Characterizing Modulators of Protease-Activated Receptors with a Calcium Mobilization Assay Using a Plate Reader

Summary

An improved protocol for a calcium mobilization assay with endothelial cells, used to identify ligands of protease-activated receptors (PARs), has been developed. The new protocol reduces total assay time by 90-120 min and yields reproducible concentration-response curves.

Abstract

Changes in calcium concentration in cells are rapidly monitored in a high-throughput fashion with the use of intracellular, fluorescent, calcium-binding dyes and imaging instruments that can measure fluorescent emissions from up to 1,536 wells simultaneously. However, these instruments are much more expensive and can be challenging to maintain relative to widely available plate readers that scan wells individually. Described here is an optimized plate reader assay for use with an endothelial cell line (EA.hy926) to measure the protease-activated receptor (PAR)-driven activation of G_αq signaling and subsequent calcium mobilization using the calcium-binding dye Fluo-4. This assay has been used to characterize a range of PAR ligands, including the allosteric PAR1-targeting anti-inflammatory "parmodulin" ligands identified in the Dockendorff lab. This protocol obviates the need for an automated liquid handler and permits the medium-throughput screening of PAR ligands in 96-well plates and should be applicable to the study of other receptors that initiate calcium mobilization.

Introduction

Protease-activated receptors (PARs)^1,2,3 are a subfamily of class A G protein-coupled receptors (GPCRs) that are expressed in a variety of cell types, including platelets and endothelial cells^4,5,6,7. Unlike the majority of GPCRs, PARs have a unique intramolecular mode of activation. Most GPCRs are activated by soluble ligands interacting with a distinct binding pocket, but PARs are activated by the proteolytic cleavage of the N-terminus, which results in a new tethered ligand that can interact with the extracellular loop 2 domain on the surface of a cell^6,8,9. This interaction activates the receptor and can initiate several signaling pathways, promoting effects such as inflammation and platelet activation^4,10,11,12. Different proteases can activate PARs through cleavage at unique sites on the N-terminus, revealing different tethered ligands (TL) that stabilize receptor conformations, which initiate different signaling pathways^9,13,14,15. For example, in the most well-studied member of the subfamily, PAR1, cleavage by thrombin is used to support numerous biological processes, including platelet activation and leukocyte recruitment to the endothelium, but can lead to deleterious effects when the receptor is overexpressed or overactivated^{4,16,17,18,19,20,21}. Conversely, cleavage by activated protein C (aPC) can promote anti-inflammatory effects and maintenance of endothelial barriers^{15,22,23,24,25,26,27,28,29}. PARs can also be activated by peptide analogs of the TLs in an intermolecular fashion^13,30,31. These peptides are routinely used to measure PAR inhibition (modulation) in place of PAR-targeting proteases, and they are used in this protocol.

Numerous disorders are associated with pathological PAR1 signaling, including sepsis^22,32, cardiovascular disease^{33,34,35,36,37,38}, kidney disease^39,40,41,42, sickle cell disease⁴³, fibrosis⁴⁴, osteoporosis and osteoarthritis^45,46, neurodegeneration^{47,48,49,50,51}, and cancer^{52,53,54,55,56,57,58,59}. Antagonists of PAR1 have been studied since the 1990s as antiplatelet agents for cardiovascular disease, and the growing list of diseases associated with the receptor necessitates the identification of novel ligands for use as biological probes (tool compounds) or as potential therapeutics. Currently, there is only one FDA-approved PAR1 antagonist, vorapaxar, which is used to treat coronary artery disease in high-risk patients^34,36,37,60. An alternative PAR1 antagonist, the pepducin PZ-128, completed a successful phase II study to prevent thrombosis in cardiac catheterization patients³⁸. The Dockendorff group has focused on the medicinal chemistry and pharmacology of a separate class of small molecules, PAR1 ligands known as parmodulins^61,62. Unlike reported PAR1 antagonists such as vorapaxar, parmodulins are allosteric, biased modulators of PAR1 that selectively block the G_αq pathway while promoting cytoprotective effects similar to aPC. Unlike potent orthosteric PAR1 antagonists such as vorapaxar, published parmodulins are also reversible^63,64,65.

Initially, parmodulins were identified by Flaumenhaft and coworkers for their ability to inhibit P-selectin expression or granule secretion in platelets^61,66. However, an alternative method was required to study the effects of parmodulins on endothelial cells. One common method to monitor GPCR-related signaling is to measure intracellular Ca²⁺ mobilization, an important secondary messenger that can be measured using a suitable intracellular calcium-binding dye^67,68. Substantial evidence has been provided showing that calcium mobilization induced by PAR1 is through the activation of G_αq^69,70. Once activated by its tethered ligand (or a suitable exogenous ligand), PAR1 undergoes a conformational change which causes guanosine diphosphate (GDP) bound to the G_αq subunit to be replaced by guanosine triphosphate (GTP)⁶⁸. The G_αq subunit then activates phospholipase Cβ (PLC-β), which catalyzes the hydrolysis of phosphatidylinositol 4,5 bisphosphate (PIP2), forming 1,4,5-inositol triphosphate (IP₃) and diacylglycerol (DAG). Finally, IP₃ binds to IP₃-sensitive Ca²⁺ channels in the membrane of the endoplasmic reticulum, allowing Ca²⁺ to be released into the cytoplasm, where it can bind to Ca²⁺-dependent fluorescent dyes, such as Fluo-4, that are added to the cells⁷¹. This process occurs within seconds and can increase the concentration of Ca²⁺ 100-fold, leading to a drastic change in the amount of calcium-bound dye and a robust fluorescence signal.

