Supercharging the calcium pump: Identification of an activation hotspot on SERCA by cryo-EM

Abstract

The sarco-endoplasmic reticulum Ca²⁺-ATPase (SERCA) is a ubiquitous P-type ATPase that restores cytosolic Ca²⁺ to the sarco-endoplasmic reticulum. SERCA is essential for cardiac Ca²⁺ cycling and cellular energy metabolism. Several small molecules enhance SERCA function and show promise in models of metabolic and cardiovascular diseases. However, the structural basis for SERCA activation has remained unknown, hindering mechanism-driven lead optimization. Here we present cryo-EM structures of SERCA bound to two chemically distinct activators: the quinoline derivative CDN1163 (2.6 Å resolution) and a benzofuran derivative UM-52 (3.1 Å resolution). Biochemical assays show that both compounds stimulate Ca²⁺-dependent ATPase activity of SERCA without altering the apparent Ca²⁺ affinity. The structures reveal a previously unrecognized “activation hotspot” in the transmembrane domain, a shallow groove formed by helices M3 and M4 and capped by M1. Despite low chemical similarity, both activators occupy the same pocket and share conserved interactions with Ser²⁶⁵, Trp²⁷², and Phe²⁹⁶. These residues are unique to SERCA and help explain selectivity relative to other P-type ATPases. Activator binding stabilizes a catalytically competent conformation, shifting SERCA toward an E1-like state poised for ATP binding and coordinated movements of the M1-M4 bundle and the cytosolic domains. Notably, density consistent with a detergent acyl chain bridges an otherwise open cavity adjacent to the compound, suggesting that altered protein-lipid interactions may contribute to activation. Together, these findings define a structural framework for SERCA activation and provide a blueprint for rational design of next-generation SERCA activators.

INTRODUCTION

The sarco-endoplasmic reticulum calcium ATPase (SERCA, EC 7.2.2.10) is an integral membrane protein that is responsible for pumping calcium back into the sarco- or endoplasmic reticulum (SR/ER) from the cytosol. SERCA is found in all eukaryotic cells, where it plays a major role in intracellular calcium homeostasis. Due to its prevalence in the ER of all cell types and the SR of cardiac and skeletal muscles, dysregulation of SERCA-dependent calcium homeostasis leads to a variety of disease states including metabolic disorders, type 2 diabetes, cardiomyopathies, and muscle-wasting diseases. Recent efforts to improve SERCA-dependent calcium homeostasis have focused on small molecules that increase SERCA’s activity. CDN1163 is the most extensively studied small molecule activator of SERCA, demonstrating positive outcomes on metabolic efficiency including reduced ER stress, blood glucose levels, and hepatic lipid accumulation, as well as improved mitochondrial function (1). Additional efforts have identified small molecule activators that increase SERCA activity as much as ~1.6 fold (2-4). Recently, small-molecule activators, including compound UM-52, were shown to significantly reverse methylglyoxal-induced inhibition of SERCA, revealing a novel therapeutic approach to preserve ER calcium homeostasis in diabetes (5). Despite these advances, there are no structures of SERCA bound to small molecule activators, limiting our understanding of how small molecules increase SERCA’s activity and hindering lead compound optimization.

RESULTS

We determined cryo-EM structures of SERCA bound to two chemically distinct activators, CDN1163 bound at 2.6 Å resolution and a recently developed small-molecule activator UM-52 (3) bound at 3.1 Å resolution (Table 1). The two structures reveal a novel binding site on SERCA that appears to be unique among the P-Type ATPase family, thereby explaining their selectivity for SERCA. Upon activator binding, SERCA adopts a conformation that is poised for ATP binding, likely shifting the conformational equilibrium in favor of the calcium- and ATP-bound E1 state of SERCA.

Table 1:

Cryo-EM data collection, refinement and validation statistics.

