Evidence for an enantioselective pumiliotoxin 7-hydroxylase in dendrobatid poison frogs of the genus Dendrobates

Abstract

Dendrobatid poison frogs readily accumulate alkaloids from diet into skin, where such compounds serve as a chemical defense against predators. Arthropods seem to be the source of decahydroquinolines (DHQs), several izidines, coccinellines, spiropyrrolizidines, pumiliotoxins (PTXs), and allopumiliotoxins (aPTXs). A DHQ iso-223F, and PTX (+)-251D were fed to poison frogs of the dendrobatid genera Dendrobates, Epipedobates, and Phyllobates. The two alkaloids were accumulated in skin unchanged except for the three species of Dendrobates, where ≈80% of accumulated PTX (+)-251D was stereoselectively hydroxylated to aPTX (+)-267A. The unnatural enantiomer PTX (-)-251D was accumulated efficiently when fed to Dendrobates auratus, but was not hydroxylated. The enantiomers of PTX 251D and their desmethyl analogs were synthesized from N-Boc-protected (-)- and (+)-proline methyl esters. Both PTX (+)-251D and aPTX (+)-267A proved to be potent convulsants in mice, with (+)-267A being ≈5-fold more toxic than (+)-251D. Both alkaloids were hyperalgesic at the site of injection. The unnatural PTX (-)-251D caused no overt effect in mice. Thus, the evolutionary development of a pumiliotoxin 7-hydroxylase would have provided frogs of the genus Dendrobates with a means of enhancing the antipredator potency of ingested PTXs.

Amphibian skin has over the past century been found to contain a variety of biologically active compounds (1). Such compounds seem to have evolved for chemical defense against infection and/or against predators. The many alkaloids, discovered in skin extracts of poison frogs of the neotropical family Dendrobatidae (2, 3), were at first thought to be elaborated by such frogs. Almost all of the “dendrobatid alkaloids” were unknown elsewhere in Nature. However, when dendrobatid frogs were raised in captivity they completely lacked alkaloids (4, 5). But when fed fruit flies dusted with alkaloid-containing powder or fed alkaloid-containing myrmicine ants, the frogs readily accumulated the alkaloids into skin (6, 7). It seemed that the frogs had no enzymes associated with alkaloid synthesis or metabolism but instead had developed or overexpressed an uptake system or systems designed to capture dietary alkaloids for use in skin secretions as a chemical defense. Alkaloids sequestered in the wild were retained during several years in captivity (4, 5). Many of the alkaloids, for example the batrachotoxins, the pumiliotoxins (PTXs) and allopumiliotoxins (aPTXs), and epibatidine, were highly toxic, whereas others, such as the decahydroquinolines (DHQs), the histrionicotoxins and the izidines, were of low toxicity, but would still serve as deterrents to predators because of bitter, unpleasant effects on buccal tissue (8, 9). “Dendrobatid alkaloids” are now known to occur in one genus from each of three other anuran families, namely Mantella from the Madagascan family Mantellidae (10, 11), the South American Melanophryniscus from the family Bufonidae (12) and the Australian Pseudophryne from the family Myobatrachidae (13). The Mantella frogs, like the dendrobatid frogs, seemed dependent on a dietary source (14). The Pseudophryne frogs seemed dependent on diet for their PTXs/aPTXs (15). However, raised in captivity, the Pseudophryne frogs still contained pseudophrynamines in skin and, thus, the frogs seem able to synthesize these unusual indolic alkaloids (15). The dietary source of the many classes of alkaloids found in skin of frog taxa has become a major challenge.

In the present alkaloid-feeding experiments, a synthetic DHQ iso-223F, an isomer of a natural DHQ, code name 223F (3), was fed along with a simple PTX (+)-251D. DHQs were known to be readily accumulated unchanged into skin of dendrobatid frogs (6). The experiment was designed to determine for the first time the relative efficiency of uptake of two of the most common classes of alkaloids found in skin of dendrobatid frogs, namely DHQs of ant origin (16) and PTXs that originate from an arthropod source, as yet unidentified as to taxon (17). However, the results led to evidence for the evolutionary development in one lineage of dendrobatid frogs of an enantioselective pumiliotoxin 7-hydroxylase that converts a toxic PTX (+)-251D to a more toxic aPTX (+)-267A (Fig. 1). The nontoxic, unnatural (-)-251D enantiomer was not hydroxylated. The DHQ iso-223F was sequestered unchanged.

