Background

How organisms adapt to a new environment is one of the fundamental research questions in evolutionary biology [1]. Subterranean organisms whose ancestors originally lived in a surface environment are ideal for investigating this issue [2, 3]. Subterranean habitats are not continuously exposed to light, and can be categorized into cave habitats, interstitial habitats and superficial subterranean habitats [4, 5]. Degeneration of eyes is generally observed in various taxa colonizing these subterranean environments [6, 7].

Do the organisms having a regressed eye also have decreased ability of photoreception? Previous studies focused on various aspects of subterranean adaptation, including signatures in photoreceptor proteins [8, 9]. For example, one previous study described the expression of visual opsin genes which encode seven-transmembrane photoreceptor proteins in the binocular eye of the Mexican blind cavefish, Astyanax mexicanus [8]. Transcripts of visual opsin genes were underrepresented in the cavefish as compared with a conspecific surface population, and this could be attributed to reduction of photoreceptor cells in the cavefish [10]. In other examples, a blind mole rat has an ultraviolet-sensitive/violet-sensitive opsin gene with deleterious mutations, and fossorial snakes with reduced eyes, Scolecophidia, did not have visual opsin genes in the genome or transcripts [11, 12].

In Insecta, subterranean diving beetles (Dytiscidae), which have highly regressed or no eyes and inhabit a calcrete aquifer located 10 m underground in Western Australia, were subjected to a similar analysis [13]. Transcripts were not detected for long wavelength opsin or ultraviolet opsin at the adult stage of the diving beetles [14], and pseudogenization of long wavelength opsin, ultraviolet opsin and some phototransduction genes was observed [15, 16].

Besides calcrete aquifers, insects have also been found to colonize superficial subterranean habitats, such as rock fissures near the surface [17]. Insects living in a superficial subterranean habitat can be exposed to light due to unexpected environmental fluctuation. The colonization of a superficial subterranean habitat is inferred to reflect an intermediate stage in an evolutionary pathway to colonization of deeper and extreme environments [4]. Despite this importance, the biological features of species living in a superficial subterranean habitat remain unexplored.

The present study focused on one trechine beetle species, Trechiama kuznetsovi (Coleoptera: Carabidae: Trechinae). This species has a vestigial compound eye and inhabits the upper hypogean zone, which is a similar environment to mesovoid shallow substratum (MSS). The habitat of the type specimens was either under stones or soil deposits along narrow streams [18]. We aimed to reveal the photoreception ability in this species. We obtained genome and transcript sequences to examine photoreceptor genes and phototransduction genes and estimate selective pressure on visual opsin genes. Also, histological investigations were performed to observe the internal structure of the vestigial compound eye and a nerve bundle connecting it to the brain.

Methods

Sample collection

Trechiama kuznetsovi samples were collected at Yûbari City, Hokkaido, Japan. We collected T. kuznetsovi adults from the upper hypogean zone consisting of small rocks and clay, by digging and finding by sight in the slope of a v-shaped valley to the depth of a few to some dozen centimeters (Fig. 1).

Fig. 1
figure 1

The habitat and morphology of T. kuznetsovi. A A v-shaped valley of a mountain area at Yûbari City in Hokkaido. B The upper hypogean zone under the slope of a v-shaped valley. It consists of small rocks and clay. C Dorsal view of T. kuznetsovi. Scale bar indicates 3 mm. D Lateral view of T. kuznetsovi head. The bracket indicates the vestigial compound eye. The black part is the pigmentation of the cuticle behind a compound eye (ocular ridge). The arrowhead indicates a hole on the ocular ridge, through which the nerve bundle is presumed to pass. There is no pigmentation in cells inside the compound eye structure. Scale bar indicates 0.2 mm

Full size image

DNA and RNA sequencing

We used one adult male of T. kuznetsovi stored in 99.5% ethanol for genome sequencing. Before DNA extraction, mites adhering to the body surface were removed and the male genitalia was preserved in 99.5% ethanol for identification. Genomic DNA was extracted using a Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA). A library was constructed using a TruSeq Nano DNA Library Prep Kit (Illumina, San Diego, CA, USA) and sequenced on the NovaSeq 6000 platform (Illumina) by Macrogen Service (Macrogen, Seoul, South Korea). 2 × 151 bp paired-end reads were generated (Table S1).