In 2018, the Dockendorff group disclosed a medium-throughput Ca²⁺ mobilization assay that could be used to identify antagonists of the G_αq pathway of PAR1⁷². The assay used EA.hy926⁷³, a hybrid human endothelial cell line, which can be used for multiple passages without a noticeable change in PAR1 expression, and is established for in vitro measurements of cytoprotective effects. 

The original protocol used EA.hy926 cells in 96-well plates and loaded with Fluo-4/AM dye, which was chosen due to its intense fluorescence at 488 nm and high cell permeability. Once the dye was loaded into the cells, lengthy washing steps were performed with an automated 8-channel liquid handler (faster methods of liquid handling, such as a 96-channel washer, were inaccessible). The reproducibility of this assay was superior to that without the careful, automated robotic media changes. Antagonists were then incubated with the cells, PAR1 was activated through the sequential addition of a selective agonist (16 wells at a time), and changes in fluorescence resulting from calcium mobilization and dye binding were measured to determine activity.

While this protocol allows for the measurement of PAR1-mediated calcium mobilization, it is limited by the time required to assay each 96-well plate. Long experiment times are problematic not only because the number of compounds that can be screened each day is limited, but also because dye efflux occurs over time, narrowing the assay window by increasing the basal fluorescence. One contributor to the long experiment time is the use of an 8-channel liquid handler for plate washing, which adds over 30 min to each experiment. The required tips also became difficult to obtain due to supply chain problems. Here an updated protocol for the PAR-mediated calcium mobilization assay that does not require a liquid handler, and therefore can be run in higher throughput, is reported. This protocol should also be suitable for measuring signaling with other GPCRs that lead to intracellular calcium mobilization. This updated plate reader protocol is ideal for academic and small industrial labs that do not have the resources for expensive cell imaging instruments but have a need to rapidly screen numerous compounds. For an example of a calcium mobilization assay using a plate imager, see Caers et al.⁷⁴.

Protocol

All media exchanges/additions made in steps 1 and 2 of the following protocol are performed in a sterile hood. Unless otherwise noted, all plasticware used in the sterile hood must be purchased sterilized and sealed or sterilized appropriately via autoclave.

1. Initiation of EA.hy926 cell line

1. Acquire EA.hy926 cells.
2. Store vial(s) of cells in the vapor phase of a liquid nitrogen tank.
3. Prepare DMEM complete medium by first warming DMEM, FBS, and pen/strep (see Table of Materials for specific details) in a 37 °C water bath for 15-30 min. Add 50 mL of FBS and 1 mL of pen/strep to 500 mL of DMEM. Gently mix the solution by inversion.
   NOTE: Mixing too vigorously causes excessive amounts of bubbles.
4. Remove one vial of cells from liquid nitrogen and warm in a 37 °C water bath for 5 min.
5. Sterilize the vial of cells and bottle of DMEM complete medium by spraying with alcohol, then move to the sterile hood.
6. Use a 1,000 µL pipette to carefully transfer the cells to a 15 mL sterile centrifuge tube. Add 1 mL of DMEM to the vial to rinse and add to the centrifuge tube.
7. Make a pellet of cells using a centrifuge (150 × g, 5 min, 22 °C). Remove the medium from the centrifuge tube by aspiration; be careful not to disturb the cell pellet.
8. Add 3 mL of DMEM complete medium to the centrifuge tube and gently break apart the pellet by mixing the cell suspension using a 1,000 µL pipette.
   NOTE: Causing bubbles by mixing too vigorously should be avoided as this could activate or damage the cells.
9. Mix 10 µL of the cell suspension with 10 µL of trypan blue in a microcentrifuge tube using a pipette with either 20 µL or 200 µL tips. Add 10 µL of the cell suspension/trypan blue mixture to a hemocytometer and count the cells.
   NOTE: Using 0.1-10 µL pipette tips can shear the cells. Tips of this size should not be used while handling the cell suspension.
10. Add DMEM complete medium to the cell suspension to make a density of 66,000 cells/mL using either a 10 or 25 mL serological pipette. Pipette 15 mL of the cell suspension into T-75 culture flask(s). Ensure equal dispersion of the cell suspension by moving the flask north to south and east to west.
11. Incubate the cells until they reach confluency (typically after 4-6 days), as estimated by examination under a microscope. At this point, use the cells in the experiment (beginning with section 2) or seed them into new flasks for expansion to prepare frozen stocks (at a recommended seeding density of 1,000,000 cells in 15 mL of complete medium per T-75 flask).
    NOTE: Cells can be grown in the culture flasks for as long as 3 weeks without a noticeable difference in shape, health, or performance in the assay. Complete medium should be exchanged every 3-4 days.