	SERCA+CDN1163 (EMDB-73823) (PDB 9Z5S)	SERCA+UM-52 (bound) (EMDB-73820) (PDB 9Z5O)	SERCA+UM-52 (apo) (EMDB-73822) (PDB 9Z5R)
Data collection and processing
Microscope	Titan Krios G3i	Titan Krios G3i	Titan Krios G3i
Detector	Falcon 4i	Falcon 4i	Falcon 4i
Energy Filter	Selectris X	Selectris X	Selectris X
Magnification	165,000x	165,000x	165,000x
Voltage (keV)	300	300	300
Electron exposure (e−/Å2)	50	50	50
Defocus range (μm)	−0.6 to −1.8	−0.6 to −1.8	−0.6 to −1.8
Pixel size (Å)	0.743	0.743	0.743
Symmetry imposed	C2	C1	C1
Initial particle images (no.)	2,513,942	2,573,740	2,573,740
Final particle images (no.)	104,006	143,855	83,166
Map resolution (Å)	2.5	3.1	3.8
FSC threshold	0.143	0.143	0.143
Map resolution range (Å)	2.5 – 39.2	3.1 – 46.9	3.8 – 60.9
Refinement
Initial model used (PDB code)	2ZBD	2ZBD	2ZBD
Model resolution (Å)	2.8	3.2	4.1
FSC threshold	0.5	0.5	0.5
Map sharpening B factor (Å2)	−58.9	−86.2	−116.6
Model composition
Non-hydrogen atoms	15,614	7717	7668
Protein residues	1986	993	993
Ligands	Ca2+: 4 CDN1163: 2 C12E8: 2	Ca2+: 2 UM-52: 1	Ca2+: 2
B factors (Å 2 )
Protein	158.54	54.89	183.53
Ligand	90.43	35.85	122.34
R.m.s. deviations
Bond lengths (Å) (no. >4σ)	0.002(0)	0.002(0)	0.003(0)
Bond angles (°) (no. >4σ)	0.480(0)	0.393(0)	0.647(0)
Validation
MolProbity score	2.14	2.23	2.26
Clashscore	28.55	18.06	13.35
Poor rotamers (%)	0.95	2.50	2.86
Ramachandran plot
Favored (%)	96.77	96.97	95.96
Allowed (%)	3.13	3.03	3.94
Disallowed (%)	0.10	0.00	0.10
Correlation model versus data
CC (mask, volume)	0.87(0.85)	0.88(0.87)	0.85(0.84)
CC (ligands)	0.85	0.80	0.90

CDN1163 is a quinoline derivative first reported in 2016 as a small-molecule activator of SERCA (Figure 1A). ATPase assays of SERCA proteoliposomes showed that CDN1163 increased calcium-dependent ATPase activity by approximately 1.5-fold (Supplementary Data Figure 1). Similarly, surface-electrogenic event reader (SEER) experiments showed that CDN1163 also increased calcium transport by SERCA (6). In this study, single particle cryo-EM resolved two different populations for SERCA in the presence of CDN1163 (Supplementary Data Figure 2), a monomer and an antiparallel dimer (cytoplasmic domains oriented on opposite sides of the micelle). The presence of SERCA monomers and dimers was also observed by mass photometry (Supplementary Data Figure 3). Interestingly, density corresponding to CDN1163 was only found at the dimerization interface between SERCA molecules in the antiparallel dimer (Figure 1A). We hypothesized that the antiparallel dimer stabilized CDN1163-bound SERCA, preventing its dissociation before freezing for cryo-EM single-particle analysis. We also assumed that activator binding must be inherently distinct from SERCA inhibitors such as thapsigargin (7) and cyclopiazonic acid (8, 9), which occupy relatively deep binding pockets and lock SERCA in a calcium-free state that prevents progression through the calcium transport cycle (Supplementary Data Figure 4).

Figure 1:

(A) Cryo-EM density map (mesh) of SERCA (blue and red ribbon diagrams) in the presence of CDN1163 adopted an antiparallel dimer configuration. Density for SERCA was represented as a mesh, whereas activator density was represented as a grey volume. The chemical structure of CDN1163 is shown. (B) Cryo-EM density map (mesh) of SERCA (blue ribbon diagram) in the presence of UM-52 was predominantly monomeric, with UM-52 binding at the same site as CDN1163. Density for SERCA was represented as mesh, whereas UM-52 density was represented as a grey volume. The chemical structure of UM-52 is shown. (C) Number of contacts between CDN1163 and UM-52 and specific residues in M1, M3, and M4, obtained from molecular dynamics simulations and represented as violin plots. The residues indicated by asterisks are shown as spheres on the ribbon diagram of SERCA with CDN1163 (yellow) and UM-52 (green). Inset shows the hydrogen bond distances between Ser²⁶⁵ and the quinolone nitrogen of CDN1163 (3.0 Å) and the benzofuran oxygen of UM-52 (2.9 Å). (D) Superposition of apo SERCA (blue ribbon diagram) versus CDN1163- and UM-52-bound SERCA (grey ribbon diagrams) indicating the domain movements upon CDN1163 and UM-52 binding.