Fig. 1.

Sequestration and hydroxylation of alkaloids of the DHQ class and the PTX class fed to dendrobatid poison frogs of the genus Dendrobates. Enantioselective conversion of (+)-PTX 251D to aPTX (+)-267A.

Materials and Methods

Alkaloids. The synthetic DHQ iso-223F is the enantiomeric compound 18 of ref. 18 and was provided by A. G. Schultz (Rensselaer Polytechnic Institute, Troy, NY); the synthetic aPTX (+)-267A was provided by L. E. Overman (University of California at Irvine).

Synthesis of Pumiliotoxins. PTXs (-)-251D and (+)-251D and their desmethyl analogs were synthesized as shown in Fig. 2. The synthesis of (+)-251D (9), (-)-251D (12) and the two enantiomeric desmethyl analogs 13 and 14 followed methodology developed by L. E. Overman (19–21) and by Barger et al. (22). In the case of (+)-251D (9), the synthesis began with the N-Boc derivative of l-proline methyl ester (1), which was reacted with methyl magnesium iodide, followed by dehydration of the resultant crude alcohol with SOCl₂/pyridine to provide the alkene 2. Epoxidation with m-chloroperbenzoic acid gave a mixture of diastereomeric epoxides 3. These procedures follow those of Overman and Bell (19) as described for N-Cbz-l-proline methyl ester. R-2-Methylhexanol (4) was converted to the dibromoalkene 6 in two steps by Swern oxidation, followed by a Wittig reaction of the crude aldehyde 5 with triphenylphosphine and CBr₄. These steps follow those of Caderas et al. (21). The dibromoalkene 6 was converted to the lithium salt of the corresponding alkyne with n-butyl lithium, followed by reaction with the diastereomeric epoxides (3) in the presence of Et₂AlCl to afford 7. Deprotection of 7 by treatment with camphorsulfonic acid was followed by cyclization/iodination with NaI and paraformaldehyde. The resultant crude vinyl iodide 8 was treated with s-butyl lithium to yield after purification the hydrochloride of (+)-251D (9). This sequence of reactions was based on that of Barger et al. (22). The synthesis of (-)-251D (12) followed the same overall procedure, starting with the N-Boc derivative of d-proline methyl ester (10) and by using S-2-methylhexanol (11) as the source of the dibromoalkene. The desmethyl analog 13 of (+)-251D was synthesized from the diastereomeric epoxides (3) by reaction with the lithium salt of 1-heptyne to form the desmethyl analog of 7, followed as above with acid, NaI, and paraformaldehyde to yield a desmethyl analog of 8, which was deiodinated to yield after purification the hydrochloride of the desmethyl analog 13. The desmethyl analog 14 of (-)-251D was synthesized from the diastereomeric epoxides obtained from 10 in the same manner. Physical properties of 2, 3, 6, 7, 9, and 12-14 are provided in Supporting Text, which is published as supporting information on the PNAS web site, www.pnas.org. R- and S-2-Methylhexanol and 1-heptyne were from commercial sources, and 5 and 8 were used without purification.

Fig. 2.

Synthetic routes to (+)-251D (9), (-)-251D (12) and the desmethyl analogs 13 and 14. a, MeMgI, Et₂O, reflux 1 h; b, SOCl₂, NEt₃, THF, -45°C, 1 h; c, m-chloroperbenzoic acid, NaHCO₃, CH₂Cl₂, at 0°C, then room temperature overnight; d, Swern oxidation with oxalyl chloride and DMSO, CH₂Cl₂, at -78°C, 30 min, add NEt₃, after 1 h warm to 0°C, and quench with aqueous HCl; e, Ph₃P in CH₂Cl₂, at -15°C, add CBr₄, cool to -78°C, add 5, after 30 min warm to -20°C, 1 h; f, 6 in toluene at -78°C, add s-BuLi, 2 h, warm to -10°C, add Et₂AlCl, after 1 h add 3, warm to 0°C, 2 h, quench with aqueous NH₄Cl; g, camphorsulfonic acid, H₂O/MeCN, heat to 80°C, after 3 h add NaI, (CH₂O)_n, heat to 100°C, 5 h; a CH₂Cl₂ extract yields the crude product 8; h, crude 8 in anhydrous Et₂O, add s-BuLi at -78°, slowly warm to -20°C, cool to -78°C, add CH₃OH, then room temperature overnight; i, as in f, with 1-heptyne in place of 6 and then continue as in g and h.