We used one live adult male of T. kuznetsovi for transcript sequencing. Before RNA extraction, the beetle was washed with 99.5% ethanol, mites adhering on its body surface were removed and the male genitalia was preserved in 99.5% ethanol for identification. Total RNA was immediately extracted from the whole body using an RNeasy Micro Kit (Qiagen, Hilden, Germany) since vestigial compound eyes were too small to extract RNA and construct a library for sequencing. A library was constructed using a SMARTer Stranded RNA-Seq Kit (Illumina) and sequenced on the NovaSeq 6000 platform (Illumina) by Macrogen Service. 2 × 101 bp paired-end reads were generated (Table S1).

Assembly and mapping

Summary statistics of raw reads and adapter contamination were checked using FastQC (v0.11.9; Babraham Institute). Quality control was performed using fastp v0.20.1 [19] and Trimmomatic v0.39 [20] to trim off one base from the 3′ end, low quality sequences and adapter sequences. Then, summary statistics were rechecked using FastQC. The kmer content of reads from genome sequencing and the genome size were calculated using KmerGenie v1.7051 [21]. For the reads that passed the quality control, genome and transcript assembly were conducted with Platanus v1.2.4 [22] and Trinity v2.8.4 [23, 24]. Before the scaffolding step of genome assembly, contigs smaller than 500 bp were excluded [25]. Completeness of the assembled genome and transcript was assessed using BUSCO_v5 for insecta core gene sets and CEGMA for invertebrate core gene sets in gVolante web server [26]. Summary statistics of the assembled sequences were calculated using SeqKit v.0.16.1 [27].

RNA reads were mapped to the assembled genome scaffolds using HISAT v2.1.0 [28]. The result of the mapping was visualized using IGV v2.11.9 [29].

Homology search

We conducted a search for genes encoding three photoreceptor proteins and 15 phototransduction proteins from the assembled genome and transcripts with the tblastn program using the following queries (unless noted, protein sequences of Tribolium castaneum [Tenebrionidae] were used): Long wavelength opsin (Lw opsin) of Pogonus chalceus (Carabidae: Trechinae), Ultraviolet opsin (Uv opsin) of P. chalceus, C-opsin, Arrestin 1 (Arr1), Arrestin 2 (Arr2), G protein α-subunit 49B (Gα49B), G protein β-subunit 76C (Gβ76C), G protein γ-subunit 30A (Gγ30A), G protein-coupled receptor kinase 1 (Gprk1), Inactivation no afterpotential D (InaD), Neither inactivation nor afterpotential A (NinaA), Neither inactivation nor afterpotential C (NinaC), No receptor potential A (NorpA), Protein C kinase 53E (Pkc53E) of Drosophila melanogaster, Rab-protein 6 (Rab6), Retinal degeneration C (RdgC), Transient receptor potential (Trp), and Trp-like (Trpl). Only two types of visual opsin proteins (Uv opsin and Lw opsin) were known in Coleoptera [30]. The selection of the other proteins was based on previous studies [31, 32]. Blast-hit sequences with an e-value < 1 × e–20 were treated as having high similarity [33]. If no sequence matched this criterion, Blast-hit sequences were examined in order from the best-hit sequence. In BLAST-search for the transcripts, the presence of premature stop codons and frameshift mutations was examined.

We subsequently conducted a further analysis of lw opsin and uv opsin, which are visual photoreceptor genes in Coleoptera, while some opsin genes are known to have light-independent roles in D. melanogaster [30, 34,35,36,37]. We checked whether the blast-hit transcripts matched exon regions with mapped short reads of transcripts using HISAT. Matched transcripts were used for subsequent comparative analyses with a related surface species: P. chalceus, whose opsin amino acid sequences were already registered in NCBI protein database [30].

