Decision letter: The skin microbiome facilitates adaptive tetrodotoxin production in poisonous newts

Abstract

Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Rough-skinned newts (Taricha granulosa) use tetrodotoxin (TTX) to block voltage-gated sodium (Nav) channels as a chemical defense against predation. Interestingly, newts exhibit extreme population-level variation in toxicity attributed to a coevolutionary arms race with TTX-resistant predatory snakes, but the source of TTX in newts is unknown. Here, we investigated whether symbiotic bacteria isolated from toxic newts could produce TTX. We characterized the skin-associated microbiota from a toxic and non-toxic population of newts and established pure cultures of isolated bacterial symbionts from toxic newts. We then screened bacterial culture media for TTX using LC-MS/MS and identified TTX-producing bacterial strains from four genera, including Aeromonas, Pseudomonas, Shewanella, and Sphingopyxis. Additionally, we sequenced the Nav channel gene family in toxic newts and found that newts expressed Nav channels with modified TTX binding sites, conferring extreme physiological resistance to TTX. This study highlights the complex interactions among adaptive physiology, animal-bacterial symbiosis, and ecological context. eLife digest Rough-skinned newts produce tetrodotoxin or TTX, a deadly neurotoxin that is also present in some pufferfish, octopuses, crabs, starfish, flatworms, frogs, and toads. It remains a mystery why so many different creatures produce this toxin. One possibility is that TTX did not evolve in animals at all, but rather it is made by bacteria living on or in these creatures. In fact, scientists have already shown that TTX-producing bacteria supply pufferfish, octopus, and other animals with the toxin. However, it was not known where TTX in newts and other amphibians comes from. TTX kills animals by blocking specialized ion channels and shutting down the signaling between neurons, but rough-skinned newts appear insensitive to this blockage, making it likely that they have evolved defenses against the toxin. Some garter snakes that feed on these newts have also evolved to become immune to the effects of TTX. If bacteria are the source of TTX in the newts, the emergence of newt-eating snakes resistant to TTX must be putting evolutionary pressure on both the newts and the bacteria to boost their anti-snake defenses. Learning more about these complex relationships will help scientists better understand both evolution and the role of beneficial bacteria. Vaelli et al. have now shown that bacteria living on rough-skinned newts produce TTX. In the experiments, bacteria samples were collected from the skin of the newts and grown in the laboratory. Four different types of bacteria from the samples collected produced TTX. Next, Vaelli et al. looked at five genes that encode the channels normally affected by TTX in newts and found that all them have mutations that prevent them from being blocked by this deadly neurotoxin. This suggests that bacteria living on newts shape the evolution of genes critical to the animals' own survival. Helpful bacteria living on and in animals have important effects on animals' physiology, health, and disease. But understanding these complex interactions is challenging. Rough-skinned newts provide an excellent model system for studying the effects of helpful bacteria living on animals. Vaelli et al. show that a single chemical produced by bacteria can impact diverse aspects of animal biology including physiology, the evolution of their genes, and their interactions with other creatures in their environment. Introduction Coevolutionary interactions among species are a central force driving the origin of novel, adaptive phenotypes, yet the traits under selection are often complex and arise from multifaceted interactions among genetic, physiological, and environmental forces that are not well understood (Ehrlich and Raven, 1964; Futuyma and Agrawal, 2009; Schoener, 2011). Chemical interactions among species in the form of defensive compounds have evolved across all domains of life, and these toxins often target evolutionarily conserved proteins in potential predators (Brodie and Ridenhour, 2003; Hodgson, 2012; Whittaker and Feeny, 1971). For example, tetrodotoxin (TTX), the primary neurotoxin found in