A Structural Basis for the Reduced Toxicity of Dinophysistoxin-2

1782

Chem. Res. Toxicol. 2009, 22, 1782–1786

A Structural Basis for the Reduced Toxicity of Dinophysistoxin-2 Jason Huhn,† Philip D. Jeffrey,† Kristofer Larsen,‡,§ Thomas Rundberget,‡ Frode Rise,§ Neil R. Cox,| Vickery Arcus,|,⊥ Yigong Shi,*,†,# and Christopher O. Miles*,‡,|,∇ Department of Molecular Biology, Lewis Thomas Laboratory, Princeton UniVersity, Washington Road, Princeton, New Jersey 08544, The National Veterinary Institute, PB 750 Sentrum, NO-0106 Oslo, Norway, Department of Chemistry, UniVersity of Oslo, P.O. Box 1033 Blindern, NO-0315 Oslo, Norway, AgResearch Ltd., Ruakura Research Centre, PriVate Bag 3123, Hamilton 3240, New Zealand, Department of Biology, The UniVersity of Waikato, PriVate Bag 3105, Hamilton, New Zealand, and Marine Institute, RinVille, Oranmore, Co. Galway, Ireland ReceiVed May 11, 2009

Okadaic acid (OA), dinophysistoxin-1 (DTX-1), and dinophysistoxin-2 (DTX-2) are algal toxins that can accumulate in shellfish and cause diarrhetic shellfish poisoning. Recent studies indicate that DTX-2 is about half as toxic and has about half the affinity for protein phosphatase 2A (PP2A) as OA. NMR structural studies showed that DTX-1 possessed an equatorial 35-methyl group but that DTX-2 had an axial 35-methyl group. Molecular modeling studies indicated that an axial 35-methyl could exhibit unfavorable interactions in the PP2A binding site, and this has been proposed as the reason for the reduced toxicity of DTX-2. Statistical analyses of published data indicate that the affinity of PP2A for DTX-1 is 1.6-fold higher, and for DTX-2 is 2-fold lower, than for OA. We obtained X-ray crystal structures of DTX-1 and DTX-2 bound to PP2A. The crystal structures independently confirm the C-35 stereochemistries determined in the earlier NMR study. The structure for the DTX-1 complex was virtually identical to that of the OA-PP2A complex, except for the presence of the equatorial 35-methyl on the ligand. The favorable placement of the equatorial 35-methyl group of DTX-1 against the aromatic π-bonds of His191 may account for the increased affinity of PP2A toward DTX-1. In contrast, the axial 35-methyl of DTX-2 caused the side chain of His191 to rotate 140° so that it pointed toward the solvent, thereby opening one end of the hydrophobic binding cage. This rearrangement to accommodate the unfavorable interaction from the axial 35-methyl of DTX-2 reduces the binding energy and appears to be responsible for the reduced affinity of PP2A for DTX-2. These results highlight the potential of molecular modeling studies for understanding the relative toxicity of analogues once the binding site at the molecular target has been properly characterized. Introduction 1

Okadaic acid (OA) (1) (Figure 1) was first isolated from the sponges Halichondria okadai and Halichondria melanodocia (1) and shown to be a potent new type of protein phosphatase (PP1 and PP2A) inhibitor (2). Subsequent studies have shown that OA, the closely related analogues dinophysistoxins-1 (DTX1) (2) (3) and -2 (DTX-2) (3) (4), and their derivatives are responsible for causing a form of food poisoning known as diarrhetic shellfish poisoning (DSP) (5). DSP occurs around the world and is caused by the consumption of filter-feeding shellfish that have consumed microalgae containing okadaic analogues (6). Heterotrophic and mixotrophic dinoflagellates of the genus Dinophysis have long been known to contain okadaic * To whom correspondence should be addressed. Tel: +47 2321-6226. Fax: +47 2321-6201. E-mail: [email protected] (C.O.M.). Tel: +86 10-6279-6163. Fax: +86 10-6279-2736. E-mail: [email protected] (Y.S.). † Princeton University. ‡ National Veterinary Institute. § University of Oslo. | AgResearch Ltd. ⊥ The University of Waikato. # Present address: Department of Biological Sciences and Biotechnology, Tsinghua University, Beijing 10084, China. ∇ Marine Institute. 1 Abbreviations: DTX, dinophysistoxin; MRM, multiple reaction monitoring; PP, protein phosphatase; OA, okadaic acid; TEF, toxic equivalence factor.

