When Robots Are Good: Fully Automated Thermo LAS Robotic Assay System with Dual FLIPRTETRA and TAP SelecT Robotic Cell Culture System

TAP SelecT.

The Tap SelecT automated cell culture system (Fig. 1) has a capacity of 182 T-175 cm² flasks. In the example shown, cells harvested from five flasks were pooled together, counted, and plated. Cedex flow cytometer is used to determine density and viability of harvested cells based on trypan blue exclusion. SelecT will support a well-defined protocol of cell culture and plating with uniform precision and consistent conditions for all flasks and plates produced during the multiple weeks of a screening campaign. The flow chart (Fig. 2) illustrates the schedule for handling vials with cryopreserved cells, tissue culture flasks, and finally assay plates 7 days a week, including unattended operations on weekends, to support 5 days a week of screening. The incubator racks of plated cells produced by the SelecT are interchangeable with incubator racks used on the Thermo LAS robotic assay systems and are quickly transferred in a manual operation on the day of use.

Thermo LAS Robotic System with Dual FLIPR^TETRA.

Thermo LAS Integration of Dual FLIPR ^TETRA . Thermo LAS performed custom integration of several laboratory instruments essential for handling cell-based fluorescent assays with kinetic image-based analysis of response (Fig. 4). We otiose to use MDC FLIPR^TETRA as a kinetic plate reader. The FLIPR^TETRA uses an LED light source to excite fluorescence rather than the argon-ion laser-based technology used in previous FLIPR models resulting in reduced footprint and relaxed facility/utility requirements. A 25-position centrally located, high-speed movable belt (the LPT, linear plate transport) moves plates from one part of the system to another, and eight flip mover robotic arms move plates from the LPT to peripheral instruments. The vertical array loader (VAL), an articulated robotic arm moves plates from the plate carousels (320 plates capacity each) to the LPT. Plates can be incubated at variable temperatures in two randomly accessed Stericult incubators (147-plate capacity) with plate elevators that permitted them to be positioned under the table, freeing more table space for other peripheral equipment. A Biotek plate washer Elx 405, one Perkin-Elmer Flexdrop IV noncontact dispenser, and Perkin-Flmer Evolution EP3 pipettor were integrated into the system to dispense and deliver liquids into assay and compound plates.

OPEN IN VIEWER

Figure 3.

GPCR-induced Ca-flux detection assay. (A) Schematic illustration of GRCR-induced changes in the concentration of intracellular Ca²⁺. Ligand-dependent activation of G-protein coupled receptors triggers a cascade of events: translocation of the alpha subunit of G-pro-tein to phospholipase C; activated PLC converts phosphatidylinositol bisphosphate (PIP2) into diacylglycerol (DAG) and inositol triphosphate (IP3); free cytoplasmic IP3 binds to the IP3-gated Ca-channels in the endoplasmic reticulum membrane; and Ca²⁺ enters the cytosol. The dephosphorylation of IP3 and pumping of cytosolic Ca²⁺ out of the cell rapidly terminate the response. (B) Trace of the signal detected in one we II by FLIPR^TETRA. The increase in the cytosolic Ca²⁺ could be detected using calcium-sensitive dyes (as Fluo-4 and its derivatives) loaded into cells in the nondisruptive manner. FLIPR detects the increase in fluorescence quantum yield upon binding of calcium to dye.

Polara. Polara is the Thermo LAS scheduling and control software. The Polara scheduler optimizes the timing of each individual step to maximize overall throughput and uniformity of parallel plate processing. To develop a schedule that has consistent timing for each plate in the run, the stagger time between samples is adjusted so that the standard deviation of the total run time to process each sample is minimized. Typical limit of sample time deviations was about 10 s or less. Polara also controls runtime communication for all elements of the automated system during assay (Fig. 5). It faciliates conversion of a multistep protocol into a detailed schedule for each plate of a 100-plate batch.

Cell Culture Protocols.