The structure of CDN1163 bound to SERCA suggested two possible interpretations. CDN1163 binding to an activation site on SERCA may have provided a complementary interface for antiparallel dimer formation. Alternatively, a preformed antiparallel dimer of SERCA may have provided an interface for CDN1163 binding that was not relevant to the mechanism of activation. To discriminate between these two possibilities, we determined the structure of SERCA in the presence of a recently developed SERCA activator, UM-52, a potent activator with low chemical resemblance to CDN1163 (Figure 1B; ~24% RDKit Fingerprint and Tanimoto similarity coefficient). ATPase assays of SERCA proteoliposomes showed that UM-52 increased calcium-dependent ATPase activity by approximately 1.6-fold (Supplementary Data Figure 1). Interestingly, cryo-EM images of SERCA in the presence of UM-52 appeared mainly monomeric, which could be further separated into apo and bound populations (Supplementary Data Figures 5 & 6). The activator UM-52 was found to bind to a similar site to CDN1163 involving transmembrane segments M3 and M4 of SERCA. While UM-52 has a 2,3-dihydrobenzofuran group instead of the quinolone derivative in CDN1163, the benzofuran oxygen of UM-52 and the quinolone nitrogen of CDN1163 form a hydrogen bond with Ser²⁶⁵ on M3 (Figure 1C). Both UM-52 and CDN1163 cause SERCA to adopt the same conformation (RMSD 0.3 Å), suggesting that their modes of activation are similar (Figure 1D). We were able to resolve a SERCA monomer bound to the activator, UM-52, suggesting that it may form a more stable interaction with SERCA compared to CDN1163. This may be due to the interaction of a phenyl ring of UM-52 with Trp²⁷² of SERCA (Figure 1C). Despite these differences, the observation that two chemically distinct activators occupy the same groove on SERCA supports the existence of a genuine activator site and indicates that the antiparallel SERCA dimer did not force CDN1163 into an artificial binding mode.

The activators bind along transmembrane segment M3 of SERCA in a shallow groove lined by M4. M1, M3, and M4 play different roles in the formation of the activator binding site. M4 runs parallel to the activators and small hydrophobic residues (e.g. Ala³⁰³, Ala²⁹⁹) allow formation of a cavity that accommodates the quinoline and benzofuran groups of CDN1163 and UM-52, respectively (Figures 1C & 2A; Supplementary Data Figure 7). M3 also runs parallel and lines one face of the activator binding site, making the closest contacts among the three helices and contributing van der Waals interactions via several small non-polar residues to stabilize the activators (e.g. Cys²⁶⁶, Val²⁶⁹). M3 contains two key residues that are unique to SERCA that are important for positioning the activators in the binding pocket. Trp²⁷² stabilizes one end of the activators through CH-π interactions with the isopropyl-group of CDN1163 and π-stacking with the phenyl-group of UM-52 (Figure 1C). Phe²⁹⁶ in M4 is near Trp²⁷² and could also play a role in forming an aromatic cleft and π-network to stabilize the activators (both residues are ~4 Å away from the isopropyl group). Importantly, hydrophobicity in this region of the pocket appears to be essential for stabilizing activator binding, as polar substitutions on the aryl-ring of UM-52 derivatives yield inactive compounds (3). As mentioned above, Ser²⁶⁵ is an important residue unique to SERCA that forms a hydrogen bond with the activators, aiding in their positioning in the binding site. The specificity of the activators for SERCA is likely due to Ser²⁶⁵ as other P-type ATPases do not have a residue that can act as a hydrogen-bond donor at this position (Figure 2 & Supplementary Data Figure 8). Ser²⁶⁵ appears to play an important role in activator positioning because UM-52 sits lower in the binding pocket compared to CDN1163 to allow formation of a hydrogen bond between the 2,3-dihydrobenzofuran group and the serine (Figure 1C). Recent SERCA activator development has noted the importance of a hydrogen-bond capable atom (with hydrogen-bond accepting oxygen producing higher stimulation than hydrogen-bond donor nitrogen) in the fused-ring of the activators, supporting the role of Ser²⁶⁵ in facilitating hydrogen-bond interactions (3). Importantly, the structures support the notion that balanced polarity within the fused ring is essential for maintaining activity, consistent with experimental data showing that even modest changes in polarity convert SERCA activators into inhibitors (3). Thus, M3 interactions appear necessary for activator binding, while M4 provides a sufficient cavity for insertion of the quinoline and benzofuran ‘head groups’ of the activators.

