Molecular characterization and distribution of the voltage-gated sodium channel, Para, in the brain of the grasshopper and vinegar fly

Abstract

Voltage-gated sodium (NaV) channels, encoded by the gene para, play a critical role in the rapid processing and propagation of visual information related to collision avoidance behaviors. We investigated their localization by Immuno-staining the optic lobes and central brain of the grasshopper Schistocerca americana and the vinegar fly Drosophila melanogaster with an antibody that recognizes the channel peptide domain responsible for fast inactivation gating. NaV channels were detected at high density at all stages of development. In the optic lobe, they revealed stereotypically repeating fascicles consistent with the regular structure of the eye. In the central brain, major axonal tracts were strongly labeled, particularly in the grasshopper olfactory system. We used the NaV channel sequence of Drosophila to identify an ortholog in the transcriptome of Schistocerca. The grasshopper, vinegar fly, and human NaV channels exhibit a high degree of conservation at gating and ion selectivity domains. Comparison with three species evolutionarily close to Schistocerca identified splice variants of Para and their relation to those of Drosophila. The anatomical distribution of NaV channels molecularly analogous to those of humans in grasshoppers and vinegar flies provides a substrate for rapid signal propagation and visual processing in the context of visually-guided collision avoidance.

Introduction

A hallmark of visual processing in insects is its speed. In part, this rapidity relies on graded potential neurons. For example, large monopolar cells in the lamina perform center-surround and temporal filtering on photoreceptor signals using graded potentials (Srinivasan et al. 1982; van Hateren 1992); tangential cells in the lobula plate of dipteran flies extract behaviorally relevant optic flow parameters mostly using graded potentials (Krapp and Hengstenberg 1996; Cuntz et al. 2003). However, when urgent reactions requiring rapid long-distance propagation of information are warranted, for instance in life threatening situations such as collision avoidance with a potential predator, they are invariably activated by spiking neurons (Fotowat et al. 2011; de Vries and Clandinin 2012; von Reyn et al. 2014). Collision avoidance has been extensively studied in grasshoppers and vinegar flies (Fotowat and Gabbiani 2011; Peek and Card 2016). In grasshoppers, two identified neurons, the lobula giant movement detector (LGMD) in the optic lobe and its postsynaptic partner the descending contralateral movement detector (DCMD) in the central brain convey visual information about impending collision through their spiking activity starting from the earliest developmental stages (Simmons et al. 2013; Sztarker and Rind 2014). Yet, little is known about the localization and distribution of spike-generating NaV channels in the visual neuropils of grasshoppers, vinegar flies or other insects and, more broadly, in their central brain. In insects, most NaV channels are encoded by a single gene para (Dong et al. 2014), which in Drosophila generates more than sixty isoforms through alternative splicing and RNA editing (Loughney et al. 1989; Thackeray and Ganetzky 1994). Two different polyclonal antibodies have been previously developed against Drosophila Para, but neither of them convincingly revealed its specific localization (Amichot et al. 1993; Xiao et al. 2017). Reliable, spatially resolved antibody staining is of particular interest for grasshoppers and other non-genetic insect models. For them, there are no alternative genetic techniques to map protein expression, such as protein traps (Venken et al. 2011). Because NaV channels are highly conserved across the animal kingdom (Moran et al. 2015), we reasoned that an antibody against vertebrate NaV channels might reveal their localization in the optic lobe and central brain of the grasshopper and vinegar fly. We thus used a mouse monoclonal pan-Nav antibody known to broadly stain for NaV channels in the mammalian brain (Rasband et al. 1999). The robust immune-staining pattern of NaV channels observed in the grasshopper and vinegar fly brain prompted us to search for an ortholog of the Drosophila Para protein in grasshoppers. By analyzing RNA sequencing (RNA-seq) data from the thorax and head of the grasshopper Schistocerca americana using one of the Drosophila Para isoforms, we assembled a full-length sequence. We compared this sequence to human Nav1.1, to the Drosophila Para protein isoforms, and to sequences from three more closely related Polyneopteran species, the German cockroach, the drywood and dampwood termites.

Material and Methods

Animals

We used the grasshopper Schistocerca americana, a common species found in the southeastern United States (Kuitert and Connin 1952). Grasshoppers are hemimetabolous insects developing through several nymphal instars (stages) that are similar to their adult form except for the lack of wings and reproductive organs (Supp. Fig. 1a). This contrasts with the development of holometaboulous insects such as Drosophila that pass through several larval instars and a pupal stage prior to reaching adulthood (Supp. Fig. 1b). S. americana individuals used for immunostaining were taken from a crowded colony maintained at Baylor College of Medicine under a 12 hours light/dark cycle (32/25 °C). They originated from individuals collected near Gainesville, Alachua County, Florida, in November 2002. They were fed wheat grass and oat bran three times per week. S. americana individuals used for transcriptome sequencing were from a colony maintained at Texas A&M University under a 12 hours light/dark cycle at 30 °C. They originated from a population collected in Brooksville, Pasco County, Florida, in September 2010. Individuals were reared under isolated and crowded conditions as described in Gotham and Song (2013) for an unrelated project on density-dependent phenotypic plasticity (unpublished data). Wild-type Drosophila melanogaster were a laboratory stock of Canton S flies. GMR42F06 flies (Pfeiffer et al. 2008) expressing GAL4 in medulla/lobula T4/T5 neurons of the optic lobe were obtained from the Bloomington Drosophila Stock Center (BL#54203) and crossed to a strain expressing GFP from a 10x UAS promoter (Pfeiffer et al. 2012; a gift from Dr. K. Venken, Baylor College of Medicine). All flies were reared on yeast, cornmeal, molasses, and agar food at room temperature (22.5 ± 0.5°) on a 16-hr light/8-hr dark cycle.

Antibodies

The specific monoclonal mouse pan-Nav antibody was generated against a synthetic peptide containing the sequence TEEQKKYYNAMKKLGSKK, which corresponds to a highly conserved segment of the intracellular lll-IV loop of vertebrate sodium channels (Rasband et al. 1999), and is available from Sigma (catalog # S8809). For 5-HT (serotonin) staining we used a whole antiserum antibody produced in rabbits (Sigma-Aldrich, catalog # S5545) and for GFP staining a rabbit polyclonal antibody (Invitrogen, catalog # A11122).

The specificity of the pan-Nav antibody has been verified according to several criteria. First, it labels axon initial segments and nodes of Ranvier along myelinated axons of vertebrates, sites that are known to be enriched in Na⁺ channels (see e.g., Rasband et al. 1999; Hedstrom et al. 2007). Second, the pan-Nav antibody recognizes a protein of the correct molecular weight by immunoblot. Third and most importantly, silencing expression of Na⁺ channels in hippocampal neurons using a pan-Nav shRNA construct eliminates all immunoreactivity in the targeted cells (Hedstrom et al. 2007, Fig. 1e). These criteria are consistent with accepted standards for the validation of antibodies in nervous system tissues (Rhodes and Trimmer 2006). For additional examples of controls using this antibody, see references available at antibodyregistry.org (antibody number AB_477552).

Figure 1.

Pan-Nav protein expression pattern in the optic lobe of grasshoppers, a-f: Expression in the optic lobe of a 1^st instar (a), 2^nd instar (b), 3^rd instar (c), 4^th instar (d), 5^th instar (e) and an adult (f). Abbreviations: Me, Medulla; Lo, Lobula. In a-c ‘>‘ indicates axonal fibers going from the lobula of the optic lobe to the central brain. Scalebar: 100 μm (a-f).

Immunostaining

Schistocerca americana has five nymphal instars (Capinera 1993). We selected individuals at all developmental stages for staining with the pan-Nav antibody. Grasshoppers were immobilized and fixed to a plastic holder for dissection of the brain and optic lobes. During the dissection, the whole head was bathed in ice-cold saline. After dissection, the brains of first to fifth instar grasshoppers were initially fixed in freshly made 4% paraformaldehyde (PFA, Electron Microscopy Science, catalog # 15710) for one hour. Subsequently, we used shorter fixation times for fourth and fifth instar grasshoppers to improve staining results, as described in the next section. The tissues were then washed six times for 20 minutes using 1x phosphate buffer solution (PBS, 0.01 M PO₄³⁻; made from 10x PBS, Thermo Fisher Scientific, catalog # Am9624) containing 0.5% Triton X-100 to promote penetration of the antibodies (Sigma-Aldrich, catalog # 93443). Next, goat serum was added to block non-specific antigens (final concentration: 5%; Thermo Fisher Scientific, catalog # 16210064) and tissues were incubated for one hour. Antibodies were diluted in the same solution. The pan-Nav antibody was incubated with the tissues for 48 hours at a dilution of 1:350 to 1:800 followed by regular washing (six times, as above). Goat anti-mouse tagged with Alexa Fluor 594 was used as secondary antibody at a dilution of 1:1000 (Thermo Fisher Scientific, catalog # A21125). Tissues were incubated with this secondary antibody for 48 hours at 4 °C followed by regular washing. To further reduce non-specific binding, the tissues were washed for another 48 hours. Finally, the tissues were mounted on a glass slide, cover-slipped and sealed with nail polish.

