Cyclopiazonic acid (CPA) is a specific and potent inhibitor of the sarcoplasmic reticulum Ca2+-ATPase 1a (SERCA1a). Despite high sequence similarity to SERCA1a, Listeria monocytogenes Ca2+-ATPase 1 (LMCA1) is not inhibited by CPA. To test whether a CPA binding site could be created while maintaining the functionality of the ATPase we targeted four amino acid positions in LMCA1 for mutational studies based on a multiple sequence alignment of SERCA-like Ca2+-ATPases and structural analysis of the CPA site. The identification of CPA-sensitive gain-of-function mutants pinpointed key determinants of the CPA binding site. The importance of these determinants was further underscored by the characterization of the CPA sensitivity of two additional bacterial Ca2+-ATPases from Lactococcus lactis and Bacillus cereus. The CPA sensitivity was predicted from their sequence compared with the LMCA1 results, and this was experimentally confirmed. Interestingly, a cluster of Lactococcus bacteria applied in the production of fermented cheese display Ca2+-ATPases that are predictably CPA insensitive and may originate from their coexistence with CPA-producing Penicillum and Aspergillus fungi in the cheese. The differences between bacterial and mammalian binding pockets encompassing the CPA site suggest that CPA derivatives that are specific for bacteria or other pathogens can be developed.
Bacillus cereus Ca2+-ATPase 1
octaethylene glycol monododecyl ether
Streptococcus pneumonia Ca2+-ATPase
the half maximal inhibitory concentration
Lycopersicon esculentum Ca2+-ATPase 1
Lactococcus lactis Ca2+-ATPase 1
Listeria monocytogenes Ca2+-ATPase 1
Plasmodium falciparum Ca2+-ATPase 4
Plasmodium falciparum Ca2+-ATPase 6
plasma membrane Ca2+-ATPase
sarcoplasmic reticulum Ca2+-ATPase 1a
Cyclopiazonic acid (CPA) is a fungal secondary metabolite originally isolated from the Penicillium genus  and subsequently from the Aspergillus genus . It is biosynthesized from tryptophan and hence is composed of an indole moiety in addition to a central hydrindane moiety and a tetramic acid moiety (Fig. 1A). CPA is an excellent chelator of divalent ions and has been associated with for example iron uptake . CPA is a mycotoxin in higher vertebrates when ingested in contaminated foods. It is the specific cause of ‘kodo poisoning’, a toxic reaction characterized by nausea, vomiting, intoxication and unconsciousness after ingestion of CPA-contaminated kodo millet . The toxicity of CPA in vertebrates is caused by the specific inhibition of the sarcoplasmic reticulum Ca2+-ATPase (SERCA1a) resulting in an altered intracellular calcium flux and homeostasis . Ca2+-ATPases are integral membrane proteins responsible for the active transport of Ca2+ ions across the membrane by a mechanism coupled to ATP hydrolysis . They maintain a low intracellular Ca2+ concentration (sub-micron range), which is critical for Ca2+ homeostasis in both prokaryotes and eukaryotes and essential for proliferation, differentiation and cell death in eukaryotes . The Ca2+-ATPases are composed of three cytoplasmic domains – an actuator (A), a nucleotide binding (N) and a phosphorylation (P) domain – and a transmembrane domain composed of 10 transmembrane helices (M1–M10) . CPA binds to membranous SERCA1a with nanomolar affinity  to E2/E2P states characterized by low Ca2+ affinity . The binding site is located in the cytoplasmic membrane interface in a groove formed by M1, M2, M3 and M4 (Fig. 1B) . Residues in M1 and M2 are involved in the polar contacts to the tetramic acid moiety. The inhibition is dependent on Mg2+ binding  and the 3-acyl group of the tetramic acid moiety is involved in coordination of the divalent metal ion. The Mg2+ ion is further coordinated by glutamine 56 (Q56) and three water molecules, which are interacting with aspartic acid 59 (D59) and main chain carbonyls and amides .
Putative Ca2+-ATPases are widespread in the increasing number of sequenced prokaryotic genomes ; however, their in vivo functions are still poorly understood. The Listeria monocytogenes Ca2+-ATPase 1 (LMCA1) is upregulated during adaptation to alkaline pH . LMCA1 transports a single Ca2+ ion and most probably a single proton is counter-transported per ATP hydrolysis cycle. LMCA1 displays an alkaline pH optimum which can in part be explained by a conserved arginine residue at the intramembranous ion binding site that may act as a pH sensor . A 4.3 Å resolution structure of LMCA1 revealed an overall structure similar to SERCA1a .
