Phylogenetic Survey of Soluble Saxitoxin-Binding Activity in Pursuit of the Function and Molecular Evolution of Saxiphilin, a Relative of Transferrin

Phylogenetic survey of soluble saxitoxin-binding activity in pursuit of the function and molecular evolution of saxiphilin, a relative of transferrin

" # LYNDON E. LLEWELLYN , PETER M. BELL # $  EDWARD G. MOCZYDLOWSKI *, " Australian Institute of Marine Science, PMB 3, ToWnsille MC, Queensland 4810, Australia # * Department of Pharmacolog, Yale Uniersit School of Medicine, 333 Cedar Street, NeW Haen, CT 06520–8066, USA (EdwardIMoczydlowski!Yale.edu) $ Department of Cellular and Molecular Phsiolog, Yale Uniersit School of Medicine, NeW Haen, CT 06520, USA

SUMMARY Saxiphilin is a soluble protein of unknown function which binds the neurotoxin, saxitoxin (STX), with high affinity. Molecular characterization of saxiphilin from the North American bullfrog, Rana catesbeiana, has previously shown that it is a member of the transferrin family. In this study we surveyed various animal species to investigate the phylogenetic distribution of saxiphilin, as detected by the presence of $ soluble [ H]STX binding activity in plasma, haemolymph or tissue extracts. We found that saxiphilin activity is readily detectable in a wide variety of arthropods, fish, amphibians, and reptiles. The $ pharmacological characteristics of [ H]STX binding activity in phylogenetically diverse species indicates that a protein homologous to bullfrog saxiphilin is likely to be constitutively expressed in many ectothermic animals. The results suggest that the saxiphilin gene is evolutionarily as old as an ancestral $ gene encoding bilobed transferrin, an Fe +-binding and transport protein which has been identified in several arthropods and all the vertebrates which have been studied.

site on Na+ channels has been localized to residues 1. INTRODUCTION within a conserved sequence motif in four homologous The neurotoxin, saxitoxin (STX), and a large array of domains of the α-subunit, which forms part of the ion- STX derivatives are produced by certain species of selective pore (Terlau et al. 1991). dinoflagellates in the marine environment and cyano- While the molecular pharmacology of STX with bacteria in the freshwater ecosystem (Carmichael et al. respect to Na+ channels is well characterized, our 1990; Hall et al. 1990; Schantz 1986; Shimizu 1996). group has been attempting to uncover the biological Consumption of toxic phytoplankton by filter-feeding significance of a different high-affinity binding site for organisms results in the accumulation and dispersal of STX, which is located on a soluble protein named STX through the food chain to animals which have saxiphilin. This site was first recognized due to the $ been reported to include an ascidian, annelids, mol- finding of high-affinity binding activity for [ H]STX in luscs, crabs, fish, and ultimately mammals (Anderson tissue extracts and plasma of frogs and toads (Doyle et & White 1992; Geraci et al. 1989; Gessner et al. 1996; al. 1982; Mahar et al. 1991). A component exhibiting $ Llewellyn & Endean 1989; Nagashima et al. 1984; soluble [ H]STX binding activity was purified from Yasumoto et al. 1986). The human intoxication plasma of the bullfrog, Rana catesbeiana, and found to syndrome of paralysis and death resulting from correspond to a 91 kDa protein related to the trans- $ unwitting consumption of STX-contaminated shellfish ferrin family of Fe +-binding proteins (Li & Moczyd- is commonly known as ‘paralytic shellfish poisoning’ lowski 1991). Structural similarity of saxiphilin to (PSP). The neurotoxicity of STX is due to potent transferrin is indicated by a high level of sequence blockade of voltage-sensitive Na+ channels that me- similarity; e.g. 51% identity to Xenopus laeis trans- diate nerve and muscle action potentials (Ritchie & ferrin and 44% identity to human serum transferrin Rogart 1977). STX exerts half-maximal blockage of (Morabito & Moczydlowski 1994). Serum transferrin Na+ current at a concentration of 2–100 nM of STX, and lactoferrin have a bilobed structure owing to the depending on the particular Na+ channel isoform (Guo presence of two internally homologous domains of et al. 1987). At the molecular level, the STX binding ca. 340 residues, the N-lobe and the C-lobe, that each $+ −#! contain a high affinity site for Fe (KD ca.10 M) − and the synergistic anion cofactor, HCO$ (Baker & * Author and address for correspondence. Lindley 1992). Bullfrog saxiphilin has the same internal

Proc. R. Soc. Lond. B (1997) 264, 891–902 891 " 1997 The Royal Society Printed in Great Britain 892 L. E. Llewellyn and others Phlogenetic sure of saxiphilin-like actiit duplication as the transferrins, but has substitutions in toxication from microbial sources, it ought to be $ nine out of ten highly conserved Fe +-site residues, commonly expressed by animal species inhabiting $ thereby accounting for its lack of demonstrable Fe +- ecosystems known to harbour STX-producing phyto- binding activity (Li et al. 1993; Morabito & Moczyd- plankton. Unexpectedly, we found putative saxiphilin- lowski 1994). The single high-affinity binding site for like activity in terrestrial and aquatic arthropods and STX (KD ca. 0.2 nM) in bullfrog saxiphilin has been in a wide variety of ectothermic vertebrates, including localized to the C-lobe, as determined by assay of a fish and amphibians from the aquatic environment as recombinant form of the protein in which the N-lobe well as reptiles indigenous to semi-arid locales. The $ has been deleted (Morabito et al. 1995). [ H]STX data imply that saxiphilin has an ancient origin in binding to saxiphilin is also inhibited at low pH in a animal evolution, and that it may function in a manner reminiscent of the pH-dependent release of process(es) of broadly-based biological significance. $ Fe + by transferrin (Llewellyn & Moczydlowski 1994), a process which is important in the delivery of iron to 2. METHODS eukaryotic cells by transferrin receptor-mediated endo- cytosis (Thorstensen & Romslo 1990). (a) Materials $ Aside from the recognized role of serum transferrin [ H]STX was purchased from Amersham International, in iron transport, transferrin and lactoferrin are also purified, and standardized according to Moczydlowski et al. $+ $ responsible for maintaining low levels of free Fe in (1988). Several different lots of [ H]STX used in this $+ biological fluids, which inhibits the growth of Fe - study had specific activities in the range of 20200– " requiring microorganisms and protects against the 35100 cpm pmol− . STX was purchased from Calbiochem # $ potential toxicity of Fe +\ Fe + in the generation of (La Jolla, CA). The following STX derivatives were hydroxyl free radicals (Crichton 1991). By analogy to generously provided by Dr Sherwood Hall (US Food and this latter chemical defence function, it may be Drug Administration): decarbamoylsaxitoxin (dcSTX), hypothesized that saxiphilin functions as a defence neosaxitoxin (neoSTX), B1 and C1 (see figure 2a for structures). Stock solutions of these toxins were diluted in mechanism against STX intoxication, by sequestering 1 mM citrate buffer, pH 5.0. The common buffers, Mops, any STX that an animal might acquire from microbial Mes, Hepes, and Tris, and the anaesthetics, tricaine sources. Although comparatively little is known about methanesulphonate and sodium brevital, came from Sigma the chemical ecology of STX in the freshwater (St Louis, MO). The cation exchange resin, AG50W-X2, H+ environment, an argument for this hypothesis can be form, 100–200 mesh, was obtained from Bio-Rad (Richmond, drawn from previous observations of tadpole mortality CA). Other chemicals were reagent grade, and came from associated with STX production by the cyanobacterial commercial sources. species, Aphaniomenon flos-aquae (Ikawa et al. 1982), and the occurrence of saxiphilin in Ranid tadpoles and (b) Sources of animals and sample preparation frogs (Mahar et al. 1991). Another question concerns the molecular evolution Some species used in this study were purchased from of saxiphilin. Transferrins have thus far been identified Connecticut Valley Biological (Southampton, MA), Caro- in all classes of vertebrates, several insect species and an lina Biological (Burlington, NC), and Charles Sullivan Co. (Nashville, TN). Various animals were collected in the ascidian (Bartfeld & Law 1990; Jamroz et al. 1993; vicinity of Mount Desert Island Biological Laboratory Kurama et al. 1995; Martin et al. 1984; Welch 1990). (Salisbury Cove, ME) and the Australian Institute of Marine Sequence data imply that the gene duplication Science (Townsville, ueensland, Australia). Numerous underlying the bilobed structure of modern transferrin professional colleagues, listed in the acknowledgements, occurred before the emergence of insects (Bowman et generously donated plasma and tissue samples from animals al. 1988; Bartfeld & Law 1990). Since saxiphilin used in their own studies. Plasma samples from several species contains this same internal duplication, a transferrin of sharks, dolphins and whales were obtained from the New gene may have been the direct ancestral precursor of England Aquarium (Boston, MA) through the assistance of the saxiphilin gene. Evidence has recently emerged Dr Don Anderson at the Woods Hole Oceanographic $ that saxiphilin is not the only example of a non-Fe +- Institute (Woods Hole, MA). Lyophilized plasma from terrestrial mammals, birds, and the Thailand cobra (N. n. binding member of the transferrin family. Fierke and kaouthia), liver extract of the African lungfish (P. aethiopicus) her co-workers have identified a protein inhibitor of and horseshoe crab (L. polphemus) haemolymph were carbonic anhydrase (PICA) in the pig which is purchased from Sigma. homologous to transferrin, but has substitutions of Live animals were handled humanely according to the $+ several key Fe -site residues (Roush & Fierke 1992; guidelines of the Yale University Animal Care and Use Wuebbens et al. 1994). The examples of saxiphilin and Committee. Live amphibians and reptiles studied in the PICA suggest that transferrin-like proteins may reflect laboratory were anaesthetized with tricaine methanesulphonate or sodium brevital and exsanguinated via the a family of genes with diverse biological functions. −" To pursue the function and evolution of saxiphilin, aortic arch with a syringe containing 100 µl of 0.1 mg ml we conducted a phylogenetic survey of soluble heparin sulphate. Whole blood was centrifuged in an $ Eppendorf microfuge; the plasma was removed and stored [ H]STX binding activity. The known pharmaco- frozen at k80 mC for later assay. In our experience, $ logical characteristics of this activity in the bullfrog saxiphilin-like [ H]STX binding activity in whole plasma is provide a unique set of molecular criteria to identify very stable over the course of a year when stored frozen or the presence of saxiphilin-related proteins in other lyophilized. This is in contrast to dilute solutions of purified species. We hypothesized that if saxiphilin functions bullfrog saxiphilin which are labile to repeated freeze– primarily in a defensive capacity against STX in- thawing (Llewellyn & Moczydlowski 1994). Animals which