2. Addition of EA.hy926 cells to 96-well plates

NOTE: Cells from one confluent T-75 culture flask can be used to prepare two or three assay plates or optionally expanded into as many as 15 fresh T-75 flasks. At least five assay plates can be screened using the protocol in sections 4 and 5 in a normal workday. The following instructions describe preparing and testing one assay plate, but additional plates can be prepared by repeating steps 2.2, 2.7, 2.12, and 3.2 to prepare the desired number of assay plates. Most commonly, four assay plates are prepared per day to measure concentration-responses with up to 16 compounds, which requires two T-75 flasks with confluent cells.

1. Warm DMEM complete medium in a 37 °C water bath for 15-30 min. Sterilize the bottle of DMEM complete medium by spraying it with alcohol, then move it to the sterile hood.
2. Add 100 µL of a sterilized 0.4% gelatin solution to all wells of a black-walled, 96-well, clear-bottom plate with a lid. Incubate the plate with the gelatin solution in an incubator (37 °C, 5% CO₂) for 30 min.
3. Confirm that cells in a T-75 culture flask are 80-100% confluent using an inverted microscope. Remove DMEM complete medium from the T-75 flask via aspiration.
4. Wash the cells in the flask with 10 mL PBS and gently move the flask North to South and East to West. Remove the PBS from the flask via aspiration.
5. Add 5 mL of cell dissociation solution to the flask and incubate the flask in an incubator (37 °C, 5% CO₂) for ~12 min. Tap the flask after 6 min to help facilitate dissociation.
   NOTE: Incubating cells for longer than 15 min with the referenced cell dissociation agent causes cells to begin to clump together, which makes obtaining an accurate cell count more difficult and may interfere with the assay.
6. Add 5 mL of the prewarmed DMEM complete medium to the flask and gently mix into the cell dissociation solution/cell suspension. Using a 10 mL serological pipette, add the cell suspension into a sterile 50 mL centrifuge tube. Make a pellet of cells using a centrifuge (150 × g, 5 min, 22 °C).
7. While the pellet is being formed, remove the gelatin solution from the 96-well plate via aspiration using a Pasteur pipette connected to a vacuum. Remove the plate lid from the time of gelatin removal until the solution is ready to be plated.
   NOTE: A multichannel pipette or sterile manifold connected to a vacuum may optionally be used.
8. Remove the medium from the centrifuge tube by aspiration using a Pasteur pipette connected to a vacuum or a serological pipette; be careful not to disturb the cell pellet.
9. Add fresh DMEM complete medium to the centrifuge tube using a 5 or 10 mL serological pipette. Gently break up the cell pellet by mixing the DMEM complete medium with a 1,000 µL pipette or a 5 mL serological pipette.
   NOTE: Eight milliliters of DMEM are typically used, but higher volumes can be added to make cell counting easier.
10. Mix 10 µL of the cell suspension with 10 µL of trypan blue in a microcentrifuge tube using a pipette with either 20 µL or 200 µL tips. Add 10 µL of the cell suspension/trypan blue mixture to a hemocytometer and count the cells.
11. Add DMEM complete media to the cell suspension to make a density of 600,000 cells/mL using either a 10 or 25 mL serological pipette. This will give a final density of 60,000 cells/well in a 96-well plate (determined to be the optimal density by experimentation).
    NOTE: A normal successful cell culture will yield between 12,000,000 and 15,000,000 cells, which can seed two to three 96-well plates.
12. Mix the cell suspension by gently inverting the centrifuge tube and add it to a 50 mL multi-channel pipette reservoir. Mixing the suspension in the reservoir every 30 s by gently rocking the reservoir side to side, add 100 µL of the cell suspension to each well using a multi-channel pipette. Replace the transparent plate cover and gently shake the plate by sliding it on the surface of the sterile hood from North to South and East to West to ensure even cell distribution. Incubate the plate in the incubator (37 °C, 5% CO₂) for 16-24 h.
    NOTE: Additional cell suspension can be used to prepare additional plates, or steps 1.10-1.11 can be repeated to continue the cell line.