Frog Experiments. The following species were captive-raised to adulthood at the National Aquarium in Baltimore: Dendrobates auratus (Girard, 1855); Dendrobates galactonotus (Steindachner, 1864); Dendrobates castaneoticus (Caldwell and Myers, 1990); Epipedobates tricolor (Boulenger, 1899), and Phyllobates bicolor (Bibron, 1841). Before and during alkaloid-feeding experiments they were housed in glass terraria and fed termites and/or wingless fruit flies. Immediately before feeding the frogs, the termites or fruit flies were dusted with a vitamin-mineral powder (Nekton-Rep, Nekton, Clearwater, FL) containing ≈1% by weight of a 1:1 mixture of DHQ iso-223F and either PTX (+)-251D or the unnatural enantiomer PTX (-)-251D. After 1–6 wk of alkaloid-feeding, as indicated in the legend to Fig. 3, the frogs were killed, and skin extracts were prepared and partitioned as described (15) to yield alkaloid fractions. Alkaloids were detected with a flame-ionization detector after gas chromatography on a six-feet 1.5% OV-1 packed column (2 mm, i.d.), programmed from 150 to 280°C at 10°/min with helium as the carrier gas at ≈30 ml/min. Injection volume was 2 μl, corresponding to 2 mg wet weight frog skin. Gas chromatograms are shown in Fig. 3. The alkaloids were then analyzed by gas chromatography-mass spectrometry as described (15).

Fig. 3.

Gas chromatographic profiles for alkaloids in skin of dendrobatid frogs fed alkaloid-dusted termites and/or fruit flies. Standard 1:1 mixtures of DHQ iso-223F and PTX (+)-251D were fed to frogs on dusted fruit flies unless otherwise noted. (A) Dendrobates auratus; DHQ and (+)-251D, primarily on termites, 6 wk. (B) D. auratus; DHQ and (+)-251D, 4 wk. (C) Epipedobates tricolor; DHQ and (+)-251D, 4 wk. (D) Phyllobates bicolor; DHQ and (+)-251D, 4 wk. (E) D. galactonotus; DHQ and (+)-251D, 1 wk. Frogs ate poorly and both died in ≈1 wk. (F) D. castaneoticus; DHQ and (+)-251D, 2 wk. Frogs ate poorly. (G) D. auratus; DHQ and unnatural (-)-251D, 5 wk. In each experiment, nearly identical results were obtained with 1 or 2 additional frogs.

Toxicity. The toxicities of (+)-251D,(-)-251D, the two desmethyl analogs 13 and 14, and (+)-267A were compared on s.c. injection in male 25- to 30-g NIH Swiss mice in a 1:4 mixture of methanol and isotonic saline. The injection volume was 0.2 ml. In further experiments, anticonvulsants (phenobarbital, carbamazepine, diazepam, and dizocilpine) were injected s.c. 10 min before 10 mg/kg (+)-251D. Two or three mice were used for each treatment.

Results

Feeding Experiments. The initial feeding experiments, using mainly alkaloid-dusted termites, with D. auratus, a Central American species, provided a completely unexpected result. The major portion of the dietary PTX (+)-251D was present in skin as the 7β-hydroxy derivative aPTX (+)-267A. The vapor phase infrared spectrum (not shown) demonstrated clearly that (+)-267A had been formed, not the 7-epimer. Thus, the hydroxylation was stereospecific. The DHQ iso-223F accumulated into skin unchanged, as expected. The second experiment used alkaloid-dusted fruit flies with D. auratus and gave nearly the same result; no metabolism of the DHQ and extensive metabolism of (+)-251D to (+)-267A (compare Fig. 3 A and B). The 7-hydroxylation of PTX (+)-251D did not occur with an Epipedobates species (Fig. 3C) or a Phyllobates species (Fig. 3D). The third feeding experiment compared results with D. auratus (Fig. 3B) with those for two other species of the genus Dendrobates, namely D. galactonotus (Fig. 3E) and D. castaneoticus (Fig. 3F). Both also metabolized PTX (+)-251D to (+)-267A, but these Amazonian species ate very little, probably due to suboptimal ambient temperatures of only 20–21°C. Finally, the DHQ iso-223F and the unnatural enantiomer PTX (-)-251D were fed to D. auratus. Both alkaloids were sequestered into skin, but there was no detectable conversion of (-)-251D to a 267A (Fig. 3G).

