FRET assay for live-cell high-throughput screening of the cardiac SERCA pump yields multiple classes of small-molecule allosteric modulators

Figures

• 

  Figure 1. (A) FRET biosensor, human two-color SERCA2a (2CS), showing SERCA domains: nucleotide binding (N), phosphorylation (P), actuator (A) and transmembrane (TM). GFP is fused to N and RFP is fused to A. DFRET, measured from DFLT (fluorescence lifetime), is used to detect SERCA structural changes (B) Screening funnel describing the 5-step process undertaken in this study, involving measurements of FLT and SERCA function, with SERCA in live mammalian cells and in isolated pig cardiac SR membranes, respectively. (1) FLT changes caused by test compounds were measured using the target SERCA-specific FRET biosensor 2CS, to identify initial hit compounds. False hits were ruled out as compounds that (2) are fluorescent or affect the donor directly, (3) decrease FLT, or (4) affect FRET in a null biosensor, in which donor and acceptor are separated by a non-functional flexible peptide. (5) Concentration-dependence of FLT and SERCA function was measured to further prioritize hit compounds that could result in future lead development. Experimental details are provided in Methods.

• 

  Figure 2. The human cardiac 2-color SERCA2a biosensor (2CS) in HEK293 cells was used to screen the 50,000-compound DIVERSet library. (A) Screen precision was determined by computing % CV for each plate using DMSO control wells, with a median value of 0.4% across 40 plates. (B) Change in lifetime (Dt) was computed to find potential hits (red points) with a hit threshold set at a robust z-score of 3, resulting in 2960 FLT hit compounds for triage with the SUPR instrument and two-channel lifetime detection. DMSO controls (dark blue) were grouped in the plot to better illustrate plate boundaries. (C) The histogram of compounds not affecting 2CS (light blue, 1ps bin width) are normally distributed, similar to that of DMSO alone as shown by a fit of the populations to Gaussian distributions, which are shown in log scale. The horizontal lines in B, C illustrate the approximate cutoffs used, though actual cutoffs were determined on a plate-by-plate basis.

• 

  Figure 3. DIVERSet hit compound FRET retests. The most promising 158 hit compounds were selected (“cherry-picked”) and dispensed into 1536-well plates at 10 and 30 mM concentrations (n=3 wells for each drug concentration). Data is shown from the 30 mM screens for SERCA2a (A, C) and null biosensor (B, D). (A) Distribution of significant lifetime changes for the 2CS biosensor. (B) A null biosensor was counter-screened against the 158 hit compounds. Only five compounds displayed a lifetime change greater than 50 ps, indicating that our method for eliminating fluorescent compounds removes nearly all false positives. (C) and (D) Plots of the change in fluorescence lifetime vs the R/G ratio show excellent, reproducible correlation between the two techniques of lifetime (DFLT) and SUPR (DR/G) for the 2CS biosensor (C) and null biosensor (D).

• 

  Figure 4. A representative activator compound of both Ca2+-ATPase and Ca-transport at saturating Ca2+ (VMAX). (A) Chemical structure of Compound DS26022409 (Compound 7). (B) CRC of normalized FRET E in live HEK cells shows decreasing FRET response with increasing [compound]. (C) CRC of Ca2+-ATPase of SERCA2a in pCSR vesicles show activation at high Ca2+ (VMAX, black) and at midpoint Ca2+ (VMID, red). (D) CRC of Ca-transport shows activation at both VMAX and VMID Ca2+ (black and red, respectively). Data are presented as mean ± SEM, n = 3, *p<0.05.

• 

  Figure 5. A representative activator compound that decreases FRET and uncouples Ca2+-ATPase from Ca2+-transport activity. (A) Chemical structure of Compound DS26418355 (Compound 8). (B) CRC of normalized FRET E in 2CS biosensor in live HEK cells shows decreasing FRET response with increasing [compound]. (C) CRC shows Ca2+-ATPase activation in pCSR vesicles under both VMAX (black) and VMID (red) Ca2+. (D) Activation was less for Ca2+-transport of SERCA2a in pCSR vesicles. Mean ± SEM, n = 3, *p<0.05.

• 

  Figure 6. A representative activator compound that decreases FRET, increases Ca2+-ATPase activity, and has divergent effects on Ca2+-transport activity. (A) Chemical structure of Compound DS12165787 (Compound 6). (B) Concentration response curve (CRC) of normalized FRET E shows decreased FRET with a half maximal effect (EC50) at 7.1 ± 0.2 mM. (C) CRC shows Ca2+-ATPase activation of SERCA2a in pig CSR vesicles at VMAX (black, pCa 5.4) and VM!D (red, pCa 6.2) [Ca2+] (D) CRC of Ca2+-transport of SERCA2a in pig CSR vesicles, showing inhibition for VMAX (black) and activation for VMID (red). Grey data point was omitted from fitting. D, C10, and EC50 are reported in Table 1 for (C) and (D). Data is presented as mean ± SEM, n = 3, *p<0.05.

• 

  Figure 7. A representative inhibitor compound that severely decreases FRET and inhibits both Ca-ATPase and Ca-Transport activity. (A) Chemical structure of Compound DS27464324 (Compound 12). (B) CRC of normalized FRET E shows decreased FRET response in 2CS biosensor. (C) CRC shows inhibition of Ca-ATPase in pCSR under both VMAX (black) and VMID (red) [Ca]. (D) CRC shows inhibition of Ca-transport in pCSR under VMAX (black) and VMID (red) [Ca]. Data is presented as mean ± SEM, n = 3, *p<0.05.