3. Calcium mobilization assay preparation

1. Prepare assay reagents.
   1. Prepare probenecid solution (250 mM, aqueous) by adding 36 mg of probenecid to a microcentrifuge tube and dissolving it in 0.6 M NaOH (500 µL). Vortex the solution.
      NOTE: Probenecid improves the retention of dye in the cell by inhibiting organic ion transporters^75,76,77.
   2. To make HBSS-HEPES assay buffer, add HEPES (1.19 g) to a 500 mL bottle of Ca/Mg/phenol red-free HBSS (making a 10 mM solution of HEPES). Supplement the buffer with MgCl₂ (1 M aqueous solution, 500 µL) and CaCl₂ (1 M aqueous solution, 500 µL). Mix the solution and store in a refrigerator at 5 °C when not in use.
      1. For each 96-well plate, add 50 mL of the HBSS-HEPES assay buffer to a 50 mL centrifuge tube, supplement with 500 µL of the probenecid solution, and mix well. Warm the buffer to room temperature prior to use.
   3. Make a 10% Pluronic F-127 solution in DMSO by adding Pluronic F-127 (20 mg) to a microcentrifuge tube or HPLC vial and dissolving it with 200 µL of DMSO.
      NOTE: This solution can be stored at room temperature and is sufficient for about 30 plates. Pluronic F-127 dissolves very slowly at room temperature and should be gently heated to facilitate dissolution.
   4. Prepare the dye loading buffer by first adding 24 µL of DMSO to a vial of Fluo-4/AM (50 µg) to dissolve the dye. In a foil-wrapped 15 mL centrifuge tube, add 6 mL of HBSS-HEPES assay buffer and supplement with 6 µL of 10% Pluronic F-127 and 6 µL of Fluo-4/AM solution. Briefly vortex the solution and allow to sit at room temperature, out of light, for 10 min.
      NOTE: To prevent photobleaching of Fluo-4, keep the dye out of light by wrapping aluminum foil around the centrifuge tube containing the dye solution and the lid of the assay plate.
2. Prepare assay plates.
   1. Remove the assay plate from the incubator and confirm confluency using an inverted microscope. Manually remove the DMEM complete media by flicking it into a sink and blotting the top of the plate with a paper towel.
      NOTE: Medium is sufficiently removed by holding the plate horizontally and flicking, followed by rotating the plate 180°and repeating.
   2. Add HBSS-HEPES assay buffer plus probenecid solution to a 50 mL multi-channel pipette solvent reservoir.
      NOTE: This solution will be used for later washing steps and should be put aside and covered with aluminum foil until needed.
   3. Add 100 µL of the HBSS-HEPES assay buffer plus probenecid solution to each well with a multi-channel pipette and allow the cells to sit in the presence of probenecid for 5 min. Remove the HBSS-HEPES assay buffer by flicking into the sink in the same manner as in step 3.2.1.
   4. Add dye loading buffer to a separate 50 mL multi-channel pipette solvent reservoir. Add 50 µL of the dye loading buffer to each well of the assay plate using a multi-channel pipette. Incubate the plate for 45 min (37 °C, 5% CO₂).
   5. While the plate is in the incubator, prepare the agonist/antagonist solutions.
      1. Antagonist solutions are stored as 31.6 mM solutions in DMSO. In a PCR tube, dilute 2 µL of the stock solution with 38 µL of 0.1% BSA/water to obtain a 1.58 mM solution. 2 µL of this solution will give a final concentration of 31.6 µM in the assay plate. Dilute 4 µL of the 1.58 mM antagonist solution with 36 µL of 5% DMSO/water (no BSA addition), and prepare the remaining concentrations through serial dilutions.
      2. Prepare a 108.3 µM (16.7x) solution of TFLLRN-NH₂ in HBSS-HEPES assay buffer, which will give a final concentration of 6.5 µM when 6 µL is added to each well of the assay plate. For example, dissolve a 2 mg aliquot of TFLLRN-NH₂ with 5.244 mL of HBSS-HEPES assay buffer to give a 0.5 mM stock solution. Mix 1.083 mL of this solution with 3.917 mL of HBSS-HEPES assay buffer to give a 108.3 µM solution.
         NOTE: TFLLRN-NH₂ solutions should be stored in a -20 °C freezer when not in use.
   6. Remove the plate from the incubator and ensure dye uptake by using a fluorescence microscope at a suitable setting (typically for green fluorescent protein, see Figure 1 for an example of Fluo-4 successfully loaded into cells). Remove the dye loading buffer by flicking the plate in the same manner as step 3.2.1.
   7. Wash the cells 2x with 50 µL of the HBSS-HEPES assay buffer plus probenecid solution (from step 3.2.3) in each well (remove by flicking medium into the sink).
   8. Add 92 µL of HBSS-HEPES assay buffer to each well and again view the cells with a fluorescence microscope to ensure excess dye has been fully removed, and that cells remain adherent.
   9. Use a multi-channel pipette to add 2 µL of the antagonist solutions and vehicle to appropriate wells (see Table 1 for a representative plate map).

4. Performing the calcium mobilization assay

1. Set the plate reader as follows: temperature: 37 °C; excitation wavelength: 485 nm; emission wavelength: 525 nm; measurement height: 7.8 mm (optimized); number of flashes: 100; number of repeats: 20.
2. Incubate the plate for 15 min in the plate reader at 37 °C. Measure the background fluorescence in columns 1 and 2 (five scans per well).
3. Eject the plate from the plate reader and quickly add 6 µL of agonist solution to each well in columns 1 and 2 from an 8-tube PCR strip using a multi-channel pipette.
   NOTE: For this step, we use an electronic 8-channel pipette with 0.1-10 µL pipette tips.
4. Measure the change in fluorescence as calcium mobilization occurs in the cells. Perform 20 scans of each well according to the settings above.
   NOTE: Scanning each well in two columns 20x each takes approximately 5 min (i.e., ~15 s between scans of each well).
5. Repeat steps 4.2-4.4 for columns 3-12 in two column increments.