Proc. R. Soc. Lond. B (1997) Phlogenetic sure of saxiphilin-like actiit L. E. Llewellyn and others 893 were too small for convenient collection of blood or 3. RESULTS haemolymph were anaesthetized by hypothermia and processed for tissue extraction. Whole animal extracts were From previous biochemical characterization in the prepared by homogenization on ice with a Tissumizer bullfrog, saxiphilin may be recognized by soluble high- $ homogenizer (Tekmar, Cincinnati, OH), using two 10 s affinity binding of [ H]STX, which is specifically bursts at 8000 rpm followed by two 10 s bursts at 24000 rpm displaced by excess unlabelled STX (Mahar et al. in a buffer consisting of 10 mM of Mops-NaOH, pH 7.4, 1991; Llewellyn & Moczydlowski 1994). Binding of 0.3 M of sucrose, 5 mM of EDTA, 1 µM of pepstatin, 1 µM $ [ H]STX to saxiphilin is readily discriminated from of aprotinin, and 100 µM of phenylmethylsulphonyl fluoride. + The homogenate was centrifuged at 100000 g for 1 h to binding to voltage-sensitive Na channels by its ensure complete removal of solid debris and particulate solubility in the absence of detergents, its presence in membranes that might contain STX binding sites associated plasma and extracts of non-electrically excitable + with Na channels. The supernatant was decanted for assay tissues, the lack of binding competition by 100 µMof and stored frozen at k80 mC. tetrodotoxin, and a distinctively slow time-course of dissociation. Thus, to screen various animal species for $ (c) Measurement of [3H]STX binding expression of saxiphilin, we assayed soluble [ H]STX $ binding in plasma or haemolymph samples where All measurements of [ H]STX binding were carried out at possible, or in whole tissue extracts of small vertebrates ca.0mC by incubation of assay solutions on ice. In the $ standard survey for soluble [ H]STX binding, 50–100 µl and invertebrates. The later extracts were subjected to aliquots of plasma samples or extracts were added to a ultracentrifugation to eliminate the possibility of + solution with a final concentration of 20 mM of Mops- contamination by membrane-bound Na channels. NaOH, pH 7.4, 200 mM of NaCl, 0.1 mM of EDTA, and Species selected for this study included representatives $ approximately 5 nM of [ H]STX in a volume of 250 µl, and of all major vertebrate classes and invertebrate phyla. incubated for at least 1 h. Control samples for the deter- The survey sample consisted of animals indigenous to mination of non-specific binding also contained 10 µMof five continents (N. America, S. America, Europe, Asia, STX. Duplicate aliquots of 100 µl were processed for $ and Australia) with diverse habitats, including species separation of bound and free [ H]STX on small columns of that have been previously documented to acquire STX AG50W-X2 cation exchange resin, and quantitated by from blooms of toxic dinoflagellates. liquid scintillation counting, as previously described Table 1 summarizes the effective concentration and (Llewellyn & Moczydlowski 1994). −" $ Assay of saturable binding behaviour for selected species specific activity (mg protein ) of soluble [ H]STX was similarly performed by varying the concentration of binding sites in samples from species which tested $ [ H]STX in the assay from 0.1–26 nM. Data from such positive for saxiphilin. The level of detectable activity experiments were fitted to a one-site binding model using the ranged from 0.6 nM effective concentration (pmol sites EBDA and LIGAND equilibrium binding analysis programs per ml of plasma or g of tissue extract) in a cockroach from Biosoft (Cambridge, UK). Structure–activity relation- (Periplaneta americana)toca. 5000 nM in the plasma of ships for STX derivatives were investigated by competitive $ one individual specimen of the wood frog (Rana binding titrations in which [ H]STX was held constant at slatica). Species which tested positive include animals 4.4 nM and the concentration of various unlabelled STX belonging to the major classes of ectothermic verte- derivatives was varied. Data from such experiments were nl nl nl brates and arthropods. The list of positive vertebrates fitted to the Hill equation: f l K! &\ ([toxin] jK! &), $ . . where f is the ratio of [ H]STX bound in the presence of includes teleost fish, amphibians (frogs, toads, sala- competitor toxin compared to that without competitor toxin, manders and newts) and reptiles (lizards and snakes). K!.& is the competitor concentration at 50% inhibition, and For the arthropods, the low levels of activity measured nl is a pseudo-Hill coefficient. To perform a nonlinear least in the cockroach, Periplaneta, and the water-flea, squares fit we used the curve fitting utility of Sigmaplot Daphnia, might be questionable since these measure- (Jandel, San Rafael, CA). Error estimates for K!.& given in ments approached the lower limit of detection $ table 3 were obtained from the Sigmaplot fitting routine. (ca. 0.1 nM [ H]STX binding sites). However, the Dependence of binding on pH was studied by a similar $ finding of robust activity in an Australian centipede assay using 4.4 nM of [ H]STX in buffers ranging from (Ethmostigmus rubripes), a North American isopod pH 4–9. The pH was buffered with 20 mM of Tris, 10 mM (sowbug, Oniscus sp.), a North American orb-weaving of Mes, and 10 mM of acetic acid adjusted with tetra- methylammonium hydroxide or HCl at nearly constant spider (Araneus c. f. caaticus), and four different species ionic strength, as described by Ellis & Morrison (1982). H+ of xanthid crabs (Llewellyn 1997), firmly establish the titration data were analysed by fitting to the following occurrence of saxiphilin-like activity in arthropods. function (similar to the one used for toxin competition): f l Despite a concerted attempt (see table 2), we did not + nl + nl + nl + [H ]!.& \ ([H ] j[H ]!.&), where [H ]!.& is the proton find many examples of insects containing saxiphilin concentration at 50% inhibition. $ activity. However, in addition to the positive cockroach Dissociation kinetics of [ H]STX were followed by pre- species listed in table 1, we did observe robust activity $ " equilibration of samples with [ H]STX and assays of bound (5.2 pmol g− ) in an extract of damselfly nymphs radioligand at various times after addition of 10 µM of STX. obtained from Connecticut Valley Biological (South- Similarly, association kinetics under pseudo-first order ampton, MA). Although we have not been able to conditions were followed by assays of bound radioligand at $ various times after addition of 4.4 or 8.8 nM of [ H]STX. obtain a precise taxonomic identification, such nymphs Kinetic data were fitted either to a ‘single exponential’ are the aquatic larval form of insects belonging to the function of time or to a ‘sum-of-two exponentials’ function of family Calopterygidae. time as described in the text. Protein assays were performed Table 2 is a partial listing of species that were tested according to the method of Cabib & Polachek (1984). by the same method and found not to exhibit any

