Screening for AMPA receptor auxiliary subunit specific modulators

Abstract

AMPA receptors (AMPAR) are ligand gated ion channels critical for synaptic transmission and plasticity. Their dysfunction is implicated in a variety of psychiatric and neurological diseases ranging from major depressive disorder to amyotrophic lateral sclerosis. Attempting to potentiate or depress AMPAR activity is an inherently difficult balancing act between effective treatments and debilitating side effects. A newly explored strategy to target subsets of AMPARs in the central nervous system is to identify compounds that affect specific AMPAR-auxiliary subunit complexes. This exploits diverse spatio-temporal expression patterns of known AMPAR auxiliary subunits, providing means for designing brain region-selective compounds. Here we report a high-throughput screening-based pipeline that can identify compounds that are selective for GluA2-CNIH3 and GluA2-stargazin complexes. These compounds will help us build upon the growing library of AMPAR-auxiliary subunit specific inhibitors, which have thus far all been targeted to TARP γ-8. We used a cell-based assay combined with a voltage-sensitive dye (VSD) to identify changes in glutamate-gated cation flow across the membranes of HEK cells co-expressing GluA2 and an auxiliary subunit. We then used a calcium flux assay to further validate hits picked from the VSD assay. VU0612951 and VU0627849 are candidate compounds from the initial screen that were identified as negative and positive allosteric modulators (NAM and PAM), respectively. They both have lower IC50/EC50s on complexes containing stargazin and CNIH3 than GSG1L or the AMPAR alone. We have also identified a candidate compound, VU0539491, that has NAM activity in GluA2(R)-CNIH3 and GluA2(Q) complexes and PAM activity in GluA2(Q)-GSG1L complexes.

Fig 1. Configuration of VSD assays.

(A) Arrangement of wells in compound and glutamate plates added to cells by the FDSS. (i) Compounds (orange) are added at 10 μM concentration in initial screen and as a 40 μM– 10nM curve (decreasing color saturation) for CRC testing. (ii) Controls to determine a Z’ for each plate line the edges of the glutamate plate. Positive control = 1mM glutamate (maxGLU, red), 1X FLIPR Blue dye vehicle (VHL, green), negative control = 30 μM NBQX (NBQX, pink). EC₅₀ glutamate (3–4 μM) is added across the plate (EC₅₀GLU, blue) with columns 2 and 23 used as EC₅₀GLU controls with DMSO. DMSO control was moved to column 12, in CRC plates. (B) Normalized fluorescence data (ratio of the F/F_o) readout for the FDSS on a VSD experiment showing the compound and glutamate additions at 10 sec and 300 sec, respectively. Controls are shown in colors corresponding to their colors in the glutamate plate in A(ii). Different hit windows are shaded in dark orange (CMPDslope), light orange (CMPDmaxmin), purple (GLUslope), and violet (GLUmaxmin). (C) Definition of Tier 1–4 hits in our initial screen. Hits were determined as those compounds that deviated from the mean of the test population EC₅₀GLU by more than three standard deviations in the windows specified on a per plate basis. (D) Example of a compound (black trace) that hit on (1) A2R-stg cells but not on (2) A2R or (3) TetON cells. Controls are maxGLU (red), vehicle (green), 30 μM NBQX (pink), all normalized to EC₅₀GLU (blue).

Fig 2. Behavior of established compounds in VSD assay.

(A1) Normalized fluorescence data for VSD assay on A2R-stg cells with 50 μM fluorowillardiine (black) with maxGLU (red) and NBQX (pink) controls, all normalized to EC₅₀GLU (blue). (A2) CRC curves for FW against A2-stg and (A3) A2-C3 cell lines calculated from the CMPDslope window. %max GLUslope = (CMPDslope—mean VHLslope)/(mean maxGLUslope—mean VHLslope) is further described in methods. (B1) Normalized fluorescence data for VSD assay on A2R-stg cells with 1 mM CX-546 (black) with maxGLU (red) and NBQX (pink) controls, all normalized to EC₅₀GLU (blue). (B2) CRC curves for CX-546 against A2R-stg and (B3) A2R-C3 cell lines calculated from the GLUslope window. %max GLUslope = (GLUslope—mean VHLslope)/(mean maxGLUslope—mean VHLslope) (C1) Normalized fluorescence data for VSD assay on A2R-stg cells with 500 μM cyclothiazide (CTZ, black) with maxGLU (red) and NBQX (pink), all normalized to EC₅₀GLU (blue). (C2) CRC curves for CTZ against A2-stg and (C3) A2-C3 cell lines calculated from the GLUslope window. Compound EC₅₀ values that could be reliably calculated are in the top left corner of each graph.