5. Data analysis

1. Export data from the assay as separate spreadsheets for each two-column group.
2. Find the average of the basal fluorescence readings by using the function AVG(first scan: last scan). Do this for every well in both columns.
3. Find the maximum fluorescence after the addition of agonist by using the function MAX(first scan: last scan).
4. Calculate the change in fluorescence by subtracting the basal fluorescence from the maximum fluorescence.
5. Subtract the change in fluorescence calculated for the negative control (addition of vehicle without agonist) from each well in the same column.
6. Find the relative (normalized) change in fluorescence by dividing the change in fluorescence in each well by the change in fluorescence in the vehicle (0.1% DMSO/water) + 6.5 µM TFLLRN-NH₂ well.
7. Copy the non-control values, as percentages, into a statistics and graphing software program. If using the referenced software, choose the XY table option, with the Numbers chosen as the X-axis and replicate values in side-by-side subcolumns as the Y-axis.
8. Use the program to plot concentration-response curves (CRCs) from the data. If using the referenced software, choose the log(inhibitor) vs. response - Variable slope (four parameters) option.

Results

The purpose of this assay is generally to produce concentration-response curves (CRCs) for three to four new parmodulins. On each assay plate, an additional CRC for a known compound, such as NRD-21, is often generated that acts as a quality check for the experiment due to its known IC₅₀. To generate CRCs, a plate map such as the one depicted in Table 1 should be planned. If single-point concentration-responses are desired instead, compounds at 10 µM final concentrations (or other preferred concentrations) should be added to at least three wells.

When performing the assay, there should be a quick increase in fluorescence (the increase should begin within the first 3-4 scans) displayed by the plate reader (see Figure 2A for an example output compared to the failed output in Figure 2B). It is important to note that the agonist needs to be added quickly (within 15 s) for the plate reader to be able to accurately capture the increase in fluorescence. In a working experiment, the fluorescence signal for the vehicle well should stay flat, while the fluorescence signal in the positive control wells (vehicle + 6.5 µM TFLLRN-NH₂) should start to increase after three or four scans. Once the maximum value is reached (after ~10 scans), the fluorescence signal should start to slowly decrease. In experiments with highly efficacious antagonists, the wells with the highest concentrations of antagonist should have a signal that stays relatively flat, similar to the negative control. There should also be little difference between the signal of the positive control wells and those with low concentrations of the antagonist (Figure 2C). If the lowest concentrations of antagonist are still inhibiting calcium mobilization, screened concentrations should be adjusted to include lower concentrations. Generated CRCs should have at least one data point ~100% to fully show the sigmoidal nature of the curve.

The most common sign that an experiment did not work is if the positive control well (vehicle + 6.5 µM TFLLRN-NH₂) does not show a large increase in fluorescence (Figure 2B). This could happen for a variety of reasons: the cell count could be low (if confluency was not confirmed), the agonist solution could be compromised, or inadequate amounts of dye could have been taken in by the cells. It is also possible to mistakenly not add agonist to certain wells.

With a programmed method, data is outputted as six separate spreadsheets, each containing data for two columns. The five scans of the basal fluorescence are in their own row, with the 20 scans taken after the addition of agonist in the row immediately below (see Figure 3 for representative output). For an example calculation, a column from an experiment using NRD-21 will be used. In the exported file (Figure 3), the basal fluorescence from each well is output to columns C to G, and data from after the agonist addition (the following 20 scans) are output to columns C to V on the subsequent row. Rows 14-17 in the file contain the positive control of two separate columns. The average of the basal row for one of the positive control wells was calculated using the command "=AVERAGE(C16:G16)", which gave a value of 1,039.8. This command was repeated for the additional basal measurements in each well. The maximum of the experimental scans was calculated by using the command "=MAX(C16:V16)", which gave a value of 2,629. Again, the command was repeated for all wells. The basal fluorescence for each well was then subtracted from the maximum value. In this experiment, the data was normalized by subtracting the change in fluorescence in the negative control from each well (which was 142.6 in the example column). These values were then divided by the similarly calculated value of the positive control. In this column, the maximum value was 2,629, the average basal measurement was 1,039.8, and the value subtracted for the blank was 142.6. The adjusted positive control value is thus 1,446.6 (calculated from the equation). This equation is repeated for each well in the column, and the calculated values are divided by 1,446.6 to give their relative change in fluorescence. The ratios were then converted into percentages and were copied into the analytical software. This process was repeated for the other 11 columns.

To process the data using the referenced software, an XY table should be selected with the option to have three repeated measurements in side-by-side columns. The -log values of each concentration should be inputted into the x-values, and the calculated percentages added to the y-values. A nonlinear regression analysis should be performed on the data, which causes four concentration-response curves to be produced. The program also outputs IC₅₀ values, quantified error statistics, and goodness of fit statistics.

Figure 1: Fluo-4/AM loaded into EA.hy926 cells. Fluo-4 selectively loaded into EA.hy926 cells, after dye incubation and washing steps. All fluorescence is within cells, with no dye leakage. Image obtained using the GTP setting on an EVOS cell imager and 20x magnification. Please click here to view a larger version of this figure.

Figure 2: Potential output of assay. (A) Representative plate reader output of a working assay. Row A is the negative control and has no change in fluorescence. Row B is the positive control and shows a rapid increase in fluorescence. Rows C-H are the experimental wells, with the higher concentrations of parmodulin lower in the plate. (B) Example output of a failed experiment. No increase in fluorescence is seen in row B, which is the positive control. (C) Change in fluorescence over time induced by the PAR1 agonist TFLLRN-NH₂. Scanning two columns of a 96-well plate using the provided settings takes ~5 min. on the PerkinElmer EnSpire plate reader (i.e., ~15 s between scans of each well). Different NRD-21 solutions are as indicated, "Agonist" is 6.5 µM TFLLRN-NH₂, n = 1. Please click here to view a larger version of this figure.

