A long-read draft assembly of the Chinese mantis (Mantodea: Mantidae: Tenodera sinensis) genome reveals patterns of ion channel gain and loss across Arthropoda

Abstract Praying mantids (Mantodea: Mantidae) are iconic insects that have captivated biologists for decades, especially the species with cannibalistic copulatory behavior. This behavior has been cited as evidence that insects lack nociceptive capacities and cannot feel pain; however, this behaviorally driven hypothesis has never been rigorously tested at the genetic or functional level. To enable future studies of nociceptive capabilities in mantids, we sequenced and assembled a draft genome of the Chinese praying mantis (Tenodera sinensis) and identified multiple classes of nociceptive ion channels by comparison to orthologous gene families in Arthropoda. Our assembly—produced using PacBio HiFi reads—is fragmented (total size = 3.03 Gb; N50 = 1.8 Mb; 4,966 contigs), but is highly complete with respect to gene content (BUSCO complete = 98.7% [odb10_insecta]). The size of our assembly is substantially larger than that of most other insects, but is consistent with the size of other mantid genomes. We found that most families of nociceptive ion channels are present in the T. sinensis genome; that they are most closely related to those found in the damp-wood termite (Zootermopsis nevadensis); and that some families have expanded in T. sinensis while others have contracted relative to nearby lineages. Our findings suggest that mantids are likely to possess nociceptive capabilities and provide a foundation for future experimentation regarding ion channel functions and their consequences for insect behavior.


Introduction
The Chinese mantis, Tenodera sinensis (Mantidae), is a large-bodied mantid species native to Asia and the nearby islands.In the late 1800s, the species was accidentally introduced to the Philadelphia area (Blatchley 1920) and has become established throughout the contiguous United States (according to data available from the Global Biodiversity Information Facility, GBIF.org 2023; Fig. 1).
Tenodera sinensis is probably most well known as a study system for sexual cannibalism, where females may eat males, before, during, or after copulation, though field observations suggest feeding on the male during copulation is relatively rare (<10% of cases; Hurd et al. 1994).Food limitation during oogenesis is expected to drive female mantises to continuously attract mates which, if consumed, increase fecundity (Brown and Barry 2016;O'Hara and Brown 2021).Males may make up 63% of female diets during this crucial reproductive period (Hurd et al. 1994).Males are known to adjust their approach behaviors in response to the perceived threat level posed by a female and female encounter rates (Lelito and Brown 2006;Brown et al. 2012); unlike some other sexually cannibalistic species, male T. sinensis are not self-sacrificial (Buskirk et al. 1984).
Despite males' well-studied risk avoidance behaviors that suggest a lack of complicity (see Lelito andBrown 2006 anddiscussion in Christensen andBrown 2018), sexual cannibalism in mantids has featured prominently in discussions of the plausibility of insect pain and, thus, sentience (e.g. as in Eisemann et al. 1984; though see Gibbons andSarlak 2020 andBarrett andFischer 2024).Male mantises may be unable to perceive (minimally) mechanically noxious stimuli, which animals typically perceive using nociceptive ion channels in their peripheral sensory neurons.Eisemann et al. (1984) suggest that the lack of nociceptive ability, as demonstrated by males continuing to mate while being cannibalized (e.g.behavioral nonresponsiveness), precludes any plausible further possibility of pain experience.This would make mantids an interesting species for studies of insect pain perception, a field of growing research interest given the development of large-scale insect farming and associated welfare concerns (e.g.van Huis 2021; Gibbons et al. 2022;Barrett and Adcock 2023).
Here, we present an assembly of the Chinese praying mantis genome (T.sinensis) and combine it with existing genomic resources to study the evolution of nociceptive channel evolution across the arthropod phylum.We find that mantids have genes that encode many well-studied arthropod nociceptors dedicated to perceiving diverse noxious stimuli (mechanical, chemical, and thermal).Across the arthropods, we find that some channel families are conserved, whereas copy numbers vary greatly in others.We discuss the potential factors that may underlie this variation and the implications thereof for rearing and domestication of arthropods.