Figure 1. Structures of OA (1), DTX-1 (2), and DTX-2 (3) (10, 11). DTX-2 was previously usually depicted as 4.

analogues (3), although only recently have suitable culture methods been developed (7) that have allowed the production of these toxins by Dinophysis to be demonstrated (8). Recently, DTX-2 was found to be significantly less toxic in a mouse bioassay and a weaker inhibitor of PP2A than OA (9). This finding is of considerable public health and regulatory significance because these compounds were previously assumed to be equipotent. The stereochemistries of DTX-1 and DTX-2 have been studied in detail by NMR, and their 35-methyl groups were found to differ in their stereochemistries at C-35 (10, 11), a finding that was supported by subsequent synthetic studies (12). The availability of a crystal structure of OA bound to the core enzyme of PP2A (13) facilitated molecular modeling studies of the binding of OA, DTX-1, and DTX-2 to PP1 and PP2A (10). These studies predicted that the axial 35-methyl

10.1021/tx9001622 CCC: $40.75  2009 American Chemical Society Published on Web 10/21/2009

Structures of DTX-1 and DTX-2 Bound to PP2A

group of DTX-2 would show unfavorable interactions with the His191 and Gln122 residues in the PP2A binding site, that these interactions would not be present during binding of OA or DTX1, and that this phenomenon was responsible for the reduced toxicity of DTX-2 (10). The aim of this study was to independently verify the stereochemistries of DTX-1 and -2 and to determine whether changes in the mode of binding of PP2A to DTX-2 could be responsible for its reduced affinity for this compound.

Chem. Res. Toxicol., Vol. 22, No. 11, 2009 1783 Table 1. IC50 (PP2A Inhibition) and LD50 (ip, Mouse) Values for OA (1) and DTX-2 (3), Their Ratios (Value for OA/Value for DTX-2), and 95% Confidence Intervalsa

OA (1) DTX-2 (3) ratio OA/DTX-2 (%) a

Materials and Methods Isolation and Quantitation of DTX-1 and DTX-2. DTX-1 (10) and DTX-2 (14) were isolated as described elsewhere. DTX-2 was quantitated by NMR against a series of caffeine external standards after the method of Burton et al. (15) but used the following equipment: Bruker Avance AV-II 600 MHz NMR spectrometer equipped with a 5 mm TCI cryoprobe and Z-gradient coils; Topspin 2.1 software; the solvent was methanol-d4 99.8+ atom-% D, and NMR tubes were Wilmad 5 mm, 600 MHz, Emperor grade 7 in. long 535-PP (both from Sigma-Aldrich, Steinheim, Germany). Each NMR tube was inserted in the instrument at the same height, and automated shimming, tuning, and matching were used, seeking 50 Ω impedance at the operating frequency for maximum power transfer for each sample. A 360° pulse length for each individual sample was determined using a single scan, and the satisfactory pulse length was divided by four to determine a more exact 90° pulse length. Each sample was analyzed by a standard single pulse procedure using 16 scans, with a delay between scans of 30 s to ensure complete relaxation of spins. The sample was then ejected, before being reinserted, and the procedure was repeated two times to minimize system errors. Acquired spectra were Fouriertransformed, phased, and baseline-corrected. Peaks of known resonances were integrated from a subjectively chosen point on the baseline right before the beginning of the curvature, to a point on the baseline right after the curvature. The slope and bias of each peak of the sample were then individually corrected. The Bruker Topspin software subroutine multi_integ enabled the integration of a series of spectra, with the results of the integration listed as being calibrated against one or several samples. Because the NMR quantitation method (15) had not been published at the time of its isolation, DTX-1 was quantitated gravimetrically after drying under vacuum. Aliquots of DTX-1 or DTX-2 were prepared by dilution with MeOH, evaporated to dryness under a stream of dry N2, and stored dry at -20 °C until reconstitution in MeOH when required. LC-MS/MS Analysis. DTX-1 and DTX-2 were quantified by LC-MS/MS against a certified OA standard (NRCC, Halifax, NS). The LC-MS method (14) utilized separation on a C18 column with a linear gradient of MeCN-water containing 2 mM ammonium formate and 0.01% v/v formic acid and detection with a Quattro Ultima Pt triple-quadrupole mass spectrometer operating with an ESI interface (Waters Micromass, Manchester, United Kingdom). Quantification of the OA, DTX-1, and DTX-2 was performed in negative ionization mode with multiple reaction monitoring (MRM) of the following transitions: OA and DTX-2, 803.5f255.1; and DTX-1, 817.5f255.1. Statistical Analysis of Toxicological Data. Data for inhibition of human erythrocyte PP2A by OA and DTX-2 (9) were kindly provided by J. A. B. Aasen. The relationship between percent inhibition and log concentration was fitted using a two-parameter logistic inhibition curve in Genstat 10.2 (VSN International, Hemel Hempstead, United Kingdom). Data for the proportion of mice dying after different ip doses of OA and DTX-2 were obtained from ref 9. The relationship between the proportion dying and the log dose was fitted using the maximum likelihood method with the probit model (16) in Genstat 10.2 to estimate the LD50 and its standard error. The model was fitted separately for each toxin and also combined with a common slope; results presented in Table 1 are for the model with the common slope as the LD50 values were essentially unchanged but the standard error for OA was signifi-

