Diversity, phylogenetic distribution, and origins of venomous catfishes
© Wright. 2009
Received: 28 April 2009
Accepted: 4 December 2009
Published: 4 December 2009
Skip to main content
© Wright. 2009
Received: 28 April 2009
Accepted: 4 December 2009
Published: 4 December 2009
The study of venomous fishes is in a state of relative infancy when compared to that of other groups of venomous organisms. Catfishes (Order Siluriformes) are a diverse group of bony fishes that have long been known to include venomous taxa, but the extent and phylogenetic distribution of this venomous species diversity has never been documented, while the nature of the venoms themselves also remains poorly understood. In this study, I used histological preparations from over 100 catfish genera, basic biochemical and toxicological analyses of fin spine extracts from several species, and previous systematic studies of catfishes to examine the distribution of venom glands in this group. These results also offer preliminary insights into the evolutionary history of venom glands in the Siluriformes.
Histological examinations of 158 catfish species indicate that approximately 1250-1625+ catfish species should be presumed to be venomous, when viewed in conjunction with several hypotheses of siluriform phylogeny. Maximum parsimony character optimization analyses indicate two to three independent derivations of venom glands within the Siluriformes. A number of putative toxic peptides were identified in the venoms of catfish species from many of the families determined to contain venomous representatives. These peptides elicit a wide array of physiological effects in other fishes, though any one species examined produced no more than three distinct putative toxins in its venom. The molecular weights and effects produced by these putative toxic peptides show strong similarities to previously characterized toxins found in catfish epidermal secretions.
Venom glands have evolved multiple times in catfishes (Order Siluriformes), and venomous catfishes may outnumber the combined diversity of all other venomous vertebrates. The toxic peptides found in catfish venoms may be derived from epidermal secretions that have been demonstrated to accelerate the healing of wounds, rather than defensive crinotoxins.
The venoms produced by cnidarians, mollusks, snakes, arachnids, insects, and some mammals have been the subject of multiple studies of chemical structure [1–3], pharmacology [2–5], and toxicology [5–7], in addition to several evolutionary studies [8–12], but information regarding these aspects of fish venoms is relatively sparse [13–18]. Until recently, even reliable estimates of the number of venomous fish species have been unavailable. Morphological examinations, combined with phylogenetic analyses have suggested that 585-650 species of spiny-rayed fishes are venomous, a number which rivals the known diversity of venomous snakes and is significantly higher than previous estimates of about 200 venomous spiny-rayed fish species . We still lack estimates, however, for catfishes (Order Siluriformes), a diverse, monophyletic group with 34 recognized extant families and over 400 genera containing more than 3,000 known species . The historical lack of such basic information may be largely responsible for the paucity of research on venomous fishes in general, and venomous catfishes in particular.
The chemical nature of piscine venoms is poorly known, though the loss of toxicity seen when these venoms are subjected to common denaturing agents suggests that proteins constitute the major toxic component of these secretions . Thus far, detailed examinations of these proteins in catfishes have been limited to the venoms of Plotosus canius, a particularly toxic marine species found in Southeast Asia, and Ameiurus catus, a freshwater species found in the eastern United States. The neurotoxic and hemolytic properties of P. canius venom have been attributed solely to a 15 kDa protein, termed toxin-PC . The venom of A. catus was thought to contain anywhere from two to eight toxic proteins with approximate molecular weights of 10 kDa . Both the mechanism by which these toxins act and their physiological targets are very poorly understood. It is thought that cytolytic activity due to pore formation in cell membranes is a likely explanation, as this activity is present in other 'pain-producing' venoms, such as those produced by bees  and platypus , and reactions consistent with this mechanism have been observed in response to piscine venoms .
As a globally distributed and thus, biogeographically interesting group, catfishes have recently been a topic of interest in several phylogenetic studies [26–29]. When combined with these data, information regarding the distribution of venom glands within the Siluriformes can be examined in an evolutionary context, and we can begin to build a foundation to advance the studies of venom evolution in this group to the level seen in other venomous organisms. In this work, I use histological and toxicological techniques to elucidate the diversity and taxonomic distribution of venomous catfishes and examine these findings within the phylogenetic framework established by previous authors to provide a broad-scale hypothesis for the evolutionary origins of venom glands in catfishes. These examinations are further integrated with preliminary biochemical characterizations of venoms from several catfish species to highlight an intriguing, novel hypothesis for the evolutionary development of venom glands in catfishes.
Taxonomic distributions and estimates of venomous catfish diversity.
