Evolutionary History of Venom Glands in the Siluriformes
The order Siluriformes represents a hyperdiverse group of fishes (>3,000 currently recognized species), which has been known to contain venomous species diversity for over 250 years. In spite of this historical knowledge, scientific examinations of the basic characteristics and evolutionary history of these species’ venom glands, and their products, have been extremely sparse compared to those of terrestrial venomous organisms, or even venomous fishes in general. Here, the current state of knowledge regarding the venom glands of catfishes and their products is examined in a review of morphological, pharmacological, and chemical studies of these structures. Several hypotheses regarding the evolution of siluriform venom glands are able to be drawn from the information contained in these studies as well as the limited work that has attempted to study the evolution of these structures in detail. These include selective scenarios to explain the secondary losses of venom glands in several catfish species and families, compositional variation in siluriform venom chemistry, and the derivation of venom glands from secretory cells of the epidermis. Future work directly addressing multiple issues of venom production and composition in catfishes is necessary before investigations of the evolution of siluriform venoms and delivery structures can reach the levels of detail and sophistication seen in other venomous groups. These studies will benefit greatly from the advent of genomic, transcriptomic, and proteomic methods, which have seen wide use in examinations of venoms produced by other taxa, but have yet to be widely applied to analyses of piscine venoms.
KeywordsCatfish Defense Proteins Epidermal Crinotoxins
Species falling under the general classification of “fishes” (a paraphyletic assemblage including the classes Myxini (hagfishes), Petromyzontida (lampreys), Chondrichthyes (sharks, rays, and chimaeras), Actinopterygii (ray-finned fishes), Sarcopterygii (coelacanths and lungfishes)) represent more than half of the world’s known vertebrate species (Nelson 2006). Many species within the Chondrichthyes and Actinopterygii have long been known to utilize venoms in a natural defensive capacity as well as in interactions with bathers and fishermen (Halstead 1988). Human envenomations by fishes are a relatively common occurrence; globally, incidents involving venomous spiny-rayed fish species (superorder Acanthomorpha) alone number over 50,000 cases annually (Smith and Wheeler 2006), which, due both to unreported incidents and exclusion of several venomous groups, likely severely underestimates the actual number of cases. In one study, nearly 70 % of marine fish and 90 % of freshwater fish envenomations of humans were caused by non-acanthomorph species (Haddad and Martins 2006); when extrapolated to global estimates, this would elevate the estimated number of incidents to over 100,000 per year.
As would be expected of a venom whose putative purpose is the rapid deterrence of predators, the most common result of envenomation by fish species is intense pain that is highly disproportionate to the magnitude of the injury, suggesting that components of these venoms target nociceptive sensory neurons (Church and Hodgson 2002; Trim and Trim 2013). In addition to the elicitation of this intense pain response, fish venoms are known to cause a number of other physiological symptoms, including cardiovascular, hemolytic, and neuromuscular effects (Halstead 1988; Church and Hodgson 2002; Sivan 2009). Despite their clear ramifications for human health, fewer than a dozen toxic compounds have been characterized from this highly diverse assemblage (Halstead 1988; Church and Hodgson 2002; Smith and Wheeler 2006). In addition to medical interest in their native physiological effects, fish venoms represent an untapped reservoir of potentially pharmaceutically valuable compounds, particularly as lead compounds in the development of new analgesics, due to their possible ability to directly interact with neuronal signaling pathways (Trim and Trim 2013).
Until recently, however, even the most basic information regarding venomous fishes, such as the number and phylogenetic distribution of venomous taxa, has been unavailable to researchers interested in the evolutionary history of these compounds and the structures that produce them. In the last decade, phylogenetic analyses of acanthomorph species have estimated that 585–650 of the species in this group should be presumed to be venomous, a substantial increase from previous estimates of approximately 200 species (Halstead 1988; Smith and Wheeler 2006). When other types of venomous fishes (Chondrichthyes, Siluriformes) are included, this estimate potentially increases to over 2,500 species, or just under 10 % of all known fish species (Wright 2009). Though this level of species diversity is greater than that of all other venomous vertebrates combined, venomous fishes remain severely understudied relative to venomous terrestrial organisms, as evidenced by a recent review of venom evolution that mentions fishes only in passing, and without providing any detailed information regarding the toxic action of their venoms (Casewell et al. 2012).
