Introduction

Top Abstract Introduction Methods Results Discussion References

Protein phosphatase 2A, and more specifically its catalytic subunit, is the proposed target for three related toxicants: CA, the vesicant in blister beetles and the active ingredient in the aphrodisiac "Spanish fly"; palasonin, an anthelmintic in seeds of a medicinal tree; and endothall, a synthetic herbicide (fig. 1). Synthesis of [³H]CA allowed the identification and quantification of the target site in mouse liver cytosol. The relevance of this site was initially validated through studies establishing a correlation between the binding affinity of CA/endothall analogs with their i.p. toxicity to mice (Graziano and Casida, 1987; Graziano et al., 1987,1988). Isolation and partial sequencing then identified the binding protein as the catalytic subunit of PP2A (Li and Casida, 1992; Li et al., 1993). PP2A is a member of a family of protein phosphatases, which includes PP1, PP2A, PP2B and PP2C, each demonstrating differing substrate specificities but all important for their role in the dephosphorylation of serine/threonine residues and opposing the effects of protein kinases (Cohen, 1991). PP2A is also the target for unrelated environmental toxins including the microcystins (Lambert et al., 1994) and DSPs (MacKintosh and MacKintosh, 1994a, b). The most common DSP, okadaic acid, is a potent inhibitor of [³H]CA binding, which suggests that they act at the same or closely coupled sites on PP2A (Li and Casida, 1992).

View larger version (9K): [in this window] [in a new window]   Fig. 1.   Structures of CA, [3H]CA, palasonin and endothall (showing numbering of the oxabicycloheptane).

Human poisoning by PP inhibitors may occur on dermal exposure to blister beetles or ingestion of cantharides in "Spanish fly" (Nickolls and Teare, 1954), from ingesting toxin-contaminated drinking water (Lambert et al., 1994) or shellfish (Yasumoto et al., 1985), and by improper use of the herbicide endothall (Tomlin, 1994). The effects of these compounds are complex because PP2A is distributed throughout the body and is essential for the regulation and functions of receptors, ion channels and enzymes. Although CA induces tissue-specific changes in the phosphorylation state of several phosphoproteins (Eldridge and Casida, 1995), it is difficult to establish the initiating event in PP2A inhibitor toxicity (Eriksson and Golman, 1993).

This study examines PP inhibitors in mouse neuroblastoma cells by correlating their in vitro affinity at the [³H]CA binding site to their in vivo cytotoxicity. N1E-115 cells were selected as the model because the mouse is the principal species used in studies of the [³H]CA binding site (Graziano et al., 1987, 1988; Li and Casida, 1992; Li et al., 1993) and brain tissue is particularly rich in PP activity in both the cytosolic and microsomal fractions (Suganuma et al., 1989; Sim et al., 1994). Moreover, PP plays a major role in several aspects of neuronal function (Browning et al., 1985). Phosphorus-containing bifunctional acids and divalent cations are examined as potential modifiers of cytotoxicity because they strongly influence [³H]CA binding in liver cytosol (Graziano et al., 1988; Li et al., 1993). The findings reported here validate both the relevance of the [³H]CA binding site in the toxic action of a variety of natural and synthetic inhibitors of PP2A and the neuroblastoma cell model for studies into the mechanisms of PP inhibitor toxicity.

	Methods

Top Abstract Introduction Methods Results Discussion References

Chemicals. Sources for the chemicals were: microcystin LR, calyculin A and okadaic acid from LC Laboratories (Woburn, MA); endothall analogs from our earlier studies (Matsuzawa et al., 1987; Kawamura et al., 1990); PSCP (a protease inhibitor) from synthesis in this laboratory (Casida et al., 1961); fosamine from Chem Service (West Chester, PA); others from Sigma Chemical Co. (St. Louis, MO). [³H]CA (Graziano et al., 1988) was purified on a small silica acid column with methylene chloride for elution. The radiochemical purity (>95%) was analyzed by two-dimensional thin-layer chromatography (silica gel, R_f 0.42 on first development in chloroform and 0.76 on second development in chloroform/acetic acid 9:1) followed by autoradiography. Specific activity (34 Ci/mmol) was determined by competition in the ligand binding assay (see below) with known concentrations of unlabeled CA (Akera and Cheng, 1977). [³H]CA was diluted to 25 nM in assay buffer and held for at least 1 hr at 25°C to allow hydrolysis to [³H]CA; the anhydride and dicarboxylic acid were identical in their binding affinity.