Toxicity. At a dose of 2 mg/kg, PTX (+)-251D caused only some initial hyperactivity in the mice, but no further effects. However, at 10 mg/kg, there was apparent marked pain at the injection site and hyperactivity. In ≈1 min, running convulsions commenced, leading to death after several minutes. Remarkably, the unnatural (-)-251D had virtually no effect on mice at 10 mg/kg. The aPTX (+)-267A was ≈5-fold more toxic than PTX (+)-251D. At 2 mg/kg aPTX (+)-267A, there were symptoms similar to those evoked by 10 mg/kg of PTX (+)-251D, with death in 6 min. Even at 0.5 mg/kg (+)-267A, there was apparent pain at the site of injection and pronounced long-lasting hyperactivity, with recovery in ≈1 h. The toxic effects (hyperactivity, convulsions, and death) evoked by 10 mg/kg of PTX (+)-251D were reduced by prior administration of the anticonvulsants phenobarbital (50 mg/kg) and carbamazepine (30 mg/kg), while being little affected by diazepam (5 mg/kg) or dizocilpine (0.5 mg/kg). The desmethyl analog 13 of PTX (+)-251D at 10 mg/kg seemed to cause only slight pain at the site of injection, followed by initial hyperactivity with complete recovery in 20 min. At 20 mg/kg, marked pain at the injection site and initial hyperactivity was followed in ≈10 min by prostration and finally loss of righting reflex and minor convulsions after ≈40–60 min. There was little recovery even after 3 h. The desmethyl analog 14 at 20 mg/kg caused no apparent pain at the injection site and had minimal effects on activity.

Discussion

An alkaloid-sequestering system has been proposed to be the common evolutionary event allowing certain lineages of frogs from four different families to accumulate dietary alkaloids into secretory granular glands of skin for use in chemical defense against predators (23). It seemed likely that this sequestering system was a primitive one and had been merely overexpressed in such frogs, who then were dependent on sequestered dietary alkaloids for chemical defense. The lineages proposed as having overexpressed this system were neotropical dendrobatid frogs of genera Dendrobates, Epipedobates, Minyobates, and Phyllobates, South American bufonid toads of the genus Melanophryniscus, Madagascan mantellid frogs of the genus Mantella, and myobatrachid frogs of the Australian genus Pseudophryne. Further experiments have now shown that frogs of the genus Pseudophryne do not have the same sequestering system, but instead seem likely to have an uptake system designed to retain the indolic pseudophrynamines that are synthesized by such frogs (15). The uptake system of the Pseudophryne did accumulate a PTX, albeit poorly, but not a simple DHQ or izidines. The selectivity of the uptake system probably accounts for the lack of the ant alkaloids of the DHQ and izidine classes in frogs of this genus, frogs known to be “ant specialists” (24).

Based on alkaloid-feeding experiments with frogs of the genera Dendrobates, Epipedobates, Phyllobates, and Mantella (5–7, 14), it had been proposed that the sequestering systems in such frogs were the same or very similar. Because no significant metabolism of any fed alkaloid had been observed with these frogs, it had also been assumed that all alkaloids would be sequestered unchanged. The alkaloids that have been fed and sequestered without metabolic change into skin of dendrobatid frogs (refs. 6 and 7, and unpublished results) include batrachotoxinin-A, a steroidal alkaloid of unknown dietary origin, and the following classes of alkaloids, demonstrated or presumed to be of ant origin (refs. 25 and 26, and unpublished results): pyrrolizidines, indolizidines, decahydroquinolines, and histrionicotoxins. A decahydroquinoline, two indolizidines, a histrionicotoxin, and aPTX 267A, fed to frogs of the genus Mantella, were sequestered unchanged into skin (14). A piperidine and a pyrrolidine did not seem to be effectively sequestered when fed to either dendrobatid or mantellid frogs (6, 14). Dietary alkaloids were not accumulated into skin of dendrobatid frogs of the genus Colostethus (6); wild-caught Colostethus frogs do not have detectable lipophilic skin alkaloids.