The order Siluriformes, commonly known as catfishes, is a globally distributed, highly diverse clade containing over 3,000 currently recognized species in 36–38 families (Sullivan et al. 2006; Ferraris 2007). The order has been known to contain venomous representatives for nearly 300 years, beginning with Johann Richter’s description of Spanish fishermen’s fear of stings from marine catfishes belonging to the family Ariidae (Halstead 1988). Although the stings of most catfish species are relatively harmless, albeit very uncomfortable, fatalities have been reported as the result of envenomations by members of the families Plotosidae (Plotosus lineatus) and Clariidae (Heteropneustes fossilis) (Halstead 1988). These species undoubtedly possess notably potent venoms, but these fatalities, which occurred in the late nineteenth and early twentieth centuries, were likely due to poor medical care and/or secondary infection of the wound (a common complication of siluriform envenomations) (Halstead et al. 1953; Haddad and Martins 2006). Only one modern fatality involving a catfish sting has been recorded, a freak accident in which a fisherman’s heart was penetrated by the spine of a large individual (Haddad et al. 2008).
Taxonomic distributions and estimates of venomous catfish diversity. Estimates reproduced from Wright (2009)
# 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 – eel-tailed catfishes
Pseudopimelodidae – bumblebee catfishes
Schilbeidae – glass catfishes
Siluridae – sheat catfishes
Examinations of the evolutionary history of venoms and venom production in catfishes are currently hampered by a lack of resolution in higher-level siluriform phylogeny (Sullivan et al. 2006) and basic knowledge regarding the identity of venom components and the genetic architecture underlying their production (Wright 2009; Egge and Simons 2011) as well as selective factors driving the compositional evolution and properties of defensive venoms (Casewell et al. 2012). Nonetheless, sufficient progress has been made to be able to generate inferences regarding several aspects of catfish venom gland evolution. This chapter attempts to provide a review of our current knowledge and hypotheses regarding the evolution of siluriform venom glands, as developed through an examination of relevant literature concerning the identification and anatomy of siluriform venom glands and delivery systems; the toxicology, pharmacology, and basic chemistry of the venoms of species investigated thus far; and the few studies that have attempted to directly address the ecology and evolution of the venom systems of catfishes. Such a survey serves as an illustration of not only the surprising amount of evolutionary information that can be gleaned from the existing literature but how far the study of venomous catfishes, and venomous fishes in general, must proceed before reaching the levels of detail and sophistication seen in other groups of venomous organisms.
Siluriform Venom Gland and Delivery System Morphology
Venoms, by definition, require a method by which their bearer is able to introduce them into the body of a target organism. In all known venomous fishes (with the exception of Meiacanthus sp. and members of the deep-sea family Monogathidae), this is accomplished via spiny elements associated with the fins and/or opercular and cleithral bones (Halstead 1988; Smith and Wheeler 2006). These spiny elements contain grooves that facilitate the flow of venom along the spin; in most cases, the glandular tissue rests within the groove itself. The association of these venom glands with spiny elements led Perrière and Goudey-Perrière (2003) to name their toxic secretions acanthotoxins. In Meiacanthus sp. (saber-toothed blennies), injection is achieved by the use of enlarged fangs in the bottom jaw rather than spines, with buccal venom glands surrounding the proximal two thirds of the fang (Halstead 1988; Smith and Wheeler 2006). Venom flows toward the site of envenomation through grooves along the anterior fang margins. Monognathids , which lack upper jaws, apparently inject venom via a single, hollow rostral fang, which has paired glands at its base (Bertelsen and Nielsen 1987). These species are unique among venomous fishes, in that they appear to use their venoms to subdue their prey, shrimps that are very large relative to their own size (Bertelsen and Nielsen 1987), and which would have the potential to cause significant damage to these relatively fragile fishes.
This passive method of venom delivery appears to represent a rather primitive condition, which is found across multiple groups of venomous fishes; members of only a few families (e.g., Batrachoididae, Scorpaenidae) show a more specialized, hypodermic-style apparatus characteristic of most other venomous vertebrates (Birkhead 1972; Halstead 1988; Smith and Wheeler 2006). It also results in potentially significant damage to the integumentary and glandular tissue surrounding the spine, which can take a significant amount of time (over a week) to heal (Birkhead 1972). Nonetheless, such compromised spines still represent a potent antipredatory defense; multiple experiments presenting North American catfish species (family Ictaluridae) to largemouth bass (Micropterus salmoides), a common piscivorous species, have demonstrated that the presence of spines increases predator handling time and catfish survivorship relative to individuals in which the spines had been removed (Bosher et al. 2006; Emmett and Cochran 2010; Wright 2012a). Wright (2012a) further demonstrated, however, that the presence of venom glands significantly increases the antipredatory capabilities of spines in intact tadpole madtoms ( Noturus gyrinus) , relative to individuals in which the venom glands had been surgically removed.