Proc. R. Soc. Lond. B (1997) 894 L. E. Llewellyn and others Phlogenetic sure of saxiphilin-like actiit $ Table 1. [ H]STX binding in species found to exhibit saxiphilin-like actiit $ $ (Data are the results of a standard assay for binding of 5 nM [ H]STX as described in §2. Effective concentration of [ H]STX binding sites per ml of plasma or g of tissue is reported as the meanps.d. (n) where n is the number of individuals or determinations. Abbreviations: P, plasma; H, haemolymph; E, extract; F, freshwater species; M, marine species; n.d., not determined.) $ [ H]STX binding sites " " species (common name and geographical origin) pmol ml− plasma or g tissue pmol mg− protein teleost fish (except lungfish) Hpostomus plecostomus (catfish, S. America) E, F 20p3 (4) 6.7 Poecilia reticulata (guppy, Americas) E, F 72p9 (4) 15.8 Anguilla rostrata (eel, N. America) E, M, F 2.5p0.8 (4) n.d. Gambusia affinis (mosquito fish, N. America) E, F 72p8 (4) 3.2 Pomacentrus sp. (damsel fish, N. America) E, M 1.9p0.1 (2) 0.1 Apogon sp. (cardinal fish, N. America) E, M 29p1.3 (3) 1.0 Danio rerio (zebra fish, Asia) E, F 6.3p0.3 (3) 0.3 Protopterus aethiopicus (lungfish, Africa) E, F 0.5 (1) 0.02 amphibians Notopthalamus iridescens (Eastern newt, N. America) E 1.5p0.5 (11) 0.2 Ambstoma tigrinum (tiger salamander, N. America) P 76p2 (3) 3.3 Rana slatica (wood frog, N. America) P 1590p440 (18) 68 Rana temporaria (grass frog, Europe) P 669p18 (3) 34 Bufo marinus (cane toad, S. America) P 49p5 (8) 1.9 reptiles Varanus rosenbergii (goanna monitor lizard, Australia) P 326p25 (4) 5.0 Sceloporus poinsetti (crevice spiny lizard, N. America) P 1100p109 (6) 491 Naja naja kaouthia (Thailand cobra, Asia) P 223p20 (3) 7.6 Crotalus iridus iridus (rattlesnake, N. America) P 2.4p0.5 (3) 0.1 Thamnophis ordinoides (garter snake, N. America) P 133p89 (4) 10.7 Thamnophis sirtalis (garter snake, N. America) P 486p2322 (6) 19.8 arthropods Daphnia sp. (water-flea, N. America) E 0.8p0.2 (2) n.d. Oniscus sp. (sowbug, N. America) E 8.2p0.5 (8) n.d. Ethmostigmus rubripes (centipede, Australia) H 79p18 (8) 2.2 Araneus c. f. caaticus (orb web spider, N. America) E 2.9p0.7 (7) n.d. Periplaneta americana (cockroach, N. America) E 0.6p0.2 (3) n.d. Lophoomus pictor (xanthid crab, Australia) H 111p12 (2) 21 Liomera tristis (xanthid crab, Australia) H 97p9 (2) 24 Chlorodiella nigra (xanthid crab, Australia) H 24p5 (1) 6.3 Actaeodes tomentosus (xanthid crab, Australia) H 81p22 (2) 37 detectable saxiphilin activity. We did not find evidence Brachiopoda, (Lingula sp.). Of particular interest of saxiphilin-like activity in endothermic vertebrates, among negative species are the bivalve molluscs, as represented by various birds or mammals, including Saxidomus giganteus and Mtilus edulis, which are known dolphins and whales. Negative-testing reptiles included to accumulate STX from toxic marine dinoflagellates various turtles, crocodilians, a tiger snake, and the (Shimizu et al. 1978; Schantz 1986), but they do not primitive tuatara of New Zealand. The common appear to produce saxiphilin (table 2). $ laboratory species, Xenopus laeis (African clawed frog), To test the presumption that soluble [ H]STX is the only amphibian tested to date that does not binding activity found across this diverse phylogenetic appear to express saxiphilin. Negative vertebrate spectrum has properties similar to that of the previously marine species also include several fish, four sharks, the characterized saxiphilin from Rana catesbeiana, samples evolutionarily primitive coelocanth, and a lamprey. from the following representative species were selected Except for the arthropods, we did not detect for further characterization: plasma of Bufo marinus saxiphilin activity in other invertebrate phyla. As (marine or cane toad), plasma of Thamnophis sirtalis shown in table 2, the negative phyla include ascidians, (garter snake), lyophilized plasma of Naja naja kaouthia echinoderms, annelids and molluscs. In addition, the (Thailand cobra), an extract of whole Gambusia affinis following invertebrate species are not shown in table 2, (mosquito fish), and haemolymph of Ethmostigmus but they also tested negative: phylum Porifera (the rubripes (an Australian centipede). Appropriate dilu- sponges, Xestospongia exigua, Lanthella basta, Jaspis tions of these samples were titrated with various $ stellifera, Pericherax heteroraphis, Rhopaloides odorabile), concentrations of [ H]STX in the range of 0.1–26 nM, phylum Platyhelminthes (the flatworm, Pseudoceros sp.), and assayed for binding in the absence and presence of phylum Cnidaria (soft corals, Sarcophton elegans, Sinula- excess unlabelled STX (10 µM) to assess total and non- ria, dura; sea anemone, Actinia australis), and phylum specific binding, respectively. Samples from all five of

Proc. R. Soc. Lond. B (1997) Phlogenetic sure of saxiphilin-like actiit L. E. Llewellyn and others 895