# Presumed Venomous
Siluriformes - Catfishes
Akysidae - Asian stream catfishes
Amblycipitidae - Torrent catfishes
Anchariidae - Madagascan catfishes
Ariidae - Sea catfishes
Bagridae - Bagrid catfishes
Callichthyidae - Armored catfishes
Chacidae - Angler catfishes
Clariidae - Labyrinth catfishes
Claroteidae - Claroteid catfishes
Cranoglanididae - Armorhead catfishes
Doradidae - Thorny catfishes
Heptapteridae - Shrimp catfishes
Ictaluridae - North American catfishes
Mochokidae - Squeakers
Pangasiidae - Shark catfishes
Pimelodidae - Antennae catfishes
Plotosidae - Eeltail catfishes
Pseudopimelodidae - Bumblebee catfishes
Schilbidae - Glass catfishes
Siluridae - Sheat catfishes
The effects of several catfish species' venoms on Largemouth Bass.
Loss of Equilibrium
Arius jordani (Ariidae)
Corydoras paleatus (Callichthyidae)
Horabagrus brachysoma (incertae sedis)
Microglanis iheringi (Pseudopimelodidae)
Noturus gyrinus (Ictaluridae)
Pangasius hypophthalmus (Pangasiidae)
Pimelodus pictus (Pimelodidae)
Plotosus lineatus (Plotosidae)
Synodontis multipunctata (Mochokidae)
The evolution of venom glands within the order Siluriformes was examined by performing maximum parsimony character optimization analyses on several previously published siluriform phylogenies that were reconstructed from both morphological [26, 30] and molecular  data. Multiple phylogenies were analyzed due to the fact that the relationships of some siluriform families are either poorly resolved or vary between reported phylogenies. Given the widespread presence of venom glands in catfishes, it was expected that these previous systematic studies, in conjunction with the results presented above, would offer some insight into broader phylogenetic patterns of siluriform venom gland evolution in spite of the poor resolution of familial relationships found in these phylogenies.
Examinations of histological sections of pectoral-fin spines, in conjunction with character optimization analyses of previously published siluriform phylogenies and toxicological assays, imply that approximately 1250-1625 species of catfishes from at least 20 families are venomous. These numbers are much higher than previous estimates, based largely on anecdotal evidence, which suggested a maximum of 1000+ venomous catfish species . Of these families, 14 (Akysidae, Anchariidae, Callichthyidae, Chacidae, Claroteidae, Cranoglanididae, Doradidae, Heptapteridae, Mochokidae, Pangasiidae, Pimelodidae, Pseudopimelodidae, Schilbidae, Siluridae) are shown to contain venomous taxa for the first time; six (Amblycipitidae, Ariidae, Bagridae, Clariidae, Ictaluridae, Plotosidae) have previously been demonstrated to contain venomous representatives . The approximation of 1250 species of venomous catfishes is undoubtedly an underestimate, as many genera in siluriform families containing venomous taxa remain to be examined. New species of catfishes are also continuously being discovered and described (958 species described in the last 10 years according to the Catalog of Fishes ), with some venomous genera such as Chiloglanis (Mochokidae) containing an estimated 25 or more undescribed species [J.P. Friel, pers. comm.].
The loss of venom glands appears to be a common phenomenon within catfishes, which is not surprising given that bony fin spines have been lost in some families (Malapteruridae, most amphiliids). Genera in several families that contain venomous representatives (Heptapteridae, Pimelodidae, Siluridae) have also lost bony dorsal and/or pectoral-fin spines. Without an effective delivery system, there would seem to be no selection pressure for the maintenance of venom producing structures, leading to their reduction and eventual loss. The apparent loss of venom glands in groups that have maintained bony fin spines [Aspredinidae, Auchenipteridae, Sisoridae, some ictalurids (see Additional file 1)] is more unexpected, and explanations for these losses are not immediately apparent.
Inter- and intrageneric loss of venom glands was also found within the family Ictaluridae (Additional file 1). Both Ameiurus melas and Pylodictus olivaris lack any structures that could be identified as venom glands based on histological examination. Additionally, SDS-PAGE analysis detected no putative venom peptides in either species (Fig. 7). This finding was particularly surprising for A. melas, which had previously been considered venomous and quite virulent, based upon toxicological and histological work [13, 20]. This discrepancy may be attributable to geographic variation in venom production; A. melas is a widely distributed species and the specimens examined in the current study were collected in Michigan, while those used in the previous toxicological study came from Texas. A potentially important factor in the case of Pylodictus is that this species can reach adult sizes that would presumably prohibit predation by even the largest North American predatory fishes (all of which are gape-limited predators), possibly weakening or eliminating selection for the maintenance of venom glands through adulthood.
Taxonomic distributions and estimates of venomous vertebrate diversity.