Manual Cell Culture. Gq-protein coupled receptor overexpressed in CHO-Dux cell line was used for Ca-flux assay (provided by Dr. R. Ring, Wyeth Research, US). Cells were continuously cultured in DMFM (GIBCO), 10% dialyzed FBS (HiClone), 1% nonessential amino acids (Cellgro), and 1% penicillin/streptomycin (Cellgro) in 37 °C 5% CO₂ incubators. Then 95% confluent cells were dissociated from the flask with 0.05% trypsin/ FDTA (Cellgro) and suspended at 5 × F5 cells/ml. Cells were plated in 384-well plates (BD/Falcon) at a density of 20,000 cells per well and incubated for 12-16 h in 37 °C incubator with 5% CO₂ prior to assay.

Automated Cell Culture with TAP Select. As described in legend to Fig. 2, a vial of 6 × 10⁷ cryopreserved cells was rapidly thawed in a 37 °C water bath and resuspended in DMFM (GIBCO) supplemented with 10% dialyzed FBS (HiClone), 1% no-essential amino acids (Cellgro), and 1% penicillin/ streptomycin (Cellgro). Cells were washed in the culturing media and resuspended in 40 ml. A total of eight T-175 flasks were seeded and imported into the SelecT system. Flasks were incubated in SelecT for 3 days, split 1-5, and returned for 3 more days of incubation. Cells were grown to 90-95% confluence. Cells harvested from five flasks were pooled together by the SelecT, counted in Cedex, and plated by multifunctional cell dispenser (MFCD) at a density of 20,000 cells per well into 384-well plates (BD/Falcon). Up to 120 plates were plated per day. In 12-16 h, racks with plates were moved from SelecT plate-incubator into Stericult (Forma scientific) incubator integrated in the Thermo LAS/FLIPR^TETRA system.

FLIPR-based Ca-flux Detection Assay.

Assay Protocol to Validate FLIPR^TETRA. Ca-3 dye kit (MDC, Sunnyvale, CA) was used to detect changes in intracellular Ca²⁺ upon ligand activation of the Gq-coupled receptor (Fig. 3). CHO-Kl cell line with recombinant muscarinic receptor 1 was generously provided by Julia Heinrich (Wyeth Research, US) and used for the validation of new FLIPR^TETRA by comparison to traditional FLIPR-3. A general description of the FLIPR assay has been previously published. ³ For the validation experiments, 14,000 cells were plated in DMFM medium, supplemented with 10% dialyzed FBS (HiClone), 1% nonessential amino acids (Cellgro), and 1% penicillin/streptomycin (Cellgro) into the wells of 384 clear bottom black plates (Falcon, BD) 18-20 h before assay. On the day of assay, Packard PlateTrak aspirated media from the wells and dispensed 50 ul of diluted Ca-3 dye. One vial of express kit reagent was dissolved in 500 ml of HBSS buffer supplemented with 10 mM Hepes, pH 7.5, and 2.5 mM Probenecid. Probenecid was used to suppress pumping of the dye out from the cytosol. It was made fresh as a 500 mM stock solution in 1 N NaOH. Cells were incubated with the dye for 60 min at 37 °C in 5% CO₂.

Plates were moved into FLIPR where the agonist was dispensed according to the following description of the FLIPR protocols:

OPEN IN VIEWER

	FLIPR-3	FLIPRTETRA
Intensity	0.600 W	50%
Gain	40	80
Exposure (s)	0.4	0.4
Dispense volume (μl)	12	12
Height of dispense (μl)	40	35
Speed of dispense (μl/s)	33	25

Ml receptor dose-response to acetylcholine was used to validate FLIPR^TETRA by comparison of results to FLIPR-3. Data were exported as Max—Min value of change in raw fluorescent units. Fold induction was calculated as value of Max—Min for sample wells normalized to the average value of background signal in buffer-treated wells on the same plate (N = 24). The EC50 calculated using Graph Pad Prizm Project was similar across the two platforms and equal to 60 nM.

Protocol 1. Fluorescence Measurements Using Stacker-based FLIPR-3. One vial of Ca-3 express kit was dissolved in 500 ml of HBSS buffer supplemented with 10 mM Hepes, pH 7.5, and 2.5 mM Probenecid. Probenecid was made fresh as a 500 mM stock solution in 1 N NaOH. PlateTrak (Pakard/Perkin-Elmer) was used to remove media from wells and to dispense 50 ul of dye solution into the assay plate.