Cell culture and cytotoxicity assays. N1E-115 cells (Tissue Culture Facility, Department of Molecular and Cellular Biology, University of California at Berkeley) were cultured in 10 × 15 cm flasks containing Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 1% pyruvate, 0.06% L-glutamine, 100 µg/ml streptomycin and 100 U/ml penicillin G at 37°C under a humidified atmosphere of 5% CO₂/95% air. The cells were aliquoted into 96-well plastic plates at 10⁴ cells/well and after 24 hr they were exposed for an additional 24 hr to the test compound in the absence of antibiotics. Inhibitors were initially dissolved in dimethyl sulfoxide, which was present at <1% final concentration during assays, followed by serial dilution in media. Cell viability was determined by the MTT procedure (Mosmann, 1983) which measured absorbance at 570 nm with a microplate reader (Molecular Devices, Sunnyvale, CA).

Cytosol and microsome preparations. Neuroblastoma cells (1.4 × 10⁹) were grown until confluence in 100 flasks (10 × 15 cm), harvested by rapid agitation and pelleted by centrifugation at 300 × g. The resulting pellet of intact cells was resuspended in 10 volumes of ice-cold assay buffer [50 mM imidazole HCl (pH 7.0) containing 1 mM EDTA, 1 mM EGTA, 100 µM N-methylmaleimide and 10 µM PSCP] and sonicated (Sonic Dismembrator, model 50, Fisher Scientific, Santa Clara, CA). Homogenates were centrifuged at 15,000 × g for 15 min, the pellet was discarded and the cytosol and microsome preparations were obtained by further centrifugation at 105,000 × g for 60 min at 4°C. In some studies the low molecular weight fraction (containing EDTA and EGTA) was removed from cytosolic preparations by gel filtration (10 DG Econo-Pac desalting column, Bio-Rad, Richmond, CA). The supernatant (cytosolic) and pellet (microsomal) fractions were analyzed for protein (Bradford, 1976) and stored at 70°C. Mouse brain cytosol was prepared as above except for use of Polytron and Potter-Elvehjem homogenization procedures in place of sonication.

Competitive binding studies. The binding assays were based on those reported for mouse tissues (Graziano et al., 1987, 1988). In the standard procedure, 20 µg cytosol protein (or 15 µg microsome protein) and 5 nM [³H]CA were incubated in pH 7.0 assay buffer for 2 hr at 37°C. These conditions were varied in preliminary studies to establish optimal binding parameters. Candidate inhibitors were dissolved at 10 mg/ml in dimethyl sulfoxide or assay buffer, as appropriate, followed by serial dilution in buffer. After incubation, samples were simultaneously filtered through glass-fiber filters (Wallac Filtermat B) presoaked in 0.3% polyethylenimine (to retain charged proteins) and rapidly rinsed (3 × 1 ml) with 50 mM imidazole buffer (pH 7.0) containing 100 µM N-methylmaleimide with use of a Wallac TomTec cell harvester (Harvester 96 Mach II, Gaithersburg, MD). Filtermats were dried and impregnated with MeltiLex melt-on scintillator sheets (Wallac) and the bound radioactivity was quantified by use of a Wallac Betaplate (model 1205) flat-bed counter. Specific binding is defined as the difference between labeling with 5 nM [³H]CA only (total binding) and in the presence of 10 µM unlabeled CA (nonspecific binding).

Data analysis. IC₅₀ values, the concentrations for 50% inhibition of binding, were determined by four-parameter logistic curve-fitting. LC₅₀ values for 50% inhibition of cell viability were estimated by plotting log-concentration of test compound versus absorbance of the MTT formazan product on a probit scale. The correlation between IC₅₀ and LC₅₀ data was determined by simple linear regression. Data presented in figures represent typical single experiments.