For adult grasshoppers, the brain and optic lobes were initially fixed overnight in ice cold, freshly made 4% PFA. Subsequently, we used shorter fixation times to improve the staining results, see next section. The tissues were then washed in 10x PBS (0.1 M P0₄³⁻) for 10 min. To increase antibody penetration, adult tissues were incubated in 1 mg/ml collagenase (Sigma-Aldrich, catalog # 10269638001) diluted in 5x PBS. For permeabilization of cell membranes and blocking, we used 1x PBS with 0.3% Triton X-100 and 5% goat serum in solution (PBTGS) overnight at 4 °C. After washing out the PBTGS solution, the primary antibodies (either pan-Nav or 5-HT) were added diluted in PBTGS (1:350 to 1:800 Pan-Nav; 1:5,000 5-HT), leaving them for 4–5 days at 4 °C. The primary antibody was washed off three times with PBTGS (20 min, 20 min, and 1 h washes; 5 min between washes). The secondary antibodies were added in PBTGS (1:1000 Alexa Fluor goat anti-mouse IgG 1 for pan-Nav; 1:500 chick anti-rabbit tagged with Alexa Fluor 488 for 5-HT, Thermo Fisher Scientific, catalog # A21441) for three days at 4 °C. Secondary antibodies were washed off three times in PBTGS, followed by 10x PBS and 5x PBS (5 min between washes). The tissue was dehydrated in an ascending ethanol series (25, 50, 75, 90, 95, 100%, 15 min each), cleared in methyl salicylate (20 min; Sigma-Aldrich, catalog # M6752), embedded in Permount (Fisher Scientific, catalog # SP15–100) and cover-slipped on a slide with a concave well (3 mm thick, 16 mm in diameter concavity, 0.8 mm deep, United Scientific Supplies, catalog # CSTK02).

For Drosophila, we used adult brains and ventral nerve cords, as well 3^rd instar larvae. In our initial protocol, adults were anesthetized on ice, then dissected and fixed in freshly made 4% PFA for one hour. The brains were subsequently washed and stained as described above for the grasshopper nymphal instars following established protocols for Drosophila (Wu and Luo 2006). Drosophila brains were mounted as previously described (Gnerer et al. 2015) in SlowFade Gold antifade reagent (Molecular Probes Life Technologies, catalog # S36937).

Additional immunostaining protocols

In rats and mice, it has been observed that staining with pan-Nav yields better results with short formaldehyde fixation times, presumably because the six lysines present in the antigen make the protocol highly sensitive to this parameter (personal communication, Dr. M. Rasband; Dapson 2007). In grasshoppers, we thus tried shorter fixation times of 10, 30 or 60 min in 4% fresh PFA at 4 °C. The protocol was otherwise identical to that described above. Additionally, the optic lobes were cryo-sectioned at a thickness of 10–25 μm and the sections were transferred onto coverslips for antibody staining following the protocol described in (Rasband et al. 1999). Briefly, primary and secondary antibody staining lasted 60 min at room temperature, separated by three washes (5–10 min) in PBTGS. The antibody concentrations were identical to those reported above. In another variant, we embedded the optic lobe in agar at 42 °C and sectioned it at a thickness of 200 μm after hardening. We then followed the protocol described above for the brain and optic lobes of first to fifth instar grasshoppers. In another variant, we incubated the whole tissue in 0.5% saponin either with or without collagenase in goat serum for increasing penetration and blocking non-specific binding, respectively (see above). The final variant used 0.1% saponin (Sigma-Aldrich, catalog # 558255) and 0.5% Triton-X added to the washing solution used after tissue fixation. For the same reason, we tried 1, 2, 5, 10, and 15 min fixation times in Drosophila, with the protocol otherwise identical to that described above. We obtained the best results with short fixations times of 5 min in Drosophila and 10 min in grasshopper. It is likely that these short fixation times led to under fixation of the tissue (Dapson 2007; Ramos-Vara and Miller 2014; see Discussion).

Imaging

The brains and the optic lobes were typically imaged under an upright microscope with 10x, 20x, or 40x objectives using a structured illumination module (Zeiss Plan-Neofluar-10x/0.3, Plan-Neofluar-20x/0.5, Plan-Neofluar-40x/0.75, ApoTome and ZEN software). In two cases, Fig. 1f and 2f, we used a custom built two-photon microscope with a 10x/0.30W Nikon Plan Fluor objective (Zhu and Gabbiani 2018). To image the Alexa 488 and 594 fluorescent dyes we used EGFP and Texas Red filter sets (Kramer Scientific, catalog # KSC 295–833 and KSC 295–818). The characteristics of the excitation, dichroic and emission filters for EGFP and Texas Red can be found on the Chroma Technology web site (EGFP parts # HQ480/40X, 505DCXT, and HQ510LP; Tx Red parts # ET560/40X, T585LP, and ET630/75M, respectively). The orientation of the preparation was optimized for imaging purposes; the resulting micrographs were rotated post acquisition to standardize their orientation and the figure background was filled with black pixels, resulting in the occasional presence of sharp oblique edges (e.g., Fig. 1f). The images shown in Figs. 1,2 and 3 and Supp. Figs. 2, 3 and 4 are maximum intensity projections from ~10 to ~60 slices through the optic lobe or whole brain of the animal. To generate images of the fly optic lobe from a perspective showing the lobula and lobula plate (Supp. Fig. 4f), image stacks were imported into Vaa3D and rotated using the Neuron Annotator mode (Peng et al. 2010). After manual orientation, images were exported as TIFF format screenshots (see also Wu et al. 2016). Movies were assembled from stacks of images acquired with the following dimensions (x, y and z): 0.650 × 0.650 × 4.55 μm (Supp. Movie 1), 0.325 × 0.325 × 1.35 μm (Supp. Movies 2, 5), or 0.163 × 0.163 × 0.600 μm (Supp. Movies 3, 4).

Figure 2.

Pan-Nav protein expression pattern in the central brain of grasshoppers, a-f: Expression in the central brain of a 1^st instar (a), 2^nd instar (b), 3^rd instar (c), 4^th instar (d), 5^th instar (e) and an adult (f). Abbreviations: Ca, Calyx of the mushroom body; Pe, Peduncle of the mushroom body; CC, Central Complex; AL, Antennal Lobe; AN, Antennal Nerve. Scalebar: 100 μm (a-f).

Figure 3.

Pan-Nav protein expression pattern in the brain and ventral nerve cord (VNC) of adult Drosophila, a: Whole brain image of an adult Drosophila, b: Close-up of the optic lobe, c: Close-up image of the central brain, d: Image of the adult VNC. In a-d fixation time was 5 min. In c, d arrowheads show representative fascicles in resolved brain areas. Abbreviations: ped, peduncle of the mushroom body; LP, Lateral Protocerebrum; SEZ, Sub Esophageal Zone; fsb, fan-shaped body; VNC, Ventral Nerve Cord. Scalebar: 50 μm (a-d).

Annotation

To identify brain structures we consulted, respectively, the brain atlas of Schistocera gregraria, a species closely related to Schistocerca americana, and that of Drosophila (Strausfeld 1976; Kurylas et al. 2008; El Jundi et al. 2010; Ito et al. 2014).

Sequencing grasshopper thorax and head RNA

To standardize the developmental stage specimens, we used last nymphal instar females that molted in the morning. They were dissected between 8 and 9 AM, roughly 72 hours after molting. We preserved head and thorax tissues at −20 °C in RNAIater (Qiagen, Valencia, CA). A total of ten specimens, five isolated and five crowded individuals, were dissected. RNA was extracted using a Trizol-chloroform isolation protocol, followed by clean-up with a RNeasy mini kit, both following manufacturer’s guidelines (Thermo Fisher Scientific, Waltham, MA and Qiagen, Valencia, CA, respectively). RNA purity was quantified using a spectrophotometer (DS-11, DeNovix, Wilmington, DE). Specifically, RNA extracts were used for sequencing if the ratios of 260 to 280 nm and of 260 to 230 nm absorbance values were above 2 (Imbeaud et al. 2005). Additionally, RNA integrity was analyzed by microcapillary electrophoresis with a Fragment Analyzer and ProSize software (Agilent Technologies, Ankeny, IA). RNA Quality Number (RQN) values over 3.9 were used for RNA-sequencing. This cut-off is lower than what is used in mammalian samples (Escobar and Hunt 2017). However, RQN values can vary widely across insect species due to differences in 28S ribosomal RNA structure compared to other eukaryotic species (Winnebeck et al. 2010; Macharia et al. 2015; Fabrick and Hull, 2017, Assessing integrity of insect RNA, Agilent Application Note). The samples were processed as part of a larger RNA-seq study of density-dependent phenotypic plasticity in Schistocerca, which generated a total of 80 sequenced samples. We used lllumina’s TruSeq Stranded Total RNA Library Prep Kit for library preparation and performed paired-end sequencing (150 bp) using 8.5 lanes on an Illumina HiSeq4000 (San Diego, CA). The library preparation and sequencing were performed at Texas A&M’s AgriLife Research Genomics and Bioinformatics Service. Sequence cluster identification, quality prefiltering, base calling and uncertainty assessment were all done in real time at the genomics core using lllumina’s software (HCS 2.2.68; RTA 1.18.66.3; default parameter settings). Demultiplexing base call files and formatting them into FASTQ files was carried out using lllumina’s bcl2fastq script (version 2.17.1.14).