Mantis rearing
The specimen used for whole-genome sequencing (adult female; Fig. 2) was reared at 50% relative humidity, 27°C, 14:10 L:D from a set of 6 oothecae purchased from Carolina Biological in spring 2023.Mantises were reared collectively until the third instar, then reared in separate containers and fed flightless Drosophila melanogaster and Drosophila hydei, Acheta domesticus cricket nymphs, and Tenebrio molitor mealworm larvae.The single female individual used for analyses (Fig. 2) was flash frozen on liquid  (red, pre-2000; orange, 2000-2009; yellow, 2010-2014; green, 2015-2019; purple, 2020-2023).nitrogen and sent to the Arizona Genomics Institute (AGI, Tucson, AZ, United States) for DNA extraction and sequencing.A female specimen was chosen as they are the homogametic sex in mantises (Paliulis et al. 2022).

DNA extraction
High molecular weight DNA was extracted from ground thoracic tissue in an extraction buffer with Tris HCl buffer 0.1 M, pH 8.0, EDTA 0.1 M, pH 8, SDS 1%, and Proteinase K in 50°C for 60 min.Mixture was spun down, and aqueous phase was transferred to a new tube; 5 M potassium acetate was added, precipitated on ice, and spun down.After centrifugation, the supernatant was gently extracted with 24:1 chloroform:isoamyl alcohol.The upper phase was transferred to a new tube and DNA precipitated with iso-propanol.DNA was collected by centrifugation, washed with 70% ethanol, air dried, and dissolved thoroughly in 1× TE followed by RNAse treatment.DNA purity was measured with NanoDrop, DNA concentration was measured with Qubit HS kit (Invitrogen, Carlsbad CA, United States), and DNA size was validated by Femto Pulse System (Agilent, Santa Clara, CA, United States).

Genome sequencing
DNA was sheared to an appropriate size range (10-20 kb) using Megaruptor 3 (Diagenode, Denville, NJ, United States) followed by SMRTbell cleanup beads.The sequencing library was constructed following manufacturer's protocols using SMRTbell Prep kit 3.0.The final library was size selected on a Pippin HT (Sage Science, Beverly, MA, United States) using S1 marker with a 10-25 kb size selection.The recovered final library was quantified with Qubit HS kit (Invitrogen, Carlsbad, CA, United States) and size checked on Femto Pulse System (Agilent, Santa Clara, CA, United States).The final library was prepared for sequencing with PacBio Sequel II Sequencing kit 3.1 for HiFi library, loaded on a single Revio SMRT cells, and sequenced in CCS mode for 24 h.

Tenodera sinensis genome assembly and annotation
The initial assembly was highly fragmented by long-read genome assembly standards (number of contigs = 8,798; N50 = 1.6 Mb; total size = 3.3 Gb) and roughly 20% larger than our GenomeScope A long-read draft assembly of the Chinese mantis | 3 prediction (2.7 Gb; Supplementary Fig. 1).Blobtools analysis determined that the assembly was heavily contaminated with viral and bacterial contigs (N = 3,832, Supplementary Fig. 2; 18.81% of raw reads, Supplementary Table 1).Our contaminant-filtered assembly was still found to be fragmented and larger than expected (number of contigs = 4,966; N50 = 1.8 Gb; total length = 3.03 Gb; Supplementary Table 1).This is similar to genome sizes reported for other mantis species (Hymenopus coronatus = 2.88 Gb; Deroplatys lobata = 3.96 Gb; Huang et al. 2023).RepeatMasker found that 68.83% of the genome was composed of repetitive elements (Supplementary Table 2), predominantly retroelements (18.41%) and DNA transposons (18.08%).Inspector analysis found that our assembly had a number of structural errors (N = 593) and small-scale errors (N = 83,917; 27.7 per Mb), but polishing was able to remove most small-scale errors (after polishing N = 8,788; 2.9 per Mb; see Supplementary Table 1 for details).Despite assembly errors, BUSCO analysis determined that our assembly was highly complete [odb10_insecta complete = 98.9% (single copy = 93.5%,duplicated = 5.4%), fragmented = 0.5%, missing = 0.6%; Fig. 3].Helixer structurally annotated 24,673 genes in our assembly, 14,092 of which were functionally annotated via comparison to the eggnog-mapper database using only orthology to Arthropoda.BUSCO assessment of our annotation showed that a substantial portion of total gene content is likely not represented [odb10_insecta complete = 78.2%(single copy = 75.1%,duplicated = 3.1%), fragmented = 7.9%, missing = 13.9%;Fig. 3]; however, this was also the case (albeit less drastically) for our Helixer annotation of A. domesticus [odb10_insecta complete = 86.7%(single copy = 82.7%,duplicated = 4.0%), fragmented = 7.4%, missing = 5.9%; BUSCO complete in assembly = 94.8%;Dossey et al. 2023] suggesting there may be biases in Helixer's premade invertebrate training data set, which is dominated by insects in the orders Diptera and Lepidoptera (details regarding Helixer training data sets can be found at https://uni-duesseldorf.sciebo.de/s/lQTB7HYISW71Wi0).Thus, the lower than expected recovery of single-copy orthologs may represent this bias rather than the quality of our assembly.Summary of assembly and annotation statistics is found in Table 1, with further details provided in Supplementary Table 1.