	Results

Top Abstract Introduction Methods Results Discussion References

Binding characteristics of [³H]CA in cytosol and microsomes. [³H]CA undergoes specific and saturable binding in both the cytosol and microsome fractions (fig. 2). Specific binding is >90% under the standard assay conditions. On considering the overall postmitochondrial binding activity, ~70% resides in the cytosol and ~30% in the microsomes. Binding is essentially linear and similar for cytosol and microsomes up to 20 µg protein per tube under the standard assay conditions. At higher extract concentrations the binding continues to be linear in the microsome fraction but not in the cytosol. The linear range for cytosol is extended on removal of the low molecular weight fraction by gel filtration (data not shown) presumably by eliminating an inhibitory component. The association half-time is 28 and 39 min for the cytosol and microsomes, respectively, with maximum binding at 60 to 180 min and some decrease by 360 min. Scatchard analysis establishes saturable, high-affinity, specific binding as a single component with K_d values of 12 and 44 nM and B_max values of 22 and 69 pmol/mg protein for the cytosol and microsome fractions, respectively.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Binding characteristics of [3H]CA in neuroblastoma cytosol and microsomes. Binding assays performed with 20 µg cytosol protein or 15 µg microsome protein (equivalent to 100,000 and 250,000 cells, respectively) and 5 nM [3H]CA in 250 µl, pH 7, assay buffer incubated for 2 hr at 37°C. The panels show effect of protein level, incubation time and [3H]CA concentration (as Scatchard plots), with all other factors held constant.

Inhibition of [³H]CA binding. [³H]CA binding in neuroblastoma cytosol is inhibited by microcystin LR, cantharidin, endothall, ATP and Ca⁺⁺ with IC₅₀ values of 0.70, 10, 254, 5800 and 340,000 nM, respectively (fig. 3). The curves for these compounds, as examples of inhibitors of the various types, are consistent with a single binding site in each case except ATP which diverges at high concentrations. The same relative potencies and characteristics of the inhibition curves are evident with neuroblastoma microsomes and mouse brain cytosol (not illustrated).

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Inhibition of [3H]CA binding by microcystin LR (MC), cantharidin (CA), endothall (EN), ATP and CaCl2 in neuroblastoma cytosol. The incubation was initiated by adding cytosol to [3H]CA and test compound under the standard assay conditions.

Cytotoxicity of cantharidin. Cantharidin reduces cell viability in a time- and concentration-dependent manner with a 24-hr LC₅₀ of 4.5 µM decreasing progressively to 1.3 µM by 120 hr (fig. 4). Assays were standardized at 24 hr for convenience and considering the variable stability of the inhibitors. On comparing the in vivo and in vitro systems, the LC₅₀ for cantharidin cytotoxicity is 400-fold higher than the IC₅₀ for [³H]CA binding (fig. 5).

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Time course of cantharidin cytotoxicity in neuroblastoma cultures. Cell viability measured by the MTT method. LC50 values obtained from plots of cantharidin concentration versus cell viability at times indicated.

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Comparison of cantharidin potency in inhibiting [3H]CA binding in cytosol (IC50) and cytotoxicity in neuroblastoma cultures (LC50).

Structure-activity relationships for cytotoxicity. Microcystin LR and the DSPs range from very potent to low activity as cytotoxicants, i.e. LC₅₀ values of 3 to 31 nM for calyculin A and okadaic acid versus >5000 nM for microcystin LR. Cantharidin, palasonin as the diacid and endothall thioanhydride are quite toxic (LC₅₀ values, 2300-8900 nM); two other analogs are of moderate potency (endothall and its anhydride giving the same values of ~43,000 nM); the remaining compounds are essentially inactive (table 1).

Correlation between potency as inhibitors of [³H]CA binding and as cytotoxicants. In general, the structure-activity relationships for PP2A inhibition parallel those for cytotoxicity (table 1, fig. 6). A plot of the IC₅₀ versus the LC₅₀ data (24 hr) on a log-log scale establishes a linear relationship and high correlation for the active compounds (r² = .9, n = 8) and three analogs inactive in the binding assay are also not cytotoxic, suggesting that the binding affinity determines the cytotoxicity. Cantharidin and ETA differ only 2.3-fold in cytotoxicity but 27-fold in binding affinity. Calyculin A is the only compound with an LC₅₀ below the IC₅₀. Microcystin LR with an LC₅₀ >5000 nM is not included in calculation of the correlation coefficient.