Identification of lw opsin gene

A part of the genome sequence that had high similarity score to the Lw opsin amino acid sequence of P. chalceus in a tblastn search was divided into three scaffolds (Table S2). To join these scaffolds together, primers were designed on each scaffold using Primer3Plus (https://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi) (Table S3) and PCR was performed using PrimeSTAR Max DNA Polymerase (Takara Bio, Shiga, Japan). The sequences of the PCR products, which were determined by Sanger sequencing, were overlapped on the scaffolds. The transcript sequence that had a high similarity score to the Lw opsin amino acid sequence of P. chalceus in a tblastn search was divided into two contigs assembled by Trinity (Table S2). It is generally difficult to assemble transcripts expressed at low levels into a single contig [38]. These contigs were joined together by Sanger sequencing using the same method as above. Primers were designed on each contig (Table S3) and RT-PCR was performed using a 3′ RACE CORE Set (Takara Bio).

To specify exon and intron regions, the acquired transcript sequence of the lw opsin gene was aligned to the acquired genome sequence of the lw opsin gene with Exonerate v2.4.0 [39]. The exon and intron regions were illustrated with GenePalette [40].

Identification of uv opsin gene

There was one genome scaffold that had high similarity score to the Uv opsin amino acid sequence of P. chalceus in a tblastn search (Table S2), but no transcript contig was found. However, short read sequences originated from RNA sequencing were mapped to the scaffold of the uv opsin gene. This means that the uv opsin gene was expressed and short reads derived from mRNA were detected as a result of RNA sequencing, but not correctly assembled by Trinity, probably because of the low number of reads. We determined the transcript sequence of the uv opsin gene by Sanger sequencing. Firstly, the exon regions were predicted with Exonerate using the Uv opsin amino acid sequence of P. chalceus. Then primers were designed on the predicted exon regions with Primer3Plus (Table S3) and RT-PCR was performed using a 3′ RACE CORE Set.

To specify exon and intron regions, the acquired transcript sequence of the uv opsin gene was aligned to the acquired genome sequence of the uv opsin gene with Exonerate. The exon and intron regions were illustrated with GenePalette.

Opsin phylogeny and tests of selection

Blastp was performed using the acquired amino acid sequences of opsins in T. kuznetsovi as queries for non-redundant protein sequences in the NCBI database (Table S4). Amino acid sequences of opsins in T. kuznetsovi, four beetle species (P. chalceus, Gyrinus marinus, Thermonectus marmoratus and T. castaneum) and a honeybee (Apis mellifera) were aligned with MUSCLE in MEGA v 11 [41]. Based on the maximum likelihood method, a phylogenetic tree of nucleotide sequences was reconstructed under the best-fit GTR + G + I model with 1000 bootstrap generations.

The ancestral sequences of opsin gene sequences between subterranean T. kuznetsovi and surface P. chalceus were estimated, based on the above five beetles’ phylogenetic relationship using MEGA. Based on the maximum likelihood method [42], the ratios of nonsynonymous (Ka) to synonymous (Ks) nucleotide substitution rates were calculated between an ancestral sequence and a sequence of T. kuznetsovi and between the ancestor sequence and a sequence of P. chalceus using KaKs_Calculator v 3.0 [43]. Fisher’s exact test on a 2 × 2 contingency table was conducted using the number of synonymous and nonsynonymous sites and synonymous and nonsynonymous substitutions.

The Ka/Ks analysis is able to suggest that observed changes in a sequence have been influenced by positive selection (Ka/Ks > 1), neutral evolution (Ka/Ks = 1), or negative (purifying) selection (Ka/Ks < 1). In our study, the apparent result is expected to be that opsin genes of T. kuznetsovi were under purifying selection, because along the evolutionary branch from an ancestor to T. kuznetsovi, opsins will have been selected under surface habitat before this lineage colonized subterranean habitat. To resolve this problem, we also compared the degrees of purifying selection between opsin genes of T. kuznetsovi (test) and P. chalceus (reference), carrying out branch-by-branch analyses with RELAX in Hyphy [44]. As the result of this model, k < 1 is indicative of relaxed selection, while k > 1 is indicative of purifying selection.