poisonous pufferfishes (Tsuda and Kawamura, 1952), has been discovered across a broad phylogenetic distribution of animals (Chau et al., 2011; Hanifin, 2010). The unusual molecular structure of this toxin serves to selectively target voltage-gated sodium (Nav) channels, which are critical for generating action potentials in neurons, muscles, and other excitable cells (Hille, 2001). Thus, TTX toxicity can have substantial impacts on eco-evolutionary interactions among species, impacting both the toxic species and potential predators. Rough-skinned newts (Taricha granulosa) are among the most poisonous TTX-producing animals and serve as an excellent model system for understanding ecological influences on toxin production and predation (Figure 1A). This species is endemic to the Pacific Northwest of North America, where certain populations possess high quantities of TTX relative to other TTX-laden species including pufferfishes and blue-ringed octopuses (Hanifin, 2010; Williams, 2010). In some populations, individual newts possess enough TTX to kill several adult humans (Brodie et al., 2005; Hanifin, 2010; Hanifin et al., 1999). Variation in newt toxicity is driven in part by the evolution of TTX resistance in predatory garter snakes (Thamnophis sirtalis), as TTX toxicity and resistance in newts and snakes are strongly correlated geographically, suggesting that these two phenotypes are coevolving (Brodie et al., 2005; Brodie et al., 2002; Hanifin et al., 2008). Furthermore, TTX resistant Nav channels have evolved independently across different garter snake populations, suggesting multiple independent origins of TTX resistance in predatory snakes (Feldman et al., 2009; Geffeney, 2002; Geffeney et al., 2005). Figure 1 with 2 supplements see all Download asset Open asset Characterization of the skin-associated microbiota of rough-skinned newts. (A) Rough-skinned newt (Taricha granulosa); photo by Gary Nafis (CC by ND NC 3.0). (B) Two populations of T. granulosa previously reported to possess either high concentrations of TTX (Lincoln Co, OR; red) or no TTX (Latah Co, ID; black) were compared in our study. (C) TTX measured from dorsal skin biopsies: newts from Oregon possessed 126.5 ± 42.1 ng mL−1 TTX (n = 5) while Idaho newts possessed no detectable TTX (n = 17). (D–E) Scanning electron micrographs of host-associated bacterial communities upon the dorsal skin surface and within the ducts of TTX-sequestering granular glands. Scale bars 10 µm and 1 µm for D and E, respectively. (F–G) Comparison of bacterial community richness and diversity between these two populations, along with soil samples collected contemporaneously from the ponds in which the newts were caught. Non-toxic Idaho newts possessed higher OTU richness (Chao1 index, t unequal var. = 7.90, p<0.0001) and diversity (Simpson [1-D] index, t unequal var. = 4.11, p<0.0001) than toxic Oregon newts. (H) Mean relative abundance of bacterial OTUs present in dorsal or ventral skin of each population, as well as local soil samples. Sample sizes for panels F-H were n = 12 for Oregon (TTX+) and n = 16 for Idaho (TTX-) newts. Figure 1—source data 1 Raw data for newt skin toxicity measurements by LC-MS/MS. Estimated TTX concentrations (ng mL−1) were calculated relative to a calibration curve of pure TTX standards. https://cdn.elifesciences.org/articles/53898/elife-53898-fig1-data1-v1.docx Download elife-53898-fig1-data1-v1.docx Figure 1—source data 2 The top 20 most abundant bacterial OTUs found among toxic (Oregon) and non-toxic (Idaho) newts. The relative abundance of each bacterial OTU within a sample was calculated and averaged across all samples for each population; these data are shown below as percentages. Taxonomy was assigned using the Ribosomal Database Project (Cole et al., 2014) and a confidence threshold of 80%. OTUs shared between the two populations are in bold. https://cdn.elifesciences.org/articles/53898/elife-53898-fig1-data2-v1.docx Download elife-53898-fig1-data2-v1.docx Figure 1—source data 3 Variation in alpha diversity between newt populations and sampling sites across the bodies of individual newts. Each value is shown as mean ± SEM. https://cdn.elifesciences.org/articles/53898/elife-53898-fig1-data3-v1.docx Download elife-53898-fig1-data3-v1.docx Despite the central role of TTX toxicity in coevolutionary interactions between newts and