Variation in ion channel copy numbers in T. sinensis
Tenodera sinensis was found to have at least 1 copy of 8 of 9 genes for receptor channels of interest, with the exception being pkd2 (Table 2).Pkd2 acts as a direct cold sensor in class III md neurons in D. melanogaster and is necessary (alongside NOMPC and TRPm) for aversive responses to cold behavior (Turner et al. 2016).Tenodera sinensis was found to have an expansion in both nompC and trpm copy number (from 1 each in D. melanogaster to 3 each in T. sinensis).The lack of pkd2 could suggest an inability to directly sense aversive levels of cold in T. sinensis despite the presence of nompC and trpm, but given its function has been characterized solely in D. melanogaster and we could not detect an ortholog in many insect genomes, it is difficult to infer this with confidence.Further study of T. sinensis behavior and physiology would be needed to assess function.Other orders of insects have evolved novel thermal nociceptors following the loss of other thermosensitive nociceptive ion channels (e.g.duplication and neofunctionalization of waterwitch into HsTRPA to complement the loss of TRPA1 in the Hymenoptera; Kohno et al. 2010).Further, there is often more than 1 receptor for assessing potentially noxious thermal information (e.g. the heat sensitivity of pyrexia, painless, and TRPA1; Tracey et al. 2003;Lee et al. 2005 ;Wang et al. 2009).
Prior arguments by Eisemann et al. (1984) and others have hypothesized that behavioral nonresponsiveness to injury in insects, such as an T. sinensis sexual cannibalism, suggests insects are unlikely to perceive (minimally, mechanically) noxious stimuli and thus unlikely to have an adaptive role for pain and sentience.However, since these publications, the genetic components of nociception have been found to be ancient and highly conserved (e.g.Peng et al. 2015); it is, therefore, unsurprising that our data suggest that chemical, mechanical, and thermal nociception are plausible in T. sinensis.Still, further tests that confirm (1) these channels are expressed in multidendritic sensory neurons and (2) they have a nociceptive function in the Mantodea will be important for conclusively demonstrating nociception in mantids.
Differential expression and splicing into novel isoforms can lead to different functional roles of these receptors in different cell types (e.g.dTRPA1; Zhong et al. 2012;Gu et al. 2022).Amino acid substitution can also lead to variation in channel sensitivity to noxious stimuli, even for orthologous channels (e.g.loss of chemical sensitivity in SiHsTRPA vs AmHsTRP; Wang et al. 2018).Altogether, these various mechanisms for generating functional plasticity of these receptors in different tissue types suggest moderate caution should be employed when suggesting the presence or absence of a specific gene corresponds to the presence or absence of a specific nociceptive ability in Mantodea.Still, we may broadly expect some perception of noxious stimuli to be highly plausible given this genetic information and the ancient and highly conserved nature of many of these genes.If male mantises are, indeed, able to perceive a mechanically noxious stimulus (such as female cannibalism), this suggests mechanisms other than a lack of nociceptive perception underpin any instances of apparent behavioral unresponsiveness (see Gibbons and Sarlak 2020).Instead, our genetic data support prior behavioral evidence that shows male mantises attempt to avoid being cannibalized (Lelito and Brown 2006;Brown et al. 2012) and provides further support for the ancient and highly conserved nature of the peripheral components of pain pathways across Arthropoda.