To understand the binding determinants of CPA to Ca2+-ATPases we undertook a mutational study of LMCA1 and identified four amino acid positions that influence CPA binding. These findings were further underscored by characterization of the CPA inhibition of two additional bacterial Ca2+-ATPases from Lactococcus lactis and Bacillus cereus.
Despite a high sequence identity to SERCA1a (38%) LMCA1 is only inhibited by the classical SERCA1a inhibitor cyclopiazonic acid (CPA) at very high concentrations (Table 2). A multiple sequence alignment of previously described CPA-sensitive Ca2+-ATPases, CPA-insensitive ATPases and LMCA1 was constructed to gain insight into the residues in LMCA1 responsible for the CPA insensitivity (Fig. S1). Residues within 5 Å of CPA were identified in the SERCA1a–CPA complex structure (Fig. 1B) . The variation in these positions was compared for LMCA1, CPA-sensitive Ca2+-ATPases and CPA-insensitive Ca2+-ATPases in the alignment (Table 1). Four positions in LMCA1 displayed large variations, namely T54, M59, S240 and S295 corresponding to Q56, L61, G257 and P312 of SERCA1a (Fig. 2A). In a first round of mutational studies single mutants restored SERCA1a residues (Table 2). The T54Q mutation, restoring the residue that coordinates Mg2+ with CPA, resulted in a 2-fold reduction of the half maximal inhibitory concentration (IC50) value (487 μm) compared with wild type. M59L and S240G, both interacting with the indole moiety, did not change the IC50 value significantly. S295P, interacting with the hydrindane moiety, resulted in a 7-fold reduction in the IC50 value (123 μm). In a second round of mutational studies T54Q was combined with the other sites. A 7-fold reduction of the IC50 value was observed for T54Q, M59L (113 μm), whereas an increased IC50 value was observed for T54Q, S240G (1640 μm). A 100-fold reduction of the IC50 value (9 μm) was observed when the glutamine and proline were restored (T54Q, S295P). A slightly higher (19 μm) IC50 value was observed for the restoration of glutamine, leucine and glycine (T54Q, M59L, S240G). The IC50 value (9 μm) was not lowered further by restoration of all four positions to the residues found in SERCA1a (T54Q, M59L, S240G, S295P) compared with the T54Q, S295P mutant.
|Interaction||CPA sensitive||CPA insensitive||Unknown|
|SERCAla||LCAI||PfaATP4||PfATP6||BACCA1||LMCA1||LLCA1||Na+/K+-ATPase||H+/K+-ATPase||PMCA1||CaxP||Pcl4g01690||Put. Ca2+- ATPase||Putative Ca2+-ATPase|
|Oryctolagus cuniculus||Solanum lycopersicum||Plasmodium falciparum||Plasmodium falciparum||Bacillus cereus||Listeria monocytogenes||Lactococcus lactis||Homo sapiens||Homo sapiens||Homo sapiens||Streptococcus pneumoniae||Penicillium chrysogenum||Aspergillus flavus||Aspergillus oryzae|
|LMCA1||WT||918 ± 38|
|LMCA1||T54Q||487 ± 75|
|LMCA1||M59L||950 ± 60|
|LMCA1||S240G||1100 ± 17|
|LMCA1||S295P||123 ± 12|
|LMCA1||T54Q, M59L||113 ± 14|
|LMCA1||T54Q, S240G||1640 ± 80|
|LMCA1||T54Q, S295P||9 ± 1|
|LMCA1||T54Q, M59L, S240G||19 ± 1|
|LMCA1||T54Q, M59L, S240G, S295P||9 ± 1|
The CPA sensitivity of two additional bacterial Ca2+-ATPases was determined to validate the importance of the four positions identified in LMCA1. Two putative Ca2+-ATPases from L. lactis and Bacillus cereus were overexpressed in Escherichia coli and purified using a protocol similar to LMCA1. Both putative Ca2+-ATPases displayed Ca2+-dependent activity (Fig. S2) and hence were named L. lactis Ca2+-ATPase 1 (LLCA1) and Bacillus cereus Ca2+-ATPase 1 (BACCA1). LLCA1 has a glycine in position 240 corresponding to G257 in SERCA1a. The amino acids in the remaining three positions of interest correspond to those of LMCA1, and wild-type LLCA1 is CPA insensitive (IC50 > 1000 μm) like LMCA1 (Table 3). Restoration of either the glutamine (T54Q) or proline (S295P) resulted in a 3-fold decrease in the IC50 value (343 and 327 μm respectively). Restoration of both the glutamine and proline (T54Q, S295P) resulted in an IC50 value of 60 μm. The IC50 value was further lowered 10-fold (7 μm) by the restoration of all four positions to the residues found in SERCA1a (LLCA1 T54Q, M59L, S295P). BACCA1 has the same amino acids as SERCA1 in all four positions and displayed an IC50 value of 4 μm (Table 3).
