Quantitative Automated Assays in Living Cells to Screen for Inhibitors of Hemichannel Function

Reagents, Cell Culture, and Chemical Compounds

Opti-MEM culture media, 0.05% trypsin-EDTA, penicillin/streptomycin, Hank’s Balanced Salt Solution (HBSS), fetal bovine serum (FBS), Yo-Pro-1, iodide, and other chemical compounds were purchased from Invitrogen (Saint-Aubin, France), Sigma-Aldrich (St. Louis, MO), or GE Healthcare Life Science (Velizy-Villacoublay, France). IncuCyte Cytotox Red Reagent was purchased from Sartorius (Royston Hertfordshire, UK). Human glioblastoma U251 cells were purchased from Sigma-Aldrich and Ln215-YFP^QL cells were licensed by Yonsei University, Seoul, Korea. ²¹

An in-house selection of 2242 Food and Drug Administration (FDA)- or European Medicines Agency (EMA)-approved drugs and compounds under clinical assays was generated from two commercial sources, including 1280 compounds from the Prestwick Chemical Library (Illkirch, France) and 962 additional drugs that were selected in the tebu-bio library (Le Perray-en-Yvelines, France). Compounds were dissolved in 10 mM DMSO.

Cx43 Silencing Procedure

As presented in Picoli et al., ²⁴ silencing of Cx43 in U251 cells was efficiently achieved by stable transfection with shRNA-Cx43 TRCN0000344843 (shCx43-843) (targets 30-UTR region) from the Mission Sigma shRNA Library, according to the manufacturer’s instructions. Cells stably transfected with nonrelevant shRNA (indicated as shNE) were used as controls.

Cytotoxicity Assay

U251 and Ln215-YFP^QL cells were seeded at 10,000 and 24,000 cells/well, respectively, in 100 µL of fresh culture medium in 96-well tissue culture microplates. Cellular viability was assessed using the IncuCyte Cytotox Red reagent according to the manufacturer’s protocol, in the presence of Yo-Pro (5 µM) for U251 cells or iodide (140 mM) for Ln215-YFP^QL cells, in the presence or absence of Ca²⁺. Four images per well (covering 27% of the well surface) were automatically acquired using IncuCyte Zoom (Essen BioScience, Ann Arbor, MI) every 5 min for 1 h in phase contrast and in the red fluorescence channel with a 10× objective. A threshold based on the fluorescence intensity of Cytotox Red staining was applied for the selection of the dead cells.

Monitoring of Yo-Pro Incorporation in U251 Cells by Real-Time Imaging

A Yo-Pro incorporation assay has been adapted from Orellana et al. ²⁵ in 96-well plates. Briefly, U251 cells were seeded at 10,000 cells/well in 96-well tissue culture microplates (Greiner, Les Ulis, France, no. 655090) in Opti-MEM medium supplemented with 10% FBS and allowed to grow for 24 h. Cells were then washed with HBSS buffer with or without Ca²⁺ and treated simultaneously with 5 µM Yo-Pro and chemical compounds. The microplates were transferred into an automated imaging system (IncuCyte Zoom). Four images per well were acquired in phase contrast and in the green fluorescence channel with a 4× objective, 10, 15, and 20 min after compound addition. A threshold based on the fluorescence intensity of Yo-Pro green staining was applied for the selection of the Yo-Pro-positive (Yo-Pro⁺) cells. Yo-Pro uptake activity was subsequently calculated and expressed as a percentage of maximal uptake obtained in the absence of calcium, according to the following formula:

Y o − P r o u p t a k e = V a l u e c o m p o u n d − m e a n C o n t r o l + C a 2 + m e a n C o n t r o l − C a 2 + − m e a n C o n t r o l + C a 2 +

where m e a n C o n t r o l + C a 2 + and m e a n C o n t r o l − C a 2 + correspond to the mean number of Yo-Pro fluorescent cells for the bioactive and bioinactive controls, respectively, and V a l u e c o m p o u n d to the number of Yo-Pro⁺ cells after compound treatment.

Automated Monitoring of YFP^QL Quenching by Iodide Incorporation in Ln215 Cells

The YFP^QL quenching assay has been here adapted from Lee et al. to a 96-well plate format. ²¹ Briefly, YFP^QL-expressing Ln215 cells were seeded at 24,000 cells/well in 96-well tissue culture microplates (Greiner, no. 655090) in DMEM supplemented with 10% FBS and allowed to grow for 24 h. Cells were then washed with HBSS buffer with or without Ca²⁺ and treated with chemical compounds for 10 min at 37 °C. Immediately after the subsequent addition of 140 mM sodium iodide, the YFP^QL fluorescent signal was monitored using an Infinite M1000 multimode reader (Tecan, Switzerland) with acquisitions every 80 s for 560 s. Results were expressed in percentage of quenching activity according to the following formula:

Q u e n c h i n g a c t i v i t y = V a l u e c o m p o u n d − m e a n C o n t r o l + C a 2 + m e a n C o n t r o l − C a 2 + − m e a n C o n t r o l + C a 2 +