View larger version (19K): [in this window] [in a new window]   Fig. 6.   Correlation for diarrhetic shellfish poisons and endothall analogs between potency as inhibitors of [3H]CA binding in cytosol and as toxicants in neuroblastoma cells. Individual compounds are identified in table 1. Data points are means of 3 or more measurements. Compound 1 is not cytotoxic (LC50 < 5000 nM) and 10-13 are essentially inactive in both assays.

Inhibition of [³H]CA binding by ATP and other phosphorus-containing bifunctional acids. The binding site of neuroblastoma cytosol is much more sensitive to ATP (IC_50, 6 µM) than ADP (IC_50, 37 µM), AMP or other adenine nucleotides (IC₅₀ values > 1000 µM) (table 2). The most potent inhibitors are pyrophosphate and phosphonoformate (IC₅₀ values, 1.5 and 2.5 µM, respectively) with lower activity for four of their analogs (4-6 and 8). The same structure-activity relationships are evident with neuroblastoma and mouse brain cytosols (table 2). The phosphonoformate derivative fosamine is inactive.

View larger version (27K): [in this window] [in a new window]   Fig. 7.   Scatchard plot showing effect of pyrophosphate on the binding of [3H]CA to mouse brain cytosol. Kd and Bmax ± S.E. shown in inset. Significant differences (P  .05) are observed for the Kd values (nM) but not the Bmax values (pmol/mg protein).

View larger version (28K): [in this window] [in a new window]   Fig. 8.   Effect of 20 µM cantharidin on cell viability in the presence of pyrophosphate (POP) and methylenediphosphonate (PCP) at 0.3 µM to 1 mM. Error bars represent variation within a single experiment.

Inhibition and stimulation of [³H]CA binding by selected divalent cations. [³H]CA binding in neuroblastoma cytosol is inhibited by NiCl₂, CoCl₂, ZnCl₂ and MgCl₂ with IC₅₀ values of 20 to 65 µM, whereas with MnCl₂ and CaCl₂ they are 150 and 340 µM, with similar inhibition characteristics for mouse brain cytosol (table 3). Mn⁺⁺ is unique in stimulating [³H]CA binding at 25 to 50 µM, whereas at high levels it inhibits CA binding similar to the other divalent cations (fig. 9). Scatchard plots suggest that 50 µM Mn⁺⁺ increases the number of [³H]CA binding sites, whereas higher concentrations inhibit radioligand binding (figs. 9 and 10).

View larger version (22K): [in this window] [in a new window]   Fig. 9.   Effects of divalent cations on [3H]CA binding in neuroblastoma cytosol. Low molecular weight fraction removed by gel filtration, and EDTA or EGTA was not used.

View larger version (23K): [in this window] [in a new window]   Fig. 10.   Scatchard plot showing effect of MnCl2 on [3H]CA binding in neuroblastoma cytosol. For conditions see figure 9. Kd (nM) and Bmax (pmol/mg protein) ± S.E. are shown in inset.

	Discussion

Top Abstract Introduction Methods Results Discussion References

Protein phosphatases are present at high levels in brain and play an essential role in neuronal function (Hemmings et al., 1989; Huganir and Greengard, 1990). The microsomal site may play a special role because there are neuronal-specific isoforms of PP2A associated with synaptosomes (Mayer et al., 1991; Sim et al., 1994). The mouse N1E-115 neuroblastoma cell line is similar to mouse brain in the binding characteristics and ligand specificity of its [³H]CA binding site. B_max values (pmol/mg protein) for the [³H]CA binding site vary 10-fold in various assays of cytosolic preparations, i.e., 22 for the N1E 115 cells and 8 for the mouse brain (this study in an optimized system) compared with 1.8 to 3.4 for the mouse brain in earlier studies (Graziano et al., 1988). In each case the microsomal binding sites are very similar to those of cytosol, which suggests that both are potential targets for toxicant action.