IC50 (ng/mL)

95% confidence interval

LD50 (µg/kg)

95% confidence interval

2.87 5.96 48

2.51-3.27 5.42-6.55 40-56

194 364 53

141-266 290-456 33-73

Calculated from Aune et al. (9) and data from J. A. B. Aasen.

cantly reduced. The Ki values of OA and DTX-1 for PP2A from rabbit skeletal muscle, together with their associated confidence intervals, were obtained from ref 17. Expression and Purification of DTX-1- and DTX-2-Bound PP2A Core Enzyme. Full-length PP2A AR subunit (1-589) was cloned into the pGEX-2T vector (GE Healthcare, Piscataway, NJ) as an N-terminal GST-fusion protein and overexpressed in Escherichia coli strain BL21-Gold(DE3). The soluble fraction of the E. coli lysate was purified by glutathione sepharose column chromatography (GE Healthcare). The eluate from the glutathione sepharose column was further fractionated by anion exchange chromatography (Source 15Q, GE Healthcare). GST-AR was eluted from the Source 15Q column with a 0-500 mM NaCl gradient buffered with 25 mM Tris pH 8.0. Full-length PP2A CR subunit (1-309) was cloned into the baculovirus transfer vector pVL1392 (BD Biosciences, San Jose, CA) as an N-terminal 8 × His-tagged fusion protein. Recombinant virus was produced utilizing the BaculoGold cotransfection kit (BD Biosciences). N-His-tagged CR was overexpressed in a baculovirusinfected Hi-5 insect cell suspension culture. The soluble fraction of the Hi-5 cell lysate was purified via Ni2+-NTA column chromatography (QIAGEN, Valencia, CA) followed by anion exchange chromatography (Source 15Q, GE Healthcare). N-HisCR was eluted with a 0-500 mM NaCl gradient buffered with 25 mM Tris, pH 8.0, and fractions were assayed for active N-His-CR using a colorimetric phosphatase assay. Fractions containing N-HisCR were pooled together and mixed with 1.2 mol equiv of DTX-1 or DTX-2. Complete inhibition of phosphatase activity was verified using the colorimetric assay. Inhibitor-bound N-His-CR was digested with 2 mg of trypsin at 4 °C overnight. The trypsin-digested N-His-CR-inhibitor complex was purified by anion exchange chromatography. Fractions containing N-HisCR were pooled together and incubated with a molar equivalent of GST-AR subunit immobilized on glutathione sepharose resin. The AR-CR dimer was on-column-digested with thrombin overnight at 4 °C. Thrombin-digested AR-CR dimer was released from the GS4B sepharose column and concentrated with a 30 kDa cutoff Amicon filtration membrane. Concentrated AR-CR dimer was purified with a Superdex 200 10/30 gel filtration column (GE Healthcare) that had been equilibrated with 10 mM Tris, pH 8.0, 150 mM NaCl, 2 mM DTT, and 50 µM MnCl2. Fractions containing inhibitor-bound AR-CR dimer were pooled, aliquotted, and snap-frozen in liquid nitrogen. Purified AR-CR dimer aliquots were stored at -80 °C. Crystallization and Data Collection for DTX-1 and DTX-2 Bound to the PP2A Core Enzyme. Crystals were grown at 277 K using the hanging-drop vapor diffusion method. Crystals grew in space group I222 with typical cell dimensions of a ) 93.4 Å, b ) 197.2 Å, c ) 201.7 Å, and R ) β ) γ ) 90° and contained a single AC complex in the asymmetric unit. Because of the limited amount of protein, sparse-matrix screening was avoided. Rather, the crystallization conditions of the published OA- and microcystinLR-bound PP2A core enzyme structures were used as a starting point for screening (13). DTX-1- and DTX-2-bound PP2A core enzyme complexes were crystallized in a solution of 100 mM Tris, pH 7.4-7.7, 250 mM Li2SO4, and 1.6-1.8 M (NH4)2SO4. Crystals appeared overnight and measured roughly 0.30 mm × 0.30 mm × 0.05 mm. Crystals were cryoprotected in crystallization buffer supplemented with 20% (v/v) glycerol and flash frozen in liquid