The extent and orientation of the venom glands in relation to spine serrations, as well as grooves within the spine itself, varies significantly between different catfish species (Halstead 1988; Wright 2009; Egge and Simons 2011). Serrated spines may increase the amount of mechanical damage produced when the spine enters a potential predator, increasing the surface area exposed to the concomitantly released venom (Reed 1907; Birkhead 1972; Egge and Simons 2011). There is little evidence, however, to suggest that the venoms of species with greater levels of spine ornamentation possess significantly greater toxicity (Birkhead 1972), nor have experiments been performed to demonstrate increased predator deterrent ability in those species. In fact, Egge and Simons (2011) found that of five evolutionary changes in sting morphology in the genus Noturus , four involved decreases in morphological complexity, including the loss of spine serrations, loss of venom gland tissue associated with serrations, or, in one case, the total loss of the venom gland.
These results suggest that certain ecological and life history traits may result in the relaxation of selective pressures related to the maintenance of venom glands, leading to their eventual loss, which appears to be a relatively widespread phenomenon throughout the order (Wright 2009). Such scenarios may include ontogenetic loss of venom glands in species obtaining body sizes that effectively protect them from natural, gape-limited predators (Egge and Simons 2011) or the secondary loss of venom glands in members of families that have lost ossified fin spines (e.g., Malapteruridae, many amphiliids) and, thus, an effective delivery system for metabolically expensive venom compounds (Wright 2009). Wright (2009) also found that members of the families Sisoridae and Erethistidae have secondarily lost venom glands, while maintaining their fin spines. Many of the species in these families occupy highly rheophilic habitats (as does Ameiurus brunneus , another species that has secondarily lost venom glands), where effective foraging by large-bodied predatory species would be highly difficult, if not impossible, offering a possible explanation for the lack of venom production in these species.
The cellular morphology of venom glands in fishes is very similar across broad taxonomic categories, indicating possible widespread convergent evolution of these cells. Venom-producing cells are enclosed within an integumentary sheath composed of epithelial cells. The venom gland cells are large and polygonal, with prominent nucleoli and highly granulous cytoplasm, presumably due to high concentrations of venomous peptides (Reed 1907; Halstead et al. 1953; Halstead 1988); in catfishes, the cells of the venom gland are also binucleate (Reed 1907; Halstead et al. 1953; Halstead 1988). As the cells mature, organelles and nuclear structures are lost and only the cytoplasmic granules are visible. Venomous secretions are either held within the cells or the cells undergo holocrine secretion, whereby the secreting cells are lysed and release the venomous secretions (along with cellular fragments) into the intercellular space, where they are held until being used.
Cameron and Endean (1973) hypothesized that the venom gland cells of fishes and the acanthotoxins that they contain are evolutionarily derived from the clavate or club cells of the epidermis, which secrete proteins known as crinotoxins (Halstead 1988). While crinotoxic secretions are released into the water when the cells are ruptured, ostensibly to repel predators or fouling organisms (Cameron and Endean 1973), the direct injection of these compounds into other organisms has also been shown to have toxic effects (Al-Hassan et al. 1987; Shiomi et al. 1987, 1988). A preliminary study of the catfish Plotosus lineatus offers some support for Cameron and Endean’s hypothesis, as the club cells of this species were found to produce a substance that is similar, and possibly identical, to one of the toxic fractions found in the venom gland, based on immunological reactions (Shiomi et al. 1988).
Perrière and Goudey-Perrière (2003), however, point out that common production of a single toxic component is not sufficient evidence to prove the homology of these cell types. While certain crinotoxins and acanthotoxins produced by P. lineatus show similar histochemical and pharmacological activities, Whitear et al. (1991a) found distinct differences in the ultrastructure and histochemistry of the venom gland cells and club cells in the skin of Heteropneustes fossilis (Indian stinging catfish) that, in their estimation, precludes the homology of the two cell types. Specifically, club cells were found to contain helical filaments and a division of the cytoplasm into perinuclear and peripheral zones, both of which were lacking in the venom cells. Additionally, while a previous study (Zaccone et al. 1990) had shown a positive immunohistochemical reaction for serotonin in the club cells of this species, Whitear et al. (1991a) found that this reaction was lacking in the venom cells.
Whitear et al. (1991a) did not address why these differences should mean that the venom gland cells could not possibly have been derived from epidermal club cells. If venom glands are indeed adaptive structures, one might expect their cellular morphology and the secretions that they produce to be subject to selection pressures that differ from those experienced by secretory cells in other locations. The differences reported by Whitear et al. may simply reflect this history. Additional comparative morphological and transcriptomic studies of venom glands and secretory epidermal cells from different groups of venomous catfishes should serve to clarify these issues.
Siluriform Axillary Glands
In addition to the venom glands lining the spinous elements of the fins, many siluriform species possess secretory glands situated in the axil of the pectoral fin (Reed 1907; Halstead et al. 1953; Halstead and Smith 1954; Greven et al. 2006). These structures are known from several families, including the Akysidae, Ariidae, Callichthyidae, Ictaluridae, Mochokidae, and Plotosidae, but to date, no comprehensive survey has been performed to document the distribution of axillary glands throughout the Siluriformes. Various authors have considered the axillary glands to be part of the venom apparatus (Reed 1907; Halstead et al. 1953; Birkhead 1967; Cameron and Endean 1971), and, as such, they are briefly discussed here.