Histological study

The internal structure of a vestigial compound eye in T. kuznetsovi adults was observed using paraffin sections. Adult heads were fixed in 50% alcohol Bouin solution (ethanol:Bouin solution [for pathology, Fujifilm Wako Pure Chemicals] = 1:1) at room temperature overnight or longer. The fixed samples were rinsed in 70% ethanol, dehydrated in increasing concentrations of ethanol (90, 95 and 100%) [45, 46], and then cleared in xylene. Next, the samples were embedded in paraffin (Paraplast Plus; Sigma Aldrich, MO, USA), and transverse sections (6 µm) were serially cut with a microtome (OSK 97LF506; Ogawa Seiki, Tokyo, Japan). Sections were stained with hematoxylin, observed with a microscope (CX-43; OLYMPUS, Tokyo, Japan) and photographed with a mounted camera (EOS Kiss X9; Canon, Tokyo, Japan).

Because we could not obtain the complete series of cross sections due to their friability, the nerve bundle between a compound eye and a brain in T. kuznetsovi adults was observed with dissection. Adult heads were dissected in phosphate-buffered saline and stained with 0.5% methylene blue solution (22409-32; Nakalai Tesque, Kyoto, Japan) for 1 h. The samples were washed in phosphate-buffered saline, observed with a stereo microscope (SZX16; OLYMPUS) and photographed with a mounted camera (EOS Kiss X9; Canon).

Results

DNA and RNA sequencing

The genome size was estimated to be 554,652,206 bp based on the k-mer frequency distribution of genome reads with KmerGenie. The assembled genome contained 55,616 scaffolds with a total length of 456,726,283 bp (N50: 16,592 bp) and 96.20% BUSCO completeness (Table 1). The assembled transcripts contained 71,303 contigs with a total length of 67,456,903 bp (N50: 1,883 bp) and 89.10% BUSCO completeness (Table 1). 81.90% of RNA reads were mapped to the assembled genome with HISAT. The information of paired-end reads is summarized in Table S1.

Table 1 The summary statistics for assembly of the genome and transcripts in T. kuznetsovi
Full size table

Expression of photoreceptor genes and phototransduction genes

In the assembled genome, BLAST search found lw opsin gene, uv opsin gene and non-visual c-opsin gene (Table 2). lw opsin and c-opsin were found in the assembled transcripts, but uv opsin was not. Fifteen BLAST-searched phototransduction genes were found in the assembled genome. Out of those, transcripts for 14 BLAST-searched phototransduction genes were detected, namely, Arr1, Arr2, Gα49B, Gβ76C, Gγ30A, Gprk1, inaD, ninaA, ninaC, norpA, Pkc53E, Rab6, rdgC and trp. Most of these transcript sequences had neither a premature stop codon nor a frameshift mutation, but transcript sequences of Gα49B, Gβ76C, Gprk1, norpA and rdgC included apparent functional isoforms and those with premature stop codons or frameshift mutations. These transcripts might include primary transcripts before splicing. One phototransduction gene, trpl, was not detected in the Trinity-assembled transcripts. Short read sequences obtained by RNA sequencing were mapped to the genome scaffold of the trpl gene. We performed an additional RT-PCR experiment, but no clear PCR product was observed. This probably means that the trpl gene was expressed at a very low level.

Table 2 Photoreceptor and phototransduction genes detected in the assembled genome and transcripts
Full size table

Opsin genes

One lw opsin gene of T. kuznetsovi was identified with BLAST search. The gene was divided into three genome scaffolds and two transcript contigs (Table S2). The scaffolds and contigs were joined together using PCR and RT-PCR, and then exon and intron regions of the lw opsin gene were determined (Fig. 2A). The conceptually translated Lw opsin amino acid sequence was 379 residues and consisted of six exons. There was neither a premature stop codon nor a frameshift mutation in the coding sequence.

Fig. 2
figure 2

The structure of opsin genes in T. kuznetsovi. A The putative structure of lw opsin gene, which consists of six coding exons. B The putative structure uv opsin gene, which consists of six coding exons

Full size image

One uv opsin gene of T. kuznetsovi was identified by performing BLAST search. The gene was present within one scaffold in the genome and no contig was found in the transcripts (Table S2). The cDNA sequence of the uv opsin amino acid sequence was determined using RT-PCR, and then exon and intron regions of the uv opsin gene were determined (Fig. 2B). The conceptually translated Uv opsin amino acid sequence was 373 residues and consisted of six exons. There was neither a premature stop codon nor a frameshift mutation in the coding sequences.