OPEN IN VIEWER

Figure 4.

Fully automated Thermo LAS system with dual FLIPR^TETRA. A centrally located, high-speed movable belt (the LPT or linear plate transport) moves plates from one part of the system to another; one vertical array loader (VAL), an articulated robotic arm, moves compound source plates from the two plate carousels (320-plate capacity each) to the LPT; eight Flip Mover robotic arms move plates from the LPT to peripheral instruments; two random access Stericult incubators (147-plate capacity) with plate elevators; one Biotek Plate Washer Elx 405; one Perkin-Elmer Flexdrop IV noncontact dispenser; one Perkin-Elmer Evolution EP3 pipettor; one eight-station lidder; one carousel with lidder/delidder; and two MDC FLIPR^TETRA plate readers.

Plates were moved as a batch of 15 into at 37 °C in 5% CO₂ incubator and incubated for 60 min. Plates were moved as a batch of five to 10 to cool down to room temperature and placed into stacker on FLIPR-3 for the measurement of Ca-redistribution response. Ligand solution (12 ul) was dispensed to 50 ul of dye in the assay well to activate Gq-cou-pled receptor. Data were treated as described previously.

OPEN IN VIEWER

Figure 5.

Polara Scheduler (Thermo LAS). Screenshots of the schedule protocol for 100 plates in the agonist assay. (A) An overview of the schedule for 100 plates. Each bar represents all the steps in moving single plate to and from different locations included in the automated system. (B) Magnified view of detailed schedules for 10 plates. Each red bar is a step of incubation in Forma Scientific Stericult incubator for each assay plate. (Blue and yellow) Boxes reflect alternating use of two FLIPRs. (C) Further magnified view of the end part of one assay and one compound plates schedule.

Protocol 2. Fluorescence Measurements Using Thermo/ CRS Integrated FLIPR. Cell plates prepared in SelecT (TAP, UK) were moved in racks into Stericult incubators (147-plate capacity). Schedule was developed for each plate with consistent timing step in the protocol. Each cell plate was moved up by plate elevators and placed onto LPT to be delivered into the Perkin-Elmer Flexdrop IV. Dye (60 ul) was diluted as described previously and dispensed into wells of the cell plate without media removal. The cell plate was then moved by flip mover back onto LTP to be delivered into the incubator. After 60 min of incubation with dye, cell plate was again lifted up onto LTP to be delivered to one of the FLIPR readers. At that time, a compound plate was moved from plate carousels to Perkin-Elmer Flexdrop IV. Compound was diluted with the assay buffer to the requested intermediate concentration, and the compound plate was moved on LTP to a FLIPR where it is incubated together with a cell plate.

We have used an advantage of FLIPR^TETRA to successfully transfer smaller volumes of the ligand solution from the source plate into assay plate. Compound (4 ul) was dispensed into 80 ul in the assay well. To ensure better diffusion of the agonist, the pipettor was programmed to aspirate 10 ul more of the dye/media mixture from the assay well immediately after aspirating the agonist. Then, tips dispensed 19 ul of the mixture back into the assay well. This modification of the protocol had the following advantage: more efficient mixing effect was achieved in the assay well with less distortion of the cell layer than the traditional FLIPR mixture program.

New FLIPR ^TETRA .

We used a well-characterized Ca-flux detection assay to compare the results obtained on FLIPR^TETRA with those obtained on our standard FLIPR-3 platforms. The changes in the intracellular calcium were induced by activation of muscarinic receptor 1 subcloned in the CHO-K cell line. The protocols for both platforms were developed to be as similar to each other as possible so the difference in results would be attributed exclusively to the difference in performance of the two readers (see Materials and Methods). The plate view of FLIPR traces in Figure 8A demonstrates detection of acetylcholine dose-response on both platforms. The raw fluorescent units are different on FLIPR-3 and FLIPR^TETRA readers, however, the fold induction of the signal over background is similar (Fig. 8B). The kinetics of the response, as well as the concentration dependence on the ligand, is similar for both platforms. A volume of 12 μl of acetylcholine was dispensed into the wells by both kinetic readers for consistency between the two protocols for this comparison. However, another advantageous feature of the FLIPR^TETRA is its ability to perform a more precise aspiration and dispense, and to successfully transfer smaller volumes from the source plate into the assay plate. This feature allowed us to reduce the volume of agonist solution in target assays as described in the Materials and Methods section.