Figure 3: Example Excel output for a working assay. Basal fluorescence is output into even-numbered rows starting at row 10, and the result from the first 10 scans after agonist addition is output shown in subsequent odd-numbered rows. Please click here to view a larger version of this figure.

Figure 4: Robotic liquid handling vs "flicking." Precoating the assay plates with gelatin and manually performing media exchanges by flicking the plate and blotting it dry produces concentration-response curves with lower variability. (A) Concentration-response curves of parmodulins ML161 and NRD-21 and 5 µM TFLLRN-NH₂, n = 3 (automated liquid handling). (B) Concentration-response curves of parmodulins ML161 and NRD-21 and 5 µM TFLLRN-NH₂, n = 3 (gelatin-coated, flicked plates). Data are means ± SEM. Please click here to view a larger version of this figure.

Figure 5: Temperature studies. Performing the assay at 37 °C generates more reproducible data than at room temperature. (A) Concentration-response curves of ML161 and 5 µM TFLLRN-NH₂, n = 3 (room temperature). (B) Concentration-response curves of ML161 and 5 µM TFLLRN-NH₂, n = 3 (37 °C). Data are means ± SEM. Please click here to view a larger version of this figure.

Figure 6: Cell density studies. The optimal cell density is 60,000 cells/well when seeding 16-24 h prior to beginning an assay. (A) Concentration-response curves of TFLLRN-NH₂ at different cell densities, n = 3. (B) Concentration-response curve of TFLLRN-NH₂ at the optimal cell density of 60,000 cells/well. Data are means ± SEM. Please click here to view a larger version of this figure.

Figure 7: Assay volume. Using an assay volume of 100 µL generates more reproducible concentration-response curves than assays using a final volume of 200 µL. (A) Concentration-response curves of ML161 and 5 µM TFLLRN-NH₂, n = 3 (200 µL final volume). (B) Concentration-response curves of ML161 and 5 µM TFLLRN-NH₂, n = 3 (100 µL final volume). Data are means ± SEM. Please click here to view a larger version of this figure.

Figure 8: Generation of Z' factors to determine the ability of the assay to detect a hit compound. Calcium mobilization responses with 6.5 µM TFLLRN-NH₂ (column B) plus optional PAR1 antagonists 10 µM NRD-21 (column C) or 0.316 µM vorapaxar (column D). (A) Assay completed 16 h after cell seeding. (B) Assay completed 24 h after cell seeding. (C) Table of calculated Z'-factors using the following equation: , where σ_p is the standard deviation of the positive control, σ_n is the standard deviation for the negative control or sample, μ_p is the mean of the positive control, and μ_n is the mean of the negative control or sample. Each value is calculated by designating vehicle + 6.5 μM TFLLRN-NH₂ as the positive control and comparing to the indicated condition as the negative control (vehicle) or sample (NRD-21 and vorapaxar). Please click here to view a larger version of this figure.

Figure 9: Calcium mobilization from a PAR2 agonist. Change in fluorescence over time induced by 10 μM SLIGKV-NH₂, n = 4. Data are means ± SEM. Please click here to view a larger version of this figure.

Table 1: Example 96-well plate map for compound screening. Each concentration is measured in triplicate. Row A contains negative controls (no antagonist; no agonist), and Row B contains positive controls (no antagonist; agonist). Columns 1 to 3 contain six half-log concentrations of the parmodulin NRD-21 that can optionally be used for quality control. PM-A, PM-B, PM-C are unnamed parmodulins. Please click here to download this Table.

Table 2: Comparison of previously published protocol and new, optimized protocol. Please click here to download this Table.

Discussion

While the previously reported protocol⁷² was generally reliable and allowed us to identify a new lead parmodulin, NRD-21,⁶² a more efficient protocol was desired. The assay was further compromised during the supply shortage caused by the COVID-19 pandemic. Acquiring tips for the automated liquid handler became difficult, and attempting to wash, sterilize, and reuse the tips produced CRCs with significant variance. This facilitated an urgent series of experiments designed to streamline an improved PAR-driven calcium mobilization assay.

Identifying a valid alternative to the use of the automated liquid handler for washing steps was the main priority in this study. During previous assay development, the use of the robotic liquid handler reduced errors caused by manual media exchanges. However, this added significant time to the assay, as the media exchanges with the 8-channel liquid handler increased the total time of each assay plate by 30-45 min. Although a "flicking" method of media exchange had been tried previously, these experiments were done without coating the assay plates with gelatin prior to cell seeding. By pretreating the plates with gelatin, the cells were ensured to stay in place regardless of the force used to manually remove media. CRCs generated in an assay using gelatin-coated, flicked plates had better goodness-of-fit scores and smaller error bars at multiple data points than curves where the automated liquid handler was used (Figure 4). Root-mean-square error (RMSE) values for the plate using the automated liquid handler were 9.687 for the ML161 curve and 10.690 for the NRD-21 curve, whereas the RMSE values for the plate using gelatin and a manual media exchange were 7.658 for the ML161 curve and 6.408 for the NRD-21 curve. Lower RMSE values indicate a curve that better fits the given data. This change also greatly reduces the total assay time. Alternatively, a multi-channel automated plate washer could be used.