snakes, the origin of TTX in newts and other freshwater animals is unknown (Daly, 2004; Hanifin, 2010). In TTX-bearing marine species, toxicity is derived either from dietary accumulation from TTX-laden prey, or from symbiotic interactions with TTX-producing bacteria (Chau et al., 2011; Miyazawa and Noguchi, 2001). Pufferfishes harbor numerous TTX-producing bacteria symbionts in toxic tissues including skin, liver, intestines, and ovaries, and cultured non-toxic pufferfishes are able to sequester dietary-administered TTX under laboratory conditions (Jal and Khora, 2015). TTX-producing bacteria have also been isolated from xanthid crabs, horseshoe crabs, starfish, chaetognaths, nemerteans, gastropods, and blue-ringed octopuses (Jal and Khora, 2015; Magarlamov et al., 2017). However, the origin of TTX in rough-skinned newts has been more controversial (Hanifin, 2010). Newts raised in long-term captivity on artificial diets maintain their TTX toxicity (Hanifin et al., 2002), and captive newts forced to secrete their TTX by electric shock are able to slowly regenerate their toxicity over time (Cardall et al., 2004). Additionally, one group attempted to amplify bacterial DNA from various newt tissues using 16S rRNA gene primers, but failed to recover any PCR products aside from the intestine, which contains low levels of TTX (Lehman et al., 2004). These studies suggest that the source of TTX in newts is not dietary, but whether newts have acquired the ability to produce TTX endogenously via convergent evolution or horizontal gene transfer, or from symbiosis with TTX-producing bacteria remains unclear. Furthermore, despite the extreme toxicity of some newt populations, the molecular basis of TTX resistance in this species is not well understood. A previous study identified amino acid replacements in the highly conserved pore-loop (P-loop) region of the skeletal muscle isoform Nav1.4 and found that skeletal muscle fibers were considerably resistant to TTX (Hanifin and Gilly, 2015). Amphibians, however, possess six Nav channel isoforms that are differentially expressed across excitable tissues (Zakon et al., 2011). Unlike pufferfishes in which TTX is sequestered in certain tissues, newts possess TTX throughout their bodies (Mebs et al., 2010; Wakely et al., 1966; Yotsu et al., 1990), indicating that the central and peripheral nervous systems are exposed to TTX. Thus, the evolution of whole animal resistance should necessarily involve all Nav channel subtypes, providing an opportunity to examine molecular evolution in response to a specific selective pressure (i.e., TTX) across an entire gene family. In this study, we investigated the source of TTX toxicity and the molecular basis for TTX resistance in rough-skinned newts. We re-examined the hypothesis that newts derive their TTX from symbiosis with TTX-producing bacteria, focusing on the bacterial communities inhabiting the skin of T. granulosa, as this organ possesses specialized granular glands for storing toxins and contains the highest quantities of TTX in the animal (Daly et al., 1987; Hamning et al., 2000; Hanifin et al., 2004; Santos et al., 2016; Toledo and Jared, 1995; Tsuruda et al., 2002). We took advantage of the natural variation in TTX toxicity across newt populations to characterize the skin-associated microbiota in newts from a highly toxic and a non-toxic population and applied an unbiased cultivation-based approach to isolate numerous bacterial symbionts from the skin of toxic newts. Subsequent LC-MS/MS screening of bacterial cultivation media revealed TTX production in eleven bacterial strains from four genera: Aeromonas, Pseudomonas, Shewanella, and Sphingopyxis. Furthermore, to determine the molecular and physiological basis of extreme TTX resistance observed in this species, we cloned and sequenced five uncharacterized Nav channel paralogs (Nav1.1, Nav1.2, Nav1.3, Nav1.5, Nav1.6) from a highly toxic population of newts in Oregon. We identified amino acid substitutions in all five genes, many of which have been observed in other TTX-possessing species. To test whether Nav channel mutations impact TTX resistance in newts, we used site-directed mutagenesis to insert three newt-specific replacements identified in Nav1.6 into the TTX-sensitive Nav1.6 ortholog from Mus musculus. We found that each amino acid replacement reduced TTX