Table 2. Species that tested negatie for saxiphilin " $ (This table lists species that did not contain a detectable amount (" 0.1 pmol ml− sample) of [ H]STX binding activity in plasma (P), haemolymph (H), or soluble extracts (E). Several additional negative species are given in §3.) phylum Vertebrata (super-class Agnatha) Asterius forbesii (starfish) E Mordacia mordax (lamprey) E Asterius ulgaris (starfish) E phylum Vertebrata (super-class Gnathostomes) Solaster endeca (starfish) E Squalus acanthius (dogfish shark) P Linckia laeigata (starfish) H Ginglmostoma cirratum (nurse shark) P Cucumaria frondosa (sea cucumber) E Negaprion breirostrio (lemon shark) P Echinarachnius parma (sand dollar) E Somniosus microcephalus (Greenland shark) P Holothuria atra (holothurian) H Raja erinacea (little skate) P Stichopus chloronatus (holothurian) H Scomber scombrus (Atlantic mackerel) P phylum Annelida Cclopterus lumpus (lumpfish) P Gastrolepida claigea (annelid) E Pseudopleuronectes americanus (flounder) P Eisenia fetida (tiger worm) E Makaira indica (marlin) P Eudrilus eugenia (nightcrawler) E Arothron manilensis (puffer fish) E Reteterebella queenslandia (polychaete) E Latimeria chalumnae (coelocanth) P Spirobranchus giganteus (polychaete) E phylum Vertebrata (class Amphibia) phylum Arthropoda Xenopus laeis (African clawed frog) P Limulus polphemus (horseshoe crab) H phylum Vertebrata (class Reptilia) Penaeus monodon (tiger prawn) H Elsea dentata (snapping tortoise, Australia) P Cancer borealis (Jonah crab) H Dermochels coriacea (leatherback turtle) P Pagurus sp. (hermit crab) H Pseudems scripta (red ear turtle, N. America) P Homarus americanus (lobster) H Caretta caretta (loggerhead turtle) E Ocpode corimana (ghost crab) Sphenodon punctata (tuatara, New Zealand) P Artemia salina (brine shrimp) E Notechis scutatus (Australian tiger snake) P Uca omeris (fiddler crab) H Alligator mississippiensis (alligator, N. America) P Callianassa australiensis (marine yabby) H Crocodlus porosus (crocodile, Australia) E Sclla serrata (mud crab) H phylum Vertebrata (class Aves) Manduca sexta (tobacco hornworm moth) H Gallus gallus (chicken) P Periplenata australiensis (cockroach) E Anser anser (domestic goose) P Blaberus sp. (cockroach) H, E Anas sp. (duck) P Drosophila melanogaster (fruit fly) E Columba lies (domestic pigeon) P Cicada sp. (cicada) E Meleagris gallopao (common turkey) P Oecophlla smaragdine (green ant) E phylum Vertebrata (class Mammalia) Camponotus pennslanicus (carpenter ant) E Equus caballus (horse) P Oncopeltus fasciatus (milkweed bug) E Orctolagus sp. (rabbit) P Acheta domestica (cricket) E Ois aries (domestic sheep) P Hippodamia conergens (lady beetle) H Homo sapiens (human) P Reticulitermes flaipes (termite) E Rattus noregicus (rat) P Tenebrio molitor (mealworm larvae) E Bos taurus (cow) P Halophora cecropia (silkmoth) E Sus scrofa (pig) P Papilio polxenes (black swallowail butterfly) E Lagenorhnchus acutus (white-sided dolphin) P Dermestes sp. (dermestid beetle) E Globicephala malaena (pilot whale) P phylum Mollusca Delphinus delphis (common dolphin) P Mtilus edulis (blue mussel) H, E Balaenoptera acutorostrata (minke whale) P Saxidomus giganteus (butter clam) H, E phylum Chordata (subphylum Ascidiacea) Aplsia californica (sea hare) H, E Polcarpa sp. (ascidian) E Spisula solidissma (clam) E Didemnum molli (ascidian) E Acmaea c. f. testudinalis (limpet) E Polcarpa aurata (ascidian) E Littorina litorea (periwinkle) E Tunica mogula (tunicate) E Donax deltoides (bivalve) E phylum Echinodermata Vepricardium multispinosum (bivalve) E stronglocentrotus droebachiensis (sea urchin) E Sepia plangon (squid) E Acanthaster planci (crown-of-thorns starfish) E these species exhibited a component of saturable, high- range where the accuracy of equilibrium analysis is affinity binding, and a linear component of non-specific limited by the ability to accurately determine ex- $ $ [ H]STX binding (figure 1). The binding titration tremely low concentrations of free [ H]STX ligands. data were analysed with the EBDA and LIGAND However, the values indicate an equivalent or higher programs to derive an apparent equilibrium dissoci- affinity than that previously reported for purified $ ation constant (KD) for [ H]STX, based on the native (KD l 350p20 pM) or recombinant (KD l assumption of a single class of binding sites. The KD 220p10 pM) saxiphilin from Rana catesbeiana. values estimated by this approach were: 210p29 pM Although numerous organic molecules have been (B. marinus), 15p6pM(N. n. kaouthia), 240p60 pM tested, derivatives of STX are the only class of (T. sirtalis), 39p7pM(G.affinis), and 10p6pM(E. compounds that we have ever observed to com- $ rubripes). These KD values, all less than 1 nM, fall in a petitively displace [ H]STX binding to bullfrog saxi-

Proc. R. Soc. Lond. B (1997) 896 L. E. Llewellyn and others Phlogenetic sure of saxiphilin-like actiit

(a) (a)

(b)

(c)

(d)

(d) (e)

Figure 1. Saxiphilin-like activity in plasma, haemolymph or tissue extract of various species as demonstrated by titration $ $ of [ H]STX binding. A fixed amount of soluble protein from Figure 2. Competitive inhibition of [ H]STX binding by various animal species was assayed for binding at the unlabelled STX and four different STX derivatives. (a) $ indicated concentrations of [ H]STX as described in §2. Chemical structure of naturally occurring STX derivatives Plasma samples and extracts were pre-equilibrated with used in this study: STX (#), R" l CONH#,R#lH, R$ l $ [ H]STX for 1–2 h before binding assay, except for E. rubripes H; dcSTX (4), R" l H, R# l H, R$ l H; neoSTX (=), R" l CONH#,R#lOH, R$ l H; B1 (W), R" l haemolymph which was pre-equilibrated for 8 h. Total − − CONHSO$,R#lH, R$ l H; C1 (*), R" l CONHSO$, binding in cpm (#). Non-specific binding in the presence of − R# l H, R$ l OSO$.(b)Naja naja kaouthia plasma, 40 µg 10 µM of STX ($). Data points are the means of duplicate " " protein ml− .(c)Gambusia affinis extract, 180 µg protein ml− . determinations. Non-specific binding was fitted by a simple −" linear regression and total binding was fitted to a one-site (d) Ethmostigmus rubripes haemolymph 180 µg protein ml . model with K values given in the text. (a) Bufo marinus (cane Data points are the mean of duplicate determinations. Solid D " toad) plasma, 72 µg protein ml− .(b)Naja naja kaouthia curves are fits of the data to a Hill function given in §2. Best- " (Thailand cobra) plasma, 74 µg protein ml− .(c)Thamnophis fit values of K!.& for the various STX derivatives are listed in " sirtalis (garter snake) plasma, 103 µg protein ml− .(d) table 3. −" Gambusia affinis (mosquito fish) extract, 180 µg protein ml . $ (e) Ethmostigmus rubripes (centipede) haemolymph, 360 µg displacement of specific [ H]STX binding was very " $ protein ml− . close to the concentration of free [ H]STX in the assay (ca. 4 nM), in accordance with the relationship of philin. To investigate whether the soluble STX- Cheng & Prusoff (1973) for one-site binding. With binding sites of the five test species share a common respect to displacement by the four STX derivatives, structure–activity relationship, a competition displace- the highest and lowest affinities were observed for ment assay was performed for four STX derivatives: dcSTX and C1, respectively. For the snake, fish, and neoSTX, dcSTX, B1, and Cl (figure 2a). As illustrated centipede species, neoSTX exhibited higher affinity in figure 2, samples from all five test species exhibited than B1 (table 3). However, for the toad, B1 exhibited monotonic displacement titration curves with Hill slightly higher affinity than neoSTX. These results are coefficients close to 1.0, characteristic of a single site or in contrast to saxiphilin activity from Ranid frogs, a homogeneous class of STX binding sites. In each where neoSTX has a 550-fold lower affinity than STX case, the concentration of free STX required for 50% (table 3). It thus appears that the low intrinsic affinity

Proc. R. Soc. Lond. B (1997) Phlogenetic sure of saxiphilin-like actiit L. E. Llewellyn and others 897 $ Table 3. Saxitoxin structure–actiit relationships and pH dependence of [ H]STX binding for arious species $ + ([ H]STX binding to plasma or extracts was titrated with unlabelled STX derivatives or H as in figures 2 and 3. IC&! values l for STX derivatives, pH!.& values for 50% inhibition and pseudo-Hill (n ) coefficients were determined by fitting data to a Hill function given in §2. Values in parentheses are ratios relative to STX for each species. Data for Rana catesbeiana are taken from Mahar et al. (1991) and Llewellyn & Moczydlowski (1994).)