# Presumed Venomous
Actinopterygii - Ray-finned fishes
≈1835 - 2275 species
Siluriformes - Catfishes
Acanthomorpha - Spiny-rayed fishes
Chondrichthyes - Cartilaginous fishes
Sarcopterygii - Lobe-finned fishes and tetrapods
Cameron and Endean  hypothesized that the venom glands of fishes are derived from glandular epidermal cells that secrete toxic proteinaceous compounds (termed "ichthyocrinotoxins") when fishes are threatened or injured. While it is true that these compounds are secreted in these situations, the hypothesis that they serve in an antipredatory capacity in catfishes appears flawed. With the exception of ichthyocrinotoxins associated with the epidermis of the dorsal and pectoral fin, there is no effective delivery device for these compounds, which are produced all over the body. This is of particular importance, as all assays demonstrating toxicity of epidermal secretions of catfishes have relied on intravenous injection of these compounds as a toxicological assay [36–40]. Furthermore, the presence of epidermal secretions does not appear to be a significant deterrent to potential predators, as they will attack and feed on distressed catfishes, as well as other baits coated with catfish epidermal secretions [; pers. obs.].
That venom glands in catfishes produce similar compounds to epidermal glandular cells has been indicated by immunocytochemical assays . The results of SDS-PAGE analyses presented here offer additional support for the similarity of these secretions. The major toxic factor of the skin secretion of Arius bilineatus has been isolated and shown to have a molecular weight of approximately 39 kDa . The venom of Arius jordani clearly shows a strong band at approximately 39 kDa which is found in low concentration in the control lane (Fig. 7). The presence of this protein in the control sample is likely due to the presence of epidermal secretory cells in the tissue sample used, while the low concentration is due to the removal of most of the epidermal secretions before sample preparation. While these cells were also probably present in spine samples, the large difference in concentration indicates that venom gland cells are likely responsible for production of most of this protein band. A similar case is seen in the electrophoretic profile of Plotosus lineatus, which shows major toxin bands at 15-16 kDa and 13-14 kDa (Fig. 7). While the larger band is similar in weight to toxin-PC, as characterized by Auddy and Gomes , the lower band is very similar in weight to a toxic fraction isolated from the skin secretions of this species [37, 38], with the slight discrepancy in estimated size possibly being due to differences in sample preparation and analysis.
While the venom gland cells in catfishes (and other fishes as well) are likely to be derived from epidermal secretory cells, an alternative scenario to Cameron and Endean's antipredatory hypothesis is also able to explain their origin. Studies of the epidermal secretions of several Arius species have indicated that these compounds are able to accelerate healing of wounds and may also have some antimicrobial properties [41–43]. The spines of catfishes act to effectively increase their cross-sectional circumference when locked into place, and would likely be the first structures to contact a gape-limited predator's tissues during an attack. As such, the spines would often be damaged, and individuals with larger numbers of epidermal secretory cells surrounding the spine could gain a selective advantage due to decreased healing time and a corresponding decreased chance of infection of exposed tissues. This selection may have led to increased aggregations of these cells around the fin spines, with the toxic effects of their secretions being an epiphenomenon to their primary healing benefits. Once the toxic secretions had become associated with an effective delivery device, selection for increased toxicity, as seen in some plotosid and clariid species, could begin to operate. Explicit tests of this scenario will require more detailed structural and genetic characterizations of these compounds.
The symptoms of envenomation produced by a diverse array of catfish species' venoms are very similar and a large number of putative toxins appear to fall within a well-defined molecular weight range. The conserved molecular weight patterns and toxic effects of catfish venom peptides suggest two possible scenarios for the evolution of venoms in catfishes: widespread convergent evolution of catfish venom toxins with similar targets and thus similar molecular characteristics and effects, or common origins of toxic peptides with subsequent species-specific alterations. The widespread presence of venom glands shown by the character optimization evidence discussed above strongly suggests that the latter case is the more parsimonious and likely scenario, even in cases where phylogenetic resolution of basal siluriform divergences is lacking.
This study utilizes several lines of investigation to increase our knowledge of several poorly understood areas of the biology of venomous catfishes. These investigations have demonstrated that at least 1250, and possibly over 1600 species of catfishes may be venomous, a number far greater than any previous estimate of venomous catfish diversity. In conjunction with previous systematic studies, these findings also offer insight into the evolutionary history of venom glands in the order Siluriformes, indicating at least two independent evolutionary origins of these structures. Finally, the symptoms of catfish envenomation, along with preliminary biochemical characterizations of toxic catfish venom peptides, may suggest a novel selective explanation for the evolution of catfish venom glands and their secretions.
Finer-scale studies of venom gland evolution in fishes will require continued systematic studies of venomous fish families to elucidate the relationships of the species contained therein. Additionally, examinations of the chemical composition of fish venoms and the identities and structures of their constituents will provide valuable insight into the mechanisms and potential selective factors driving venom evolution in fishes, as well as their potential for biomedical research and pharmaceutical bioprospecting.