Post-sequencing processing

For further processing, raw reads were imported into a personalized Galaxy environment (Afgan et al. 2018) on a supercomputing cluster of the High-Performance Research Computing group of Texas A&M University (Ada, https://hprc.tamu.edu). Reads were transformed to Sanger format using FastQ Groomer (Blankenberg et al. 2010) and were subsequently filtered with Trimmomatic (Bolger et al. 2014). The filtering consisted of cutting off any bases at both ends if their quality score was lower than 30, trimming with a sliding window of 3 bases and a minimum average quality score of 30, and finally discarding all reads of less than 30 bp. Additionally, FastQ Screen (Wingett and Andrews 2018) was used to filter out reads from bacterial and other contaminating sources.

Transcriptome assembly

The filtered reads were used for transcriptome assembly using Trinity (Grabherr et al. 2011; default settings, in silico normalization enabled). Transcriptomes were assembled separately for head and thorax tissue. Similar contigs were removed using CD-hit-EST with a threshold of 0.9 (Li and Godzik 2006; Fu et al. 2012). Finally, an extra filtering step was performed with Transrate (Smith-Unna et al. 2016), and any contig scoring below the suggested cutoff-score was filtered out. The final transcriptomes had a total of 342,815 and 305,904 sequences for head and thorax, respectively. Trinitystats (Grabherr et al. 2011) was used to calculate the guanine-cytosine (GC) content (40.86% and 41.17% for head and thorax, respectively) and the contig N50 (597 and 622, respectively). The fraction of reads used for transcriptome assembly that mapped back to the transcriptome was assessed using bowtie2 (Langmead et al. 2009; Langmead and Salzberg 2012). We used the preset mode ‘very sensitive, end-to-end’ and flagstat (Li et al. 2009; Li 2011a, b; see also SAMtools manual page, available at htslib.org). The fraction of reads mapped back to the head and thorax transcriptomes was 89 and 85.2%, respectively. Lastly, the completeness of the transcriptome was assessed with BUSCO (Benchmarking Universal Single-Copy Orthologs; Simão et al. 2015) by looking for single-copy orthologs of the lineage Insecta. For both transcriptomes the amount of complete BUSCOs was over 96% and the amount of single copy BUSCOs in both transcriptomes was over 80%.

Schistocerca americana Para sequence assembly

The assembled transcriptome sequences were imported into Geneious (R10.2.6; BioMatters, Ltd., Auckland, New Zealand). We searched for the para gene in the S. americana transcriptome data using as reference different isoforms of the paralytic locus (para) of Drosophila melanogaster with the program Megablast (E-value 10⁻⁵, word size of 28, scoring of respectively 1 and −2 for Match and Mismatch, and low complexity filter on). The resulting partial sequences were manually combined and curated based on the orthologs in D. melanogaster and the cockroach, B. germanica. Where needed, sequence information was inferred from three closely related grasshopper species that were sequenced in the same RNA sequencing project (S. serialis cubense, S. piceifrons, S. nitens, unpublished data; https://orthoptera.speciesfile.org). Raw reads for each specimen were mapped back to the full transcriptome in Galaxy using bowtie2 (very sensitive end-to-end, disable no-mixed and no-discordant behavior). Mapping statistics were assessed using SAMtools’ idxstats (Li et al., 2009). Transcript expression levels were calculated using the Transcripts per Million (TPM) normalization method in R (version 3.5.2; www.r-project.org). Because the sequence from exons 1 to 17 was fragmented into several small pieces, the comparison of transcript levels between head and thorax was based on the sequence from exon 18 to 31.

Multiple sequence alignment and splice form characterization

We performed sequence alignments between human Nav1.1 and different insect Para sequences using the software program SnapGene (GSL Biotech; available at snapgene.com). The transmembrane domains and pore forming domains delimited by light blue boxes in Figs. 3 and 4 where obtained from the annotation of the Human SCN1A protein (Nav1.1) in Uniprot (www.uniprot.org, P35498). The corresponding domains in the D. melanogaster Para protein often differ by a few amino acids at each of their ends (P35500). We used the FlyBase listing of Drosophila Para isoforms to construct an exon/intron map of the para locus. To assess whether other splice forms exist in Schistocerca, we mined our transcriptome data using the reconstructed Drosophila optional and mutually exclusive exons as queries. In addition, we examined whether three closely related insects, the German cockroach (Blattella germanica), the drywood termite (Cryptotermes secundus), and the dampwood termite (Zootermopsis nevadensis) showed evidence for splice forms that contain any of these alternate exons. Cockroaches, termites (Blattodea) and grasshoppers (Orthoptera) all belong to the Polyneoptera (Wipfler et al. 2019), and are thus phylogenetically closer to each other than to Drosophila (Diptera). Several cDNA sequences corresponding to the para locus from B. germanica have been isolated and deposited in the NCBI GenBank database (Miyazaki et al. 1996; Dong 1997; Suzuki and Yamato 2018). The B. germanica, C. secundus, and Z. nevadensis genomes and transcriptomes were recently sequenced and assembled (Terrapon et al. 2014; Harrison et al. 2018).

Figure 4.

NaV channel sequence alignment in Homo sapiens, Drosophila melanogaster and Schistocerca americana from the N terminus up to the second domain of the protein’s 6^th transmembrane segment. The top multicolor-coded line identifies the degree of amino acid consensus among the three species sequences. Hot colors (e.g., carmine) signify high consensus and cold colors (e.g., deep blue) low consensus (color bar at bottom of Fig. 5). The second line provides the consensus sequence among the three species. The next line reports the human Para protein sequence (HsapNav1.1), followed by that of D. melanogaster (DmelPara_PX) and that of S. americana (SamePara). Residues highlighted in yellow are identical across the three species. The sequence portions highlighted in blue correspond to the six transmembrane segments, S1–S6, and the pore region (Pore) for the first two of four domains, Doml–DomlV, of the NaV channel protein (bottom right inset in Fig. 5). Triangle: first two nucleotides constituting the inner pore ring; star: same for outer ring. Green arrows denote the start of the obligatory, optional and mutually exclusive exons in the para gene and mRNA. Numbers on right are cumulative amino acid residue counts for each species.

Results

pan-Nav staining in the grasshopper optic lobe reveals axon fascicles

We used an antibody that recognizes a conserved epitope, pan-Nav, across all vertebrate NaV channel family members to examine whether we could stain for the major NaV channel in the optic lobes of grasshoppers. We observed specific fascicular staining patterns in the optic lobes of each instar and in the adult of S. americana (5, 5, 5, 2, 4 and 6 animals for 1^st, to 5^th instar and adults, respectively, Fig. 1 a–f). Staining was relatively diffuse in 1^st instars (compare Fig. 1a with Figs. 1b, c), but a regular arrangement of vertical fascicles converging towards the lobula was nonetheless clearly visible at the level of the medulla. In contrast, little specific staining was observed in the lamina. These fascicles likely correspond to axons of descending neurons and their regular arrangement conforms with the modular structure of the medulla, which is subdivided in cartridges processing local information originating from each ommatidium (facet) of the compound eye (Strausfeld 2009). In addition, several larger bundles of axons originating at the level of the lobula of the optic lobe are visible (Fig. 1a–c, >). In the second and third instars the pattern of fascicles was pronounced, and the convergence of these axon bundles was evident from the strong staining at the level of the lobula. In contrast, staining in fourth instars was diffuse and no clear pattern indicative of axons fascicles or axon bundles was reproducibly observed across most of the optic lobe when the brain was fixed for 60 min (Supp. Fig. 2a; 4 animals). Optic lobe staining was also diffuse in 5^th instars and adults (4 and 5 animals, respectively; Supp. Fig. 2b, d and e). After trying several variants of the staining protocol, we found that a short, 10 min fixation yielded satisfactory stainings in 4^th instars and adults, as well as particularly good ones in 5^th instars, generally reproducing the features observed at earlier developmental stages (Fig. 1 d–f).