IC&! (nM) pH dependence l species STX dcSTX NEO B1 C1 pH!.& n Rana catesbeiana 4.7p0.7 (1.0) 11.1p1.2 (2.4) 2640p180 (560) 4.5p0.4 (1.0) 730p160 (155) 5.7 1.0 Bufo marinus 5.3p0.9 (1.0) 10.8p1.6 (2.0) 32.3p2.4 (6.1) 19.0p2.1 (3.6) 7540p620 (1420) 4.4 1.1 Thamnophis sirtalis 4.0p2.1 (1.0) 1.4p0.2 (0.4) 6.0p1.7 (1.5) 253p25 (63) 3480p190 (870) 5.5 1.4 Naja naja kaouthia 4.0p0.4 (1.0) 1.7p0.3 (0.4) 20.8p1.7 (5.2) 118p5 (30) 7540p460 (1890) 5.4 2.1 Gambusia aﬃnis 5.0p0.3 (1.0) 7.7p0.7 (1.5) 11.8p0.8 (2.4) 179p10 (36) 63000p2500 (12600) 4.7 2.0 Ethmostigmus rubripes 5.0p0.3 (1.0) 6.1p0.9 (1.2) 8.5p0.4 (1.7) 20.3p1.2 (4.1) 2830p120 (566) 5.4 3.2

chemical property of saxiphilin activity is conserved in (a) $ other species, equilibrium binding of [ H]STX was measured as a function of pH for the five test species (figure 3). The pH titration curves of the two snake species, N. n. kaouthia and T. sirtalis, exhibited a similar pH dependence as that of Rana catesbeiana with half- maximal inhibition occurring near pH 5.5 (table 3). However, the Hill coefficient (nl) derived for H+ was (b) higher than that previously observed for the bullfrog protein (nl l 1.0), with n l 1.4 and 2.1 for T. sirtalis $ and N. kaouthia, respectively. [ H]STX binding activity from B. marinus and G. affinis was less sensitive to H+ inhibition. Complete titration curves for these latter two species could not be obtained due to the known limitations of the cation exchange column assay at pH values less than 4.5. The available data suggest pH!.& values of approximately 4.4 and 4.7 for B. $ Figure 3. pH dependence of [ H]STX binding. Panel (a) marinus and G. affinis, respectively. The putative " shows results for Bufo marinus plasma ($,72µg protein ml− ), saxiphilin activity of the centipede species E. rubripes " Naja naja kaouthia plasma (W,74µg protein ml− ), and exhibited a pH!.& of 5.4 and a Hill coefficient of ca. 3.2, " Gambusia affinis extract (=, 180 µg protein ml− ). Panel (b) corresponding to a steeply sensitive inhibition with + shows results for Thamnophis sirtalis plasma (4,41µg protein respect to [H ] that may indicate more than one site of " ml− ) and Ethmostigmus rubripes plasma (#, 360 µg protein H+ modulation. −" ml ). Data points and error bars are the meanpSEM of The kinetics of dissociation and association of $ three experiments. Solid and dashed curves represent fits of [ H]STX were also investigated. Figure 4 shows the data to a Hill function of [H+] given in §2. Best-fit values l representative data collected in these experiments for of pH!.& and n are listed in table 3. B. marinus, N. n. kaouthia, G. affinis and E. rubripes. In the case of the toad and the centipede, both the dissociation for neoSTX in bullfrog saxiphilin is an exception and association kinetics were well described by a rather than the rule, and that saxiphilin from many simple exponential time-course. The derived dissoci- species binds neoSTX nearly as well as STX. For ation rate constant for B. marinus from the experiment $ −& −" [ H]STX binding activity examined in this study, in (figure 4a) at pH 7.4 was k l 1.1p0.1i10 s . $ off no case did we observe competitive displacement by The rates of [ H]STX dissociation measured for plasma 100 µM tetrodotoxin, which is an Na+ channel blocker samples from two other specimens of B. marinus were −& −" −& −" with a very different structure from STX (Ritichie & 1.2p0.1i10 s and 1.3p0.2i10 s , illustrating Rogart 1977). This emphasizes the unambiguous the consistency of these kinetic determinations from pharmacological discrimination of saxiphilin from individual to individual. The association rate of the B. STX-binding sites of voltage-sensitive Na+ channels, marinus sample was measured in the presence of 4.4 nM $ which are well known to exhibit competitive binding of [ H]STX under near pseudo-first order conditions, between STX and tetrodotoxin. where the concentration of ligand was about seven-fold Previous studies showed that the pH dependence of greater than the number of total sites. Under these $ [ H]STX binding to bullfrog saxiphilin resembles the conditions, the time-course was so rapid that the early $ pH dependence of Fe + binding to serum transferrin portion could not be resolved with the present (Llewellyn & Moczydlowski 1994). We previously methodology. However, the data can be used to suggested that this coincidence may reflect a common calculate a lower limit for the bimolecular association ' −" −" structural–functional relationship or a shared aspect of rate constant of kon & 9.8i10 s M . The ratio of cellular physiology. To investigate whether this bio- koff to kon provides an upper limit estimate for the

Proc. R. Soc. Lond. B (1997) 898 L. E. Llewellyn and others Phlogenetic sure of saxiphilin-like actiit

(a)(e)

(b)

) ( f ) Ð2 10 ×

(g)

(d) (h)

$ Figure 4. Time-course of dissociation and association for [ H]STX binding compared for different species. (a,e) Bufo marinus plasma. (b,f) Naja naja kaouthia plasma. (c,g) Gambusia affinis extract. (d,h) Ethmostigmus rubripes haemolymph. $ Dissociation (a,b,c,d) or association (e,f,g,h) time-course of specific [ H]STX binding was measured as described in §2. Data were obtained at pH 7.4 except for a, b and c, where dissociation at pH 7.4 (4) is compared to that at various acidic pH values (#) as follows: (a) pH 4.3, (b) pH 5.3, and (c) pH 4.5. Data points are fitted (solid lines) to a single exponential function (a,d,e,f,h) or to a sum-of-two exponentials (b,c,g) as described in the text. equilibrium KD for B. marinus of 1.1 pM, which is an process comprised 77–92% of the decay for these three order of magnitude lower than that estimated (15 pM) species, corresponding to the following koff rate −' −" from the equilibrium titration data of figure 1. As constants: 9.8p1.5i10 s for G. affinis, 1.1p0.1i −% −" −& −" noted above, equilibrium KD measurements in the 10 s for N. n. kaouthia, and 1.3p0.1i10 s for range of 1 pM by standard Scatchard analysis with the T. sirtalis. These latter slow components correspond $ available specific activity of [ H]STX are technically to dissociation halftimes of 19.6, 1.7 and 14.8 h, re- unattainable. Thus, the KD values estimated here by spectively, all slower than the halftime of 1.3 h the kinetic approach are more likely to be closer to the previously measured for bullfrog saxiphilin (Llewellyn true value. & Moczydlowski 1994). The association time-course The corresponding rate constants for E. rubripes were for the two snake species conformed well to a single −' −" ' −" koff l 4.3p0.3i10 s and kon l 5.2p0.7i10 s exponential process, but the fish sample was better −" M , giving an estimated KD of 0.8 pM. Figure 4h described by two components (figure 4g). For the shows that the time-course of association for the cobra, N. n. kaouthia, and the fish, G. affinis, the centipede sample was fairly well resolved, being slower association time-course was faster than a system ( " " than that of the other species in figure 4. The governed by a bimolecular k of 10 s− M− . Curiously, $ on dissociation rate of [ H]STX from the centipede the association rate observed for the garter snake, T. haemolymph sample corresponds to a halftime of 1.9 sirtalis, was very well resolved since it was slower than days, which is extraordinarily slow for the dissociation that of any of the other species (data not shown), & −" −" of a small organic molecule from a protein acceptor corresponding to a kon of 3.2p0.2i10 s M .Ifwe site. use a one-site equilibrium to approximate the be- The kinetic behaviour of the fish and snake samples haviour of these systems, and consider only the major was more complex than that of the toad and centipede. slow component of the dissociation time-course, then $ For these species, the time-course of [ H]STX the ratios of koff to kon for the two snake species and the dissociation was better fitted by a sum-of-two exponen- fish species yield the following effective KD estimates: tials rather than one exponential (figure 4). However, 6 pM, N. n. kaouthia; 40 pM, T. sirtalis; and 0.6 pM, a major slow kinetic component of the dissociation G. affinis. Surprisingly, the approximate KD values for