In view of the results of earlier feeding experiments, it was entirely unexpected to find that PTX (+)-251D was metabolized extensively when fed to a dendrobatid frog, D. auratus (Fig. 3 A and B). Further experiments revealed that another frog, D. galactonotus, which, like D. auratus, is a member of the Dendrobates tinctorius group (27), also extensively metabolized PTX (+)-251D to aPTX (+)-267A (Fig. 3E). In addition, D. castaneoticus, a member of the Dendrobates quinquevittatus group (28), also hydroxylated PTX (+)-251D (Fig. 3F). Thus, it would seem that the expression of a pumiliotoxin 7-hydroxylase may prove to be common to all poison frogs of the genus Dendrobates. However, no frog of the third group of Dendrobates, namely the Dendrobates histrionicus group (29), has been tested, nor have frogs of a proposed dendrobatid genus Minyobates (30). No metabolism of PTX (+)-251D occurred in frogs of two other dendrobatid genera, namely Phyllobates and Epipedobates (Fig. 3 C and D).

Examination of our data on occurrence of PTX 251D and aPTX 267A in extracts from dendrobatid, mantellid, and bufonid anurans supports the conclusion that the presence of a pumiliotoxin 7-hydroxylase is probably associated as a synaptomorphy with frogs of the dendrobatid genus Dendrobates. In frogs of the genus Dendrobates, including members of the D. histrionicus group (29), the presence of significant amounts of PTX 251D always was accompanied by significant amounts of the putative hydroxylation product aPTX 267A. Representative examples are presented in Table 1. In some cases, aPTX 267A was present, whereas PTX 251D was not detected. In such cases, the aPTX 267A may have been of dietary origin or dietary PTX 251D may have been completely converted to aPTX 267A. Certainly, the question remains unanswered as to how much of the frog skin aPTX 267A is formed by 7-hydroxylation of ingested PTX 251D and how much is obtained directly from a dietary source. The aPTXs probably do originate in some cases from dietary sources, a possibility indicated by the recent discovery of aPTX 323B in mixed collections of Panamanian leaf-litter arthropods that were microsympatric with Dendrobates pumilio frogs containing aPTX 323B (17).

Table 1.

Occurrence of pumiliotoxin 251D and allopumiliotoxin 267A in skin extracts of frogs of the dendrobatid genera Dendrobates, Epipedobates, Minyobates, and Phyllobates^*

Dendrobatidae	PTX 251D	aPTX 267A
Dendrobates
D. arboreus	Minor	Major
D. auratus (5A)†‡	Minor	Major
D. azureus	ND§	Major
D. granuliferus	ND	Major
D. histrionicus (14A)†‡	Minor	Major
D. histrionicus (14C)†	ND	Major
D. lehmanni	Trace	Major
D. leucomelas (16A)†	ND	Major
D. occultator	ND	ND
D. pumilio (25A)†‡	ND	Trace
D. pumilio (25F)†	Minor	Major
D. quinquevittatus	ND	Minor
D. reticulates	ND	Minor
D. speciosus (29A)†	Minor	Minor
D. tinctorius	ND	Major
D. truncates	ND	ND
D. ventrimaculatus (35A)†	ND	Minor
D. vicentei¶ (8A)†	Major	Major
Epipedobates‡
E. erythromos	ND	Major
E. espinosai	Minor	Minor
E. parvulus (22A)†	ND	ND
E. petersi	ND	ND
E. pictus (24 A)†	ND	ND
E. silverstonei	Major	ND
E. tricolor	Major	Trace
E. trivittatus	ND	ND
Minyobates
M. abditus	ND	ND
M. altobueyensis	ND	Major
M. bombetes (7A)†	Major	ND
M. claudiae∥ (17A)†	Minor	Minor
M. fulguritus	ND	Major
M. viridis (36A)†	ND	ND
M. minutus (18A)†	Trace	Major
M. opisthomelas	ND	ND
M. steyermarki	Trace	Major
Phyllobates‡
P. aurotaenia	Trace	ND

Data primarily from table 5 of ref. 1.