The right pectoral-fin spine was removed from 158 catfish specimens (see Additional file 1), housed in the fish collection of the University of Michigan Museum of Zoology. Spines were decalcified in CalEx® according to the manufacturer's instructions, after which segments from the distal third of the spine of an appropriate size for histological preparation were removed. These segments were subjected to automated dehydration and paraffin infiltration and embedding at the Tissue Core Facility of the University of Michigan Comprehensive Cancer Center. Serial sections of 0.7 microns were then obtained from each spine sample. Sections were stained with hematoxylin and eosin and mounted on glass slides.
Spines were examined for the presence of venom glands using a Nikon YS2-T compound microscope. Morphological confirmation of the presence of venom gland cells was achieved by comparisons with previously published photomicrographs of venom glands in catfishes and spiny rayed fishes [20, 35, 44, 45], descriptions of piscine venom gland cellular anatomy , and sections obtained from the spines of catfish species that have been shown to secrete venomous substances by previous studies [13, 20]. When a representative of a particular genus was found to possess venom glands, all members of that genus were presumed to be venomous, except in the case of the ictalurid genus Ameiurus, where the examination of multiple species within the genus indicated otherwise. These generic counts of venomous species formed the basis for the minimum estimate of venomous catfish species (Table 1). The number of species contained in unexamined genera from families containing venomous representatives was added to the minimum estimate to give a maximum estimate of venomous catfish species (Table 1).
Representatives of the catfish families Ariidae, Bagridae, Callichthyidae, Ictaluridae, Mochokidae, Pangasiidae, Pimelodidae, Pseudopimelodidae, and Plotosidae were obtained either from field collections (Ictaluridae) or the aquarium trade (other families). Specimens were euthanized using MS-222 at a concentration of 300 mg/L in fresh water. All further preparations were carried out either on ice or under refrigeration at 4°C. Spines and caudal fin tissue were removed from each specimen, rinsed in physiological saline and gently scraped with a microspatula in order to remove any excess epidermal secretions, and weighed to the nearest 0.001 g using a GeneMate digital balance. Spines were minced and then further homogenized in a 2 mL Dounce homogenizer along with either marine (Plotosidae) or freshwater (other families) euteleost physiological saline at a volume of 2 mL/g of tissue . The homogenate was then centrifuged at 6,000 rpm at 4°C for 20 minutes and the supernatant collected. The supernatant served as the crude venom extract. Control extracts prepared from caudal fin tissue were prepared in the same manner.
Largemouth Bass were collected from Boyden Creek, Washtenaw Co., MI in October of 2008. Bass were anesthetized in MS-222 at a concentration of 75 mg/L of fresh water and weighed to the nearest 0.1 g. They were then placed in 10 G experimental aquaria in a room with natural light and allowed to acclimate for a period of 72 hours. After the 72 hour acclimation period, bass were injected in the caudal peduncle at a depth of 2 mm with 2 μL/g body weight of either crude venom extract or control extract. Individuals were then observed at one minute, one hour, and 24 hours after injection for symptoms consistent with envenomation (Table 2). For each species of catfish tested, two bass were injected with venom extract and two were injected with caudal fin control extract.
Several previously published phylogenetic hypotheses for the order Siluriformes [26, 28, 30] were examined using MacClade 4.0 PPC . Presence and absence of venom glands was traced onto the trees using the criterion of maximum parsimony. Specific taxa that were present in the phylogenetic reconstruction but which were not examined in the current study were coded as ambiguous (?) within the data matrix.
Crude extracts were prepared for SDS-PAGE analysis by reduction with NuPAGE® reducing agent and loading buffer, according to manufacturer's instructions. Reduced samples were subjected to electophoresis in NuPAGE® precast 4-12% Bis-Tris polyacrylamide gels in 1× MES running buffer for 35 minutes, at 200 V in an x-Cell SureLock™ Mini Cell. Reduced peptides were visualized using SimplyBlue™ SafeStain according to manufacturer's instructions. Molecular weights of venom and caudal fin extracts were estimated by comparison with Novex® Sharp Protein Standard. Proteins unique to venom extracts (relative to caudal-fin extracts) were treated as putative toxins, pending further characterization.
I thank the staff at the Tissue Core Facility of the University of Michigan Comprehensive Cancer Center for consultation regarding and performance of steps involved in histological preparation of samples. Thomas Duda kindly provided lab space for performance of SDS-PAGE analyses. I also wish to thank W. Fink, G. Smith, T. Duda, D. Nelson, J.M. Wright, H. Ng, P. Chakrabarty, R. Oldfield, and K. Birkett for helpful discussions and suggestions for the improvement of this manuscript. All animal care was performed in accordance with University of Michigan Committee on the Use and Care of Animals (UCUCA) regulations and guidelines. Experimental procedures involving live animals were approved under UCUCA protocol # 09713. Financial support was provided by the University of Michigan Museum of Zoology and the University of Michigan Rackham Graduate School.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.