pan-Nav staining in the grasshopper central brain clusters along major axonal tracts

Next, we looked at the pattern of pan-Nav staining in the central brain. Fig. 2 shows examples of 1^st to 5^th instars and of an adult. Similar results were observed in the same animals used for optic lobe staining. Fig. 2a shows the localization of pan-Nav staining in the central brain of a 1^st instar. Heavy staining is immediately apparent in the mushroom bodies, a higher center for learning and memory of the olfactory system (Strausfeld 2009). The staining pattern was consistent with the anatomy of the calyx and peduncle of the mushroom body, which contain the axons of an estimated 50,000 Kenyon cells (Jortner et al. 2007). Faint staining was also visible in the central complex. Additional axon bundle tracts were visible running down towards the tritocerebrum. In the 2^nd instars (Fig. 2b) the peduncle and calyces of the mushroom bodies were stained heavily, and the fan-shaped body was more clearly outlined than in the 1^st instar brain. Supp. movie 1, assembled from an image stack acquired across the central brain, sharply illustrates the three-dimensional structure of additional bundle tracts in the central brain. A similar pattern of staining was observed in the 3^rd instars (Fig. 2c) but here we observed a clear outline of the antennal lobes based on a series of stained fascicles inside and around them. In the 4^th instars staining could still clearly be seen in the mushroom bodies and the antennal lobes, although axon tracts were not as well resolved (Fig. 2d). Additionally, the olfactory antennal nerves were strongly stained, consistent with the presence of a large number of olfactory receptor neuron axons. This resolved pattern of staining contrasted with the diffuse pattern observed in the optic lobe of the 4^th instars when fixation lasted 60 min (see above). In the 5^th instars, only diffuse staining was observed in the central brain when fixing for the same duration (Supp. Fig. 2c; 4 animals). Similar results were observed in adults (5 animals; not shown). The lack of specific staining in later instars and adults might have been caused by decreased penetration of the antibody in thicker brain tissue (i.e., decreased diffusion into the fixed tissue, Thurber et al. 2008). However, this appears unlikely given the differential staining of the central brain and optic lobes in the 4^th instars. To further rule out this possibility, we carried out simultaneous pan-Nav and 5-HT (serotonin) immunofluorescence staining in adult grasshoppers (2 animals). While the optic lobes showed diffuse staining to pan-Nav (see above), we could resolve staining to 5-HT (Supp. Fig. 2d, e). In contrast, a short fixation time of 10 min largely reproduced in 5^th instars and in adults the results observed with longer fixation times in 1^st to 4^th instars (Fig. 2e, f; 4 animals).

Short fixation time reveals matching Para localization pattern in vinegar flies

As in adult and 5^th instar grasshoppers, staining either Drosophila adults (Supp. Fig. 3a; 6 animals) or larvae (Supp. Fig. 3d; 5 animals) with pan-Nav following 60 min fixation showed predominantly diffuse patterns, although part of the mushroom body calyces and motor neuron axons could be resolved in L3 larvae. Based on the observations made in grasshoppers, we tested successively shorter fixation times. With 15 min fixation, a slight improvement was observed in adults (Supp. Fig. 3b; 3 animals). The same held true for L3 larvae (Supp. Fig. 3e; 3 animals). In contrast, we observed in both cases well resolved neurite projections when staining for 5-HT using the same fixation times (Supp. Fig. 3c, f; 3 and 3 animals, resp.). An even shorter, 5 min fixation resulted in a large number of well resolved anatomical features in adults (Fig. 3a; 11 animals, Supp. Movie 2). Imaging the optic lobes at higher magnification, revealed the presence of fascicles similar to those observed in grasshoppers at the level of the medulla but none in the lamina (Fig. 3b; c.f. Fig. 1). Individual images from acquired stacks showed different types of transmedullary neuron stained in the optic lobes (Supp. Movie 3). To further resolve the identity of the stained neurons, we stained brains expressing GFP in medullary T4 and lobula T5 neurons using GMFt42F06-GAL4>UAS-GFP flies (Jenett et al. 2012; Maisak et al. 2013; Thurmond et al. 2019). Although we saw some overlap between the GFP signal and pan-Nav, we also observed Para localization outside of the T4/T5 neurons (Supp. Fig. 4a, b, Supp. Movie 2, 3). In the central brain, the mushroom body peduncles were apparent as well as a large number of thick fascicles and commissures (Figs. 3a, c). In the maximum intensity projection of the brain some fascicles belonging to the central complex could be resolved, such as the fan-shaped body (Fig. 3c). In a movie of stacked projections these fascicles outlining the entire central complex were very well resolved as well as large fascicular bundles and commissures throughout the entire central brain (Supp. Movie 4). Fascicles were also clearly visible in the lateral protocerebrum and the sub esophageal zone. The overall pattern of these bundles in the central brain was reminiscent of the intricate pattern documented in Musca and Drosophila by classical anatomical methods (Strausfeld 2009; Ito et al. 2014). We also stained the ventral nerve cord of the adult and could visualize fascicles throughout although they are less pronounced than in the brain (Fig. 3d, Supp. Movie 5). In L3 larvae, fascicular bundles became visible in the ventral nerve cord and the central brain when fixation times were reduced to 5 min although they were less well resolved than in the adult brain (Supp. Fig. 3g). Shortening the fixation time further to 2 min did not improve the results in larvae (Supp. Fig. 3h, 5 animals). In all of the larval and adult Drosophila brain preparations, we observed staining in cell bodies localized to the cortex surrounding each neuropil. We did not observe such somatic signals in the grasshopper brains. The fascicular stainings described above closely match those recently reported in an endogenously GFP-tagged Para reporter strain (Ravenscroft and Bellen, personal communication, 2019 Neurobiology of Drosophila Cold Spring Harbor Meeting; Venken et al. 2011).

High level of homology between Drosophila and Schistocerca sequences

Given the robust staining observed in grasshoppers and vinegar flies, we investigated the molecular composition of NaV channels in grasshoppers. We used the Para isoform sequences of Drosophila to identify potential orthologs in the head and thorax transcriptome of S. americana. This allowed us to assemble a complete S. americana Para sequence depicted in Figs. 4 and 5. The sequence was aligned with that of Drosophila melanogaster’s closest isoform, Para PX, and with that of Nav1.1, one of Homo sapiens’ NaV channel orthologs expressed in central neurons. We used the known location of the transmembrane segments S1-S6 for each of the four domains of the channel (Doml-DomlV) in Drosophila and Homo to identify their location in Schistocerca. Fig. 4 shows that segments S1-S6 of the first two domains are highly conserved between S. americana and D. melanogaster, with only 6 out of 233 amino acids differing (2.6%). Similarly, segments S1-S6 of the last two domains differed from those of D. melanogaster by only 9 out of 230 amino acids (3.9%; Fig. 5). In addition, the Na⁺ selectivity filter amino acid sequence ‘DEKA’ which forms an inner ring distributed across the four pore domains was conserved across all three species (Heinemann et al. 1992; Catterall 2000; Figs. 4, 5 arrowheads). In contrast, the methionine of the vertebrate outer ring sequence ‘EEMD’ involved in channel permeability (Chiamvimonvat et al. 1996) and sensitivity to TTX (Terlau et al. 1991) was substituted by an isoleucine in S. americana, as is the case in D. melanogaster (Figs. 4, 5, star). In vertebrates, a short hydrophobic amino acid sequence ‘IFM’, located in the intracellular loop between the last two transmembrane domains, plays a critical role in fast inactivation of the channel through interactions with domain IV (Fig. 5, horizontal square bracket; (West et al. 1992; Catterall 2000; Capes et al. 2013; Pan et al. 2018; ClairfeuilIe et al. 2019). The equivalent insect amino acid sequence is MFM in Drosophila and several other insects (Dong et al. 2014). In contrast, the first amino acid is substituted by leucine in Schistocerca, which is closely related to the above-mentioned isoleucine (vertebrates) and methionine (insects). Immediately downstream of this tripeptide lies the peptide sequence used to generate the pan-Nav antibody, which was identical in S. americana and D. melanogaster and differed only by two of eighteen amino acids from that in H. sapiens (Fig. 5). Outside these transmembrane and pore regions, the sequence was less conserved, with the highest variability observed at the amino and carboxyl termini, the segment between domains I and II, and the segment between domains II and III, as is also the case in vertebrates (Soderlund 2005). On average, there was 75% identity between D. melanogaster and S. americana, which is on the lower end of that observed in other insects that have been compared to D. melanogaster (e.g., Anopheles gambiae 82%, Davies et al. 2007; Periplaneta americana 81%, Moignot et al. 2009; Bombyx mori 74%, Shao et al. 2009). Similarly, homology between S. americana Para and H. sapiens Nav1.1 was on average 43.2%, while that of D. melanogaster was 41.6%.

Figure 5.

