Ecologically mediated differences in electric organ discharge drive evolution in a sodium channel gene in South American electric fishes

Abstract

Active electroreception—the ability to detect objects and communicate with conspecifics via the detection and generation of electric organ discharges (EODs)—has evolved convergently in several fish lineages. South American electric fishes (Gymnotiformes) are a highly species-rich group, possibly in part due to evolution of an electric organ (EO) that can produce diverse EODs. Neofunctionalization of a voltage-gated sodium channel gene accompanied the evolution of electrogenic tissue from muscle and resulted in a novel gene (scn4aa) uniquely expressed in the EO. Here, we investigate the link between variation in scn4aa and differences in EOD waveform. We combine gymnotiform scn4aa sequences encoding the C-terminus of the Na_v1.4a protein, with biogeographic data and EOD recordings to test whether physiological transitions among EOD types accompany differential selection pressures on scn4aa. We found positive selection on scn4aa coincided with shifts in EOD types. Species that evolved in the absence of predators, which likely selected for reduced EOD complexity, exhibited increased scn4aa evolutionary rates. We model mutations in the protein that may underlie changes in protein function and discuss our findings in the context of gymnotiform signalling ecology. Together, this work sheds light on the selective forces underpinning major evolutionary transitions in electric signal production.

1. Introduction

Active electroreception is the remarkable ability to detect and produce an electric field and has convergently evolved in two weakly electrogenic freshwater fish lineages: South American Gymnotiformes and African Mormyroidea. In Gymnotiformes, a specialized electric organ generates electric organ discharges (EODs) of varying complexity, allowing these fishes to navigate and communicate [1]. Variation in gymnotiform EODs can be used distinguish between different species [2–4]. Closely related sympatric gymnotiform species may have divergent EODs, promoting prezygotic reproductive isolation and the identification of con- versus heterospecifics [3,5]. Moreover, male and female Gymnotiformes may exhibit divergent signals [3,6,7] and also modulate their signal depending on the season [8] or reproductive value [9]. Electric signal generation poses a challenge when electroreceptive predators (e.g. the electric eel) co-occur with weakly electric fish. In these instances, evolution of more complex electric signals to mask conspicuous signal components and deter ‘eavesdropping' is likely favoured [4,10,11].

Within an electric organ, special cells called electrocytes are packed with voltage-gated sodium channel proteins that enable coordinated electric discharges. Previous work provided compelling evidence that the neofunctionalization of the scn4aa gene (coding for the Na_v1.4a protein) contributed to the evolutionary transition of skeletal muscle tissue to a highly specialized electric organ. This transition has occurred in disparate lineages of African mormyrid and South American gymnotiform electric fishes [12–16], as well as electrogenic catfish [14].

Gymnotiform fishes exhibit a wide array of electric organ and EOD specializations [1,4,17], and the relationship between EOD and Na_v1.4a function makes them an ideal system to investigate the relationship between genotype and phenotype [13,16,18]. The gymnotiform electric organ is developmentally derived from spinal motoneurons (‘neurogenic' organs in the ghost knifefish family Apteronotidae) or hypaxial muscles (‘myogenic' organs, in the remaining four families, Gymnotidae, Hypopomidae, Rhamphichthyidae and Sternopygidae [19]). Gymnotiform EODs also fall into two physiological categories: pulse-type EODs, which consist of a train of relatively short pulses with one to six phases separated by silence; and wave-type EODs comprising a higher-frequency, continuous, periodic wave [20] (figure 1a). Among Gymnotiformes with pulse-type EODs, signals span a wide range of complexity: many species produce a complex multiphasic signal, and others a simple monophasic pulse [1,4,20]. At the physiological and neuroanatomical level, variation in electrocyte size and innervation patterns, the spatial distribution of ion channels across the electrocyte membrane, diversity of ion channels and finally ion channel protein structure likely mediate this diversity in EOD types [4,19,21–23].

Figure 1.

(a) In Gymnotiformes, the electric organ is derived from either skeletal muscle tissue (myogenic) or neural tissue (neurogenic). Electric organ discharge (EOD) type also varies between wave-type and pulse-type. Within Gymnotidae and Hypopomidae, there are multiple reversions from a multiphasic to monophasic EOD. (b) Shifts in the strength of selection on scn4aa from fishes with different types of EOD. Asterisks indicate significance relative to a null model with no selective shift. Tree topology based on the scn4aa C-terminus maximum-likelihood gene tree.