The potency of DSPs and endothall analogs in competing for the [³H]CA binding site is predictive of their cytotoxicity in neuroblastoma cells, which suggests that these two classes of compounds act at the same or closely coupled sites on PP2A in disrupting cell functions. Two DSPs (calyculin A and okadaic acid) are the most active compounds in both binding and cytotoxicity assays, perhaps reflecting very slow rates of dissociation as established earlier with okadaic acid (Takai et al., 1992). The higher potency of calyculin A in cytotoxicity than in [³H]CA binding assays may be attributable to dual targets, i.e., PP1 as well as PP2A (Ishihara et al., 1989) with [³H]CA binding only estimating the PP2A site (Li et al., 1993). Endothall, endothall anhydride and endothall thioanhydride have identical affinities for the binding site in vitro, but the thioanhydride is 5-fold more toxic than the others to neuroblastoma cells and 13- to 45-fold more toxic to mice (Kawamura et al., 1990), which suggests possible differences in toxicant distribution or the potential interaction of endothall thioanhydride with additional targets such as PP1 (Erdödi et al., 1995).

Microcystin LR, despite its high inhibitory potency at the [³H]CA binding site (IC₅₀ < 1 nM), is not cytotoxic in neuroblastoma cells. This may be related to poor cellular uptake dependent on a bile acid transport mechanism as noted earlier with hepatocytes and enterocytes (Hooser et al., 1991; Runnegar et al., 1991,1995; Falconer et al., 1992).

Cantharidin cytotoxicity develops slowly over 48 to 120 hr (this study), whereas the DSPs alter protein phosphorylation, cell ultrastructure and cell viability within minutes. This difference in speed of action is probably caused by their relative uptake, potency and reversibility (Eriksson and Golman, 1993; Hardie et al., 1991). These factors are also likely to play a role in the 400-fold higher LC₅₀ value for CA cytotoxicity than the IC₅₀ for [³H]CA binding.

Several endogenous compounds inhibit [³H]CA binding including ATP, ADP, pyrophosphate, Ca⁺⁺ and Mg⁺⁺, in some cases at normal cellular levels (Graziano et al., 1988; Li et al., 1993), which suggests that these or similar endogenous substances may be important in regulating PP2A function. The action of ATP and pyrophosphate is not caused by phosphotransferase reactions because nonphosphorylating analogs (with P---C---P bonds replacing the P---O---P required for phosphorylation) are also effective, and Scatchard analysis suggests that pyrophosphate competes at the toxicant binding site. Despite the high potency of pyrophosphate and methylenediphosphonate in vitro, they are ineffective in vivo as assayed in alleviating cantharidin-associated cytotoxicity.

Divalent cations inhibit [³H]CA binding including Ca⁺⁺ and Mg⁺⁺ (Graziano et al., 1988) and Mn⁺⁺, Co⁺⁺, Ni⁺⁺ and Zn⁺⁺ (this study). Mn⁺⁺ at 25 to 50 µM increases [³H]CA binding, a unique observation relative to other divalent cations. Interestingly, ATP-induced inhibition of cardiac phosphorylase phosphatase activity is reversed by Mn⁺⁺ or Co⁺⁺ but not other divalent cations (Hsiao et al., 1978). Additional evidence of a functional role for Mn⁺⁺ is the observation that expressed PP1 catalytic subunit (related to PP2A gene) requires Mn⁺⁺ for full activity in recombinant insect or bacterial expression systems (Mumby and Walter, 1993). These observations indicate that endogenous divalent cations, known to be required for PP2B and PP2C activity (Cohen, 1991), are modulators of [³H]CA binding and may therefore affect the toxicity of PP2A inhibitors.

Neuroblastoma cells are validated here as a model system for predicting whole-animal toxicity of PP inhibitors (Graziano et al., 1988; Li and Casida, 1992) and for establishing the relationship between the toxicant, ATP and Mn⁺⁺ binding sites of PP2A and its cellular functions.