1784

Chem. Res. Toxicol., Vol. 22, No. 11, 2009

Huhn et al. Table 2. Summary of Structure Refinement for Complexes of the PP2A Core Bound to DTX-1 (2) and DTX-2 (3)a data set

DTX-1 (2) data collection 1.08 100-2.85 (2.95-2.85) 42933 4.5 (4.5) 99.6 (99.4) 14.6 (2.9) 69.2 0.064 (0.399)

wavelength (Å) resolution (Å) unique reflections redundancy completeness (%) 〈I〉/〈σI〉 Wilson B factor (Å2) Rsym

refinement 30-2.85 (2.87-2.85) reflections (work/free) 40799/2113 completeness (%, work + free) 99.5 no. of atoms (waters) 6968 (43) R-work 0.217 (0.381) R-free 0.262 (0.466) rmsd bond lengths (Å) 0.009 rmsd bond angles (°) 1.52 rmsd B factors over 2.7 bonds (Å2) average B factor 57.2 (38.7) (Å2) (waters) average B factor 43.7 of ligand (A2) resolution (outer shell) (Å)

Figure 2. Quantitative LC-MS/MS analysis of a standard of OA (1), DTX-1 (2), and DTX-2 (3).

nitrogen. Crystals diffracted to approximately 2.8 Å resolution at a synchrotron light source. Data were collected at BNL-NSLS beamline X29A. Data were indexed and integrated using DENZO and scaled using SCALEPACK (18). Structure Determination of DTX-1- and DTX-2-Bound PP2A Core Enzyme. The structures of DTX-1- and DTX-2-bound PP2A core enzyme were determined through rigid body refinement of the existing OA-bound PP2A core enzyme structure. Reflections in the test set were inherited from the test set selected for the OAbound PP2A X-ray data (13), based on Miller index, to reduce potential bias of R-free from the starting model. Models of DTX-1 and DTX-2 were built into prominent difference density in the active site of the catalytic CR subunit. The stereochemistry of the methyl group attached at C35 was confirmed by simulated annealing difference maps (Supporting Information) of DTX-1 and DTX-2 data refined against a model containing OA (which lacks the methyl group at C35). Occupancies for the ligands (DTX-1 and DTX-2) were kept at 1.0, based on them having comparable B factors to the rest of the core enzyme structure. Structurally conserved water molecules were included in the difference density where corresponding water molecules were present in the OA-bound PP2A core enzyme structure. The structure of each inhibitor-bound PP2A core enzyme structure was refined with CNS and rebuilt with O and COOT. A simulated annealing omit map was calculated to verify ligand presence and the location of Gln122 and His191, in which the occupancies for side chain atoms of Gln122 and His191 and the ligand were set to zero and refined against the DTX-2 data. Surface areas were calculated using ProtorP (http://www.bioinformatics.sussex.ac.uk/protorp/, accessed 27 Nov 2008) (19). The atomic coordinates of the PP2A core enzyme bound to DTX-1 and DTX-2 have been deposited in the Protein Data Bank with the ID codes 3K7V and 3K7W, respectively.