Variation in EOD type has important ecological implications. The Predator Avoidance Hypothesis [10] posits that complex multiphasic signals evolved as a result of predation pressure—these signals are more cryptic to electroreceptive predators [10,24]. Crucially, several monophasic species in the family Gymnotidae live in habitats where electroreceptive predators are absent (or are themselves apex predators of other Gymnotiformes, in the case of the electric eel) and may therefore be released from the evolutionary forces promoting increased EOD complexity [4,10,11,25].

We hypothesized that the electric organ and EOD diversity outlined above would be reflected in differential selection pressure on scn4aa. Here, we conduct a molecular evolutionary analysis of the Na_v1.4a C-terminus—a crucial functional region known to contribute to the protein's fast inactivation and calmodulin binding properties [26]—from over 100 gymnotiform species. The C-terminus was selected because channel activity is likely modulated in part through this domain [27], and therefore variation in this region may reflect variable EOD properties across Gymnotiformes (the electronic supplementary material contains additional details concerning the role of the C-terminus in Na_v1.4a function [28]). Moreover, mutations in the C-terminus have been linked to abnormal channel function and neuromuscular disease phenotypes in humans, highlighting its functional relevance [29]. We tested whether EOD type (wave versus pulse), as well as the developmental origin of the electric organ (derived from neural versus muscle tissue), may affect selection pressures on scn4aa. We also tested whether transition from multiphasic to monophasic (less complex) EODs in pulse-type species is accompanied by a relaxation of selection on the scn4aa gene, and investigated amino acid substitutions in the encoded Na_v1.4a protein that may contribute to this dramatic reduction in EOD complexity.

2. Material and methods

Gymnotiform specimens spanning all five families were collected throughout South America (electronic supplementary material, figure S1 and table S1). EODs were recorded from specimens in the field (described in [3,7,9]) and collected from the literature. Classification of EOD waveforms is detailed in [4]. The portion of the scn4aa gene encoding the C-terminus of the Na_v1.4a protein was sequenced from 105 species (561 nucleotides in length), and we used molecular evolutionary models including Clade Model C (CMC) [30] in PAML [31] and RELAX [32] in HYPHY [33] to investigate shifts in selection (estimated by changes in the rate of non-synonymous to synonymous amino acid substitutions; d_N/d_S or ω) associated with differences in electric organ developmental origin (myogenic versus neurogenic) and EOD type (pulse versus wave). Among fishes with pulse-type EODs, we tested whether differences in pulse complexity (monophasic versus multiphasic) mediated selection on scn4aa. We used ChimeraX [34] and MODELLER [35] to assess the structural relevance of unique mutations in the Na_v1.4a protein from monophasic Gymnotiformes (models based on [36,37]). PolyPhen was used to infer the impact of these mutations [38] and their functional impact was also evaluated in light of previous mutagenesis and disease studies, where available. Additional details on specimens, sequencing and analyses are provided in the electronic supplementary material, and all alignments, trees and PAML results are on Dryad [39].

3. Results

The region of scn4aa that encodes the Na_v1.4a C-terminus was under positive selection in Gymnotiformes (electronic supplementary material, table S2). CMC analyses supported a shift to purifying selection in scn4aa in fishes with neurogenic electric organs (table 1 and figure 1b). Pulse-type fish have a class of scn4aa codon sites under strong positive selection relative to wave-type fish (table 1 and figure 1b). Monophasic Gymnotiformes did not exhibit a significant shift in selection in scn4aa relative to all other electric fish (i.e. both multiphasic pulse and wave-type fishes; table 1 and figure 1b).

Table 1.

Results of selection tests on gymnotiform scn4aa datasets. Np, number of parameters; lnL, ln likelihood; k, transition/transversion ratio; d.f., degrees of freedom; FG, foreground or test branch. PAML random sites results are shown in electronic supplementary material, table S2 and RELAX results in electronic supplementary material, table S3. AIC values for all models are listed in electronic supplementary material, table S4.

dataset	model	np	lnL	ΔAIC	K	parameters			null	LRT	d.f.	p-value
						ω0	ω1	ω2/ωd
all	M2a_rel	214	−6551.6	23.8	214	(72%) 0.13	(23%)1	(5%)2	M1a	8.4	2.0	0.01
	FG: wave	215	−6542.1	6.8	215	(72%) 13	(20.5%) 1	(7.4%) pulse:2.48 wave:0.62	M2a_rel	19.0	1.0	0.00
	FG: neurogenic	215	−6538.7	0	215	(73%) 12.9	(11%) 1	(17%) myogenic:1.29 neurogenic:0	M2a_rel	25.8	1.0	0.00
	FG: monophasic	215	−6550.2	23	215	(72%) 13.2	(24%) 1	(4.5%) multiphasic:1.8 monophasic:5.35	M2a_rel	2.8	1.0	0.09
pulse only	M2a_rel	128	−4364.4	9.4	3.0	(74%) 0.11	(19%) 1	(6%) 2.2	M1a	17.1	2.0	0.00
	FG: all phase types	131	−4356.7	0	2.9	(73%) 0.11	(23%) 1	(4%) monophasic:6.6 2 phases:0.4 3 phases:0.01 4 + phases:3.7	M2a_rel	30.2	3.0	0.00
	FG: monophasic	129	−4362	6.6	3	(72%) 0.11	(21%) 1	(6%) multiphasic:2.2 monophasic:5.3	M2a_rel	4.2	1.0	0.03