Opsin phylogeny and selective pressure

A molecular phylogenetic tree of opsin genes of T. kuznetsovi, P. chalceus, G. marinus, T. marmoratus and T. castaneum and A. mellifera was reconstructed (Fig. 3, Table S4). The lw opsin and uv opsin of T. kuznetsovi were clustered with those of P. chalceus, in accordance with their taxonomic relationship. The branch length of opsin genes in T. kuznetsovi was not extended long.

Fig. 3
figure 3

A maximum likelihood tree for visual opsin genes. Bootstrap probabilities are provided on nodes. OTU names consist of the abbreviated species names, gene names and accession numbers: Tkuz, Trechiama kuznetsovi; Pcha, Pogonus chalceus; Gmar, Gyrinus marinus; Tmar, Thermonectus marmoratus; Tcas, Tribolium castaneum; Amel, Apis mellifera

Full size image

In lw opsin genes, Ka/Ks ratio was 0.158131 between the sequences of the ancestor and T. kuznetsovi (p = 1.12862e–013, Fisher’s exact test), and Ka/Ks ratio was 0.0275072 between the sequences of the ancestor and P. chalceus (p = 2.70467e–109, Fisher’s exact test) (Table 3). In uv opsin, Ka/Ks ratio was 0.138044 between the sequences of the ancestor and T. kuznetsovi (p = 1.34285e–024, Fisher’s exact test), and Ka/Ks ratio was 0.0695552 between the sequences of the ancestor and P. chalceus (p = 7.58434e–092, Fisher’s exact test). In all of these cases, Ka/Ks ratios were far below 1.0, indicating that opsin genes have been under negative (purifying) selection in both the lineage leading to T. kuznetsovi and that leading to P. chalceus.

Table 3 The test of selection in opsin genes of T. kuznetsovi
Full size table

Subsequently, the difference in the degree of the purifying selection between the lineages from the ancestor to T. kuznetsovi and to P. chalceus was tested. According to the result of Relax analysis in Hyphy, k value was 0.61 (p = 0.225) in lw opsin and 2.14 (p = 0.539) in uv opsin (Table 3). Because this analysis did not show a statistical significance, we were unable to conclude whether the selection on opsin genes was relaxed or intensified in the lineage leading to T. kuznetsovi compared to the control lineage.

Internal structure of a compound eye

Putative photoreceptor cells stained by hematoxylin were observed in the internal structure of a vestigial compound eye in a T. kuznetsovi adult (Fig. 4A). The surface was covered by a transparent cuticle, a cornea. By observation from the outer surface of the head, we could see a transparent cornea and ocular ridge with black pigmentation (Fig. 1D). There was no pigmentation in cells within the eye structure, unlike compound eyes of other carabid beetles [47]. No crystalline cones or any similar structure were found [48]. An optic stalk, which is a nerve bundle connecting a compound eye and a brain, was observed [49] (Fig. 4B).

Fig. 4
figure 4

The internal structure and nerve tissue of T. kuznetsovi head. A Transverse section of a compound eye, dorsal is up. Scale bar indicates 20 µm. CO, cornea; R, putative photoreceptor cells; OR, ocular ridge. B Dorsal view of dissected head. Nerve tissue was stained with methylene blue solution. Scale bar indicates 0.5 mm. CE, compound eye; NB, nerve bundle; BR, brain (supraesophageal ganglion)

Full size image

Discussion

To understand the process of subterranean colonization of organisms, the question of whether shallow subterranean habitats are a gateway to colonizing deep zones has been featured in subterranean biology [3, 4]. In the present study, we focused on a trechine beetle, T. kuznetsovi, which inhabits the upper hypogean zone and has a vestigial compound eye [18]. We evaluated the ability of photoreception in T. kuznetsovi by genomics and histological observation.