Results and Discussion LC-MS analyses indicated that the DTX-1 and DTX-2 were of high purity. The MS response and hence the peak area for the DTX-2 (Figure 2), which had been quantitated by NMR, was 95% that of the OA standard. However, the MS response and hence the peak area for DTX-1 (Figure 2), which had been quantitated gravimetrically, was ca. 81% that of the OA standard, presumably due to the presence of a solvate and possibly traces of water, salts, and other minor contaminants. It is not unusual for purified algal toxins quantitated by gravimetry to contain significantly less than the stated toxin content when measured by LC-MS against standards that have been certified on the basis of quantitative NMR (J. A. B. Aasen, personal communication). Thus, these results indicate that OA, DTX-2, and probably DTX-1 also have essentially identical response factors under the LC-MS/MS conditions used here for analysis.

within favored (%) within allowed (%) outliers (%)

Ramachandran plot 92.8 98.6 1.4

DTX-2 (3) 1.08 100-2.95 (3.06-2.95) 39083 4.0 (3.9) 99.8 (99.9) 13.6 (2.5) 74.5 0.082 (0.439) 30-2.95 (2.98-2.95) 37128/1918 98.6 6949 (25) 0.221 (0.358) 0.263 (0.419) 0.009 1.51 2.4 65.6 (37.0) 58.6

92.1 98.3 1.7

a Rsym ) ΣhΣi|Ih,i - Ih|/ΣhΣiIh,i, where Ih is the mean intensity of the i observations of symmetry-related reflections of h. R ) Σ|Fobs - Fcalcd|/ ΣFobs, where Fobs ) FP and Fcalcd is the calculated protein structure factor from the atomic model (Rfree was calculated with 5% of the reflections). rmsd in bond lengths and angles are the deviations from ideal values.

Table 3. Toxic Equivalency Factors (TEFs) for OA, DTX-1, and DTX-2 Based on PP2A Inhibitiona

a

toxin

TEF

95% confidence interval

OA DTX-1 DTX-2

1.00 1.58 0.48

1.12-2.03 0.40-0.56

Calculated from data in Table 1 and from Takai et al. (17).

Statistical analysis of published data (9) indicated that the inhibitory potency of DTX-2 toward PP2A was about 50% that of OA and that DTX-2 was only about half as toxic toward mice as OA (Table 1). Thus, the available PP2A inhibition data indicate a significant reduction in the affinity of PP2A for DTX2, relative to OA. Although there appear to be no published quantitative comparisons of the LD50 values for OA and DTX1, the available data indicate (17) that DTX-1 has an inhibitory potency toward rabbit muscle PP2A that is ca. 1.6-fold that of OA (Table 3). Structures of DTX-1 and DTX-2 Bound to PP2A Core Enzyme. DTX-1 (35R-methyl-OA) only differs from OA by having an additional equatorial methyl group at the 35-position (Figure 1). DTX-2 is an isomer of OA such that the methyl group at the 31-position in OA is absent, and in its place, an axial methyl group is located at the 35-position in DTX-2 (Figure 1). The structures of the DTX-1 and DTX-2 complexes were refined at 2.85 and 2.95 Å resolution, respectively (Table 2). The global architecture of the protein fold and the arrangements of secondary, tertiary, and quaternary structures have been

Structures of DTX-1 and DTX-2 Bound to PP2A

Figure 3. Structures of DTX-1 and DTX-2 bound to the PP2A core enzyme. (A) DTX-1 and (B) DTX-2 bound to the core enzyme. The surface of the active site of the catalytic subunit is colored light blue, the cofactor manganese atoms are shown as blue spheres, and the ligand is shown in yellow (carbon) and red (oxygen). Magnified cartoon representations are shown of DTX-1 (C) and DTX-2 (D) in the hydrophobic cage of the binding site. The structural integrity of the hydrophobic cage formed by His191 and Gln122 in the DTX-1 complex is altered in the DTX-2 complex. Images were generated using PyMol (22).