The earliest mention of axillary glands in catfishes was made by Günther (1880). He assumed that secretions issuing from the axillary pore anoint the pectoral fin spine, allowing them to be injected along with secretions from the pectoral venom glands. Many works that followed (i.e., Reed 1907) accepted this statement without experimental confirmation. More recently however, several additional, more likely, hypotheses have been proposed for the function of these structures. While later studies showed that axillary gland extracts are toxic when injected into other organisms (Cameron and Endean 1971; Birkhead 1967), the water-soluble nature of axillary pore secretions is difficult to reconcile with the venomous scenario envisioned by earlier authors. Current hypotheses regarding the function of the axillary gland secretions include antimicrobial (Kiehl et al. 2006), ichthyotoxic (Greven et al. 2006), pheremonal, and ionoregulatory roles, though only the first two are supported by empirical evidence.
While it appears that the axillary glands of catfishes do not function as part of the venom delivery apparatus, their true function and the action of their products remains a potentially fruitful area for future research. Fairly simple procedures, such as comparative electrophoresis, HPLC, or mass spectrometry of venom and axillary gland extracts, could be used to more conclusively rule out the presence of axillary gland secretions on the pectoral spine. Further investigations of the antimicrobial and ichthyotoxic hypotheses that have thus far received preliminary support are also warranted.
Pharmacology and Toxicology of Siluriform Venoms
Wright (2009, 2011, 2012a, b) has demonstrated that the crude venom extracts of a phylogenetically diverse group of catfish species produce a wide array of symptoms when injected into a model predatory species (largemouth bass), most notably the rapid loss of color pattern throughout the body, which has been observed from the venoms of nearly every catfish species studied thus far. This suggests the presence of conserved venom function that acts in some way on the nervous system, which controls chromatophore and melanophore activity . Suites of additional envenomation symptoms observed by Wright (2009, 2011, 2012a, b) were highly species specific and included the expansion of melanophores at the injection site, rapid loss of color pattern elsewhere on the body, muscle spasms of varying degrees of intensity and duration, hemorrhage, loss of equilibrium, and, in the case of Plotosus lineatus , rapid mortality. Earlier work by Birkhead (1967, 1972) examined the venoms of several species from the North American family Ictaluridae that were also studied by Wright (2012b) and found that they produced some of the same symptoms, including melanophore expansion and hemorrhage. Several additional symptoms were found, however, including notable edema, necrosis, and death. Some of these differences may be attributable to Birkhead’s use of a different assay organism ( Gambusia affinis) in his assessments of venom toxicity. It must be noted, however, that the effects demonstrated by both Birkhead (1967, 1972) and Wright (2009, 2011, 2012a, b) were likely elicited by the injection of much higher doses of venom than would be encountered in a natural situation. These encounters usually result in violent ejection of the catfish from the buccal cavity of the bass, accompanied by rapid gaping of the mouth and flaring of the gills (Wright 2011, 2012a).
As naturally occurring substances which are able to elicit potent responses in vertebrate physiological systems, the venoms of fishes have come under increased scrutiny as possible sources of future biomedical compounds. Studies of the toxic effects elicited by fish venoms in other organisms have revealed a high degree of similarity in these effects and the mechanism of their production, providing an additional example of apparent convergent evolution of fish venom glands and the substances they produce. The most common sites of human envenomation are the hands or feet, and in many cases, the pain has been known to travel up the entire length of the affected appendage (Halstead et al. 1953; Calton and Burnett 1975; Halstead 1988; Church and Hodgson 2002; Sivan 2009). The pharmacological actions of the venoms of a select few catfish species have been studied and have been shown to have cardiovascular, neuromuscular, and general cytolytic effects in various assays (Church and Hodgson 2002; Sivan 2009).
The widespread elicitation of cardiovascular effects by piscine venoms in experimental tissue preparations indicates convergence in venom target systems, although the nature of these effects and the mechanisms by which they are produced vary between species and taxonomic groups. The venoms of Plotosus canius and Heteropneustes fossilis are thought to either contain or cause the release of prostaglandins, contributing to their production of smooth muscle contractile responses in a number of tissue preparations (Church and Hodgson 2002; Sivan 2009). In contrast, the smooth muscle contraction produced by the venom of Arius thallasinus appears to be produced through effects on muscarinic acetylcholine receptors (Church and Hodgson 2002; Sivan 2009). Effects on cardiac muscle preparations are similarly variable, with the venom of H. fossilis producing inotropic increases in guinea pig and toad hearts, while toxin-PC isolated from P. canius causes cessation of heartbeat in guinea pig preparations (Auudy and Gomes 1996; Church and Hodgson 2002; Sivan 2009). The combined effects of siluriform venoms on blood vessel and cardiac function have also produced alternate results in in vivo preparations. The venom of P. canius has been shown to produce a hypertensive response, while that of H. fossilis produces a hypotensive effect (Auddy and Gomes 1996; Church and Hodgson 2002; Sivan 2009).