We isolated pulse-type Gymnotiformes to test whether independent transitions from a complex (multiphasic) to less complex (monophasic) EOD may have relaxed selective constraints on scn4aa. CMC and RELAX support elevated ω, rather than relaxed selection, in monophasic electric fish lineages (table 1; electronic supplementary material, tables S3,S4). We tested whether fish with one, two, three and four phases were each experiencing separate shifts in selection in scn4aa. This was the best-fitting model, with monophasic species exhibiting the highest ω (table 1 and figure 2b). Monophasic Gymnotidae had unusual amino acid substitutions in Na_v1.4a that do not occur elsewhere in our dataset (electronic supplementary material, tables S5,S6). We found four radical amino acid substitutions: E1606D, D1621H, G1671K and a convergent E1702K mutation in two independent Gymnotus lineages (figure 2; electronic supplementary material, table S6). Sites 1606, 1671 and 1702 were also under positive selection within the genus Gymnotus (electronic supplementary material, table S4). These substitutions are plotted on the phylogeny (figure 2a), and protein structure in figure 2c. Evidence supporting the functional importance of these substitutions is summarized in electronic supplementary material, table S6.

Figure 2.

(a) Maximum-likelihood phylogeny of the portion of scn4aa that encodes Na_v1.4a C-terminus from pulse-type Gymnotiformes depicting EOD (filled circles). Amino acid substitutions found in monophasic lineages are plotted on the tree. A tree with node support values is shown in electronic supplementary material, figure S2. (b) Differences in ω_d in scn4aa across different degrees of pulse complexity. (c) Na_v1.4a protein from electric eel [36]. Inset: Na_v1.4a C-terminus from a multiphasic pulse-type gymnotiform (purple) overlaid with a C-terminus from a monophasic gymnotiform (yellow) with unique substitutions in monophasic lineages highlighted [37]. Electronic supplementary material, table S6 provides additional details on the mutations. Numbering follows human Na_v1.4 numbering.

4. Discussion

Selective shifts in the portion of scn4aa that encodes Na_v1.4a C-terminus were associated with differences in both electric organ developmental origin and EOD structure among a diverse sample of electric fishes. Scn4aa was under purifying selection in Gymnotiformes with neurogenic electric organs (family Apteronotidae). In this group, scn4aa is not expressed in the electric organ and does not contribute to the production of an EOD; rather, it retains its function and expression in muscle tissue [40]. This scenario makes a plausible case for purifying selection, as the Na_v1.4a protein has likely not accumulated substantial variation for a repurposing of function in the electric organ.

Scn4aa is under stronger positive selection in pulse-type Gymnotiformes than in wave-type fishes. This is consistent with physiological and neuroanatomical studies of EOD as well as structural investigations of the Na_v1.4a protein. Wave-type species require a sustained sodium current to produce high-frequency waves, and have a fixed EOD rate [40], while pulse-type species modulate the rate of EOD discharge [20], and many species can switch their EOD off and on. This distinction may impose differential demands on the Na_v1.4a protein: in wave-type fishes, mutations that impair the fast inactivation of the protein may enable the production of a persistent sodium current [40], and once the protein can generate this current, the need for additional amino acid variation modulating signal may be unnecessary. Positive selection in pulse-type fishes may in part be due to greater complexity in signal form and intensity, arising from selective pressures for species recognition in diverse electric fish assemblages [3]. Pulse-type fish also tend to occupy high complexity habitats compared to wave-type fishes, which may lead to increased EOD complexity associated with electrolocation performance [41,42].