Figure 2:

(A) Surface representation of SERCA with CDN1163, UM-52, and a C12E8 molecule represented as sticks within the binding site of SERCA. A ribbon diagram is also shown as an inset. (B) Surface representation and ribbon diagram of the Na⁺,K⁺-ATPase aligned with SERCA (RMSD 3.7 Å; PDB code 7E21) indicating the relative positions of the activators. Surface representations and ribbon diagrams of the (C) plasma membrane Ca²⁺-ATPase aligned with SERCA (RMSD 2.9 Å; PDB code 9GSH) and (D) secretory pathway Ca²⁺-ATPase aligned with SERCA (RMSD 4.0 Å; PDB code 8IWW) indicating the relative positions of the activators. Steric clashes with the other P-type ATPases likely prevent binding of the SERCA activators.

M1 appears to play a relatively minor role, with Leu⁶⁰ and Ile⁶⁴ forming the cap of the activator head group binding site (Supplementary Data Figure 7). Interestingly, there is a large gap in the binding cavity (~10 Å distance) between the activator head groups and M1, where van der Waals interactions were absent. However, there was a long density that could be fit by a C₁₂E₈ detergent molecule in the SERCA-CDN1163 map that tunneled from the aqueous environment into the space between CDN1163 and M1 (Figure 2A). This suggests that an acyl chain of a detergent or lipid may stabilize CDN1163 and provide a mobile element that links activator binding to M1. A lipid acyl chain occupying the space between activator and M1 may allow the movements of M1, M2 and the A domain that facilitate ATP-hydrolysis and calcium transport while still activator-bound. The presence of an acyl chain could explain why small-molecule activators have planar head groups, as the M1 gap and lipid binding may be necessary for activation. We hypothesize that a potential activator having a chemical moiety that fills this gap could immobilize M1 and lead to SERCA inhibition, effectively turning the activator into an inhibitor.

DISCUSSION

Several potential binding sites have been proposed for activator binding, including both the cytoplasmic and transmembrane domains of SERCA (4), but these studies did not predict the binding site presented here. Interestingly, the activator binding site coincides with an alternative binding mode of phospholamban (PLN) and sarcolipin (SLN) to SERCA (10, 11). PLN and SLN were found to interact with transmembrane segment M3 and modulate the maximal activity of SERCA, which is consistent with the binding of CDN1163 and UM-52 to M3 of SERCA. Among the P-type ATPases, SERCA possesses a novel binding site for small-molecule activators in the groove formed by M1, M3, and M4 (Figure 2A). Surface representation of other P-type ATPases whose structures are known, such as the plasma membrane calcium ATPase, the Na⁺, K⁺ ATPase, and the secretory pathway calcium ATPase, reveal that these family members lack a cavity along M3 and M4 that could accommodate CDN1163 or UM-52 (Figure 2). While there are cavities in the M3/M4 region of PMCA2 and SPCA1 that may be suitable for structure-guided drug development (Figure 2C, D), none of the other P-type ATPases can accommodate the chemical structures of the SERCA activators, suggesting that this effector site is unique to SERCA.

The structure of SERCA in the bound complexes is similar for both CDN1163 and UM-52 (RMSD 0.3 Å). When compared to the apo (unbound) structure of SERCA in the presence of UM-52, they reveal the mechanism behind SERCA activation that has remained elusive, where both small molecules cause SERCA to adopt a conformation that closely resembles the E1, calcium- and ATP-bound state of SERCA. The activator-bound structures of SERCA are most similar to SERCA2b in the E1·2Ca state (RMSD 1.3 Å; PDB code 7E7S (12)), indicating that CDN1163 and UM-52 promote a conformation of SERCA with bound calcium that is poised to bind ATP without significant changes in the cytoplasmic domains. We suggest that activator binding in the presence of calcium causes SERCA to adopt a catalytically competent conformation that will readily hydrolyze ATP, thus shifting the equilibrium towards E1~P·ADP·2Ca conformation. In comparison to the unbound apo structure of SERCA (Figure 1D), M1-M4 undergo a ~10° rotation toward the cytoplasmic domains and the A domain undergoes a ~26° upward rotation and contacts the N-domain to poise SERCA for ATP binding. With movement of the A domain, the cytoplasmic domains are in a more compact conformation compared to the E1·2Ca conformation, which aligns with previous TCSPC studies showing that CDN1163 induced closure of the cytoplasmic headpiece (13). The apo structure of SERCA is most similar to the E1-like state in the presence of SLN (14, 15) and PLN (16) (e.g. RMSD 2.5 Å; PDB code 3W5B), though calcium is bound in this case rather than magnesium (14, 15).