Proc. R. Soc. Lond. B (1997) Phlogenetic sure of saxiphilin-like actiit L. E. Llewellyn and others 899 the five test species investigated most thoroughly in this 1993; Kurama et al. 1995). The occurrence of saxiphilin study are all substantially lower than the KD of ca. in arthropods is consistent with the possibility that it 200 pM previously measured for bullfrog saxiphilin. arose directly from an ancestral bilobed transferrin With respect to the pH dependence discussed above, precursor and later acquired an insertion of 144 STX binding to bullfrog saxiphilin has also been residues that is present in bullfrog saxiphilin (Morabito shown to undergo allosteric modulation by H+ as & Moczydlowski 1994; Morabito et al. 1995). Alterna- manifested by an increased rate of ligand dissociation tively, it is possible that bilobed transferrin and at low pH (Llewellyn & Moczydlowski 1994). To saxiphilin both arose independently by gene dupli- investigate whether this functional property is con- cation from a single-lobed, ca. 40 kDa transferrin served, we have also measured the dissociation time- precursor, such as that previously described in the course at pH 4.3–4.5 for three species. As shown in urochordate ascidian, Pura stolonifera (Martin et al. figure 4a–c, samples from B. marinus, N. n. kaouthia, 1984). and G. affinis all exhibited a substantially faster The most striking finding of our characterization of $ dissociation time-course at the lower pH. This effect [ H]STX binding activity from evolutionarily diverse is equivalent to a 3.6-fold, 5.6-fold and 9.1-fold en- species is the extraordinarily high affinity of the soluble hancement of the slowest component of the dissociation STX binding site in some animals. While K ’s in the "# D reaction for G. affinis, B. marinus and N. n. kaouthia, range of 10− M are not uncommon for bioactive respectively. peptides, binding affinity in this range is relatively rare for small organic molecules; e.g. one of the strongest known interactions is that of biotin-avidin, K ca. "& D 4. DISCUSSION 10− M (Gitlin et al. 1987). The binding energy for such protein–ligand interactions is generally derived (a) Saxiphilin has an ancient origin in animal from multiple non-covalent interactions between the evolution functional groups of protein residues and atoms of $ The properties of soluble [ H]STX binding activity ligand substituent groups. The STX molecule, with its characterized for the cane toad (B. marinus), garter divalent positive charge and six guanidino nitrogen snake (T. sirtalis), Thailand cobra (N. n. kaouthia), atoms, offers a rigid scaffold with the potential ability mosquito fish (G. affinis), and centipede (E. rubripes), to form multiple electrostatic and hydrogen-bonding leave little doubt that all five of these species contain a interactions within a protein binding site. Another protein homologous to bullfrog saxiphilin. Although factor that may contribute to this picomolar affinity is there are some subtle differences in binding kinetics, the mechanical ability of transferrin proteins such as $ $ pH dependence, and structure–activity relationships lactoferrin to capture their ligands (i.e. Fe +\HCO −) of STX derivatives among the five species, these within a cavity that closes, in the bound state, like a parameters have the characteristic biochemical and hinged jaw (Anderson et al. 1990), thus utilizing a pharmacological signature of the purified saxiphilin protein conformational change to stabilize the bound protein from Rana catesbeiana (Mahar et al. 1991; ligand. Whatever the mechanism, the chemical specifi- Llewellyn & Moczydlowski 1994). The results thus city and evolutionary conservation of the STX– imply that an STX binding site like that present in saxiphilin interaction makes it difficult to argue that bullfrog saxiphilin is conserved in phylogenetic groups this association is a completely fortuitous molecular as diverse as arthropods and reptiles. This leads us to affiliation unrelated to biological function. In this conclude that a gene coding for the saxiphilin protein regard, it is evident that the STX–saxiphilin in- is present and functionally active in both the arthropod teraction in some species may be at least three orders of and chordate phyla. magnitude stronger than the typical nanomolar affinity Of the major vertebrate classes, saxiphilin activity is of STX binding to the Na+ channel (Ritchie & Rogart readily detected in numerous fish, amphibians, and 1977), the site that mediates biological toxicity of STX. reptiles. The biogeographic distribution of animals found to express saxiphilin includes representatives (b) Constitutive expression of saxiphilin is from diverse habitats in Asia (e.g. N. n. kaouthia), exhibited by certain ectothermic animal species Australia (e.g. V. rosenbergii), Europe (e.g. R. temporaria), Africa (e.g. P. aethiopicus), and the Americas (R. The inference that a saxiphilin gene is widely slatica), indicating that this phenotypic characteristic distributed in the arthropod and vertebrate genomes is not confined to a particular climatic zone. The leads to a series of questions raised by the negative probable existence of saxiphilin in a myriapod (centi- observations of table 2, which imply the apparent pede, E. rubripes), an isopod (sowbug, Oniscus sp.), an absence of saxiphilin in sister taxa and possibly in arachnid (orb web spider, Araneus c. f. caaticus), whole classes of the vertebrate subphylum. For ex- crustaceans (water-flea, Daphnia; xanthid crabs, ample, among the anuran subclass of Amphibia, why Llewellyn (1997)) and some insects (cockroach, damsel- is saxiphilin readily detected in particular frogs and fly nymph) further suggests that it may have first toads (e.g. Rana catesbeiana, R. slatica, R. temporaria, appeared during the emergence of invertebrates some Bufo marinus), but not in the African frog (Xenopus 800 million years ago. Evolutionary speculations on its laeis)? Similarly, among the class Reptilia, why is origin may be considered in light of the fact that saxiphilin present in the subclass (Lepidosauria) $ bilobed, Fe +-binding, transferrin-like proteins are also containing the orders of lizards and snakes, but present in insects (Bartfeld & Law 1990; Jamroz et al. apparently absent in both the subclass (Archosauria)