^*The generic classifications are those of Myers (30). It has been suggested, based on very limited molecular data, that the frogs of the proposed genus Minyobates should be returned to the genus Dendrobates (31, 32). The present results suggest that further molecular data are required

^†Representative populations of this species are given. Code designations are from table 5 of ref. 1. For other populations see that table

^‡In certain populations of D. auratus, D. histrionicus, and D. pumilio and in Epipedobates anthonyi, E. femoralis, E. myersi, Phyllobates bicolor, Phyllobates lugubris, Phyllobates terribilis and Phyllobates vittatus neither PTX 251D nor aPTX 267A were detected (see tables 3 and 5 of ref. 1)

^§ND, not detected

^¶Described in ref. 33, previously cited as D. species (8A, 8B) in table 5 of ref. 1

^∥Described as D. claudiae of the minutus group in ref. 34, previously cited as D. species (17A, 17B) in table 5 of ref. 1

Both PTX 251D and aPTX 267A have been found to occur in skin from dendrobatid frogs of the polyphyletic genus Epipedobates (Table 1). In two species of Epipedobates, aPTX 267A was absent or a trace, whereas PTX 251D was a major alkaloid (Table 1). In the most striking example, the major alkaloid in skin extracts from one population of the Ecuadoran E. tricolor was PTX (+)-251D (≈300 μg per frog; ref. 35) whereas only trace amounts of aPTX 267A were detected. In the present study, E. tricolor did not convert PTX 251D to aPTX 267A (Fig. 3C). In E. erythromosus, aPTX 267A was a major alkaloid whereas PTX 251D was not detected. In E. espinosai, both PTX 251D and aPTX 267A were present as minor alkaloids (Table 1). The aPTX 267A presumably must have a dietary origin, if these species, like E. tricolor, do not have a pumiliotoxin 7-hydroxylase.

In species of the proposed dendrobatid genus Minyobates (ref. 30; see footnote *, Table 1), for which feeding experiments were not conducted, the data from extracts are difficult to interpret for this probably polyphyletic genus. Two populations of Minyobates bombetes had PTX 251D as a major alkaloid whereas aPTX 267A was not detected, indicating lack of the 7-hydroxylase (Table 1). In different populations of Minyobates minutus, a species suggested on the basis of molecular data to group with Dendrobates (32), aPTX 267A was always a major alkaloid whereas PTX 251D varied from not detectable to being a major alkaloid. In Minyobates claudiae and Minyobates viridis, PTX 251D was a minor alkaloid and was accompanied by aPTX 267A as a minor or major alkaloid, respectively. In Minyobates fulguritus extracts, a species suggested on the basis of mitochondrial and ribosomal RNA analyses to group with Dendrobates (31), aPTX 267A was a major alkaloid, whereas PTX 251D was not detected. The data from Minyobates bombetes with major amounts of PTX 251D and no aPTX 267A in skin extracts indicate the lack of a pumiliotoxin 7-hydroxylase in this species, but conclusions regarding other Minyobates are not yet warranted. Such tiny frogs are “mite specialists” (36).

In poison-dart frogs of the genus Phyllobates, PTX 251D was absent except for a trace amount in one species (Phyllobates aurotaenia), and aPTX 267A was not detected (Table 1). PTX 251D was readily taken up by Phyllobates bicolor, but was not hydroxylated (Fig. 3D).

Madagascan mantellid frogs of the genus Mantella contain a variety of PTXs, some aPTXs, and some homopumiliotoxins (10, 11). PTX 251D was detected as a major alkaloid in populations of Mantella cowani and Mantella expectata, and as a minor or trace alkaloid in populations of Mantella betsileo and Mantella baroni. However, aPTX (+)-267A was not detected in Mantella frogs, although the 7-epimer was present as a trace alkaloid in extracts of one population of Mantella cowani, where PTX 251D was a major alkaloid. Thus, it seems that Mantella frogs do not have a pumiliotoxin 7-hydroxylase.

In bufonid toads of the South American genus Melanophryniscus,PTX 251D has been detected as a minor or trace alkaloid, but aPTX 267A has never been detected in the six species from Argentina, Uruguay, and Brazil that have been studied (ref. 12, and unpublished results). Thus, such toads apparently do not have a pumiliotoxin 7-hydroxylase.