The method of agonist addition was an additional factor that was studied. This was done to determine if agonist could be added to all plate wells simultaneously followed by a scan of the entire plate at once; this would eliminate the need for researchers to stay near the plate reader for the duration of the time-sensitive assay for numerous separate agonist additions, which could increase throughput significantly. Unfortunately, adding agonist to all wells and scanning the entire plate was problematic, because the response is too fast to reliably measure the transient maximum calcium concentration in every well, giving highly variable CRCs. Reducing the number of columns scanned from 12 to 6 slightly improved results, but the best results were generated from scanning only two columns (16 wells) at one time. One could hypothesize that reducing the temperature of the assay from 37 °C to room temperature could slow the change in calcium concentration enough to allow for reliable measurements. Although multiple calcium mobilization protocols with different dyes or cell lines have been validated at room temperature^78,79, in this specific assay, improved results are observed at 37 °C. Measured antagonist concentration-response curves at room temperature (~23°C) were not as uniform as at 37 °C (Figure 5A vs. Figure 5B). Because of these results, it was determined that scanning two columns at a time at 37 °C is optimal.

Another variable that was scrutinized to reduce assay time was the antagonist incubation time. In the previous protocol, antagonists were incubated for 40 min prior to agonist addition, but it was determined here that 15 min is sufficient time for parmodulins to interact with PAR1. This was tested by adding parmodulin solutions to multiple columns at the same time followed by adding agonist at differing time points (agonist was added at t = 15 min, 30 min, and 45 min). The activity of parmodulins was retained, but not improved, by incubating the antagonists longer. As a result of this finding, in the new protocol, antagonist solutions are added to the entire plate at one time, and agonist solutions are added to the first columns after an incubation of 15 min. This change further reduces the total experiment time by 30 min. It should be noted that antagonists with slow binding kinetics, such as vorapaxar, may require longer incubation times^80,81.

While reducing the assay time and eliminating the reliance on the liquid handler were the main factors necessitating an update to the assay protocol, an alternative cell culturing method was also tested. Previously, plates were seeded with cells directly from frozen stocks and allowed to grow in the plates for 40-48 h after seeding. However, the extended time between seeding plates and performing assays limited the number of experiments that could be performed each week. Using cells grown in culture flasks drastically reduces this time. Cells were grown in tissue culture-treated T-75 flasks and were removed from the flask surface by using a non-enzymatic cell dissociation agent. This may be important to avoid damaging certain cell surface receptors, in particular PAR2, with trypsin. Once added to the assay plate, the cells need adequate time to settle and stick to the gelatin. Because the flicking method of media removal can also displace cells, enough time is needed to allow the cells to become embedded in the gelatin. An experiment allowing the cells to settle for two hours caused a significant loss of cells, resulting in an unusable assay (data not shown). Settling times between 2 h and 12 h limit the number of plates that can be analyzed in a single day, so longer overnight incubation is preferred. It was found that incubating the newly cultured cells anywhere between 16 h and 24 h allowed the cells to fully stick to the gelatin, and reproducible data can be obtained.

Changing from frozen to culture flask-grown cells also changed the optimal cell seeding density. Predictably, seeding wells at the 25,000 cells/well density that was standard in the previous protocol did not allow cells to reach confluency with a shorter incubation time. Agonist CRCs were generated with eight separate cell densities to determine optimal conditions (Figure 6). Previously, TFLLRN-NH₂ was used as the PAR1 agonist at 5 µM, as this permitted reasonable antagonism with lower solubility parmodulins. There is not a significant difference in the background or maximal signal when seeding with 45,000-65,000 cells/well, but there is a decrease in variability with 5 µM TFLLRN-NH₂ when adding 60,000 cells/well. Interestingly, there is not a significant increase in maximal fluorescence with a higher number of cells, so 60,000 cells/well was chosen as the optimal density due to the decreased variability. Agonist CRCs also produced an EC₅₀ of 6.5 µM for TFLLRN-NH₂, so this was chosen as the new standard agonist concentration.

Another factor that was analyzed was the volumes used in the assay. In the previous protocol, 2 µL of agonist was added to each well to give a final assay volume of 200 µL. It was hypothesized that adding higher volumes of agonist solution—but keeping the final concentration constant—could help aid in mixing, causing a faster and more reliable signal. To test this hypothesis, the addition of 2 µL, 6 µL, and 10 µL TFLLRN-NH₂ into either 100 µL or 200 µL of HEPES-HBSS in an assay plate containing different concentrations of NRD-21 was compared. CRCs of NRD-21 from each condition were generated to determine which combination of agonist volume and final assay volume gave the most consistent data. While both 2 µL and 6 µL additions resulted in generated CRCs with little variability, the same could not be said about curves produced from the addition of 10 µL TFLLRN-NH₂ (data not shown); the increases in fluorescence were not consistent in these wells in either magnitude or timing, which produces CRCs that should be identical but are instead drastically different. Though 2 µL and 6 µL additions produced similar results, the change in fluorescence occurred much quicker with the addition of 6 µL of TFLLRN-NH₂, which allowed the total number of scans in each well to be cut in half, from 40 scans to 20 scans. It was also determined that using a final volume of 100 µL produces more uniform CRCs compared to those generated with a final volume of 200 µL (Figure 7).