|LCCA1||WT||1012 ± 61|
|LCCA1||T54Q||343 ± 44|
|LCCA1||S295P||327 ± 17|
|LCCA1||T54Q, S295P||60 ± 9|
|LCCA1||T54Q, M59L, S295P||7 ± 1|
|BACCA1||WT||4 ± 1|
CPA is a specific and potent inhibitor of SERCA1a with an IC50 value of 10 μm for the detergent-solubilized enzyme . Despite 38% sequence identity LMCA1 is only inhibited by CPA at much higher concentrations. Therefore, we constructed a multiple sequence alignment of CPA-sensitive and CPA-insensitive P-type ATPases. The distribution of amino acids in the positions within 5 Å of CPA in the structure of the CPA–SERCA1a complex  was investigated in the multiple sequence alignment. Four positions corresponding to Q56, L61, G257 and P312 of SERCA1a displayed variations between the CPA-sensitive and CPA-insensitive Ca2+-ATPases.
A requisite for CPA sensitivity was the restoration of the divalent metal ion coordination site, and the T54Q mutation of LMCA1 resulted in a 2-fold reduction of the IC50 value. M59L and S240G mutations had no effect. Restoration of the proline (S295P) resulted in 7-fold reduction of the IC50 value underscoring the importance of the interaction between the aliphatic proline side chain and the hydrindane group of CPA. An IC50 value in the range of octaethylene glycol monododecyl ether (C12E8)-solubilized SERCA1a was observed for restoration of the glutamine and proline (T54Q, S295P). The proline is involved in the interaction with the hydrindane moiety by forming a hydrophobic pocket together with P308 of the PEGLP motif in M4. The two prolines are responsible for the unwinding of M4 and the pocket formed between the two has previously been shown to be responsible for the interactions with one of the t-butyl groups of the SERCA1a inhibitor 2,5-di-(tert-butyl) hydroquinone [17, 18]. An IC50 value in the same range as SERCA1a was observed for restoration of the glutamine combined with the leucine and glycine restorations (T54Q, M59L, S240G). The leucine and glycine are involved in the interaction of the indole moiety of CPA. The indole moiety of CPA is vital to its inhibitory action as deletion of the indole moiety abolished Ca2+-ATPase inhibition . The indole pocket was only re-established when the leucine and glycine residues were restored as neither the T54Q, M59L mutant nor the T54Q, S240G mutant was CPA sensitive. Hence, restoring either the hydrindane or indole pocket in LMCA1 is sufficient for CPA binding. An additive effect of the restoration of both indole and hydrindane pockets was not observed, however, as the T54Q, M59L, S240G, S295P mutant displayed the same CPA sensitivity as the T54Q, S295P. Restoration of the hydrindane pocket allows substitution with the bulkier serine instead of glycine (99 vs. 66 Å3 ) in the G257 position in the indole pocket (Fig. 2A). However, only smaller amino acids are allowed as substitution with the cognate isoleucine (I292, Fig. 3) of the Na+/K+-ATPase resulted in a 250-fold reduction in affinity [21, 22]. A tilting of M2 towards the centre of the CPA pocket was observed in the 4.3 Å resolution structure of LMCA1  and rearrangement of this helix may be induced by CPA binding.