Neither PTX 251D nor aPTX 267A have been detected in skin extracts of myobatrachid frogs of the Australian genus Pseudophryne. However a side-chain hydroxylated analog of 251D, namely PTX 267C, does occur in Pseudophryne, often as a major alkaloid (13). An aPTX that would result from 7-hydroxylation of PTX 267C has not been detected in Pseudophryne. PTX 307A, when fed to Pseudophryne semimarmorata, was accumulated to a limited extent, accompanied by apparent conversion to reduced and/or side-chain hydroxylated congeners (15). Thus, it is unlikely that Pseudophryne frogs have a pumiliotoxin 7-hydroxylase.

Until the structural requirements for 7-hydroxylation of PTXs are defined, the importance of this pathway for conversion of other PTXs, such as PTX 307A to aPTX 323B, and of PTX 323A to aPTX 339A, remains uncertain. Both PTX 307A and aPTX 323B have been reported from mixed extracts of leaf-litter arthropods from Panamanian sites, where the dendrobatid frog D. pumilio has both alkaloids in its skin (17). As yet, neither PTX 251D nor aPTX 267A has been detected in arthropod extracts.

In addition to defining the remarkable evolutionary development of a pumiliotoxin 7-hydroxylase in frogs of the genus Dendrobates, the present results provide evidence suggesting that the alkaloid-sequestering systems of frogs is not rigorously conserved across genera. Thus, the DHQ iso-223F was accumulated quite efficiently relative to 251D/267A in D. auratus (Fig. 3 A and B) and E. tricolor (Fig. 3C), compared with Phyllobates bicolor (Fig. 3D). Frogs of the genus Dendrobates have been characterized as “ant specialists” (37–39). Thus, development of a sequestering system with enhanced efficiency toward putative ant alkaloids, such as the DHQ class, might have been advantageous to the “ant specialists.” However, DHQ iso-223F was not efficiently accumulated in the two Amazonian Dendrobates spp. (Figs. 3 E and F).

The mechanism and target underlying the unexpectedly high toxicity and hyperalgesic effects of PTX (+)-251D and aPTX (+)-267A are of great pharmacological import. Such high toxicity does not seem to correlate with the only known mechanism of action of PTXs, namely as positive modulators of sodium channels (40, 41), where PTX 251D and aPTX 267A seemed ineffective and only weakly active, respectively (42–44). It is noteworthy that, in initial experiments with anticonvulsants (45), the toxic effects of PTX (+)-251D were reduced by carbamazepine, an anticonvulsant that does target sodium channels, whereas diazepam, which targets γ-aminobutyric acid type A (GABA_A) receptors, and dizocilpine, which targets N-methyl-d-aspartate (NMDA) receptors, were relatively ineffective. Phenobarbital, which targets both GABA_A receptors and calcium channels, did reduce the toxic effects of PTX (+)-251D. Whatever the target is, it must be highly selective because the unnatural enantiomer PTX (-)-251D, administered at a comparable dose to PTX (+)-251D, had no observable toxic effects in mice. In addition, the side-chain methyl substituent of (+)-251D contributes to toxicity, in particular to the running convulsions, because the desmethyl analog 13 was less toxic and elicited prostration rather than marked running convulsions. The enantiomeric desmethyl analog 14, like unnatural PTX (-)-251D, was relatively inactive. The effects of PTXs/aPTXs on sodium channels and inositol trisphosphate formation have been proposed to account for the cardiotonic effects of such alkaloids (43), but other sites of action with marked stereoselectivity now must be considered to account for the convulsant activity of PTX 251D and aPTX 267A.

Further investigation needs to focus on the following: (i) characterization of the alkaloid-sequestering systems and their possible derivation from primitive transporters of unknown functional significance; (ii) characterization of the unique enantiospecific pumiliotoxin 7-hydroxylase and its relationship to other hydroxylases; (iii) the presence or absence of the hydroxylase in other species of dendrobatid frogs as a defining taxonomic feature indicative of synaptomorphy; (iv) the advantage to frogs of the conversion of PTXs to aPTXs (the increase in toxicity to predators may be the sole advantage to frogs); (v) the biological role of PTXs and aPTXs in arthropods, where both occur; and (vi) the molecular target(s) underlying the toxic effects of PTX 251D and aPTX 267A.