NaV channel sequence alignment in H. sapiens, D. melanogaster and S. americana from the second domain of the protein’s 6^th transmembrane segment up to the C terminus. The data are presented as in Fig. 4. From top to bottom: the multicolor-coded consensus, the consensus sequence, the human, Drosophila and Schistocerca sequences. As in Fig. 4, the sequence portions highlighted in blue correspond to S1-S6 and Pore, Doml-DomlV of the protein. The sequence highlighted in pink is the peptide sequence used to generate the Pan-Nav antibody. The bottom right inset illustrates schematically the secondary structure of the channel from the N-terminus of the first domain to the C-terminus of the fourth domain. Triangle, star symbols are last two nucleotides of inner and outer pore rings, respectively. Horizontal square bracket indicates inactivation ‘lid’. Green arrows and numbers on right as in Fig. 4.

As we had transcriptomes for both head and thorax tissue, we compared their para expression levels. The expression of the para gene was 5.2 times higher in head than in thorax tissue in last instar nymphs (TPM values of 0.344 and 0.067, respectively). These values show that para is a lowly expressed gene: in comparison, the actin gene has TPM values of 104.83 and 181.45, respectively, for head and thorax tissue. This is likely due to specific expression of para in neural tissue, while our transcriptomes originated from whole head and whole thorax tissue.

para splice variants in Schistocerca, Blatella, Cryptotermes, and Zootermopsis

In Drosophila, a large number of para splice variants have been reported (Loughney et al. 1989; Lin et al. 2009; Dong et al. 2014; Thurmond et al. 2019). To characterize those in grasshoppers, we started by compiling a list of the complete amino acid sequences of obligatory, optional and mutually exclusive exons of the Drosophila para locus using the Para isoform sequences available in FlyBase (Thurmond et al., 2019). Fig. 6a shows the structure of the Drosophila para locus. Obligatory coding exons are numbered from 3 to 31 (the first two exons are non-coding). In addition, the locus contains 7 optional exons denoted by the letters j, i, a, b, e, f and h (Lin et al. 2009; Dong et al. 2014) as well as two pairs of mutually exclusive exons denoted by the letters c, d (pair 1) and k, L (pair 2), respectively. Table 1 lists the amino acid sequence of all these exons and their corresponding base pairs in the para locus. The Drosophila Para isoform PX depicted in Figs. 4 and 5 contains all numbered obligatory exons and alternative exons j, i, a, c, h and L. In these two figures, a right-angled green arrow indicates the start location of each exon by its associated label. For example, immediately after exon 3 at the start of the sequence, the second coding exon of Para PX is exon j, which starts with the amino acid phenylalanine (Fig. 4 top line and Table 1). The mapping between the 60 Para isoforms documented in FlyBase and the alternatively spliced exons of para is given in Table 2. FlyBase shows isoforms that vary in exon 7, encoding part of transmembrane segments 2 and 3 in domain I of the channel (Fig. 4, second line). It also contains five isoforms of the channel truncated after exon 6, exon d or exon 18 (PAR, PALI, PAV, PBC and PBE, Table 2). Fig. 6b maps Para domain transmembrane segments and the pan-Nav sequence onto specific exons of the Drosophila para locus.

Figure 6.

a: Diagram of the intron-exon structure of the para gene in Drosophila (boxes represent coding exons). Optional exons are shown in yellow and mutually exclusive exons are shown in orange, b, c: Transcript corresponding to a splice form containing all the optional exons in Drosophila (b) compared to the para splice form in Schistocerca (c) assembled from our RNA-seq data. The localization of the transmembrane domains and pan-Nav sequence are depicted below the exons in blue boxes and a pink box, respectively, d: The secondary Para protein structure of the different domains corresponding to the transcripts above is shown in cartoon form, depicting in color the optional and mutually exclusive exons as yellow boxes and orange barrels, respectively. The blue boxes represent optional exons in Drosophila that are obligatory in Polyneoptera (exons i, a and h). The green boxes represent optional exons that are species specific (exon g in Polyneoptera and exon f in Drosophila). Of the mutually exclusive exons c/d, only exon c is present in Polyneoptera and no isoforms with exon d were detected (denoted by an asterisk in the diagram). The inactivation gate and adjacent pan-Nav sequences as well as the selectivity filter and outer ring are also shown on the cartoon (see labels). For detailed sequence information see also Table 1.

Table 1:

Drosophila melanogaster Para coding exons

	Exon Sequence	Start	End	Size	Intron Size
3	MTEDSDSISEEERSLFRPFTRESLVQIEQRIAAEHEKQKELERKRAEGEV	17,349	17,498	191	2,795
j	PQYGRKKKQKE	20,284	20,316	33	4,620
5	IRYDDEDEDEGPQPDPTLEQGVPIPVRLQGSFPPELASTPLEDIDPYYSNVL	24,937	25,092	156	109
6	TFVVVSKGKDIFRFSASKAMWMLDPFNPIRRVAIYILVHPLFSLFIITTILVNCILMIMPTTPTVESTE	25,201	25,406	206	4,011
7	VIFTGIYTFESAVKVMARGFILCPFTYLRDAWNWLDFVVIALA	29,417	29,545	129	3,135
8	YVTMGIDLGNLAALRTFRVLRALKTVAIVP	32,682	32,773	92	100
9	GLKTIVGAVIESVKNLRDVIILTMFSLSVFALMGLQIYMGVLTQKCIKKFPLDGSWGNLTDENWDYHNR NSSNWYSEDEGISFPLCGNISGAG	32,874	33,150	277	155
10	QCDDDYVCLQGFGPNPNYGYTSFDSFGWAFLSAFRLMTQDFWEDLYQLVLRAAGPWHMLFFIVIIFL GSFYLVNLILAIVAMSYDELQKKAEEEEAAEEEAIR	33,305	33,614	310	1,209
11	EAEEAAAAKAAKLEERANAQAQAAADAAAAEEAALHPEMAKSPTYSCISYELFVGGEKGNDDNNKEK MSIRSVEVESES	34,824	35,059	237	0
i	VSVIQRQPAPTTAHQATKVRKVST	35,060	35,131	71	3,421
a	TSLSLPGSPFNIRRGSRSSHK	38,552	38,614	63	1,610
13	YTIRNGRGRFGIPGSDRKPLVLSTYQDAQQHLPYADDSNAVTPMSEENGAIIVPVYYGNL	40,225	40,405	181	58
14	GSRHSSYTSHQSRISYTSHGDLLGGMAVMGVSTMTKESKLRNRNTRNQSVGATNGGTTCLDTNHKL DHRDY	40,465	40,676	212	74
15	EIGLECTDEAGKIKHHDNPFIEPVQTQTVVDMK	40,752	40,851	100	5,754
16	DVMVLNDIIEQAAGRHSRASDRG	46,606	46,674	69	2,569
b	VSVYYFPT	49,244	49,267	24	0
17	EDDDEDGPTFKDKALEVILKGIDVFCVWDCCWVWLKFQEWVSLIVFDPFVELFITLCIVVNTMFMAMD HHDMNKEMERVLKSGNY	49,268	49,521	254	83
18	FFTATFAIEATMKLMAMSPKYYFQEGWNIFDFIIVALSLLELGLEGVQGLSVLRSFRL	49,605	49,778	174	194
c	LRVFKLAKSWPTLNLLISIMGRTVGALGNLTFVLCIIIFIFAVMGMQLFGKNYT	49,972	50,134	163	941
d	LRVFKLAKSWPTLNLLISIMGRTMGALGNLTFVLCIIIFIFAVMGMQLFGKNYH	51,076	51,238	163	976
20	DHKDRFPDGDLPRWNFTDFMHSFMIVFRVLCGEWIESMWDCMYVGDVSCIPFFLATVVIGNLV	52,216	52,403	188	132
21	VLNLFLALLLSNFGSSSLSAPTADNDTNKIAEAFNRIGRFKSWVKRNIADCFKLIRNKLTNQISDQPS	52,537	52,741	205	0
e	GERTNQISWIWSE	52,742	52,780	39	2,452
f	GKGVCRCISA	55,232	55,261	30	0
22	EHGDNELELGHDEILADGLIKKGIKEQTQLEVAIGDGMEFTIHGDMKNNKPKKSKYLNNAT	55,262	55,443	182	569
h	MIGNSINHQDNRLEHELNHRGLSLQ	56,013	56,087	74	0
23	DDDTASINSYGSHKNRPFKDESHKGSAETMEGEEKRDASKEDLGLDEELDEEGECEEGPLDGDIIIHA HDEDILDEYPADCCPDSYYKKFPILAGDDDSPFWQGWGNLRLKTFQLIENKYFETAVITMILMSSLAL	56,088	56,495	408	1,127
24	ALEDVHLPQRPILQDILYYMDRIFTVIFFLEMLIKWLALGFKVYFTNAWCWLDFVIVM	57,622	57,795	174	523
k	LSLINLAAVWSGADDVPAFRSMRTLRALRPLRAVSRWEGMK	58,319	58,441	123	3,107
L	VSLINFVASLVGAGGIQAFKTMRTLRALRPLRAMSRMQGMR	61,550	61,672	123	1,698
26	VVVNALVQAIPSIFNVLLVCLIFWLIFAIMGVQLFAGKYFK	63,371	63,493	123	986
27	CEDMNGTKLSHEIIPNRNACESENYTWVNSAMNFDHVGNAYLCLFQVATFKGWIQIMNDAIDSRE	64,480	64,674	195	619
28	VDKQPIRETNIYMYLYFVFFIIFGSFFTLNLFIGVIIDNFNEQKKKAGGSLEMFMTEDQKKYYNAMKKMG SKKPLKAIPRPR	65,294	65,539	246	2,089
29	WRPQAIVFEIVTDKKFDIIIMLFIGLNMFTMTLDRYDASDTYNAVLDYLNAIFWIFSSECLLKIFALRYHYF IEPWNLFDWWILSIL	67,637	67,907	271	4,414
30	GLVLSDIIEKYFVSPTLLRVVRVAKVGRVLRLVKGAKGIRTLLFALAMSLPALFNICLLLFLVMFIFAIFGM SFFMHVKEKSGINDVYNFKTFGQSMILLFQ	72,322	72,626	305	1,702
31	MSTSAGWDGVLDAIINEEACDPPDNDKGYPGNCGSATVGITFLLSYLVISFLIVINMYIAVILENYSQATE DVQEGLTDDDYDMYYEIWQQFDPEGTQYIRYDQLSEFLDVLEPPLQIHKPNKYKIISMDIPICRGDLMY CVDILDALTKDFFARKGNPIEETGEIGEIAARPDTEGYEPVSSTLWRQREEYCARLIQHAWRKHKARG EGGGSFEPDTDHGDGGDPDAGDPAPDEATDGDAPAGGDGSVNGTAEGAADADESNVNSPGEDAA AAAAAAAAAAAAGTTTAGSPGAGSAGRQTAVLVESDGFVTKNGHKVVIHSRSPSITSRTADV	74,329	75,333	1,005