OPEN IN VIEWER

Figure 7.

Cell culture conditions. Overlay of FLIPR-3 traces from multiple wells on a 384-well plate. Cells for agonist Ca-flux assay were cultured as suspension (A) or in adherent monolayer (B), harvested, cryopreserved, and used in experiments. Cells grown as suspension show less activation by reference agonist. Signal to background ratio is only four-fold with very high level of noise (Z‘ =0.08; N well-s = 192). However, the same cell line expanded as adherent monolayer and plated in the same assay shows signal to background ratio about 14-fold and very small deviation, Z‘ =0.72; N wells = 192.

OPEN IN VIEWER

Figure 8.

Comparison of FLIPR-3 and FLIPR^TETRA. (A) Plate view of the traces from FLIPR-3 and FLIPR^TETRA. Acetylcholine was titered with two-fold step dilution from row C down to row N. The range of raw fluorescent unit response is different for the two readers but can be normalized by calculating the ratio of signal to background, shown below. (B) Results of the dose-response to acetylcholine calculated as Fl of the Max—Min raw fluorescent units of treated wells over that of control wells. EC50 for acetylcholine was 60 nM for both curves.

The FLIPR^TETRA produced comparable results to the FLIPR-3. However, it has reduced footprint and support requirements because there is no laser and laser cooling system. The good performance and rich feature set of FLIPR^TETRA made it a model of our choice for integration with LAS robotic system. It can be conveniently rolled away from the automated platform for service or used in the manual mode for assay development and validation.

The Advantage of the Thermo LAS Automated System as Compared to Batch Treatment of Plates.

Analysis of Stacker-based Batch Treatment of Multiple Cell Plates. Assay plates with cells were taken from the incubator. Those plates were dye loaded as described in the Materials and Methods section for the FLIPR-3 protocol. Two parallel experiments were carried out to address potential artifacts caused by batch treatment. In the first experiment, the plates were taken out as a batch of five plates. In the second parallel experiment, the plates were taken from the incubator one at a time for assay. It was essential to cool down the plates to room temperature for 10 min to ensure a consistent sample temperature in all plates before placing them in the FLIPR-3. It is important to emphasize that when the plates were treated as a batch only the first plate was placed into the reader without a delay, in the same interval as the individually treated plate. The second plate from the batch experienced a 4-min delay. This delay accumulates for sequential plates, so the 10th plate in the stacker receives agonist with a delay of 44 min compared to the first plate. The scheme of this process is illustrated in Figure 9A. During this delay, the cells are exposed to room temperature conditions, which may affect the response to the activator. Indeed, as depicted in Figure 10A, the cellular response was very sensitive to the extra exposure to ambient conditions. The induction of Ca-flux promoted by agonist was dramatically reduced in each consecutive plate in the stacker. However, the plates from the parallel experiment, which were placed into the FLIPR-3 individually, demonstrated a consistent level of induction. The extra exposure to room temperature in the stacker-based protocol also affected the variability of the signal. The Z‘ factor, which reflects the statistical validity of the assay, was acceptable for the individually treated plates, but was dramatically reduced for each consecutive plate in the stacker (Fig. 10B).

The Thermo LAS Automation System. The Thermo LAS automation system was designed with the major goal of treating each plate individually, and scheduling every plate to be treated identically. The ideal scheme will be similar to the work that is usually done with a single plate in a manual protocol and is visualized in Figure 9A.

The stagger time between samples in the Polara protocol was adjusted so that the standard deviation of the total run time for each sample was minimized and did not exceed 10 s. Each red bar represents the schedule of all events for a single plate or the timeline for each plate. Part A is an overview of the schedule for 100 plates in the assay protocol for the CRS-automated FLIPR system described in Materials and Methods. The delays or discrepancies in the treatment of each plate would be reflected as disruptions in the stagger of the plate timelines. The similarity between the schematic model of manual single plate processing and schedule of the current protocol calculated by Polara (Fig. 4) reflects the success of this attempt. The robust induction of signal in the CRS-automated system is illustrated in Figure 11. The agonist response to the control ligand on each plate (high signal) was consistent for five consecutive plates. The tight clustering of the values for the high signal from well to well, as well as for the background signal, illustrates a very low variability in the assay.