Variation in ion channel copy number across Arthropoda
All the gene families except the ppk/rpk/ppk26 family saw reasonably similar ranges of copy number across the surveyed arthropods (from 0 up to 6-9 copies).However, greater variation (0-31 copies) was found in the ppk/rpk/ppk26 family, with roles in mechanical and chemical nociception.This included substantial intraorder variation (e.g. 2 copies in H. illucens, 11 copies in Musca domestica, and 31 copies in Culex quinquefasciatus, all dipterans).Prior research on mosquitoes has found that additional ion channel gene copies play a role in nonnociceptive sensory processes, specifically heat perception involved with prey location (Wang et al. 2009).Our results suggest that ion channel copy numbers are highly variable and that they are likely to play an important role in the evolution of Arthropod sensory systems, but more precise studies of expression dynamics and cellular functions are necessary to better understand this process.
Our search also specifically focused on insects that are mass reared (such as the black soldier fly, H. illucens) or even A long-read draft assembly of the Chinese mantis | 5 domesticated (such as the silkworm moth, Bombyx mori) to see if there were any patterns of ion channel loss or duplication associated with the novel ecological conditions posed by rearing at scale on farms under significant human management.Generally, farmed species retained at least 1 copy of all genes except pkd2, with only T. molitor and M. domestica having a copy of this gene.This could suggest, as in T. sinensis, some loss of cold nociception capabilities that should be confirmed functionally (and considered alongside the aforementioned caveats).
The only other gene loss event was painless in H. illucens; however, H. illucens had significant duplication of trpa1 (6 copies), where no other arthropod species in our study had more than 2 copies of this gene.It is possible the additional copies of trpa1 may serve to recover some of the lost functions of painless as both are TRPA proteins that have roles, for instance, in thermal nociception in other Diptera (Neely et al. 2011;Hwang et al. 2012).Tenebrio molitor had significant duplication of painless (8 copies) and ppk/rpk/ppk26 (19 copies).Acheta domesticus also had significant duplication of ppk/rpk/ ppk26 (12 copies).Bombyx mandarina had significant duplication of piezo (8 copies) and trpm (5 copies).These genes were both highly expressed in the 2 lepidopterans-B.mandarina and Manduca sextabut not in any other order we reviewed.Overall, these results suggest that domestication and/or mass rearing do not result in nociceptive ion channel loss in insects, which means it will still be important to monitor and reduce the occurrence of noxious stimuli in mass-rearing facilities as part of ethical mini-livestock husbandry and slaughter.

Conclusion
In summary, we have produced a high-quality draft genome of the Chinese mantis (T.sinensis) and used it to identify multiple putative nociceptive ion channels.Comparative analyses suggest that some of these genes have undergone multiple duplications that are shared with Blattodea, whereas others have been lost in these closely related clades.These findings suggest that mantids are likely capable of perceiving different types of noxious stimuli (extreme heat/cold, mechanical damage, etc.).Future research into their precise functional roles will elucidate the evolution of nociception in an iconic sexually cannibalistic species.

Data availability
A male nymph and adult female from the same set of oothecae were vouchered at Purdue Entomological Research Collection (PERC 0154978 and PERC 0154979).The entire individual used for This table includes gene counts of species included in the database and those obtained from eggnog-mapper outputs of additional species.
sequencing was destroyed during sampling.Raw reads and our assembly are available under BioProject PRJNA1036567.No new code was written for this study, but scripts used to run existing programs are found at https://github.com/caterpillar-coevolution/Tsinensis_genome_V1 alongside annotation files.Supplemental material available at G3 online.

Fig. 2 .
Fig. 2. Photos of adult female T. sinensis specimen that was used for genome assembly.Scale bar in the left photo represents approximately 1 cm (the mantid stands at an angle and thus changes depth in the photo).

Fig. 3 .
Fig. 3. BUSCO assessment of T. sinensis: our filtered/polished assembly and its annotation.Annotation assessment was performed in proteome mode.

Fig. 4 .
Fig. 4. Gene tree showing the relatedness of painless genes across our sample of arthropods.Tenodera sinensis genes cluster most closely with those of Z. nevadensis (Blattodea).Genes with resolved species names were present in the eggnog5 database (Orthofamily 41TMZ), whereas those with only gene IDs at branch tips are those of our additional species (Adom, A. domesticus; Boma, B. mandarina; Msex, M. sexta; Tmol, T. molitor; Tsin, T. sinensis).

Table 1 .
Summary statistics of the T. sinensis draft assembly and annotation.

Table 2 .
Ion channel gene families obtained from the eggnog5 database.