Overall, the structures presented here suggest a novel mechanism where chemically distinct small-molecule activators bind to a similar site in the transmembrane domain of SERCA and promote the transition from the E1-like state to the catalytically competent E1·2Ca·ATP state of the calcium pump. The question remains as to how the small molecules promote this transition. The location of the binding site and the presence of an acyl chain in the SERCA-CDN1163 complex (Figure 2A) suggest that activator binding may alter lipid interactions as a mechanism for increasing SERCA activity.

METHODS

Purification of SERCA from rabbit hind leg muscle

Purification of SERCA was carried out as previously described (17, 18). Briefly, rabbit hind leg, fast-twitch muscle was harvested, homogenized, and sarcoplasmic reticulum (SR) membranes were collected by differential centrifugation. SR membranes were solubilized in 1% C12E8 (Sigma-Aldrich) and SERCA was purified by affinity chromatography using Reactive Green-19 cross-linked agarose (Sigma-Aldrich). Fractions containing SERCA were concentrated using a 100 kDa MW cut-off Amicon centrifugal filter to approximately 30 μL and incubated with 10 μM CDN1163 (MedChemExpress). Sample was loaded onto a Superdex 200 column pre-equilibrated with 50 mM MOPS pH 7.0, 5% glycerol, 100 mM KCl, 1 mM CaCl₂, 10 μM CDN1163, 1 mM DTT, and 0.006% C₁₂E₈. Fractions were pooled and concentrated to approximately 20 μL, flash frozen in liquid N₂ and stored at −80°C. Concentrated SERCA samples were analyzed by SDS-PAGE (19) to assess purity. Sample preparation for SERCA-UM-52 complex followed the same procedure.

ATPase activity assays

C₁₂E₈-solubilized SERCA was reconstituted into proteoliposomes as previously described (20). Briefly, SERCA was added to a thin-lipid film of egg yolk phosphatidylcholine and egg yolk phosphatidic acid (EYPC & EYPA; Avanti Polar Lipids) and Biobeads SM-2 (Bio-Rad Laboratories) were slowly added over a four-hour time course to promote proteoliposome formation through detergent adsorption. Proteoliposomes were isolated by ultracentrifugation at 137,000 RCF with a 20/50 % sucrose step-gradient. Proteoliposomes were collected from the sucrose step-gradient interface, flash frozen in liquid N₂ and stored at −80°C.

Calcium-dependent ATPase activity assays were performed as previously described using a coupled enzyme assay adapted to a 96-well format (21, 22). A Biotek Epoch2 (Agilent) plate reader was used to measure the decrease in absorbance at 340 nm due to NADH depletion coupled to ATP hydrolysis over a range of calcium concentrations in the presence or absence of 100 μM CDN1163 or UM-52. The purity of the activators was >95% by HPLC. Specific activities of SERCA were calculated at each calcium concentration range, and SigmaPlot (Grafiti LLC) was used to fit the data to the Hill equation.

Sample quality control by mass photometry

Mass photometry (TwoMP, Refeyn Ltd. (23)) was used to monitor sample quality for cryo-EM by determining the oligomerization behavior and relative detergent micelle concentrations of candidate samples following gel-filtration chromatography and sample concentration. Mass calibrations were conducted using bovine serum albumin and apoferritin as standards in PBS buffer. For SERCA-containing samples, buffer (20 mM MOPS pH 7.0, 5% glycerol, 100 mM KCl, 1 mM CaCl₂, 1 mM DTT, 0.004% C₁₂E₈) was used to focus the TwoMP (20 μL) followed by the addition of 0.2 μL of SERCA sample. One-minute recordings were collected and processed using the DiscoverMP software. Samples that generated distinct SERCA peaks with relatively small peaks for free detergent micelles were selected for preparation of frozen-hydrated grids, cryo-EM, and single particle analysis.