To determine if the new protocol is suitable for higher throughput assays, Z' factors for the assay were calculated (Figure 8). A Z' factor is a statistical method measuring the quality of a screening assay and determines if an assay will reliably differentiate active and inactive compounds⁸². Z' factors were calculated for assays performed between 16 and 24 h after seeding plates, and at different cell passages. In all cases—for both agonist only and agonist plus antagonist modes—the calculated Z' factors were greater than 0.5, indicative of an excellent assay⁸². The calculated Z' factors also were similar to those calculated for the previous assay, indicating that the changes made to the protocol to produce a faster assay did not negatively impact the quality.

The protocol for PAR-driven calcium mobilization was updated through experiments generating IC₅₀ values from CRCs, but the assay can also be used to measure the efficacy of PAR1 antagonists at single concentrations. The assay can also be used to measure the activity of PAR2 ligands; for example, the PAR2 agonist SLIGKV-NH₂ can replace TFLLRN-NH₂ in the updated protocol (Figure 9); the previous protocol was also used with PAR2 agonists and antagonists⁸³. Though not explicitly tested here, this protocol should also be applicable to Human Umbilical Vein Endothelial Cells (HUVEC), which were used with the previous protocol, and potentially other adherent cell lines. A limitation to the use of a scanning plate reader with protocols such as this one is that they are not suitable for studying fast (sub-second) responses, at least not in parallel as is possible with a plate imager.

In conclusion, reported here is a faster and simpler protocol for the endothelial PAR-driven calcium mobilization assay using a plate reader. By eliminating the need for an automated liquid handler, reducing the antagonist incubation time, and decreasing the number of scans for each well, the total assay time was lowered by at least 90-120 min, and assay costs were decreased (for a comparison between the previously reported and current protocols, see Table 2). This not only will permit the screening of more compounds each day, but also eliminates the risk of a decreased assay window from dye efflux, which could lead to unreliable results. The newly optimized assay allows for the faster screening in endothelial cells of PAR1 or PAR2 ligands, or potentially other GPCR ligands that activate intracellular calcium mobilization, in labs limited to the use of plate readers.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Cell Culture Reagents
Adherent EA.hy926 cells	ATCC	CRL-2922
CellStripper cell dissociation reagent	Corning	25-056-CI	Trypsin can optionally be used, but should definitely be avoided with PAR2 assays.
Dulbecco's Modified Eagle Medium (DMEM) w/phenol red	Corning	10-013-CV
Fetal Bovine Serum (FBS)	Avantor	97068-091
Gelatin from porcine skin	MilliporeSigma	G2500	Use to make an aqueous 0.4% (w/v) solution with deionized water. Autoclave before use to sterilize.
Pen/Strep (100X)	Corning	30-002-CI
Phosphate-buffered saline (PBS)	Corning	21-040-CV
Trypan Blue (0.4% w/v)	Corning	25-900-CI
Calcium Mobilization Reagents
4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)	Thermo	172571000
Bovine serum albumin (BSA)	Avantor	97061-420
Calcium chloride dihydrate	Thermo	42352-0250
Dimethyl sulfoxide	Thermo	J66650-AD
Fluo-4/AM	Invitrogen	F14201
Hank's balanced salt solution (Ca/Mg/phenol-red free)	Corning	21-022-CV
Magnesium chloride hexahydrate	MilliporeSigma	M2393
Pluronic F-127 (Poloxamer 407)	Spectrum Chemical	P1166
Probenecid	TCI America	P1975
Sodium hydroxide	VWR International	BDH9292
TFLLRN-NH2 (TFA salt)			Prepared by Trudy Holyst at the Versiti Blood Research Institute
Materials
96-well culture-treated, black-walled, clear bottom assay plate	Corning	3603	with transparent lids
Centrifuge tube, 15 mL	Avantor	89039-664
Centrifuge tube, 50 mL	Avantor	89039-656
Culture flask, T-75	Corning	353136	tissue culture treated
Disposable reagent reservoir, 50 mL	Corning	RES-V-50-S
Enspire plate reader	Perkin Elmer		Discontinued
Microcentrifuge tube, 1.5 mL	Avantor	20170-038
Pasteur pipette, 9"	Fisher	13-678-6B	must be sterilized
PCR tube strip with separate flat cap strips	Avantor	76318-802
Pipette tips, 20 µL	Biotix	63300042	sterile, filtered tips
Pipette tips, 200 µL	Biotix	63300044	sterile, filtered tips
Pipette tips, 1250 µL	Biotix	63300047	sterile, filtered tips
Prism	GraphPad		volume 6 used
Serological pipette, 5 mL	Tradewinds Direct	07-5005
Serological pipette, 10 mL	Tradewinds Direct	07-5010
Serological pipette, 25 mL	Tradewinds Direct	07-5025