The first coding exon in the para locus is exon 3. All exons are numbered up from there except for the optional and mutually exclusive exons that are designated by letters. Adjacent mutually exclusive exons are in red. Exons i and e use differential splice donors, while exons b, f and h use differential splice acceptors and their sequences are part of a numbered exon. The other optional or mutually exclusive exons are fully independent exons and replace a numbered exon. Red colored residues in exons c and k show the differences with their mutually exclusive counterpart (d and L, respectively). Start and end refer to nucleotide number in the D. mel. para locus SnapGene file (Supplementary Data File). Size is the nucleotide length of the exon. Intron Size is the length of the Intron immediately following the exon (see Fig. 6A).

Table 2:

Alternatively-spliced exons in Drosophila melanogaster Para isoforms^†

Isoform		Isoform		Isoform
PA	j i a b d e f L	PAT	no-7 i b d e f L	PG	j i b d L
PAA	j a d L	PAU	no-7 i b d truncated	PH	i a b d L
PAB	j a c f L	PAV	no-7 i b d truncated	PI	j i d L
PAC	j a b d L	PAW	b d f L	PJ	j i a c L
PAD	i b d f k	PAX	i a d f L	PK	j a c L
PAE	j i a b d k	PAY	i a d f h k	PL	j i b c L
PAF	i a b d e L	PAZ	i b d f h k	PM	j i a d L
PAG	i b d e L	PB	j i a b d e f k	PN	j i b d e L
PAH	i a b d f L	PBA	i a b c f h L	PO	j i d e L
PAI	i b d f L	PBB	i a b d f	PP	j i a b d e L
PAJ	i a b d f h L	PBC	truncated after 18	PQ	j a d L
PAK	i a b d L	PBD	i a b d	PR	j i c L
PAL	i b d f h L	PBE	i a truncated after 18	PS	j i a b c e L
PAM	a b d f L	PBF	j i a b d e f L	PT	j i a d f L
PAN	j i a b d f h L	PBG	j i a b d e f L	PU	j i a b d f L
PAO	j i a d e L	PBH	i a b d	PV	j i b d f L
PAP	j i b d f k	PC	j i a b c e f L	PW	j i d e f L
PAQ	j i d f L	PD	i a c h k	PX	j i a c h L
PAR	3 5 6 truncated	PE	j i a b d L	PY	j i a c e h L
PAS	no-7 i a b d e L	PF	j i a b c L	PZ	j a c e L

^†Lin et al. (1994) describe isoforms lacking exon 8 although none are not found in FlyBase.

By combining this information with the most common reads of our RNA-seq data, we assembled a primary Para isoform in Schistocerca that contains some of the optional and mutually exclusive Drosophila exons (Fig. 6c). Specifically, it contains the 4 optional exons, j, i, a and h and exons c and L of the mutually exclusive pairs (c,d) and (k,L) in Drosophila. In Schistocerca, our data indicates that exon L is mutually exclusive with exon k. However, we found no evidence for the existence of exon d (see below) and it thus seems that exon c is obligatory in Schistocerca and not mutually exclusive with exon d, in contrast to the situation in Drosophila. We also compared these exon sequences with those of three additional species of Polyneoptera, Blattella germanica, Cryptotermes secundus, and Zootermopsis nevadensis, which are phylogenetically closer to each other and to Schistocerca than to Drosophila (see Material and Methods). In general, the sequences exons differ in their amino acid composition from those found in Drosophila, as summarized in Table 3 and in alignments between the different species (Supp. Fig. 5).

Table 3:

Alternatively-spliced coding exons in Drosophila, Schistocerca, Blatella, and Cryptotermes Para

	D. melanogaster	S. americana	B. germanica	C. secundus
Exon j	PQYGRKKKQKE	QGDFGRRKKKE	GDFGRRKKKKE
Exon i*	VSVIQRQPAPTTAHQ ATKVRKVST	LSEHRGRTIGPNGKVRKV SA	ISEHKGRVGANGTAIRKVSA	ISEHRGRIAGANGSMVRK VSA
Exon g		VPCQYRDYTATKRQLTFS YQENLMK	VPQFRDTKTATKSQFTFAY QENLVK	VPHCQYRDSNTATKSQFT FTYPESLVK
Exon a*	TSLSLPGSPFNIRRGS RSSHK	ASLSLPGSPFNLRRGSR GSHQ	ASLSLPGSPFNHRRGSQGS HH	ASLSLPGSPFNLRRGSRG SHQ
Exon b	VSVYYFPT		VSIYYFPT	VSIYYFPT
Exon c*	LRVFKLAKSWPTLNLL ISIMGRTVGALGNLTF VLCNIFIFAVMGMQLF GKNYT	LRVFKLAKSWPTLNLLISI MGRTVGALGNLTFVLCIII FIFAVMGMQLFGKNYT	LRVFKLAKSWPTLNLLISIMG RTVGALGNLTFVLCIIIFIFAV MGMQLFGKNYY	LRVFKLAKSWPTLNLLISIM GRTVGALGNLTFVLCIIIFIF AVMGMQLFGKNYG
Exon d	LRVFKLAKSWPTLNLL ISIMGRTMGALGNLTF VLCNIFIFAVMGMQLF GKNYH
Exon e	GERTNQISWIWSE	GEGPSSSWKE		GEGPSSSWKE
Exon m				DAHEHDTDLDL
Exon f	GKGVCRCISA
Exon h*	MIGNSINHQDNRLEH ELNHRGLSLQ	LIGNSIKENHQDNRIENEY YKQRL	VIGNSLNHKDNRIESGDYLH NRQ	VIGNSFQGNHKDNRIENEY LHNRQ
Exon k	LSLINLAAVWSGADD VPAFRSMRTLRALRP LRAVSRWEGMK	LSLINLAAIWAGAADIPAF RSMRTLRALRPLRAVSR WEGMR		LSLINLTAVWTGAADIPAF RSMRTLRALRPLRAVSRW EGMR
Exon L	VSLINFVASLVGAGGI QAFKTMRTLRALRPL RAMSRMQGMR	VSLINFVASLCGAGGIQAF KTMRTLRALRPLRAMSR MQGMR	VSLINFVASLVGAGGIQAFK TMRTLRALRPLRAMSRMQG MR	VSLINFVASLAGAGGIQAF KTMRTLRALRPLRAMSRM QGMR

Adjacent exons in red font are mutually exclusive exons. Other exons are optional. Red colored residues in Drosophila Para exons c and k show the differences with their mutually exclusive counterpart (d and L, respectively). Empty cells indicate there is no evidence to support the existence of that exon in the species corresponding to that column. Exons denoted with an asterisk (*) are obligatory in Polyneopterans.

