can be profiled in just a few weeks, but further improvements are in progress (eg, 1536 well sampling). Other important, platform independent, factors require consideration, such as assay costs and cell supply, particularly the availability of primary human cells.

Despite the significant advances in sample throughput, managing and interpreting complex multi-parametric data remains a challenge. Traditional flow cytometry data analysis tools are not designed to meet this problem, as the typical software package is a complex tool designed for universal use, not to handle high capacity plates, eg, 384 or more wells. Even with 96-well plates, most flow-cytometry analysis approaches are inefficient. Better computational analysis tools are required to handle the quantity and complexity of the data. Combining the increased quantity of data from simultaneously measured biomarkers with data collected following cell population perturbation after exposure to drugs or other factors (growth factors, etc) may require a multifactorial approach to data analysis, where complex response patterns are generated from heterogeneous populations. Speciality custom-tailored informatics solutions designed with flow cytometry in mind will be required, such as the use of Cytobank and advanced visualisation tools such as spanning-tree progression analysis of density-normalised events (SPADE)16 or PlateAnalyzer17.

The authors would like to acknowledge Poonam Shah and Metul Patel for their scientific contributions and Stuart Baddeley (Director, Screening & Compound Profiling, UK) for expert opinion and input.


Dr Steve Ludbrook is currently a Section Head in Screening & Compound Profiling, Platform Technologies and Science at GlaxoSmithKline in Stevenage, UK. He has worked for more than 15 years in various areas in the target validation through to candidate selection phase, with a current focus on the use of phenotypic assays in drug discovery. Dr Rob Jepras is an Investigator in the Screening and Compound Profiling Department at GlaxoSmithKline in Stevenage, UK. He has more than 20 years’ experience at GSK, with a particular focus on high content technologies and primary cell biology in drug discovery. 

1 Moldovan, A. Photo-electric technique for the counting of microscopical cells. Science 80: 188-189, 1934.

2 Brind’Amour, J and Lansdorp, PM. Analysis of repetitive DNA in chromosomes by flow cytometry. Nature Methods, 8: 484-486, 2011.

3 Fattaccioli, J, Baudry, J, Emerard, J-D, Bertrand, E, Goubault, C, Henryb, CN and Bibettea, J. Size and fluorescence measurements of individual droplets by flow cytometry. Soft Matter, 5: 2232–2238, 2009.

4 Morgan, E, Varro R, Sepulveda, H, Ember, JA, Apgar, J, Wilson, J, Lowe, L, Chen, R, Shivraj, L, Agadir, A, Campos, R, Ernst, D and Gaur, A. Cytometric bead array: a multiplexed assay platform with applications in various areas of biology. Clin Immunol, 110(3): 252-66, 2004.

5 Litwin, V (Editor) and Marder, P (Editor). Flow Cytometry in Drug Discovery and Development, Wiley, 2010.

6 Ramanathan, M. Flow Cytometry Applications in Pharmacodynamics and Drug Delivery. Pharm Res, 14(9):1106-1114, 1997.

7 Taylor, DL, Haskins, JR and Giuliano, KA, Ed. Humuna Press. High content screening: A powerful approach to systems cell biology and drug discovery. Method in Molecular Biology, Vol. 356.

8 Stewart, CC and Nicholson, KJA (ed). Wiley-Liss, Immunophenotyping, 2000.

9 Feng, Y, Mitchison, TJ, Bender, A, Young, DW and Tallaricon. JA. Multi-parameter phenotypic profiling: using cellular effects to characterize small-molecule compounds. Nature reviews, Drug Discovery, (8): 566-578, 2009.

10 Howard, T and Meyer, M, Chemotactic peptide modulation of actin assembly and locomotion in neutrophils. The Journal of Cell Biology, 98: 1265-1271, 1984.

11 Edwards, BS, Yound, SM, Saunders, MJ, Bologa, C, Oprea, TI, Ye, RD, Prossnitz, ER, Graves, SW and Sklar, LA (2007). High-Throughput Flow cytometry for drug discovery, Expert Opinion In drug Discovery 2(5), 1-12.

12 Edwards, BS, Kuckuck, F and Sklar, LA. Plug flow cytometry. An automated coupling device for rapid sequential flow cytometric sample analysis. Cytometry, 39: 156-159, 2001.

13 Solovjov, D, Pluskota, Plow, E. Distinct roles for the alpha and beta subunits in the functions of integrin alphaMbeta2, J Biol Chem, 280 (2): 1336-45, 2005.

14 Lee, R, Tran, M, Nocerini, M and Liang, M. A highthroughput hybridoma selection method using fluorometric microvolume assay technology. Journal of Biomolecular Screening 13(3): 210-217, 2008.

15 Krutzik, PO and Nolan, GP. Fluorescent cell barcoding in flow cytometry allows high-throughput drug screening and signaling profiling, Nature Methods, 3: 361- 368, 2006.

16 Qiu, P, Simonds, EF, Sean, C, Bendall, SC, Gibbs, KC Jr, Bruggner, RV, Linderman, MD, Sachs, KS, Nolan, GP and Plevritis, SK. Extracting a cellular hierarchy from highdimensional cytometry data with SPADE. Nature Biotechnology, 29: 886-891, 2011.

17 Robinson, PR, Rajwa, B, Patsekin, V and Davisson, VJ. Computational analysis of high throughput flow cytometry data. J Expert Opin Drug Discov, 7(8): 679-93. 2012.

18 Traves, SL, Smith, SJ, Barnes, PJ and Donnelly, LE. Specific CXC but not CC chemokines cause elevated monocyte migration in COPD: a role for CXCR. Journal of Leukocyte Biology, 76 (2): 441-500, 2004.