Potent neuromuscular activities have been reported from several catfish venoms, in addition to the systemic, neurologically mediated color loss and muscle spasms observed in toxicity assays using living predators. The crude venom of Plotosus canius has been shown to irreversibly inhibit electrically induced muscle contractions in rat and chick muscle preparations, as has an isolated preparation of toxin-PC, the lethal component of that species’ venom (Church and Hodgson 2002; Sivan 2009). It is thought that toxin-PC prevents neurotransmitter release presynaptically, as it produces sustained muscular contraction without affecting muscular preparations’ responses to acetylcholine or carbachol, although its blockage of neuromuscular activity apparently does not result from K+-channel or cholinesterase modulating abilities (Auddy and Gomes 1996; Church and Hodgson 2002; Sivan 2009). The venom of a related species, P. lineatus , has also been shown to produce neurotoxic symptoms upon intraperitoneal injection into mice (Fahim et al. 1996). In another case of interspecific divergence in siluriform venom effects, however the venom of Heteropneustes fossilis has been found to display no appreciable neuromuscular effect (Church and Hodgson 2002; Sivan 2009).
Nearly all piscine venoms exhibit cytolytic properties, and the venoms of catfishes are no exception. In fact, local necrosis is one of the most common clinical symptoms of piscine envenomations (Sivan 2009) and has also been documented in Birkhead’s (1967, 1972) envenomations of Gambusia with ictalurid species’ venoms. The lack of such symptoms in Wright’s (2009, 2011, 2012a, b) experiments, however, may indicate that these necroses are largely due to secondary bacterial infections. Nonetheless, the venoms of several siluriform species have produced hemolysis in rabbit ( Plotosus canius) , rat (P. canius, P. lineatus), human ( Arius thallasinus) , mouse (P. canius), cow (A. thallasinus, P. canius), and sheep (A. thallasinus) erythrocytes (Church and Hodgson 2002). The venom of P. lineatus has additionally been shown to be cytotoxic to cultured Ehrlich ascites tumor cells as well as a number of other cell types (Fahim et al. 1996). The cytolytic action of these venoms is thought to contribute to other negative effects of envenomation, through forming pores in the plasma membranes of target cells, allowing the influx of Ca2+ which triggers the release of several biologically active compounds from the cell (Church and Hodgson 2002). Such an action is also known from bee (Pawlak et al. 1991) and platypus venoms (Kourie 1999), both of which are primarily pain-producing venoms, like those of catfishes.
Chemistry of Siluriform Venoms
Additional putatively toxic peptides, generally falling within the size range of 10–20 kDa (Calton and Burnett 1975; Church and Hodgson 2002; Auddy and Gomes 1996; Wright 2009), have been identified in the venoms of many siluriform species, although putative toxins of 40–50 kDa have been indicated in some ariid species (Junqueira et al. 2007; Wright 2009). The lethal fraction of the venom of Plotosus canius (toxin-PC) is one such protein, having a molecular weight of approximately 15 kDa (Auddy and Gomes 1996; Wright 2009). These smaller venom components show significant variation in number and size over interspecific, intergeneric, and interfamilial scales, identifying them as likely candidates underlying the variation observed in the effects elicited by different species’ venoms in toxicological assays (Wright 2009, 2012b). This high degree of variation, even between relatively closely related species, would seem to strongly indicate that selective forces associated with different habitats and/or predatory regimes have contributed to the establishment of differing levels of venom protein identity and complexity between different siluriform lineages. Our current lack of information regarding the genes coding for these proteins, as well as their structure and physiological targets, precludes the testing of further hypotheses regarding the evolution of these compounds. The molecular weight data obtained thus far for siluriform venoms is nonetheless valuable, as it offers an independent check on the identities of the potentially novel toxin-related sequences that will undoubtedly be uncovered by future evolutionary studies utilizing omics-scale technologies and analytical methods to identify catfish venom toxin genes, transcripts, and proteins.