reviewed in the introduction and therefore will not be discussed here. The focus will be on toxin binding to the catalytic CR subunit. DTX-1 bound to the same surface pocket above the active site of the catalytic subunit as microcystin-LR and OA. The interactions between DTX-1 and CR are very similar to those between OA and CR. The carboxyl group at the end of DTX-1 points down into the active site and coordinates the manganese atoms via a bridging water molecule (Figure 3A). The carboxyl group also engages in hydrogen bonding with the side chain of Tyr265. Near the carboxyl moiety, Arg89 forms hydrogen bonds with two distinct oxygen atoms of DTX-1. Across the active site from Arg89, the side chain of Arg214 forms a hydrogen bond with an oxygen atom of DTX-1. The hydrophobic end of DTX-1 extends along the surface pocket and settles in a hydrophobic cage formed by CR residues Gln122, Ile123, His191, and Trp200. The stereochemistries of DTX-1 and DTX-2 observed in their respective crystal structures were in complete agreement with those proposed on the basis of NMR analysis (10). The 35methyl group of DTX-1 is positioned in the hydrophobic cage of the binding pocket, and no steric clash occurs between the methyl group and the residues lining the hydrophobic cage (Figure 3A,C). The 35-methyl group of DTX-1 is angled away from CR and is therefore easily accommodated by the surface pocket. The equatorial 35-methyl group of DTX-1 is partially buried upon binding, and the close proximity of this methyl group to the aromatic π-orbitals of His191 should give rise to a favorable hydrophobic interaction. The observed difference in Kd for OA binding as compared to DTX-1 binding (17) corresponds to a difference in ∆∆Gbind of 1.1 kJ mol-1 and is consistent (20) with the modest increase in the burial of hydrophobic surface area upon binding observed from the structures determined here for PP2A-OA and PP2A-DTX-1. DTX-2 bound in the active site of the catalytic subunit in the same way as DTX-1 (Figure 3B), but a key difference in binding was observed in the region near to the 35-methyl group (Figure 3D). While the 35-methyl group of DTX-1 is equatorial, that of DTX-2 is axially oriented. This difference in stereochemistry leads to a steric clash between DTX-2 and His191

Chem. Res. Toxicol., Vol. 22, No. 11, 2009 1785

in the hydrophobic cage of the CR binding pocket (Figure 3B,D), such that the side chain of His191 is pushed out and away from the binding pocket. Consequently, the distance between His191 and Gln122 in the DTX-1 structure is 3.97 Å (Figure 3C), but this increases to 7.60 Å in the DTX-2 structure (Figure 3D). Weakening of the structural integrity of the hydrophobic cage combined with the energy cost of displacing His191 would be expected to lead to a reduced affinity of PP2A for DTX-2, relative to DTX-1, thereby explaining the lower toxicity of DTX-2 (9) (Table 1). The displacement of the His191 side chain by a rotation of 140° opens the hydrophobic cage and significantly affects the burial of surface area for the ligand. Binding of DTX-2 buries 27.1 Å2 less surface area relative to binding of DTX-1, and this is consistent with the observed ∆∆Gbind value of 2.9 kJ mol-1 calculated from the Kd values for DTX-1 and DTX-2. This reduction in buried surface area along with the disruption of the hydrophobic cage by the displacement of His191 gives an explanation for both the reduced binding affinity of DTX-2 and its reduced toxicity. OA and DTXs are 5000-10000 times more potent against PP2A than PP1 (17). Consequently, as Aune et al. (9) have pointed out, PP2A is widely regarded as the molecular target at which these toxins exert their toxic effects. The relative values of the murine LD50 and the PP2A IC50 for OA and DTX-2 (Table 1) are consistent with this proposal. Unfortunately, the high level of statistical uncertainty inherent in the LD50 measurements means that they can only provide a very approximate measure of the toxic equivalence factors (TEFs) used to convert the measured concentrations of toxin analogues in samples into total toxicity (in OA equivalents). Because PP2A inhibition measurements have a much lower level of uncertainty associated with them (Table 1) and use less toxin than LD50 determinations, these may provide a more convenient and statistically more reliable estimate of the TEFs of the major chemically similar OA analogues. Takai et al. (17) have measured the Ki values and their associated 95% confidence intervals for OA and DTX-1, allowing their relative potencies to be compared directly with the IC50 data for OA and DTX-2 in Table 1. Analysis of the available PP2A inhibition data therefore suggests (Table 3) that appropriate TEFs for OA, DTX1, and DTX-2 could be 1.0, 1.6, and 0.5, respectively. This information may be helpful in protecting consumers from toxic shellfish, as Albano et al. (21) recently showed that an additive model based on the TEFs of individual OA analogues was appropriate for estimating the total activity of multicomponent mixtures in unknown samples. In light of the potentially lower toxin content in some gravimetrically quantitated toxins, valid and accurate TEFs should (whenever possible) be estimated from direct comparison in toxicological experiments performed using high purity toxins with concentrations established relative to certified analytical standards. Unfortunately, such data are not currently available for OA, DTX-1, and DTX-2. The results presented here provide an explanation for the relative affinity of PP2A for OA, DTX-1, and DTX-2 (Table 3) and thus for the toxicity of mixtures these compounds to animals (21). The observed crystal structures are also consistent with the results of molecular modeling studies that predicted that the axial 35-methyl group of DTX-2 would give rise to unfavorable steric interactions. This suggests that molecular modeling has the potential to be used in predicting the approximate relative toxicities of those algal toxins that are available in quantities too small to allow toxicological evaluation. Such methods must be interpreted with caution and are