Cryo-EM grid preparation

Quantifoil^® Cu 300 R 2/1 grids were glow-discharged in a Pelco easiGlow^™ for 30 seconds at 15 mV. SERCA (3 μL at 10 mg/mL) in the presence of activators (10 μM) was applied to the grids, blotted for 7-10 s with zero blot force, then plunge-frozen in liquid ethane using a FEI Vitrobot mark IV (Thermo Fisher Scientific) with 100% humidity at 4°C. Frozen grids were screened at the Stanford-SLAC cryo-EM centre (S²C²) on a Titan Krios operated at 300 kV equipped with a Selectris X energy filter and a Falcon 4i detector (Thermo Fisher Scientific). Data was collected using EPU in fast acquisition mode on candidate grids at a nominal magnification of 165k with a pixel size of 0.743 Å. Images were collected with a 6.2 s exposure over 43 frames, resulting in 1.16 e⁻ /Ų per frame with a total dose of 50 e⁻. The defocus range of the data was −0.6 μm to −1.8 μm.

Data Processing and Model Building

Micrographs were processed using CryoSPARC (Structura Biotechnology Inc. (24)), with workflow summarized in the Extended Data Figures. Density maps for a SERCA monomer and antiparallel dimer in the presence of CDN1163 were determined at 3 Å and 2.6 Å, respectively. Unsharpened maps were used for subsequent model building. Density maps for unbound (apo) and bound SERCA monomers in the presence of UM-52 were determined at 3.8 Å and 3.1 Å, respectively. The atomic coordinates for SERCA were obtained from a previously published structure (PDB ID: 2ZBD), with individual domains of SERCA (A-, N-, P- and TM-domains) isolated as separate PDB files, and each domain was sequentially fitted into the cryo-EM density maps using Phenix dock in map program (25). Manual model-building along with ligand placement was performed in COOT (26), with eLBOW (27) used to generate CDN1163 and UM-52 restraints from the SMILES input. Models were refined in Phenix using real-space refinement after each manual model-building stage, with Molprobity (28) used for model validation. Model statistics can be found in Extended Data Table 1. Figures were created using USCF Chimera (29) and ChimeraX (30).

Molecular dynamics simulations

The structures of SERCA bound to CDN1163 or UM-52 were inserted in a pre-equilibrated 120×120 Å bilayer that contained POPC and POPE (2:1 ratio) lipids to mimic the lipid composition of the SR. For lipids, we used the LIPID21 force field (31) and a cutoff distance of 10 Å for non-bonded interactions. We used the replacement method to generate lipid packing around the protein-ligand complex. We solvated each system using the OPC water model with a minimum margin of 20 Å between the protein and the z-axis edges. K⁺ and Cl⁻ ions were added to neutralize the system and to produce a KCl concentration of 100 mM. The systems were prepared using the CHARMM-GUI web server (32, 33). Energy minimization and equilibration were performed as follows: two 25-ps restrained canonical ensemble (NVT) simulations, one 25-ps restrained isothermal-isobaric ensemble (NPT) simulation, three 250-ps restrained NPT simulations, and a single unrestrained NPT simulation for 5 ns. The Langevin thermostat was used to keep the temperature at 25°C and the Monte Carlo barostat to maintain a constant pressure of 1.0 bar. Bonds involving hydrogen atoms were constrained using the SHAKE algorithm. We performed three independent 1-μs MD simulations of the complexes using AMBER24 on Tesla V100 GPUs (34) and the AMBER ff14SB force field (35). Data analysis was performed using VMD 2.0 (36).

SIGNIFICANCE STATEMENT.

SERCA pumps Ca²⁺ into the sarco-endoplasmic reticulum, enabling muscle relaxation and shaping calcium signals across tissues. Small-molecule SERCA activators improve cardiac and metabolic phenotypes in animal models, but drug development has been limited by the absence of a defined binding site and activation mechanism. We determined cryo-EM structures of SERCA bound to two distinct activators, CDN1163 and UM-52. Both compounds occupy a groove formed by transmembrane segments M3-M4, anchored by a hydrogen bond to the SERCA-specific Ser²⁶⁵ and aromatic contacts near Trp²⁷² and Phe²⁹⁶. An acyl chain bridges a gap in the binding pocket toward M1, suggesting that protein-lipid coupling may play a role in SERCA activation. These results directly enable structure-mechanism guided design of next-generation selective SERCA activators.

Data Availability

Cryo-EM maps and atomic coordinates were deposited into the EMDB and RCSB PDB, respectively. SERCA-CDN1163 bound maps and coordinates were assigned EMD-73823 and PDB 9Z5S. SERCA-UM52 bound maps and coordinates were assigned EMD-73820 and PDB 9Z5O, whereas the apo SERCA was assigned EMD-73822 and PDB 9Z5R. The coordinates for the initial models used are available in the PDB under accession codes 2ZBD (SERCA E1 Ca²⁺ ADP-AlF4).