The importance of the four positions identified as pivotal for CPA binding was confirmed by the characterization of the CPA sensitivity of two additional bacterial Ca2+-ATPases. The L. lactis Ca2+-ATPase (LLCA1) has a glycine in position 240 corresponding to G257 in SERCA1a. Restoration of the glutamine and proline (T54Q, S295P) resulted in a higher IC50 value than the same LMCA1 (T54Q, S295P) mutant. However, the endogenous glycine (G240) could influence the interaction with the indole. Restoring all four positions resulted in an IC50 value comparable with SERCA1a. The Bacillus cereus Ca2+-ATPase has the same amino acids as SERCA1 in all four positions and displayed an IC50 value in the same range as observed for SERCA1a . The IC50 values were determined under steady-state kinetics; thus during turnover CPA most probably dissociates in the E1/E1P states. We assume that the E1/E2 equilibria are not significantly affected for any of the mutated forms as they show similar levels of activity compared with wild type. Hence, we find it safe to assume that the changes in the IC50 values are directly correlated to changes in the CPA affinity.
The CPA–M2+ complex has been previously suggested to mark a transient Ca2+ binding site in the Ca2+ entry pathway . However, there seems to be no coupling between CPA and Ca2+ binding as all the bacterial Ca2+-ATPases and mutants retained Ca2+-stimulated ATPase activity. This is also in good agreement with mutational studies of SERCA1a at the same positions showing retained ATPase activity [23-25]. P312 has previously been mutated to alanine, glycine or leucine all associated with increased Ca2+ affinity but abolished Ca2+ transport, except P312L . The structure of the P312A mutant displayed unwinding of M4 and binding of Ca2+ as observed for native SERCA1a .
Insights into the CPA insensitivity of Na+/K+-ATPase, H+/K+-ATPase and plasma membrane Ca2+-ATPase are also obtained. In the case of the Na+/K+-ATPase the glycine in the indole pocket is exchanged with a leucine. The equivalent substitution in SERCA1a (G257I) has previously been shown to reduce the affinity 250-fold [21, 22]. The proline forming the hydrindane pocket is exchanged with the bulkier leucine (129 vs. 168 Å3 ). Superposition of the two structures of the Na+/K+-ATPase available (PDB entry 3B8E and 2ZXE respectively) [28, 29] with the SERCA1a–CPA complex revealed a movement of helix M4 and hereby the leucine (L337) towards the centre of the CPA pocket further restricting the size of the pocket (Fig. 3). The CPA insensitivity of the Na+/K+-ATPases has previously been suggested to originate from the positioning of the kinked loop of helix M1 in the centre of the pocket in the 3B8E structure , whereas the loop is located similarly to SERCA1a in the 2ZXE structure (Fig. 3). Hence we speculate that the isoleucine and leucine substitutions in the indole and hydrindane pockets respectively could be responsible for the CPA insensitivity of the Na+/K+-ATPases as well as the H+/K+-ATPases where the same compositions of amino acids are found. In the case of the plasma membrane Ca2+-ATPases (PMCA) the glutamine responsible for M2+ coordination is substituted with an alanine and the leucine in the indole pocket is substituted with the more hydrophilic threonine.
Lactic acid bacteria including the Lactococcus genera are extensively used in the production of fermented dairy products including cheese , and CPA-producing fungi are at the same time found in a variety of dry-ripened cheeses . We speculated whether the Lactococcus genera has developed CPA insensitive by coexistence and compared the amino acid compositions of the four positions identified as pivotal for CPA binding (Fig. S3). None of the Ca2+-ATPases from the Lactococcus genera have the glutamine responsible for Mg2+ coordination. Several have a leucine in the G257 pocket of the indole pocket, which is too bulky for CPA binding. In addition, a third of the Ca2+-ATPases have the P312 exchange to a serine hampering the hydrindane interaction. To prevent endogenous inhibition we would expect the Ca2+-ATPases from the CPA-producing Penicillium and Aspergillus strains to harbour substitution in the four identified positions. We identified three putative Ca2+-ATPases in the genomes of Penicillium chrysogenum, Aspergillus flavus and Aspergillus oryzae (Table 1). They all harboured an alanine in the position corresponding to P312 of SERCA1a, thereby hampering interaction with the hydrindane moiety and rendering the strains CPA insensitive.
A Ca2+-ATPase from Streptococcus pneumonia (CaxP) has previously been shown to be vital for the survival of the pathogen in the high extracellular Ca2+ concentrations of the infected host and the pathogen continued growth in vitro in the presence of 100 μm CPA . The indole pocket of CaxP is disrupted by exchange of the leucine and glycine to methionine and serine respectively. These differences in the CPA pocket between mammalian and bacterial Ca2+-ATPase suggest that a bacterial-specific CPA derivative could be developed, hence making the bacterial Ca2+-ATPase a potential drug target.