Setup of a Real-Time Microscopy-Based Assay to Quantify HC Function by Yo-Pro Uptake

Yo-Pro incorporation by U251 cells was analyzed in a 96-well format from 10 to 20 min after calcium depletion ( Fig. 1A ). The seeding density was optimized for image analysis and levels of Yo-Pro incorporation in the bioactive and bioinactive controls, that is, in the presence and absence of Ca²⁺ respectively, by monitoring the Z′ factor (see below) in different cell density conditions (from 6000 to 12,000 cells/well). As expected, in the presence of Ca²⁺, only a few cells incorporated the fluorescent dye due to the closed conformation of the HCs. In contrast, removal of calcium triggered a massive entry of Yo-Pro ( Fig. 1B ). Over the time course of the assay, the number of dead cells in the absence of Ca²⁺ is similar to the number of dead cells in the presence of Ca²⁺, with around 10 dead cells ( Suppl. Fig. S1 ), that is, a negligible value compared with the number of seeded cells (10,000 cells/well). This result indicates that the Yo-Pro uptake in the absence of Ca²⁺ is not attributable to cell death. The Z′ factor was calculated according to Zhang et al. ²⁶ for each time point in the presence or absence of Ca²⁺ used as bioactive and bioinactive controls, respectively (n = 24 for each condition). The Z′ factors were 0.46, 0.46, and 0.51 and signal-to-noise (S/N) ratios reached 5.66, 6.15, and 8.01 at 10, 15, and 20 min, respectively, indicating robust differences between bioactive and bioinactive controls (S/N ratio > 5) and sufficient robustness for automated HCS at 20 min (Z′ > 0.50) ( Fig. 1C ). As anticipated, calcium depletion resulted in a dose-dependent activation of the HC activity ( Fig. 1D ). Importantly, in the range of concentration used in the assay, DMSO had no effect on the entry of Yo-Pro ( Fig. 1E ). Since U251 glioblastoma cells express high levels of Cx43, Cx43 was assumed to be the major channel for Yo-Pro incorporation. The contribution of Cx43 to Yo-Pro entry has been validated in silencing experiments using a previously validated shRNA. ²⁴ In Cx43 knockdown cells, Yo-Pro incorporation showed a ~60 % inhibition of uptake compared with the control cells at 20 min. This result clearly indicates that Yo-Pro incorporation mainly depends on Cx43 expression ( Suppl. Fig. S2 ). Using this assay, the previously described HC modulator meclofenamic acid inhibited Yo-Pro entry up to 55% at 25 and 50 µM ( Fig. 1F ), whereas carbenoxolone, another known inhibitor, had no effect at any concentration tested in the specific conditions of the assay ( Fig. 1G ).

OPEN IN VIEWER

Figure 1.

Cell-based assay setup to follow the incorporation of the Yo-Pro fluorescent marker in U251 cells by real-time imaging. (A) Scheme of the procedure. (B) Example of image acquisition and image analysis processing of cells treated with 0.25% DMSO with or without Ca²⁺ at 20 min. A threshold was applied on the green channel, allowing the detection of Yo-Pro⁺ cells (pink mask); scale = 200 µm. (C) Scatterplots representing the number of Yo-Pro⁺ cells at 10, 15, and 20 min in the absence or presence of Ca²⁺. Z′ factors and S/N ratios are indicated for each time point (n = 24 for each control). (D,E) Dose effect of Ca²⁺ (2.5–1260 µM) and DMSO (0.01%–0.4%) on the incorporation of Yo-Pro at 20 min. Values are mean ± SD (n = 4). (F,G) Dose-dependent effect of meclofenamic acid (F) and carbenoxolone (G) tested from 6.3 to 40 µM and 1.5 to 40 µM, respectively, on the incorporation of Yo-Pro at 20 min in the absence of Ca²⁺. Values are mean ± SD (n = 5 for F or n = 3 for G).

Setup of an Automated Assay to Quantify HC Function through the Rapid YFP^QL Fluorescence Quenching after Iodide Incorporation

As described previously, HC function can be monitored through the HC-dependent passive uptake of iodide revealed by the quenching of the YFP^QL, stably expressed in Ln215 cells. ²¹ We set up the conditions to monitor the quenching of YFP^QL fluorescence following Ca²⁺ depletion in 96-well microplates ( Fig. 2A ). The adequate number of cells was determined using different concentrations of cells per well (from 10,000 to 24,000 cells) in the presence or absence of Ca²⁺, with an optimal condition of robustness and reproducibility obtained with 24,000 cells. Gain of the multimode reader was adjusted to avoid saturation of the fluorescent signal (65,000 AU of fluorescence) and finally set at less than 50% of the maximum (value maximum of <25,000 in Fig. 2B ). Furthermore, in comparison to the number of cells seeded initially (24,000/well), the number of dead cells in the absence of Ca²⁺ is comparable to the number of dead cells in the presence of Ca²⁺ and negligible when there are fewer than 20 dead cells over the time course of the assay ( Suppl. Fig. S3 ). Using this setup, calculated Z′ factors (n = 20 for each control) ranged from 0.3 to 0.5. Interestingly, S/N ratios were all strictly above 2, ranging from 2.8 to 3.1, indicating at least twofold differences between bioactive and bioinactive controls ( Fig. 2B ). Furthermore, as described above with the Yo-Pro incorporation assay, a dose-dependent effect of Ca²⁺ was observed on the YFP^QL fluorescence quenching ( Fig. 2C ), whereas DMSO had no significant effect ( Fig. 2D ). In this assay, meclofenamic acid had a light inhibitory effect on the quenching of YFP^QL fluorescence with a maximal inhibitory effect of 55% observed at 15.6 and 25 µM, while carbenoxolone inhibited the quenching in a dose-dependent manner with a graphically estimated IC₅₀ of 9.1 µM ( Fig. 2E,F ).