Proc. R. Soc. Lond. B (1997) 900 L. E. Llewellyn and others Phlogenetic sure of saxiphilin-like actiit that includes alligators and crocodiles and in the or lactoferrin, in vertebrates. The expression and subclass (Anapsida) that includes turtles? One ex- secretion of these latter proteins in humans is mostly planation may be that saxiphilin is a non-essential limited to the surfaces of melanoma cells and several protein or that its function is readily served by other other cell types in the case of melanotransferrin (Rose proteins. Alternatively, the pattern of disparate ex- et al. 1986; Kennard et al. 1995); or to neutrophils and pression could indicate selective loss of a once func- particular fluid secretions such as milk in the case of tional gene. Another possible interpretation of the data lactoferrin (Lo$ nnerdal & Iyer 1995). The curious of tables 1 and 2 is that saxiphilin expression is a pattern of saxiphilin expression reflected in our results character specific to certain ectothermic animals. One merits further investigation to determine what tran- might speculate that the saxiphilin gene is prefer- scriptional and translational regulatory mechanisms entially utilized by cold-blooded animals that do not may control its synthesis. have metabolic control of their body temperature and was deactivated or lost in the evolutionary lineages (c) What is the biological function of saxiphilin? leading to birds and to mammals. Along these lines, it may be noted that the highest The simplest hypothesis for the function of saxiphilin level of saxiphilin-like activity was observed in the is that it plays a defensive role against STX in- wood frog, Rana slatica (having a mean activity of toxication. However, this idea must be questioned " 1600 pmol ml− plasma). This small frog species has since the most dramatic example of microbial STX the northernmost habitat range in North America and production in nature occurs in conjunction with has developed special mechanisms to survive prolonged sporadic blooms of toxic marine dinoflagellates, and we periods hibernating under a superficial layer of ground did not find a consistent association of saxiphilin with cover in a frozen state during the winter (Storey 1990). species that may directly or indirectly be exposed to Although we certainly have not surveyed a sufficient dinoflagellate toxins. In particular, molluscs such as number of species to permit a general conclusion, the Mtilus edulis and Saxidomus giganteus are known to present data lead us to wonder whether there may be bioaccumulate various STX derivatives (Shimizu et al. a functional correlation between high levels of saxi- 1978; Schantz 1986), but do not exhibit evidence of philin activity and particular species of ectothermic saxiphilin-like activity. The Atlantic mackerel, Scomber animals that tolerate a wide variation in body scombrus, has been found to contain STX in its viscera, temperature. and such mackerel have previously been linked to the On the other hand, an apparent lack of saxiphilin deaths of humpback whales (Clifford et al. 1993; activity as measured by the present methodology does Geraci et al. 1989). However, neither plasma from not necessarily prove that the saxiphilin gene is absent mackerel nor two whale species contained detectable in the negative species of table 2. Indeed, it is possible saxiphilin activity (table 2). Similarly, most of the that a functional gene is present and that we have other marine invertebrates, fish, and mammals that we $ failed to detect soluble [ H]STX activity for a number sampled (albeit a limited selection) tested negative for of reasons. For example, our survey results may reflect saxiphilin-like activity. Two interesting exceptions are the underlying biochemical pharmacology of the STX- high levels of saxiphilin activity found in the eel, binding site. If, for instance, there existed an en- Anguilla rostrada (table 1), which can adapt from dogenous ligand that binds to saxiphilin in a com- saltwater to freshwater, and the finding of saxiphilin- petitive manner with STX, the failure to detect like activity in certain species of marine xanthid crabs $ [ H]STX binding in a crude extract could be due to from Australia (Llewellyn 1997). Aside from these pre-existing saturation of the STX binding site with latter two possible cases, the present data do not such a ligand. Alternatively, it may be that tran- indicate that saxiphilin plays a universal or even a scription of the gene is inducible and turned off in most common role in animal encounters with toxic marine animals under normal physiological conditions. Other dinoflagellates. possibilities are that saxiphilin is present in amounts In the freshwater ecosystem, STX and various STX " below the detection limit of our assay (ca. 0.1 pM ml− ), derivatives have so far been found to be produced by or that it is expressed only in certain tissues or cell types three genera of cyanobacteria: Aphaniomenon, Anabaena that are not well sampled by our typical assay of and Lngba (Ikawa et al. 1982; Carmichael et al. 1990; plasma or whole animal extracts. Negri et al. 1995; Carmichael et al. 1995). The Previous work showed that saxiphilin mRNA is occurrence of saxiphilin in many amphibians and small transcribed in bullfrog liver and several other tissues freshwater fish suggests that it may directly function or (Morabito & Moczydlowski 1994). However, from the may be secondarily recruited as a mechanism of STX present results, saxiphilin does not appear to be a detoxification in some of these species. ubiquitous component of animal plasma-like serum In summary, this study has determined that soluble $ transferrin in the vertebrates, which is found at a [ H]STX binding activity characteristic of saxiphilin relatively constant concentration of approximately has a broad phylogenetic distribution among arthro- " 2.5 mg ml− (ca.30µM) in all adult specimens of pods and vertebrates whose nervous systems rely on vertebrate species that have been examined, excepting STX-sensitive Na+ channels that function prominently humans with the debilitating mutation of atransfer- in electrical excitability. The curious existence of the rinanaemia (Welch 1990,1992). Saxiphilin expression two distinct high-affinity binding sites for STX in in some animals may be more like that of the locally many organisms raises the suspicion of a possible expressed transferrin homologues, melanotransferrin physiological relationship. Many questions regarding

Proc. R. Soc. Lond. B (1997) Phlogenetic sure of saxiphilin-like actiit L. E. Llewellyn and others 901 the function, expression and molecular evolution of Clifford, M. N., Walker, R., Ijomah, P., Wright, J., Murray, saxiphilin as a transferrin-related protein remain to be C. K., Hardy, R., Martlbauer, E. P., Usleber, E. & resolved. Further exploration of the evolutionary and Terplan, G. 1993 Do saxitoxin-like substances have a role biological significance of saxiphilin may unravel this in scrombrotoxicosis? Food Add. Contam. 9, 657–667. biological mystery. Crichton, R. R. 1991 Inorganic biochemistr of iron metabolism. New York: Ellis Horwood. We thank John Lynch for his help and contributions to this Doyle, D. D., Wong, M., Tanaka, J. & Barr, L. 1982 study as a Yale undergraduate student, and our laboratory Saxitoxin binding sites in frog myocardial cytosol. Science colleagues for their comments on the manuscript. This work 215, 1117–1119. was supported by grants to E.M. from NIH (GM-51172) Ellis, K. J. & Morrison, J. F. 1982 Buffers of constant ionic and USAMRMC (DAMD-17–93C-3069) and to L.L. from strength for studying pH-dependent processes. Methods the Australian Lions Foundation for Medical Research. Part Enmol. 87, 405–426. of this work was carried out at Mount Desert Island Geraci, J. R., Anderson, D. M., Timperi, R. J., St Aubin, Biological Laboratory at Salisbury Cove, Maine, during the D. J., Early, G. A., Prescott, J. H. & Mayo, C. A. 1989 tenure of a 1996 New Investigator award to E.M., as Humpback whales (Megaptera noaeangliae) fatally poisoned supported by the MDIBL Milbury Fund and NIH grant by dinoflagellate toxin. Can. J. Fish. Aquat. Sci. 46, NIEHS-P30 ESO3828 to the MDIBL Center for Membrane 1895–1898. Toxicity Studies. We are indebted to numerous colleagues Gessner, B. D., Bell, P., Doucette, G. J., Moczydlowski, E., who provided us with samples and specimens, and helped Poli, M. A., Van Dolah, F. & Hall, S. 1997 Hypertension with taxonomic identification: Craig Moritz (Department of and identification of toxin in human urine and serum Zoology, University of ueensland), Don Anderson (Biology following a cluster of mussel-associated paralytic shellfish Department, Woods Hole Oceanographic Institution), Hans poisoning outbreaks. Toxicon. (In the press.) Fricke (Max Planck Institute fu$ r Verhaltensphysiologie), Gitlin, G., Bayer, E. A. & Wilchek, M. 1987 Studies of the Kenneth Storey (Institute of Biochemistry and Department biotin binding site of avidin. Biochem. J. 242, 923–926. of Biology, Carelton University, Ottawa), Kent Vliet Guo, X., Uehara, A., Ravindran, A., Bryant, S. H., Hall, S. (Department of Biological Sciences, University of Florida), & Moczydlowski, E. 1987 Kinetic basis for insensitivity to Timo Nevalainen (Department of Pathology, University of tetrodotoxin and saxitoxin in sodium channels of canine Turku, Finland), Franklin Epstein (Division of Nephrology, heart and denervated rat skeletal muscle. Biochemistr 26, Harvard Medical School), Larry Renfro (Department of 7546–7556. Physiology and Neurobiology, University of Connecticut), Hall, S., Strichartz, G., Moczydlowski, E., Ravindran, A. & Peter Speares and Elizabeth Evans-Illidge (Australian Reichardt, P. B. 1990 The saxitoxins: sources, chemistry Institute of Marine Science), Carl George (Union College), and pharmacology. In Marine toxins: origin, structure and and Kenneth Welch (Connecticut Agricultural Experimental molecular pharmacolog (ed. S. Hall and G. Strichartz), pp. Station, New Haven). 29–65. Washington, DC: American Chemical Society. Ikawa, M., Wegener, K., Foxall, T. L. & Sasner, J. J. 1982 Comparison of the toxins of the blue-green alga REFERENCES Aphaniomenon flos-aquae with the Gonaulax toxins. Toxicon 20, 747–752. Anderson, D. M. & White, A. W. 1992 Marine biotoxins at Jamroz, R. C., Gasdaska, J. R., Bradfield, J. Y. & Law, J. H. the top of the food chain. Oceanus 35, 55–61. 1993 Transferrin in a cockroach: molecular cloning, Anderson, B. F., Baker, H. M., Norris, G. E., Rumball, S. V. characterization and suppression by juvenile hormone. & Baker, E. N. 1990 Apolactoferrin structure demon- Proc. Natn. Acad. Sci. USA 90, 1320–1324. strates ligand-induced conformational change in trans- Kennard, M. L., Richardson, D. R., Gabathuler, R., Ponka, ferrins. Nature 344, 784–787. P. & Jeffries, W. A. 1995 A novel iron uptake mechanism Baker, E. N. & Lindley, P. F. 1992 New perspectives on the mediated by GPI-anchored human p97. EMBO J. 14, structure and function of transferrins. J. Inorg. Biochem. 47, 4178–4186. 147–160. Kurama, T., Kurata, S. & Natori, S. 1995 Molecular Bartfeld, N. S. & Law, J. H. 1990 Isolation and molecular characterization of an insect transferrin and its selective cloning of transferrin from the tobacco hornworm, Manduca incorporation into eggs during oogenesis. Eur. J. Biochem. sexta. J. Biol. Chem. 265, 21 684–216 791. 228, 229–235. Bowman, B. H., Yang, F. & Adrian, G. S. 1988 Transferrin: Li, Y. & Moczydlowski, E. 1991 Purification and partial evolution and genetic regulation of expression. Ad. Genet. sequencing of saxiphilin, a saxitoxin-binding protein from 25, 1–38. the bullfrog, reveals homology to transferrin. J. Biol. Chem. Cabib, E. & Polacheck, I. 1984 Protein assay for dilute 266, 15481–15487. solutions. Methods Enmol. 104, 415–416. Li, Y., Llewellyn, L. & Moczydlowski, E. 1993 Biochemical Carmichael, W. W., Evans, W. R., Yin, . ., Bell, P. & and immunochemical comparison of saxiphilin and trans- Moczydlowski, E. 1995 Evidence for paralytic shellfish ferrin, two structurally related plasma proteins from Rana poisons in the freshwater cyanobacterium Lngba Wollei catesbeiana. Mol. Pharmacol. 44, 742–748. (Farlow ex Bomont) Comb. Int. Conf. Toxic Phtoplankton 7, Llewellyn, L. E. 1997 Haemolymph protein in xanthid A140. crabs: its selective binding of saxitoxin and possible role in Carmichael, W. N., Mahmood, N. A. & Hyde, E. G. 1990 toxin bioaccumulation. Mar. Biol. (In the press.) Natural toxins from cyanobacteria (blue-green algae). In Llewellyn, L. E. & Endean, R. 1989 Toxicity and paralytic Marine toxins: origin, structure and molecular pharmacolog (ed. shellfish profiles of the xanthid crabs, Lophoomus pictor S. Hall and G. Strichartz), pp. 87–106. Washington, DC: and Zosimus aenus, collected from some Australian coral American Chemical Society. reefs. Toxicon 27, 596–600. Cheng, Y. & Prusoff, W. H. 1973 Relationship between the Llewellyn, L. E. & Moczydlowski, E. 1994 Characterization inhibition constant (Ki) and the concentration of an of saxitoxin binding to saxiphilin, a relative of the inhibitor that causes a 50% inhibition (I&!) of an enzymatic transferrin family that displays pH-dependent ligand reaction. Biochem. Pharmacol. 22, 3099–3108. binding. Biochemistr 33, 12312–12322.