1786

Chem. Res. Toxicol., Vol. 22, No. 11, 2009

predicated on reliable knowledge of the mode of action and of the molecular structures of both the ligand and its binding site. Acknowledgment. We thank A. Saxena and H. Robinson at BNL NSLS beamlines for help and J. A. B. Aasen for providing PP2A inhibition data for OA and DTX-2. This work was supported by grants from the U.S. National Institutes of Health to Y.S. This research was supported by the BIOTOX project (partly funded by the European Commission through sixth Framework Programme contract no. 514074, topic Food Quality and Safety), by the New Zealand Foundation for Research Science and Technology contract C10X0406 (International Investment Opportunities Fund), and a Marie Curie International Incoming Fellowship within the seventh European Community Framework Programme (FP7/2007-2013) under grant agreement no. 221117. The Research Council of Norway (NFR) financed the Bruker AV-II 600 MHz instrument used in this study. Supporting Information Available: Tabulated values for accessible surface area for OA, DTX-1, and DTX-2; simulated annealing difference maps for DTX-1 and DTX-2 bound to PP2A; plots of PP2A inhibition and ip toxicity data for OA and DTX-1; Genstat analysis for LD50 for DTX-2 with independent slope (σ value); Genstat analysis for LD50 for OA with independent slope (σ value); Genstat analysis for LD50 for OA and DTX-2 with common slope (σ value); and QuickTime movie prepared in PyMol (22) using eMovie (23), showing the effects of binding of OA and DTX-1 and -2 on the structure of PP2A core enzyme. This material is available free of charge via the Internet at http://pubs.acs.org.

References (1) Tachibana, K., Scheuer, P. J., Tsukitani, Y., Kikuchi, H., Van Engen, D., Clardy, J., Gopichand, Y., and Schmitz, F. J. (1981) Okadaic acid, a cytotoxic polyether from two marine sponges of the genus Halichondria. J. Am. Chem. Soc. 103, 2469–2471. (2) Holmes, C. F. B., Luu, H. A., Carrier, F., and Schmitz, F. J. (1990) Inhibition of protein phosphatases-1 and -2A with acanthifolicin: comparison with diarrhetic shellfish toxins and identification of a region on okadaic acid important for phosphatase inhibition. FEBS Lett. 270, 216–218. (3) Murata, M., Shimatani, M., Sugitani, H., Oshima, Y., and Yasumoto, T. (1982) Isolation and structural elucidation of the causative toxin of the diarrhetic shellfish poisoning. Bull. Jpn. Soc. Sci. Fish. 48, 549– 552. (4) Hu, T., Doyle, J., Jackson, D., Marr, J., Nixon, E., Pleasance, S., Quilliam, M. A., Walter, J. A., and Wright, J. L. C. (1992) Isolation of a new diarrhetic shellfish poison from Irish mussels. J. Chem. Soc., Chem. Commun. 39–41. (5) Yasumoto, T., and Murata, M. (1993) Marine toxins. Chem. ReV. 93, 1897–1909. (6) Toyofuku, H. (2006) Joint FAO/WHO/IOC activities to provide scientific advice on marine biotoxins (research report). Mar. Pollut. Bull. 52, 1735–1745.