sensitivity compared to wild-type M. musculus channels, but these mutations had the greatest effect when combined together. Overall, our results suggest that host-associated bacteria may underlie the production of a critical defensive compound in a vertebrate host, impacting predator-prey coevolution and potentially shaping the evolution of TTX resistance in a well-known ecological model system. Results Characterization of newt-associated microbiota from toxic and non-toxic newts To investigate whether bacterial symbionts produce TTX in newts, we leveraged natural variation in toxicity across newt populations to characterize skin-associated microbiota and determine whether highly toxic newts harbored candidate TTX-producing bacteria. We compared two populations previously reported to differ substantially in TTX levels (Hanifin et al., 2008; Hanifin et al., 1999), a toxic population in Lincoln County, OR and a non-toxic population in Latah County, ID (Figure 1A–B). As expected, skin biopsies collected from the dorsal skin of adult newts confirmed that Oregon newts (n = 5) possessed high TTX concentrations (126.5 ± 42.1 ng mL−1 per mg skin) while Idaho newts (n = 17) lacked detectable levels of TTX (Figure 1C and Figure 1—source data 1). T. granulosa have particularly enlarged granular glands, which amphibians use to store and secrete noxious or toxic compounds (Hanifin, 2010; Santos et al., 2016; Toledo and Jared, 1995). Interestingly, scanning electron micrographs of dorsal granular glands revealed the presence of bacteria inhabiting the surface and outer pore of TTX-sequestering granular glands, and mixed bacterial communities including rod and coccus-shaped bacterial cells were present within the ducts of these glands (Figure 1D–E). We therefore focused our sequencing and cultivation efforts on skin microbiota as a potential source of TTX in this species. Bacterial communities inhabiting the dorsal and ventral skin, cloacal gland, and submandibular gland of T. granulosa were compared by culture-independent 16S rRNA gene sequencing targeting the V4 hypervariable region. Bacterial samples were collected by swabbing wild newts captured in the field at each site (Oregon n = 12 and Idaho n = 16). Bacterial samples were collected from the Oregon and Idaho populations at separate times, in June 2013 and September 2016, respectively. However, all bacterial DNA samples were extracted, amplified, and sequenced together on the same run. In total, we identified 4160 operational taxonomic units (OTUs) with an average Good's coverage (Good, 1953) of 0.9454 ± 0.0067 (mean ± SEM) across samples from both populations. 614 OTUs were unique to toxic newts, 1943 were unique to non-toxic newts, and 1603 were shared between the two populations. Among the 20 most abundant OTUs, 8 OTUs were shared between both populations while 12 were present in only one population (Figure 1—figure supplement 1 and Figure 1—source data 2). These highly abundant and conserved OTUs may represent core skin microbiota of T. granulosa. Idaho newts possessed a greater number of distinct bacterial types with a more even distribution across their microbiota, reflected in a higher number of observed OTUs (t unequal var. = 7.70, p<0.0001) and higher OTU richness (Chao1 index, t unequal var. = 7.90, p<0.0001) on average than in Oregon newts. Bacterial alpha diversity was also significantly greater in Idaho than Oregon newts (Simpson 1-D index [t unequal var. = 4.11, p<0.0001]) (Figure 1F–G and Figure 1—source data 3). Phylum-level divisions show that newt microbiota consists primarily of Proteobacteria, Bacteroidetes, and Firmicutes, which together comprise 76.2–83.5% of the average bacterial community across all four body sites in both populations (Figure 1—figure supplement 2). At the genus level, the relative abundance of each bacterial OTU differed markedly between the two populations, as well as from soil samples collected from their respective habitats (Figure 1H). The composition and relative abundances of OTUs (i.e. beta diversity) also differed significantly between the two newt populations (Figure 2). Principal coordinates analysis (PCoA) shows a distinct clustering based on geographic location in both composition (Jaccard index) and structure (Bray-Curtis index) of skin microbiota from each population (Figure 2A–B). Permutational multivariate