Chemical Complexity of Siluriform Venoms
In contrast to the venoms of organisms that utilize these secretions in prey capture, which can potentially contain hundreds of toxic components per species, the venoms of fishes and other organisms that utilize venom in a strictly defensive capacity appear to contain only one or a few toxic components (Church and Hodgson 2002; Wright 2009; Casewell et al. 2012). This has been confirmed for several catfish species using comparative electrophoresis of extracts prepared from fin spines and associated tissues, which showed only one to three unique peptides being expressed in spine extracts relative to control extracts prepared from histologically similar fin tissues (Wright 2009, 2011, 2012a, b). The toxic nature of these peptides was confirmed using toxicity assays performed in largemouth bass ( Micropterus salmoides) , which showed marked, species-specific effects associated with injection of fin spine extracts, but no toxicity associated with the injection of fin tissue extracts. An interesting parallel to this condition of reduced venom toxin diversity is found in the venoms of sea snakes, which have also been shown to contain a highly reduced number of toxic components relative to other venomous snakes (Fry et al. 2003). The similarities become even more striking when one considers that venomous marine snakes represent two evolutionary radiations that have independently arrived at a state of reduced venom complexity (Scanlon and Lee 2004), while compositionally simple venoms have been independently derived in acanthomorph fishes no fewer than 11 times (Smith and Wheeler 2006), and at least twice in catfishes (Wright 2009).
The white catfish ( Ameiurus catus) may represent an exception to the generalization that piscine venoms exhibit low toxin diversity. The venom of this species was found to contain two to eight fractions that showed lethal activity in mice (Calton and Burnett 1975). The additional finding that A. catus venom lost little to no activity following treatment with trypsin and elevated temperature indicates that additional, non-proteinaceous compounds may be present in the venomous secretions of this species. These results are questionable however, as different methods of analysis yielded proteinaceous fractions of varying weights and biological activities. Wright (2012b) also examined the venom of this species, using the comparative electrophoretic methods discussed above, and found evidence for only two putative toxic peptides in prepared fin spine extracts; these produced a significant toxic effect in injected largemouth bass. While this study could not speak to the possible presence of non-proteinaceous toxins in A. catus venom, it clearly supports the lower value from Calton and Burnett’s (1975) estimate of the number of lethal fractions in the venom of A. catus and is much more consistent with what is known from other species. It is possible, however, that siluriform venoms show intraspecific regional variation and that the conflicting results of these studies resulted from drawing individuals from geographically distant populations (Chesapeake Bay tributaries in the case of Calton and Burnett, an inland North Carolina lake in the case of Wright).
The low number of toxic compounds found in fish venoms would appear to be an asset to studies of their evolution, as the problems of homology inherent in evolutionary studies of species that produce many different toxins should be easily addressed. It is tempting to suggest that the parallel streamlining of these species’ venoms is due to selection associated with a common target: piscine physiological systems. Little empirical evidence exists to support this hypothesis however, as few studies of the action of sea snake and piscine venoms on their (presumed) natural targets exist. The few studies of sea snake venoms performed in this context have indicated that likely prey species possess high levels of resistance to sea snake venoms (Heatwole and Powell 1998). This would appear to run counter to a selective streamlining hypothesis, as one might expect these species of sea snakes to possess more complex venoms to overcome prey resistances to particular toxic compounds. Preliminary results from studies on ictalurid catfishes (Wright 2012b) indicate that the venoms of bullheads have little effect on potential predators with which they share a habitat type. These results may indicate that coevolution between predator and prey is occurring in these systems, leading to these somewhat counterintuitive results. Further studies are clearly necessary to examine possible correlations between the low number of toxic compounds in catfish venoms and the selective factors influencing siluriform venom evolution.
Evolutionary Origins of Siluriform Venom Glands
Our lack of knowledge regarding basic characteristics of siluriform venoms and their targets represents a significant obstacle to the study of their evolution, which is compounded by our incomplete understanding of siluriform phylogeny. Current classifications divide modern catfishes into two monophyletic suborders, the Loricarioidei (South American armored, sucker-mouthed, pencil, and parasitic catfishes) and the Siluroidei (all remaining catfish families). The morphologically primitive family Diplomystidae has been alternatively recovered as the sister group to all catfishes, or the sister group of the Siluroidei, with the Loricariodei sister to all other catfishes. Multiple higher-level phylogenetic analyses of catfishes, using both morphological and molecular data (e.g., Diogo 2004; Hardman 2005; Sullivan et al. 2006), are available, but these studies nearly universally suffer from poorly supported resolution of the early evolutionary relationships between catfish families, particularly within the Siluroidei.
Evolution from Epidermal Secretory Cells
There is strong evidence that the venom glands of several previously studied catfish species produce similar compounds to epidermal glandular cells. Immunocytochemical assays of epidermal cells taken from Plotosus lineatus have indicated that these cells produce a highly similar protein to one of the toxic fractions identified from the venom gland of that species (Shiomi et al. 1988). Further evidence for this similarity is provided by the results of SDS-PAGE analyses performed by Wright (2009). This study indicated the presence of major toxin bands in the venom of P. lineatus, at 15–16 kDa and 13–14 kDa, in addition to the conserved 110 kDa putative toxin found in the venoms of most catfishes. The larger peptide is likely to represent toxin-PC, which showed a similar molecular weight in previous characterizations by Auddy and Gomes (1996) in the related species P. canius . The smaller peptide, however, is very similar in molecular weight to the toxic fraction isolated from epidermal secretions of P. lineatus by Shiomi et al. (1987, 1988). Wright (2009) also identified a ~39 kDa putative toxin in the electrophoretic profile of the venom of Arius jordani . This corresponds closely with the major toxic factor of the skin secretion of the congeneric A. bilineatus, which has been isolated and shown to have a molecular weight of approximately 39 kDa (Thomson et al. 1998).