We identified one lw opsin gene and one uv opsin gene in the genome and in the transcripts in the adult. No frameshift mutation or premature stop codon was found in these exon regions, Ka/Ks ratios were significantly less than 1.0, and there was no significant difference in the selective pressure between evolutionary lineages of subterranean T. kuznetsovi and surface P. chalceus. These analyses implied that Lw opsin and Uv opsin are under functional constraint. Transcripts of 14 phototransduction genes without deleterious mutations (premature stop codons or frameshift mutations) were detected in the assembled transcripts. One phototransduction gene, trpl, was found in RNA short-read sequences. These results suggested the ability of photoreception and phototransduction of T. kuznetsovi. In our preliminary study using LED light, we observed that adults of T. kuznetsovi showed clear negative phototaxis to UV light and probably also to green light (data not shown).

In subterranean diving beetles in Western Australia, the lw opsin gene became a pseudogene due to frameshift mutations, and neither lw opsin nor uv opsin transcripts were observed [14,15,16]. Frameshift mutations and premature stop codons occurred in some phototransduction genes: Arr1, Arr2, ninaC, trp and trpl [16]. There are two possible causes for these differences between T. kuznetsovi and the subterranean diving beetles. The first possibility is the difference in their ecological niches. The calcrete aquifer, in which subterranean diving beetles live, is located at a depth of 10–30 m underground [13, 50]. In contrast, the upper hypogean zone, in which T. kuznetsovi adults live, is a few or some dozen centimeters below the slope surface of a v-shaped valley. Trechiama kuznetsovi adults would occasionally be exposed to the surface due to landslides occurring as a result of precipitation or earthquakes [51,52,53]. This temporary light stimulus could function as a selective pressure to retain the function of opsin genes and other phototransduction genes. For example, vestigial compound eyes could function in the response to the temporary light. Friedrich et al. [32] also suggested such a function for the retention of photoreceptor genes and phototransduction genes in Ptomaphagus hirtus (Leiodidae) inhabiting the Frozen Niagara cave entrance of Mammoth Cave in Kentucky, USA. Beetles in shallow subterranean environments (the upper hypogean zone and cave entrance) could tend to maintain the photoreception. The second possibility is that opsin genes and other phototransduction genes of T. kuznetsovi have not had enough time to become pseudogenes. However, we think this scenario is less likely because they had enough time to degenerate the eye itself. Niemiller et al. [54] showed that some cave lineages in Amblyopsidae still possess functional rhodopsin, although they inhabit an aphotic environment. This retained functionality is thought to be due to insufficient accumulation of mutations during recent subterranean colonization. As in these cave lineages, pseudogenization of opsin genes in T. kuznetsovi could not be observed because divergence between T. kuznetsovi and related terrestrial species occurred recently. To further examine this possibility, the divergence time of Trechiama species needs to be studied.

By performing paraffin sectioning and dissection, we observed the cells inside the compound eye and the optic stalk connecting the compound eye and the brain in T. kuznetsovi. These observations suggested that photoreception is structurally possible even with the vestigial compound eyes of this species. Complete loss of compound eyes and optic lobes was observed in Sinaphaenops wangorum (Trechinae) inhabiting the deep area of a cave in Guangxi Autonomous Region, China [55]. Thus, the existence of the optic stalk is thought to be due to the retention of photoreception ability by T. kuznetsovi, not to an inability of the visual system to degenerate any further.

Collectively, the results of genomics and histology analyses performed here suggested the ability of photoreception in T. kuznetsovi. This species is thought to possess both a surface trait (photoreception) and some subterranean traits (vestigial compound eye, underdeveloped body pigmentation and other morphological adaptations) [56, 57]. These characteristics would reflect an intermediate phase toward colonizing a deeper subterranean niche. Further understanding of the visual degeneration process will be achieved by clarifying the phylogenetic relationship between subterranean species and surface species of trechine beetles.

Conclusions

By de novo assembly of genome and transcript sequences, we identified photoreceptor genes and phototransduction genes of a trechine beetle, Trechiama kuznetsovi, which inhabits the upper hypogean zone. The encoded amino acid sequences of lw opsin and uv opsin had neither a premature stop codon nor a frameshift mutation, and appeared to be subject to purifying selection. We identified potential photoreceptor cells in the compound eye and nerve bundle connected to the brain. The present findings suggest that T. kuznetsovi has retained the ability of photoreception.