Furthermore, analysis of our sequence reads identified transcripts missing exon j in the Schistocerca transcriptome from thorax tissue, while all transcripts from head tissue contain exon j. In Blattella one of the 5 known isoforms contains exon j, in the dampwood termite approximately half of its 22 known isoforms do, while in the drywood termite none of its 14 known isoforms have this exon. Together these results suggest that exon j is also optional in Polyneoptera. The Para exon j encoded sequence is highly conserved across the Polyneopteran species but only about half of the residues are identical to those of the vinegar fly (Table 3 and Supp. Fig. 4). While exons i and a are optional in Drosophila (Table 2), we found no transcript reads without these exons in grasshoppers or the other Polyneopteran species, suggesting that they are not optional in these insects. However, in a subset of the cockroach and termite isoforms, exons i and a are interrupted by a novel optional exon not present in Drosophila, that we have termed exon g (Table 3, Supp. Fig. 4). This exon is weakly homologous to an optional exon in the same location In B. mori Para (Shao et al. 2009). Although we found reads in S. americana that correspond to exon g in frame with the exon i sequence, they were either truncated downstream or in frame spliced midway to the next exon following exon a. For optional exon b, we found no corresponding reads in grasshopper, but about half the cockroach and termite isoforms contain exon b, with only one residue difference compared to the Drosophila sequence (Table 3, Supp. Fig. 4). We also found reads corresponding to optional exon e in our grasshopper data. While we found no cockroach isoforms with this sequence, about half the termite isoforms contain this optional exon. Its sequence is identical across grasshoppers and termites, but very degenerate compared to the Drosophila sequence (Table 3, Supp. Fig. 4). None of the Polyneoptera shows evidence for optional exon f, but two of the annotated Cryptotermes isoforms lack the first 12 residues encoded by exon 22, and we tentatively termed this alternative splice form as exon m (Table 3). The final optional exon in Drosophila is exon h. Like exons i and a, it is present in all isoforms of grasshoppers, cockroach and termites. Even within Polyneoptera its sequence only shows about 50% identity (Table 3, Supp. Fig. 4). The first set of mutually exclusive exons in Drosophila, exons c and d, encode the second half of transmembrane segment S4 through S5 of domain II and the first half of the loop that makes up the pore domain of the channel (Fig. 4, last two rows; Fig. 6c). In Drosophila, these exons only differ by 2 of 54 residues (Table 3, red letters V and T). In Polyneopteran species, we found evidence for a single exon c that differs from that of Drosophila only in the last residue (Table 3, Supp. Fig. 4). The second set of mutually exclusive exons in Drosophila, exons k and L, encode part of transmembrane segment S3 through the complete S4 of domain III and part of the subsequent intracellular loop of the channel (Fig. 5, third row; Fig. 6b). In contrast to the mutually exclusive exons c and d, exons k and L differ from each other more substantially in Drosophila (1/3 of the 41 residues are different). We found reads in grasshopper corresponding to both alternative exons. While the GenBank Blattella isoforms all use exon L, the isoforms from both termite species either contain exon k or L. The exon L sequences of the different species only differ in 1 of the 41 residues, while the exon k sequences have 6 residue differences (Table 3, Supp. Fig. 4). We found 1 transcript in our grasshopper reads that contains both exon k and L. One deposited Z. nevadensis isoform shares this k plus L configuration. Finally, in Drosophila several isoforms are annotated as missing exon 7 (Table 2). We found no isoforms missing this exon in Schistocerca but two of the Cryptotermes isoforms lack exon 7. In summary, the splice structure of Polyneoptera is relatively consistent across species but differs substantially from that of Drosophila.

Discussion

Voltage-gated sodium (NaV) channels play a critical role for information propagation in the central nervous system. Up to now, no information was available on their spatial localization in the brain of insects. In vertebrates, sodium channel antibodies stain bundles of unmyelinated axonal fibers in the central and peripheral nervous system (Gong et al. 1999; Henry et al. 2012; Su et al. 2013). Our immunostainings reveal the existence of similar well-resolved fiber tracts across the brain and the optic lobes of Schistocerca and Drosophila. In the central brain of grasshoppers, the most striking localization of NaV channels was in the peduncle of the mushroom body (MB). The MB peduncles contain the axons of Kenyon cells projecting to the MB lobes where they contact MB output neurons (MacLeod et al. 1998) expected to drive olfactory learning (Simões et al. 2011; Saha et al. 2013). Kenyon cells fire sparsely in response to olfactory input and are thought to use a temporal code to discriminate olfactory stimuli (Perez-Orive et al. 2002). The high density of NaV channels in their axons evidenced in our stains would help maintain the reliability of such a temporal code. In Drosophila, the MB peduncles were not as clearly resolved in single projection images but could be followed in movies assembled from stacks of images acquired through the central brain. In the optic lobes of Schistocerca and Drosophila, NaV channels were distributed along fascicles in the medulla that we interpret as axon bundles of transmedullary neurons projecting to the lobula complex. In grasshoppers, axon bundles originating in the lobula complex were also stained, suggesting that part of the information exiting the lobula complex rapidly propagates across sizable distances, e.g., toward thoracic motor centers. In contrast to Schistocerca, we detected pan-Nav staining in the somata of Drosophila neurons. This was unexpected as the somata of insect neurons are generally not excitable. While the cell body staining may represent an artefact, somatic staining has been reported with other ion channel antibodies (Hu et al. 2015; Xiao et al. 2017) and may result from immature protein precursors (Redondo et al. 2013).

In both Schistocerca and Drosophila, pan-Nav immunostaining offered sufficient contrast to resolve neuropils such as the antennal lobes, the central complex and its fan-shaped body. It is possible that the lighter stains observed in these structures and across the brain reflect in part a lower density of sodium channels in neuronal dendrites. However, in the hippocampus the density of sodium channels in dendrites was estimated to be 35 to 80 times lower than in axons and could only be reliably resolved by electron microscopical techniques (Lorincz and Nusser 2010). In CA1 pyramidal cells, these dendritic sodium channels contribute to dendritic action potential backpropagation in the context of synaptic plasticity. In the grasshopper LGMD neuron, action potentials backpropagating from the spike initiation zone (SIZ) along the main neurite and within the dendrites have been documented (Dewell and Gabbiani 2019). Biophysical modeling suggests that NaV channels located along the main LGMD neurite and decreasing in density as distance from the SIZ increases are required to explain such action potential backpropagation (Dewell and Gabbiani, 2018; Dewell and Gabbiani 2019). More generally, invertebrate neurons including those of insects are known to possess one and sometimes multiple SIZs as evidenced by a variety of techniques (e.g., O’Shea 1975; Perón and Gabbiani 2009; Trunova et al. 2011; Kuehn and Duch 2013; Günay et al. 2015). Backpropagating action potentials are also expected to play a role in spike-timing dependent plasticity in insects (Feldman 2012).

Several earlier investigations have indicated the presence of sodium channels in the insect brain. In S. americana antibodies against the same peptide sequence targeted in this study detected NaV channels by immunoprecipitation using a radiolabeled probe (Gordon et al. 1988). In Drosophila, in situ hybridization has shown the presence of para mRNA in parts of the central brain (Hong and Ganetzky 1994) and more recently a polyclonal antibody was able to detect coarse changes in Para protein expression levels (Xiao et al. 2017). However, none of these studies had sufficient spatial resolution to identify the fine structure of Para distribution in the brain, particularly the bundles of central brain axons and the stereotypically repeated optic lobe fascicles evidenced here. Thus, our study is the first to identify NaV channel localization in grasshoppers and vinegar flies. In vinegar flies our results are in close agreement with and complement those obtained by genetic techniques (Ravenscroft and Bellen 2019, personal communication).

In future work we see at least two ways in which our immunostaining results could be improved. First, it would be beneficial to combine immunostaining for NaV channels with simultaneous neuropil staining. This would allow unambiguous identification of the axon bundles revealed here by matching their localization relative to known neuropils with axon tract information available from anatomical atlases of the Drosophila and Schistocerca brains (Kurylas et al. 2008; El Jundi et al. 2010; Ito et al. 2014). We did not carry out such dual stains because the pan-Nav antibody requires a secondary anti-mouse antibody as do most neuropil markers. However, this Is a technical issue that can be resolved. Second, the short fixation times required to visualize sodium channels indicate that formaldehyde fixation masks the antigens targeted by the pan-Nav antibody, a well-known issue (Dapson 2007; Lorincz and Nusser 2008). Various possibilities exist to correct this problem, including antibody retrieval techniques (Lorincz and Nusser 2008; Ramos-Vara and Miller 2014), the use of low pH fixation conditions (Lorincz and Nusser 2010), and zinc formalin fixation (Ott 2008; Dapson 2010). Identifying the best method is likely to be time consuming but might open the possibility of mapping the subcellular localization of NaV channels in single cells such as the LGMD, at least in axons. In Drosophila, clonal analysis with driver lines that show clear expression in neuropil regions stained with pan-Nav may also resolve subcellular regions of NaV channel distribution.