Two scenarios have been proposed to explain the evolutionary origins and derivation of venom glands and their products in catfishes, both of which theorize that these structures and their products are derived from epidermal secretory cells. The first of these, developed by Cameron and Endean (1973) and outlined in the above discussion of siluriform venom gland cellular morphology, hypothesizes that the venom glands of all fishes, including catfish species, are derived from crinotoxin-producing epidermal cells. In this scenario, the early stages of venom gland development would consist of a thickening of crinotoxin-producing epidermal tissue surrounding the spines of early venomous catfish species, offering a selective advantage to these individuals in deterring potential predators. Subsequent evolutionary changes, including further increased concentrations of toxic protein-secreting cells and their segregation from the epidermal tissue by an integumentary sheath; the suppression of other epidermal cell types from being produced within this tissue; and the movement of this glandular tissue closer to the fin spines, thereby achieving efficient delivery of cellular products during envenomations, then established the morphology of siluriform venom glands as they are seen today. Modifications to fin spines facilitating the delivery of venom into wounds, such as the grooves found in the spines of akysid, amblycipitid, and some ictalurid catfishes (Wright 2009, 2012a, b; Egge and Simons 2011), are hypothesized to have occurred secondarily to the development of venom glands in ancestral species.
It is true that crinotoxins are released when epidermal cells are damaged during predation attempts on catfishes, which is evocative of the manner in which venom is released when the spines of a catfish enter a potential predator. The actual function of these toxins appears to be in the deterrence of fouling organisms, however, as ichthyocrinotoxic species are characteristically sedentary and possess decreased or absent squammation (Cameron and Endean 1973). Crinotoxins have also never been shown to have any appreciable predator deterrent effects, and, in fact, predatory species will readily attack and feed on damaged and distressed catfishes (Bosher et al. 2006; Emmett and Cochran 2010; Wright 2012a) as well as baits coated with stress-related epidermal secretions (Al-Hassan et al. 1985). Studies of the skin secretions of several Arius species have indicated that compounds contained therein are able to accelerate healing of wounds and may also have antimicrobial properties (Al-Hassan et al. 1983, 1985, 1987; Robinette et al. 1998). Antimicrobial capabilities have also been demonstrated from the axillary gland secretions of callichthyid catfish species, which are also thought to be derived from epidermal stress-related secretions (Greven et al. 2006; Kiehl et al. 2006), suggesting that secretory cells producing antimicrobial products have already been co-opted into other siluriform secretory structures.
This information led Wright (2009) to propose an alternative selective scenario to the one proposed by Cameron and Endean, centering on the apparent healing and antimicrobial properties of catfish epidermal secretions (see also Venom as a Component of External Immune Defense in Hymenoptera, this volume). The epidermal tissue covering the spines of catfishes is frequently damaged during interactions with predators and with their physical environment. It is therefore conceivable that higher concentrations of epidermal secretory cells surrounding the spine could confer a selective advantage, through improved healing times and decreased opportunities for infection of compromised tissues. This selection would lead to increased aggregations of these cells around the fin spines, with the toxic, antipredatory effects of their secretions being either an epiphenomenon to their primary healing benefits in catfish species or secondarily developed to augment existing defensive structures. Once venomous secretions had been established and associated with their delivery devices, lineage-specific selective regimes could then act on venom toxicity and composition in catfishes to produce the compositional and toxicological variation found in the venoms of modern siluriform species as well as the conservation of the primary, pain-producing peptides that form the basis of their predator deterrent abilities.
Conclusion and Future Directions
The evolutionary study of siluriform venom glands and their products represents an important but understudied area of inquiry, due both to their human impacts and potential untapped benefits as well as the fact that they represent an important antipredatory trait in a globally ubiquitous group of organisms, which often represent a significant portion of a given region’s aquatic vertebrate biodiversity. Though these structures have been shown to provide a formidable defense against predators in several cases, even influencing other aspects of morphological evolution in some genera, secondary losses of venom glands are evident in several groups, most likely due to relaxation of predation pressures due to different aspects of life history and habitat choice. Venomous catfishes comprise a highly diverse group of organisms, possibly outnumbering all other venomous vertebrates combined, and display a correspondingly high degree of variation in venom delivery apparatus morphology and venom effects. In natural predators and laboratory organisms, the venoms of catfishes have been shown to elicit symptoms consistent with cardiovascular, neurotoxic, hemolytic, and/or lethal effects, with a high degree of taxonomic variation in the suite of effects induced by different species’ venoms. Despite this, siluriform venoms appear to be quite simplified, consisting of only a few toxic venom proteins per species. Very few siluriform venoms have been studied in any detail, however, and future examinations of inter- and intrafamilial variation in venom toxicity and composition would have great potential to uncover additional venom toxin diversity within catfishes, as well as to generate insights into ecological differences influencing species-specific venom characteristics.