analysis of variance (PERMANOVA) tests revealed that different skin sites across the animal harbored similar communities within a population (Jaccard, p=0.375; Bray-Curtis, p=0.065), but that community composition (Jaccard index, F = 18.12, p<0.0001) and structure (Bray-Curtis index, F = 40.40, p<0.0001) differed significantly between populations. The skin communities of Idaho newts were also more variable than those of Oregon newts (Permutational test for multivariate dispersion, PERMDISP, p=0.0053) (Figure 2—figure supplement 1). Interestingly, toxic Oregon newts maintain a high relative abundance of Pseudomonas OTUs relative to non-toxic Idaho newts (Figure 2C). Three Pseudomonas OTUs (00042, 00224, and 00485) were present in greater relative abundance in toxic newts, and OTU00042 was a significant driver of the beta diversity differences observed between these two populations. Indeed, linear discriminant analysis effect size (LEfSE) indicates that Pseudomonas OTU00042 is among the top 10 most differentially abundant OTUs in Oregon newts (Figure 2D). In subsequent non-targeted cultivation of newt skin bacteria, we isolated numerous culturable TTX-producing Pseudomonas spp. strains from toxic newts (below). Figure 2 with 2 supplements see all Download asset Open asset Comparison of skin microbiota from toxic and non-toxic newt populations reveal distinct host-associated bacterial communities and enrichment of TTX-producing bacteria. (A–B) Principal coordinates analysis of bacterial community composition (OTU presence/absence; Jaccard index) and community structure (OTU relative abundance; Bray-Curtis index) of skin microbiota across four body sites from toxic Oregon and non-toxic Idaho newts reveal distinct clustering of bacterial communities within each population. Non-parametric multivariate analysis of variance (PERMANOVA) test indicates a significant effect of location on both the composition (F = 18.12, p<0.0001) and structure (F = 40.40, p<0.0001) of skin-associated bacterial communities. Newt-associated communities also cluster separately from soil communities sampled from each location. (C) Representative comparison of highly abundant OTUs from individual toxic Oregon and non-toxic Idaho newt microbiota samples (on x-axis) reveals increased relative abundance of Pseudomonas OTUs in toxic newts. (D) Linear discriminant analysis effect size (LEfSE) comparing the top 10 differentially abundant OTUs between toxic and non-toxic newt microbiota indicates that Pseudomonas OTU00042 is highly enriched in toxic newt microbiota (black arrow). Pseudomonas isolated from newt skin produced TTX (Figure 3). Sample sizes were n = 12 and n = 16 for Oregon and Idaho newts, respectively. Culturable newt microbiome and TTX production in bacterial isolates To determine whether newt skin microbes produce TTX, we employed an unbiased, cultivation-based small molecule screen to examine bacterial culture media for the presence of TTX production in vitro. This approach was necessary because the genetic basis of TTX biosynthesis is unknown, preventing application of metagenomic or other sequencing approaches to determine whether newts or their microbiota possess the genetic toolkit for TTX production. Mixed bacterial communities were collected by swabbing the dorsal skin of toxic newts and cultured on nutrient-limited minimal media (Reasoner's 2A agar) or blood agar supplemented with defibrinated sheep's blood (10% v/v). Individual colonies were re-streaked, isolated in pure culture, and taxonomically identified by 16S rRNA gene sequencing. In total, we generated a culture collection of 355 strains representing 65 bacterial genera (summarized in Figure 3—figure supplement 1). Isolated strains were subsequently grown in 10% Reasoner's 2 broth (R2B) for 2 weeks at 20 °C; 1 mL of each culture was then removed and for TTX using mixed for sample and for molecular screening (Figure Figure 3 with 2 supplements see all Download asset Open asset Bacterial isolates cultured from newt skin produce TTX in vitro. (A) of for and screening newt bacteria for TTX production. Bacterial samples were collected from dorsal skin of toxic newts, taxonomically identified by 16S rRNA gene grown in culture for 2 and the was by were screened against TTX by LC-MS/MS. (B) Representative ion to ion to and to for TTX in cultures from four