Huhn et al. (7) Park, M. G., Kim, S., Hyung, S. K., Myung, G., Yi, G. K., and Yih, W. (2006) First successful culture of the marine dinoflagellate Dinophysis acuminata. Aquat. Microb. Ecol. 45, 101–106. (8) Kamiyama, T., and Suzuki, T. (2009) Production of dinophysistoxin-1 and pectenotoxin-2 by a culture of Dinophysis acuminata (Dinophyceae). Harmful Algae 8, 312–317. (9) Aune, T., Larsen, S., Aasen, J. A. B., Rehmann, N., Satake, M., and Hess, P. (2007) Relative toxicity of dinophysistoxin-2 (DTX-2) compared with okadaic acid, based on acute intraperitoneal toxicity in mice. Toxicon 49, 1–7. (10) Larsen, K., Petersen, D., Wilkins, A. L., Samdal, I. A., Sandvik, M., Rundberget, T., Goldstone, D., Arcus, V., Hovgaard, P., Rise, F., Rehmann, N., Hess, P., and Miles, C. O. (2007) Clarification of the C-35 stereochemistries of dinophysistoxin-1 and dinophysistoxin-2 and its consequences for binding to protein phosphatase. Chem. Res. Toxicol. 20, 868–875. (11) Larsen, K., Petersen, D., Wilkins, A. L., Samdal, I. A., Sandvik, M., Rundberget, T., Goldstone, D., Arcus, V., Hovgaard, P., Rise, F., Rehmann, N., Hess, P., and Miles, C. O. (2007) Clarification of the C-35 stereochemistries of dinophysistoxin-1 and dinophysistoxin-2 and its consequences for binding to protein phosphatase. Chem. Res. Toxicol. 20, 2020. (12) Forsyth, C. J., and Wang, C. (2008) Synthesis and stereochemistry of the terminal spiroketal domain of the phosphatase inhibitor dinophysistoxin-2. Bioorg. Med. Chem. Lett. 18, 3043–3046. (13) Xing, Y., Xu, Y., Chen, Y., Jeffrey, P. D., Chao, Y., Lin, Z., Li, Z., Strack, S., Stock, J. B., and Shi, Y. (2006) Structure of protein phosphatase 2A core enzyme bound to tumor-inducing toxins. Cell 127, 341–353. (14) Rundberget, T., Sandvik, M., Larsen, K., Pizarro, G. M., Reguera, B., Castberg, T., Gustad, E., Loader, J. I., Rise, F., Wilkins, A. L., and Miles, C. O. (2007) Extraction of microalgal toxins by large scale pumping of sea water in Spain and Norway, and isolation of okadaic acid and dinophysistoxin-2. Toxicon 50, 960–970. (15) Burton, I. W., Quilliam, M. A., and Walter, J. A. (2005) Quantitative 1 H NMR with external standards: use in preparation of calibration solutions for algal toxins and other natural products. Anal. Chem. 77, 3123–3131. (16) Finney, D. J. (1952) Probit Analysis, 2nd ed., Cambridge University Press, Cambridge, United Kingdom. (17) Takai, A., Murata, M., Isobe, M., Mieskes, G., and Yasumoto, T. (1992) Inhibitory effect of okadaic acid derivatives on protein phosphatases. A study on structure-affinity relationship. Biochem. J. 284, 539–544. (18) Otwinowski, Z., and Minor, W. (1997) In Methods in Enzymology (Carter, C. W., Jr., and Sweet, R. M., Eds.) pp 307-326, Academic Press, San Diego. (19) Reynolds, C., Damerell, D., and Jones, S. (2009) ProtorP: a proteinprotein interaction analysis server. Bioinformatics 25, 413–414. (20) Al-Lazikani, B., Gaulton, A., Paolini, G., Lanfear, J., Overington, J., and Hopkins, A. (2007) In Chemical Biology. From Small Molecules to Systems Biology and Drug Design (Schreiber, S., Kapoor, T. M., and Wess, G., Eds.) pp 804-824, Wiley-VCH, Weinheim, Germany. (21) Albano, C., Ronzitti, G., Rossini, A. M., Callegari, F., and Rossini, G. P. (2009) The total activity of a mixture of okadaic acid-group compounds can be calculated by those of individual analogues in a phosphoprotein phosphatase 2A assay. Toxicon 53, 631–637. (22) DeLano, W. L. (2002) The PyMol Molecular Graphics System, DeLano Scientific, San Carlos, CA, http://www.pymol.org/ (accessed October 13, 2008). (23) Hodis, E., Schreiber, G., Rother, K., and Sussman, J. L. (2007) eMovie: a storyboard-based tool for making molecular movies. TIBS 32, 199–204.

TX9001622