Fig 3. Workflow for identifying AMPAR-auxiliary subunit modulators.

(A) 39,202 compounds were initially screened using the VSD assay against A2R-stg cells. (B) 1,184 hits from (A) were counter-screened against A2R, TetON, and A2R-C3 cells. (C) 116 compounds were identified from counter-screening in (B) as being stargazin or auxiliary subunit specific (i.e. they did not hit on A2R or TetON cells). These were tested for full compound CRCs against A2R-stg and A2R-C3 cells using the VSD assay. These CRCs identified 90 hits that fit to sigmoidal dose response curves with potency under 10 μM. (D) We identified 39 stargazin specific PAMs, 2 CNIH3 specific PAMs, and 36 PAMs that had activity in both A2R-stg and A2R-C3 cells. We also found 1 stargazin specific NAM and 9 compounds with NAM activity on both cell lines. Three compounds gave opposite effects in the two cell lines. Hits were discarded for reorder if they showed activity in the compound only window. Hits with activity in the CMPD only windows were discarded. (E) 57 of the 90 compounds in (D) were re-screened with new batch samples as compound CRCs in the VSD assay. (F) 57 hits were tested in the glutamate potency fold-shift calcium flux assay and 28 were subjected to a full compound CRC calcium flux assay to study their effects using an orthogonal approach.

Fig 4. Controls and experimental setup for calcium flux assays.

(A) CRC curves for CTZ in the presence of 1 mM glutamate from (1) A2Q-stg, (2) A2Q-C3, (3) A2-GSG, (4) A2Q cell lines in the calcium flux assay. These CRCs are calculated from the GLUslope1 (t = 122-125s) window. Calculated EC₅₀ values are included in the top left of the graph. %max GLUslope = (GLUslope–mean VHLslope) / (mean maxGLUslope–mean VHLslope) as described in methods. (B) Compound and glutamate plates added to cells by the FDSS in our Fluo-8 calcium flux compound CRC assay. (i) Compounds (orange) are added as a 30 μM– 30 nM CRC (decreasing color saturation). (ii) Controls to determine a Z’ for each plate line the edges of the glutamate plates. Positive control = 250 μM glutamate (maxGLU, red), high calcium buffer vehicle (VHL, green), and negative control = 30 μM NBQX (NBQX, pink). 1 mM glutamate is added across the plate (blue) with columns 12 and 23 used as a 1mM glutamate and DMSO control. (C) Vehicle subtracted, normalized fluorescence data readout for the FDSS on A2Q-stg cells in a calcium flux experiment showing the compound and glutamate applications at 10s and 120s, respectively. Controls are shown in colors corresponding to their colors in the glutamate plate in (D). Different hit windows are shaded in blue (GLUslope1, 122-125s), red (GLUslope2, 126-132s), yellow (GLUslope3, 140-150s), green (GLUmaxmin). The orange window is a reference baseline (CMPD baseline) prior to the glutamate addition used to determine if the compound shows activity in the absence of glutamate. (D) Compound and glutamate plates added to cells by the FDSS in our Fluo-8-based glutamate potency fold-shift assay. (i) Compounds (orange) are first added at 30 μM with DMSO controls in columns 1,24, and also in K1-K12, F13-23, P13-23, overlapping with glutamate concentration curves for a per plate comparison to compound. (ii) Controls are loaded on the edge as in (Bii) and a glutamate CRC is loaded horizontally ranging from 4 mM to 10 pM (decreasing color saturation). (E) An example of the readout for our glutamate potency fold-shift assay. A rightward shift of the NBQX pretreated cells (red) as compared to glutamate alone (black) indicates NAM activity. The leftward shift of CTZ pretreated cells (grey) indicates PAM activity.