Venom glands have arisen independently at least twice within the order Siluriformes, with the potential for a third origin in the South American family Doradidae. Additional higher-level analyses of siluriform phylogeny are also required to provide greater resolution of basal siluroid relationships and a well-supported consensus of internal relationships within this suborder, which will allow stronger conclusions regarding the developmental history of catfish venom glands to be drawn. Histological, toxicological, and electrophoretic evidence all suggest that the venom glands of catfishes are evolutionarily derived from epidermal secretory cells. Whether catfish venoms are derived from crinotoxins or healing and antimicrobial substances produced by epidermal cells in the skin is unclear, however. Further studies of both epidermal secretion types and the venoms of catfishes are required at a proteomic and genetic level in order to determine the relationships between these different substances. It is entirely possible that these crinotoxins and antimicrobial agents are one and the same, leaving little hope for possible resolution to the debate regarding which defensive selective force, increased predator deterrence, or rapid healing and infection defense initiated the process of venom evolution in catfishes.
Studies making use of next-generation proteomic, transcriptomic, and genomic technologies and analyses have great potential to generate desperately needed data regarding the identity and structure of siluriform venom toxins, their physiological targets, their genetic origins, and the selective forces driving their evolution. Continued studies of catfish venoms have the potential to greatly increase our understanding of the general ecology and evolution of this hyperdiverse order of fishes as well as to generate insights into the evolution of venoms as defensive traits. The chemical complexity of the venoms of species utilizing these secretions in prey capture makes it exceedingly difficult to determine whether and how selection for prey capture or predator defense has influenced any particular venom component, as these differing selective forces have the potential to be non-complimentary. The relatively simple composition of catfish venoms, and fishes in general, as well as their use in a strictly defensive capacity, therefore presents an outstanding opportunity to study the selective factors influencing defensive venom evolution, examinations of which are largely absent from the literature.
- Bertelsen E, Nielsen JG. The deep sea eel family Monognathidae (Pisces, Anguilliformes). Steenstrupia. 1987;13(4):141–98.Google Scholar
- Diogo R. Morphological evolution, adaptations, homoplasies, constraints, and evolutionary trends: catfishes as a case study on general phylogeny and macroevolution. Enfield: Science Publishers; 2004.Google Scholar
- Ferraris CJ. Checklist of catfishes, recent and fossil (Osteichthyes: Siluriformes), and catalogue of siluriform primary types. Zootaxa. 2007;1418:1–628.Google Scholar
- Fine ML, Sismour EN, Newton SH, Bosher BT, Sullivan ADH, Miano JP, Ghahramani ZN, Mohajer YJ, Nellis S. A primer on functional morphology and behavioral ecology of the pectoral spine of the channel catfish. In: Michaletz, PH, Travnichek VH, editors. Conservation, ecology, and management of catfish: the second international symposium. American Fisheries Society Symposium 77. Bethesda: American Fisheries Society; 2011.Google Scholar
- Greven H, Flasbeck T, Passia D. Axillary glands in the armoured catfish Corydoras aeneus. Verh Ges Ichthyol Band. 2006;5:65–9.Google Scholar
- Halstead BW. Poisonous and venomous marine animals of the world. 2nd rev ed. Princeton: The Darwin Press: 1988.Google Scholar
- Kiehl E, Rieger C, Greven H. Axillary gland secretions contribute to the stress-induced discharge of a bactericidal substance in Corydoras sterbai (Callichthyidae, Siluriformes). Verh Ges Ichthyol Band. 2006;5:111–5.Google Scholar
- Nelson JS. Fishes of the world. 4th ed. New York: Wiley; 2006.Google Scholar
- Perrière C, Goudey-Perrière F. Poisonous catfishes: venom apparatus, acanthotoxins, crinotoxins and other skin secretions. In: Arratia G, Kapoor BG, Chardon M, Diogo R, editors. Catfishes, vol. 1. Enfield: Science Publishers; 2003.Google Scholar
- Wright JJ. The evolutionary ecology of venomous catfishes, with a focus on members of the north American family Ictaluridae (Teleostei: Siluriformes). Ann Arbor: University of Michigan; 2012b.Google Scholar