We hypothesized that within pulse-type fishes, monophasic species would exhibit relaxed selection on scn4aa due to release of selective constraint related to EOD complexity. We found the opposite, observing positive (or intensified) selection in monophasic species, likely driven by several unique amino acid substitutions. Why would reduction in EOD complexity be accompanied positive selection? One explanation may be the high energetic costs of maintaining a complex EOD postulated by Stoddard et al. [24], Lovejoy et al. [25] and Markham et al. [43]. Positive selection associated with trait reduction, and presumably reduced energetic costs, has been observed in other sensory systems (e.g. vision [44]). The predator avoidance hypothesis predicts that cryptic multiphasic signals evolve in regions of high predation by electroreceptive predators [4,10]. When electric fishes colonize a region of low predation, the putative energetic costs associated with producing a complex signal could drive the rapid accumulation of novel mutations in scn4aa that are related to the evolution of a monophasic EOD.

Mechanisms of monophasic signal production likely differ across taxa and across different levels of biological organization. For example, at the neuroanatomical level, differential innervation of electrocytes plays a role: in monophasic Electrophorus and Gymnotus obscurus, electrocytes are only electrically excitable on one face, while species with complex EODs exhibit excitability on both electrocyte faces [4,22,45,46]. By contrast, despite expressing voltage-gated sodium channels on both electrocyte faces, Brachyhypopomus bennetti produces a monophasic discharge and its electrocytes are innervated only on the posterior face [23]. At the molecular level, B. bennetti does not appear to have unique mutations in the C-terminus compared to its non-monophasic sister taxon (electronic supplementary material, tables S5,S6). However, monophasic Gymnotus species do have several non-shared and non-conservative mutations in the C-terminus that we hypothesize may be related to the evolution of monophasy. This suggests that the monophasic signals in Gymnotidae are achieved via different molecular and physiological routes than in Brachyhypopomus. In particular, E1702K is convergent in two monophasic Gymnotus lineages and E1702Q occurs in the electric eel (figure 2). Based on this convergence, evidence of positive selection at this position, and the demonstrated functional impact of changes at this site (mutations cause impaired fast inactivation in humans [47]), we regard it as the most likely substitution to be associated with monophasy. The possibility that scn4aa C-terminus evolution is more important for the evolution of monophasic EODs in some lineages than others is an interesting avenue of future research on EOD diversity.

Correlating C-terminus mutations with EOD parameters has caveats. First, while the C-terminus modulates Na_v1.4a function with respect to inactivation and calmodulin signalling [26], Na_v1.4a is a large, multidomain ion channel consisting of several other functional regions [48]. Therefore, experimental validation of C-terminus mutations in the context of the entire Na_v1.4a protein is needed because modifications in other domains could affect how individual amino acid substitutions affect function. Second, much like other sensory systems, the EOD is a highly complex phenotype determined by variation across multiple levels of biological organization. EOD variation is also modulated by EO structure [4,46], electrocyte innervation and morphology [22], ion channel repertoire and spatial distribution [49,50] and post-translational modifications of ion channels [23]. Cross-disciplinary investigation combining these variables with taxonomically extensive comparative genomics and transcriptomics [14,50–52], model-based analyses of selection on genes [53] and analyses of abiotic and biotic conditions [54], represents a powerful way to understand gymnotiform electric signal diversity moving forwards.

Ethics

This article does not present research with ethical considerations.

Data accessibility

Sequences representing the portion of scn4aa that encodes Na_v1.4a C-terminus are deposited from the GenBank repository: (accession nos. OR838936–OR839040). The alignment and gene tree used for all analyses, as well as all PAML control and outfiles are available from the Dryad digital repository: http://dx.doi.org/10.5061/dryad.7m0cfxq1x [39].

The data are provided in electronic supplementary material [28].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors' contributions

F.E.H.: conceptualization, data curation, formal analysis, investigation, methodology, visualization, writing—original draft, writing—review and editing; D.X.: data curation, formal analysis, investigation, methodology, writing—review and editing; A.V.N.: conceptualization, data curation, formal analysis, methodology, visualization, writing—original draft; K.K.B.-D.L.: conceptualization, data curation, formal analysis, investigation, writing—original draft; T.R.: data curation, formal analysis, investigation, methodology, writing—original draft; A.E.E.: data curation, writing—original draft; H.S.: data curation, writing—original draft; P.S.: data curation, writing—original draft; W.G.R.C.: conceptualization, data curation, funding acquisition, investigation, methodology, writing—original draft, writing—review and editing; N.R.L.: conceptualization, funding acquisition, investigation, methodology, project administration, supervision, writing—original draft, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

Funding was provided by UTSC and NSERC postdoctoral fellowships to F.E.H., Sigma Xi Grant-in-Aid of research and Society of Systematic Biologist Graduate Research Award to D.X., MITACS postdoctoral fellowship to A.V.N., NSF DEB-1146374 to W.G.R.C., and an NSERC Discovery Grant to N.R.L.