OPEN IN VIEWER

Figure 9.

Schematic diagrams of two protocols: batch-based and individual. (A) Protocol for batch of plates using FLIPR-3 plate reader with stacker. Plates are handled in a batch of several (five in this example): from the incubator plates are manually moved as a batch to the Packard PlateTrack for dye dispensing. Then, the whole batch was returned into 37 °C 5% CO₂ incubator for dye loading. After 0-60 min of incubation, the batch of plates is removed from the incubator and cooled down to ambient temperature for 10 min to ensure equal temperature of all assay plates coming into FLIPR. The batch of plates is then manually loaded into the stacker of FLIPR-3. Assay plates and compound plates were intercalated and loaded in the same stacker for sequential transfer into the FLIPR-3 for agonist addition and detection of intracellular calcium flux. (B) Protocol for a single plate handled manually or for sequential handling of individual plates by the automated Thermo LAS system.

OPEN IN VIEWER

Figure 10.

Detection of the signal using FLIPR-3 with stacker. (Top) A parallel set of experiments with two stacks of plates, treated as described in the Materials and Methods section. In the left panel, five plates entered the FLlPR sequentially after residing in the stacker for a variable interval that increased for each plate; in the right panel, each of nine plates was treated independently without variable delays. Mean was calculated for Max—Min values of raw fluorescent units detected over 3 min by FLIPR-3, N = 96 wells in 384-well plate. (Bottom) Variability was calculated as Z‘ ⁵ factor for each plate shown above.

The schemes in Figure 9 emphasize the difference between a stacker-based approach and the CRS integrated system. The integrated system has two crucial advantages. First, is the maintenance of cell culture conditions throughout most of the protocol. Second, the absence of variable delays between steps in the protocol, which increase from plate to plate in a stacker-based automation system; if the stack is large, the delay could accumulate to over an hour. The Thermo LAS/FLIPR system allowed us to support this fragile and sensitive assay for compound screening with a high throughput of up to 120 plates per day.

Results of HTS Using Thermo LAS FLIPR^TETRA.

This high-throughput screen was targeted to find agonists for an over-expressed Gq-coupled receptor. Small molecules, 407,448, were tested at 10 μM concentration in 28 working days. Each 384-well assay plate had a column with positive control agonist and a column treated with buffer for background controls. The changes in fluorescent signal were analyzed as fold of induction (FI) above background, normalized to each plate. Calculations were done using median values for the change in raw fluorescent units in each well. The statistical analysis of control wells over the course of the screening campaign is presented as a histogram in Figure 12. The tight clustering of each group of controls and nice separation between positive controls and background illustrate robust performance of the assay throughout the whole campaign.

OPEN IN VIEWER

Figure 11.

Results of the assay done using Thermo LAS FLIPR^TETRA. Median calculated for Max-Min value for wells with agonist (high signal) and buffer-treated wells (low signal). Median value of buffer-treated wells was used to normalize values for each well (n = 192) and to calculate FI.

A histogram of the results from our entire library of test compounds over the whole campaign (Fig. 13) has a normal distribution with slightly extended right tail. An activity cutoff of Fold Induction ≥ 2 was selected based on three interquartile range (IQR) standard deviations from the median value for the population, using log transformation of the Fold Induction values. Of approximately 400,000 compounds tested, 13,804 wells had an FI value higher than 2, resulting in a primary hit rate of about 3.4%.

Of approximately 14,000 primary hits retested, 4877 hits were confirmed in a five-dose titration in triplicate, with the high dose replicating the primary screening concentration of test compound. Figure 14 demonstrates the correlation in a scatter plot of the triplicate confirmation runs with each other. However, 3633 confirmed hits also showed above 2 FI in an assay with the parental cell line, lacking the target GPCR. Thus, only 593 hits were confirmed to have statistically significant, target-specific activity.