Fig 5. Table of results from VSD and calcium flux assays.

EC₅₀ or IC₅₀ values determined by CRC fits from GLUslope. Boxes highlighted green indicate an EC₅₀ or a positive trend, red indicate an IC₅₀ or a negative trend. Estimated EC₅₀ values are added in italics, but are estimated due to incomplete CRC curves or insufficient differences in %max GLUslope across the CRC. Glutamate potency fold-shift assays indicate how much fold-change occurred in the glutamate EC₅₀ when cells were pretreated with 30 μM compound. Values greater than 2 indicate PAM activity and less than 1 indicate NAM activity.

Fig 6. Chemical structures of our candidate hits.

(A) Structure of VU0612951 highlighting the 1,3-triazole group in red. (B) Structure of VU0627849 highlighting the isoxazole group in red. (C) Structure of VU0539491 highlighting the 1,2,4-oxadiazole group in red.

Fig 7. Characterization of VU0612951.

(A) Raw data for compound CRCs in the VSD assay. A2R-stg (orange) and A2R-C3 (blue). EC₅₀GLU traces are represented by dashed lines. Compound concentrations are indicated in the top left corner. (B) CRCs calculated from GLUslope in the VSD assay for A2R-stg and A2R-C3 cells. Error bars are standard deviations. (C) Raw data for compound CRCs in the calcium flux assay. A2Q (red), A2Q-stg (orange), A2Q-GSG (green), and A2Q-C3 (blue). Dashed lines are signal of 1mM glutamate without compound. (D) Compound CRCs in calcium flux assay for A2Q-stg, A2Q-C3, A2Q-GSG, and A2Q cell lines. These are derived from the GLUslope1 window. (E) Compound CRCs calculated from the AUC in the GLUmaxmin window of the calcium flux assay plotted as %max AUC in the GLUmaxmin window (see Fig 4C) vs. log [compound].

Fig 8. Characterization of VU0627849.

(A) Raw data for compound CRCs in the VSD assay. A2R-stg (orange) and A2R-C3 (blue). EC₅₀GLU traces are represented by dashed lines. Concentrations of compound are indicated in the top left corner. (B) CRCs calculated from GLUslope in the VSD assay for A2R-stg and A2R-C3. (C) Raw data for compound CRCs in the calcium flux assay. A2Q (red), A2Q-stg (orange), A2Q-GSG (green), and A2Q-C3 (blue). 1 mM glutamate traces are represented by dashed lines. (D) CRCs calculated from GLUslope1 in calcium flux assay for A2Q-stg, A2Q-C3, A2Q-GSG, and A2Q cell lines. Plotted as %max GLUslope vs. log [compound].

Fig 9. Characterization of VU0539491.

(A) Raw data for compound CRCs in the VSD assay. A2R-stg (orange) and A2R-C3 (blue). EC₅₀GLU traces are represented by dashed lines. Compound concentrations are indicated in the top left corner. (B) CRCs calculated from GLUslope in VSD assay for A2R-stg and A2R-C3. (C) Raw data for compound CRCs in the calcium flux assay. A2Q (red), A2Q-stg (orange), A2Q-GSG (green), and A2Q-C3 (blue). Traces obtained from applying 1 mM glutamate are represented by dashed lines. (D) Compound CRCs calculated from the GLUslope1 window in our calcium flux assay for A2Q-stg, A2Q-C3, A2Q-GSG, and A2Q cell lines. These show negative and positive trends but no curve fits. (E) Compound CRCs calculated from the AUC in the GLUmaxmin window of the calcium flux assay plotted as %max AUC vs. log [compound] as in (7E).

Fig 10. Electrophysiology with VU0627849.

(A) Whole-cell recordings of A2R cell line with (blue) and without (red) VU0627849 (40μM). (B) Whole-cell recordings of A2R-stg cell line with (blue) and without (red) VU0627849 (40μM). Recording after washout of drug is in black. In these experiments, glutamate (1mM) is applied for 100 ms and 20 ms pulses with or without VU0627849.