OPEN IN VIEWER

Figure 2.

Cell-based assay for automated monitoring of YFP^QL quenching by iodide incorporation. (A) Scheme of the procedure. (B) Ln215-YFP^QL cells were incubated with or without Ca²⁺ for 10 min; YFP^QL fluorescence was measured every 80 s immediately after iodide injection for 600 s. Values are mean ± SD (n = 20). #Calculated Z′ factors above 0.5. (C,D) Evaluation of the effect of a concentration range of Ca²⁺ (2.5–1260 µM) and DMSO (0.01%–0.40%) on the quenching of YFP^QL fluorescent protein at 160 s after iodide addition. Values are mean ± SD (n = 4). (E,F) Effect of meclofenamic acid and carbenoxolone tested from 1.5 to 40 µM on the quenching of YFP^QL fluorescence at 160 s in the absence of Ca²⁺. Values are mean ± SD (n = 3).

Screening of an In-House Collection of FDA-Approved Drugs

To identify putative modulators of HCs, an in-house collection of 2242 compounds was screened using the Yo-Pro incorporation cell-based assay by real-time imaging ( Fig. 3A ). A subset of 148 compounds reducing Yo-Pro uptake to or below 35% of bioinactive controls (fixed at 100%) were selected as primary hits ( Fig. 3B ). A set of 23 cytotoxic compounds leading to a decrease of cell confluence was discarded. Visual inspection of green channel images was then performed to eliminate an additional set of 38 false-positive hits, mainly due to fluorescence interference or compound precipitation. Finally, 50 compounds were considered as nonrelevant for the field of CNS drug discovery. This final selection led to a set of 37 FDA-approved drugs, thus representing a final hit rate of 1.65% ( Fig. 3A ).

OPEN IN VIEWER

Figure 3.

Screening workflow and results. (A) Scheme of the screening and selection process. (B) Scatterplot representing the percentage of Yo-Pro incorporation compared with controls (DMSO) at 20 min for the 2242 compounds tested at a final concentration of 25 µM (n = 2 per condition; each dot corresponds to the mean of two replicates). Compounds presenting less than 35% incorporation activity were selected for further validation. (C) Scatterplot representing the lowest value (i.e., the highest inhibitory activity) of iodide-induced YFP^QL quenching activity for the 37 compounds selected after the primary screen out of the 10 concentrations tested (1.04–40.0 µM) at a 160 s time point. The 11 compounds that also inhibited the quenching of YFP^QL by iodide incorporation are highlighted in red (B,C). (D,E) Two compounds (circles in C) were tested in dose response from 1.5 to 40 µM on either the Yo-Pro incorporation assay (dark circle) or the iodide-induced YFP^QL quenching assay (empty circle). Yo-Pro uptake or quenching activity percentages are relative to the controls. Values are mean ± SD (n = 3).

Comparison of the Activity of Primary Hits on Yo-Pro and Iodide Incorporation Assays

The selected 37 primary hits were tested at 10 concentrations in simplicate, together with meclofenamic acid and carbenoxolone, used here as controls on both the Yo-Pro incorporation assay and the iodide-induced YFP^QL quenching assay. The ability to inhibit Yo-Pro incorporation was confirmed for 25 compounds, with at least one efficient concentration (not shown). In this assay, meclofenamic acid displayed an IC₅₀ above 5 µM (not shown), a value consistent with our previous results obtained during the setup of the assay (see Fig. 1F ). Interestingly, among the 25 previously validated hits, 11 compounds also inhibited the incorporation of iodide ( Fig. 3C ). Two molecules of interest displaying different profiles of activity that are representative of those obtained with the 37 primary hits were chosen and replicated. These two molecules were hit-picked in the original library and tested in triplicates at eight doses (from 1.5 to 40 µM). Molecule 1 induced a dose–response inhibitory effect on Yo-Pro incorporation (IC₅₀ = 8.6 µM) but not on iodide entry; molecule 2 repeatedly inhibited both Yo-Pro incorporation and iodide entry (IC₅₀ = 30 and 25 µM, respectively) ( Fig. 3D,E ).