bacterial The time for each was and that of both TTX and culture media supplemented with ng mL−1 TTX. were in culture Estimated TTX concentrations in bacterial cultures are shown to each strains of Pseudomonas spp. were found to produce TTX and three TTX-producing genera were Aeromonas, Shewanella, and Sphingopyxis. this we TTX in cultures from four distinct genera: Aeromonas, Pseudomonas, Shewanella, and 1 and Figure LC-MS/MS screening confirmed the presence of at and to the primary and by TTX et al., 2008). For we of and ng mL−1 pure TTX and used linear to TTX concentrations in bacterial culture media from the in each LC-MS/MS run. all TTX concentrations produced were on average ± ng mL−1 (mean ± n = TTX was in samples with and our of but below our of ng These samples were not in our but this suggests that TTX production could be with of bacterial culture 1 of TTX-producing bacteria isolated from rough-skinned newts. of each TTX-producing isolate was by 16S rRNA gene sequencing and taxonomic by the Ribosomal Database Project at (Cole et al., TTX production was by screening 1 mL culture media using LC-MS/MS. Overall, we cultured TTX-producing 16S gene suggest that some isolates may represent the same bacterial In most TTX was in cultures the of but not the of Mean TTX production ± is shown for samples the Some of these bacterial genera TTX-producing strains that have previously been identified in other toxic animals in et al., 2011). (ng in other or blood ± octopus, or blood ± TTX was in independent isolates of Pseudomonas spp. cultured from newt of 16S rRNA gene suggests that these isolates may represent four bacterial strains (Figure 3—figure supplement 2). and shared across and and also shared but these two to be distinct from each other is these two were also isolated on different cultivation blood agar and respectively. 16S rRNA gene for the two Pseudomonas strains and were unique from each other and the other two Furthermore, we identified two TTX-producing spp. strains that were isolated on different media and shared 16S Thus, it several strains of Pseudomonas and can produce TTX in We also identified one individual each of and that produced TTX under these culture conditions 1). of numerous TTX-producing symbionts from distinct genera present on newt skin is with in other toxic animal as the and blue-ringed octopus, from which numerous distinct TTX-producing strains have been isolated et al., 2017). However, we that the of bacterial isolates screened in this study did not produce TTX under our culture basis of TTX resistance in newt Nav channels Rough-skinned newts are the most toxic of TTX-producing animals (Hanifin, but the molecular basis of their TTX resistance has not been To determine the basis of TTX resistance in T. granulosa, we sequenced the Nav channel gene family and investigated the the pore to determine they possessed adaptive mutations that TTX binding and We generated from two excitable tissues and from a toxic newt and for five genes that encode Nav1.2, Nav1.3, Nav1.5, and Nav1.6 We then cloned and sequenced the of each gene for including the Nav1.6 channel of both toxic and non-toxic newts. was from for comparison number pore-loop (P-loop) are highly conserved across the vertebrate Nav gene several amino acid substitutions were present in the across all six Nav channels in T. granulosa (Figure In was independently across four of the six this a replacement from an or to a amino either or In the this site is also with a and has been shown to underlie the TTX resistance of the et al., was found at in this site is also in but not in frogs, indicating that these mutations may be In only one is present at in Nav1.3, which is to the that the group of TTX et al., This region is across suggesting that the is under Three sites differed in including and in and respectively. The is present in at five Nav channel paralogs in TTX-laden pufferfishes and TTX resistance at et al., T. granulosa and have on the molecular to TTX The other two differences have not been previously but we subsequently the effects of Nav1.6 on TTX binding (below). replacements in across four sites, including a of in the in Nav1.2, in Nav1.4 and in and in Nav1.2, and Figure with 1 supplement see all Download asset Open asset of Nav channels across of from newts and other amino acid substitutions relative to the