Identifying Para in S. americana required us to update and synthesize information about Drosophila Para scattered across the literature. The S. americana Para protein finally identified from transcriptome data exhibited a high degree of homology with that of D. melanogaster and the human Nav1.1 sequence over the transmembrane domains critical to its function. Outside those domains, divergence from D. melanogaster’s Para sequence was higher, in agreement with the evolutionary distance of these species, thought to have diverged around 380 million years ago (Misof et al. 2014). We also compared the Para sequence from Schistocerca to other Polyneopteran species focusing on the Drosophila optional exons j, i, a, b, e, f and h, which all reside in intracellular loops and whose sequences are thought to play a role in NaV channel modulation (Lin et al, 2009; Dong et al, 2014). We found no evidence for exon f in any of the Polyneopterans we examined. Conversely, exons i, a and h were present in all isoforms and thus do not appear to be optional. Among splice variants in intracellular loops, only exons a and b are highly conserved with Drosophila, consistent with the fact that their sequences are thought to play a role in protein kinase mediated phosphorylation and modulation of the channel (Dong et al, 2014). With respect to the mutually exclusive exons in Drosophila, c/d and k/L, which encode parts of two transmembrane domains in Domain II and III of the channel, we found no evidence of a d variant in any of the Polyneopteran species. One other sequence that is highly conserved in the Intracellular loops of the protein is the sequence targeted by the pan-Nav antibody. This sequence is also highly conserved across vertebrate NaV channels, conferring to the antibody its broad staining spectrum. Antibodies against this segment were shown to slow down the inactivation of NaV channels (Vassilev et al. 1988). This segment has been proposed to move along with the ‘IFM’ segment to block the channel during inactivation according to a ‘hinged-lid’ model (Kellenberger et al. 1997b, a; Catterall 2000; Pan et al. 2018). Notably, our results reveal that this sequence is also highly conserved in insects, explaining the staining observed in Schistocerca and Drosophila. Conservation across humans, grasshoppers and vinegar flies is consistent with an important role in NaV channel inactivation and opens the possibility that pan-NaV will be effective against NaV channels in a broad range of invertebrate species.

In summary, while much remains to be learned about how NaV channels influence visual processing in the context of collision avoidance behaviors and in olfactory processing, our results highlight the important role and broad localization of NaV channels in the grasshopper and vinegar fly brain as well as their relationship with homologous insect and vertebrate channels.

Supplementary Material

359_2019_1396_MOESM8_ESM

Supplementary Figure 2. a: Pan-Nav protein expression in the optic lobe of a 4^th instar grasshopper, b, c: Pan-Nav protein expression in the optic lobe and brain, respectively, of a 5^th instar grasshopper, d: Pan-Nav (red) and 5-HT (green) expression in an adult grasshopper optic lobe, e: Close up view of the lobula in the same grasshopper illustrating the strong 5-HT staining pattern observed there. Abbreviations: Ca, Calyx of the mushroom body; Me, Medulla; Lo, Lobula.

359_2019_1396_MOESM5_ESM

Supplementary Movie 4. Animation of slices through the adult central brain at twice the magnification as in Supp. Movie 2, highlighting GFP staining of large connecting bundles between the two optic lobes and finer projections and cell bodies in the dorsal protocerebrum, followed by pan-Nav staining of the central brain and ending in the merged slices. The pan-Nav signal clearly resolves the Mushroom Body lobe branches in sections of the peduncle and shows strong fascicles around the antennal lobes, central complex, and throughout the dorsal, ventral protocerbrum and sub esophageal zone. Scalebar: 20 μm.

359_2019_1396_MOESM4_ESM

Supplementary Movie 3. Animation of slices through the medulla in the Drosophila optic lobe highlighting GFP and pan-Nav staining of transmedullary axons at twice the magnification of Supp. Movie 2. The movie is structured as Supp. Movie 2, running from anterior to posterior. Stained transmedullary neurons are visible at different depths from front to back in the medulla. Fascicles are also clearly visible at the base of the optic lobe and throughout the lateral protocerebrum. Scalebar: 20 μm.

359_2019_1396_MOESM2_ESM

Supplementary Movie 1. Animation of slices through the central brain of a 2^nd instar grasshopper corresponding to the image in Figure 2b, highlighting pan-Nav staining. The movie runs from anterior to posterior and shows strong fascicular staining throughout the central brain. In the most anterior slices the antennal nerves stain strongly and converge onto the antennal lobes. The MB peduncles and calyces are visualized in deeper slices. In between, the animation resolves parts of the central complex with individual fascicles fanning out of the fan-shaped body. Additional fascicles are also clearly visible at the base of the optic lobe and throughout the protocerebrum.

359_2019_1396_MOESM11_ESM

Supplementary Figure 5. a: Alignment of the optional exons in Drosophila compared to several Polyneopteran species. Included is an alignment of optional exon g from the different Polyneopteran species but not present in Drosophila, b: Alignment of the mutually exclusive exons in Drosophila compared to Polyneopteran species. Data presented as in Fig. 4.

359_2019_1396_MOESM1_ESM

Supplementary Data File. SnapGene file corresponding to the para locus in Drosophila. All Drosophila splice variants are annotated. The start and end of each exon sequence in Table 1 correspond to the numbering in this sequence file. The file can be opened with the SnapGene Viewer, available at https://www.snapgene.com/snapgene-viewer.

359_2019_1396_MOESM6_ESM

Supplementary Movie 5. Animation of slices through the adult ventral nerve cord at 20x magnification highlighting the GFP and pan-Nav staining patterns as in the other Drosophila movies. Fascicles are again clearly resolved but they are generally thinner than in the adult central brain. Scalebar: 50 μm.

359_2019_1396_MOESM7_ESM

Supplementary Figure 1. a: Life cycle of grasshoppers. Adult females lay eggs in pods that hatch after several weeks. The grasshopper grows over successive molts into an adult after five instar stages. The adults have fully developed wings and are able to fly. b: Life cycle of Drosophila. The fertilized egg grows into a blastoderm and embryo stage followed by 3 instar stages and a pupa stage, from which the adult emerges.

359_2019_1396_MOESM10_ESM

Supplementary Figure 4: Pan-Nav and GFP protein expression in the brain, optic lobe and VNC of adult Drosophila that express GFP in the medullary T4 and lobula T5 neurons, a. Maximum intensity projection of a whole brain image of adult Drosophila, b. Close-up image of the optic lobe. c. Close-up image of the central brain, d. Image of the adult VNC. e. Image of whole brain and partial VNC. f. Same image as in B but projected horizontally. Bottom panel shows the GFP staining of the T4/T5 cell bodies and their processes in the medulla, lobula and lobula plate, middle panel shows the pan-Nav staining, and the top panel shows the merge of the bottom two panels. Abbreviations: Me, medulla; LP, Lobula Plate; Lo, Lobula. Scalebar: 50 μm (a-f).

359_2019_1396_MOESM3_ESM

Supplementary Movie 2: Animation of slices through the adult brain and optic lobes of Drosophila corresponding to the image in Figure 3a and supplementary Figure 4a, highlighting GFP and pan-NaV staining. The first set of slices show the GFP staining of the T4/T5 neurons in the optic lobes and a large commissural bundle connecting the two optic lobes through the central brain and less easily visible cell bodies and fine neurite projections in the dorsal part of the protocerebrum. The next set of slices show the pan-Nav staining pattern, followed by the merged staining of GFP and pan-Nav. In the optic lobe, transmedullary neurons at different depths are easily resolved. In the central brain, large fascicular bundles are visible throughout, particularly in the peduncles, the central complex and around the antennal lobes. Scalebar: 50 μm.

359_2019_1396_MOESM9_ESM

Supplementary Figure 3. a: Pan-Nav immunostaining in an adult Drosophila brain after 60 min fixation, b: Pan-Nav immunostaining in an adult Drosophila brain after 15 min fixation, c: 5-HT immunostaining in an adult Drosophila brain after 15 min fixation, d: Pan-Nav immunostaining in a Drosophila larval 3^rd instar after 60 min fixation, e: Pan-Nav immunostaining in a Drosophila larval 3^rd instar after 15 min fixation, f: 5-HT immunostaining in a Drosophila larval 3^rd instar after 15 min fixation, g: Pan-Nav staining in a Drosophila larval 3^rd instar CNS after 5 min fixation, h: Pan-Nav staining in a Drosophila larval 3^rd instar CNS after 2 min fixation. Abbreviations: Ca, Calyx of the mushroom body; MN, Motor Nerve. CB, Central Brain; VNC, Ventral Nerve Cord. Scalebar: 50.μm (a-h).