Proc. R. Soc. Lond. B (1997) 902 L. E. Llewellyn and others Phlogenetic sure of saxiphilin-like actiit

Lo$ nnerdal, B. & Iyer, S. 1995 Lactoferrin: molecular transferrin) deduced from the mRNA sequence. Proc. Natn. structure and biological function. A. Re. Nutr. 15, 93–110. Acad. Sci. USA 83, 1261–1265. Mahar, J., Lukacs, G. L., Li, Y., Hall, S. & Moczydlowski, Roush, E. D. & Fierke, C. A. 1992 Purification and E. 1991 Pharmacological and biochemical properties of characterization of a carbonic anhydrase II inhibitor from saxiphilin, a soluble saxitoxin-binding protein from the porcine plasma. Biochemistr 31, 12536–12542. bullfrog (Rana catesbeiana). Toxicon 29, 653–671. Schantz, E. J. 1986 Chemistry and biochemistry of saxitoxin Martin, A. W., Huebers, E., Huebers, H., Webb, J. & Finch, and related toxins. Ann. NY Acad. Sci. 479, 15–23. C. A. 1984 A mono-sited transferrin from a representative Shimizu, Y. 1996 Microalgal metabolites: a new per- deuterostome: the ascidian Pura stolonifera (subphylum spective. A. Re. Microbiol. 50, 431–465. Urochordata). Blood 64, 1047–1052. Shimizu, Y., Hsu, C., Fallon, W. E., Oshima, Y., Miura, I. Moczydlowski, E., Mahar, J. & Ravindran, A. 1988 & Nakanishi, K. 1978 Structure of neosaxitoxin. J. Am. Multiple saxitoxin-binding sites in bullfrog muscle: tetro- Chem. Soc. 100, 6791–6793. dotoxin-sensitive sodium channels and tetrodotoxin-in- Storey, K. B. 1990 Life in a frozen state: adaptive strategies sensitive sites of unknown function. Molec. Pharmacol. 33, for natural freeze tolerance in amphibians and reptiles. 202–211. Am. J. Phsiol. 258, R559–R568. Morabito, M. & Moczydlowski, E. 1994 Molecular cloning Terlau, H., Heinemann, S. H., Stu$ hmer, W., Pusch, M., of bullfrog saxiphilin: a unique relative of the transferrin Conti, F., Imoto, K. & Numa, S. 1991 Mapping the site family that binds saxitoxin. Proc. Natn. Acad. Sci. USA 91, of block by tetrodotoxin and saxitoxin of sodium channel 2478–2482. II. FEBS Lett. 293, 93–96. Morabito, M. A., Llewellyn, L. E. & Moczydlowski, E. G. Thorstensen, K. & Romslo, I. 1990 The role of transferrin 1995 Expression of saxiphilin in insect cells and local- in the mechanism of cellular iron uptake. Biochem. J. 271, ization of the saxitoxin-binding site to the C-terminal 1–10. domain homologous to the C-lobe of transferrins. Bio- chemistr 34, 13027–13033. Welch, S. 1990 A comparison of the structure and properties Nagashima, Y., Noguchi, T., Maruyama, J., Kamimura, S. of serum transferrin from 17 animal species. Comp. Biochem. & Hashimoto, K. 1984 Occurrence of paralytic shellfish Phsiol. 97B, 417–427. poisons in an ascidian, Holocnthia roreti. Bull. Jap. Soc. Sci. Welch, S. 1992 Transferrin: the iron carrier. Boca Raton, FL: Fish. 50, 331–334. CRC Press. Negri, A. P., Jones, G. J. & Hindmarsh, M. 1995 Sheep Wuebbens, M. M., Roush, E. D., De Castro, C. M. & mortality associated with paralytic shellfish poisons from Fierke, C. A. 1994 Cloning and sequencing of the porcine the cyanobacterium Anabaena circinalis. Toxicon 33, 1321– inhibitor of carbonic anhydrase. FASEB. J. 8, A1288. 1329. Yasumoto, T., Nagai, H., Yasumura, D., Michishita, T., Ritchie, J. M. & Rogart, R. B. 1977 The binding of Endo, A., Yotsu, M. & Kotaki, Y. 1986 Interspecies saxitoxin and tetrodotoxin to excitable tissue. Re. Phsiol. distribution and possible origin of tetrodotoxin. Ann. NY Biochem. Pharmacol. 79, 1–50. Acad. Sci. 479, 44–51. Rose, T. M., Plowman, G. D., Teplow, D. B., Dreyer, W. J., Hellstro$ m, K. E. & Brown, J. P. 1986 Primary structure of the human melanoma-associated antigen p97 (melano- Receied 10 Februar 1997; accepted 19 Februar 1997

Proc. R. Soc. Lond. B (1997)