Materials and methods
CPA was purchased from Sigma Aldrich (St Louis, MO, USA) (C1530) and C12E8 from Nikko Chemicals (Tokyo, Japan). Site-directed mutagenesis was performed with the Quikchange Lightning kit (Agilent Technologies, Horsholm, Denmark). The ORF of LL1366 from L. lactis (strain IL1403, Uniprot Q9CFU9) and the putative Ca2+-ATPase from Bacillus cereus (strain ATCC10987, Uniprot Q73E41) were amplified from genomic DNA with the primer pairs LCCA1_for (GGCATATGCAGCCTTACAATCAATCCG), LCCA1_rev (GGCTCGAGGTGATGTTTTTCAA-AAACGCCTTTAATTGAC) and BACCA1_for (GGCATATGAGCAATTGGTACAGTAAGACG), BACCA1_rev (GGCTCGAGGTTTTTCTTCACTAATTTAATGATTTCATTCACAACAAGCGG) respectively. The PCR products were ligated into the pCR®4-TOPO® vector (Invitrogen) and sub-cloned into pET-22b (Novagen) with NdeI and XhoI resulting in LCCA1-pET-22b and BACCA1-pET-22b. LMCA1, LCCA1 and BACCA1 were expressed and purified as previously described for LMCA1 . However, LCCA1 and BACCA1 were purified by a single-step IMAC procedure.
CPA inhibition of the ATPase activities of LMCA1, LMCA1 mutant forms, LLCA1, LLCA1 mutant forms and BACCA1 was measured by determining the liberation of inorganic phosphate by the Baginski method . Thus 1–5 μg of Ca2+-ATPase was pre-incubated with 1 mm of ethylene glycol-bis(2-aminoethylether)-N,N,N,N-tetraacetic acid and different CPA concentrations in a final reaction volume of 50 μL including 20 mm MOPS/KOH pH 7.4, 20% v/v glycerol, 200 mm KCl, 3 mm MgCl2, 5 mm β-mercaptoethanol, 5 mm NaN3, 0.2 mm NaMoO4, 50 mm KNO3 and 0.25 mg·mL−1 C12E8) in 96-well microtitre plates maintained at 20 °C. CPA was dissolved in dimethyl sulfoxide (DMSO) and a dilution series was prepared resulting in equal volumes of DMSO-dissolved CPA supplemented to all reactions (2.5 μL). Control experiments were supplemented with identical volumes of DMSO. Reactions were initiated by the addition of ATP and CaCl2 resulting in final concentrations of 3 mm and 1 mm respectively. The reactions were stopped after 10 min by addition of 50 μL ascorbic acid solution (140 mm ascorbic acid, 5 mm ammonium heptamolybdate, 0.1% w/v SDS and 0.4 m HCl) and incubated for 10 min at 20 °C. The reduced heteropolymolybdate–phosphate complex (molybdenum blue) was stabilized by the addition of 75 μL arsenate solution (150 mm sodium arsenate, 70 mm sodium citrate and 350 mm acetic acid). The absorbance of the molybdenum blue was measured at 860 nm with a Victor X multi-label plate reader (Perkin Elmer). The ATPase activity at a given CPA concentration was determined a minimum of three times.
The multiple sequence alignment was generated from 54 Ca2+-ATPase sequences in the program muscle , followed by manual inspection to assure correct alignment. The homology model of LMCA1 in the CPA-stabilized E2P state was generated applying PDB entry 3FGO  as a template structure in the program modeller . Twenty models were generated and the model with the lowest discrete optimized protein energy score was compared with the remaining models to confirm model consistency. All structural figures were prepared with pymol .
We are grateful to Anna Marie Nielsen, Tetyana Klymchuk and Annette Damsgaard for the excellent technical assistance. Bacillus cereus genomic DNA was kindly provided by Niels Bohse Hendriksen (Department of Environmental Science – Environmental Chemistry and Microbiology, Aarhus University). The work was supported by a Center of Excellence grant from the Danish National Research Foundation. J.P.M. was supported by Lundbeckfonden and Carlsbergfondet.