Complete Mitochondrial Genomes of Two Corydoras (Siluriformes, Callichthyidae) and their Phylogenetic Implications

Fish comprise the most primitive and dominant group of vertebrates in terms of the number of species and genera (Compagno, 1990; Lévêque et al., 2008). They include a wide range of species with widespread distribution and a complex origin. Studying their genetic differentiation and clarifying their evolutionary paths have always been interesting topics (Kelsh, 2004). In recent years, with the widespread adoption of molecular biology technology in various research fields, studying the genetics and evolution of fish at the molecular level has become increasingly attractive (Glasauer and Neuhauss, 2014; Hauser and Carvalho, 2008). At the molecular level, it is important to select appropriate molecular markers when studying the genetics and evolution of fish. DNA molecules contain a large amount of information on genetic variation, from which we can obtain a more objective understanding of the evolution of organisms. The biological characteristics of mitochondrial DNA render its haplotype tree more consistent than nuclear autosomal gene and species trees, and mitochondrial DNA is often used to estimate the evolutionary history of biological groups (Avise, 2009; Moore, 1995). Fish mitochondrial DNA, similar to that of many other vertebrates, comprises covalently closed, circular, and double-stranded molecules that are closely arranged (Boore, 1999; Hurst et al., 1999; Liu et al., 2015; Sun et al., 2021). The mitochondrial DNA of fish is generally 15–20 kb in size. Mitochondrial genomes vary considerably among different species. Tandem repeats and a few scattered repeats are present in their sequences, similar to those in other vertebrates (Boore, 1999). The mitochondrial genome of fish is composed of 13 protein-coding genes (PCGs), two ribosomal RNA genes (rRNAs), and 22 transfer RNA genes (tRNAs), as well as the control region (D-loop) and light chain replication initiation region related to the non-coding region of heavy chain replication initiation. With the gradual maturation of DNA sequencing technology, fish mitochondrial genomes have been widely used as molecular markers for fish germplasm resource protection, population polymorphism analysis, and phylogenetic development (Jia et al., 2020; Saha et al., 2021; Zheng et al., 2021).

Corydoras are members of the genera Aspidoras Ihering, 1907, Brochis Cope, 1871, and Corydoras Lacépède, 1803 in the subfamily Corydoradinae (Bernt et al., 2013). Owing to two small barbells on the sides of their mouths, the fish resemble mice swimming in water, hence the name Corydoras. The dwarf species Corydoras hastatus Eigenmann and Eigenmann, 1888 has a silvery white body and a black tail handle with a white frame, which is of great ornamental value (Britto, 2003; Menni et al., 1992). The fish are mostly located in the Mato Grosso Plateau in Brazil. Corydoras hastatus is not a benthic fish. It is characterized by an obvious cross-shaped black spot on its tail handle, and it often swims in groups of other fish species with a very similar appearance, such as Serrapinnus kriegi (Serra et al., 2018), in native waters, forming a symbiotic relationship. It is as small as a lampfish and the least hungry species of Corydoras (max length: 2.4 cm). Corydoras cruziensis is characterized by a bright orange head and back, metallic green body, short snout, and a round figure (Knaack, 2002).

Building upon the study on C. aeneus and C. paleatus (Sevilla et al., 2007; Sun et al., 2022), we sequenced, assembled, and annotated the complete mitochondrial genomes of C. hastatus and C. cruziensis. Using the newly sequenced genomes and 12 complete mitochondrial genomes of the genus Corydoras available in the NCBI database, we aimed to conduct a comprehensive analysis that will provide a reference for the taxonomy, evolutionary genetics, and interspecific identification of the genus Corydoras.

Single fresh specimens of the two target species were collected from a wholesale flower, bird, and fish market in Mudanjiang City, Heilongjiang Province (44°35′20.08″N, 129°36′31.87″E), in January 2022. After euthanization, specimens were immersed in absolute ethanol and stored in a freezer at −80 °C until use. Total DNA was extracted from muscle tissue using a Magen Hi Pure Inspect DNA Micro Kit following the manufacturer’s instructions. The quality and purity of the extracted DNA were tested using agarose gel electrophoresis and a NanoDrop 2000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). Species identification was verified using morphological characteristics (Alexandrou et al., 2011; Burgess, 1992), Cytb (Sevilla et al., 2007), and the 16S rRNA gene (Alexandrou et al., 2011) combined. The animal study protocol was approved by the Ethics Committee of Mudanjiang Normal University (Jan 12, 2022).

The qualifying total genomic DNA was sent to Wuhan Benagen Biotechnology Co., Ltd., where the whole-genome shotgun method was used to build the library and next-generation sequencing technology was used to conduct high-throughput sequencing on the Illumina NovaSeq 6000 sequencing platform. SPAdes v3.11.1 (Bankevich et al., 2012) was used to assemble high-quality second-generation sequencing data from scratch. Contigs and scaffolds were constructed using the default parameters. Using C. aeneus MZ571336 as the reference sequence, we conducted the collinearity analysis, determined the positional relationship between segment overlapping groups, and filled in the missing sequences between the overlapping groups in MUMmer v3.1 (Kurtz et al., 2004). Pilon v1.18 (Walker et al., 2014) was used to correct the results and obtain the final mitochondrial sequence. The complete mitochondrial genome sequence obtained through splicing was functionally annotated using the MITOS web server (http://mitos.bioinf.uni-leipzig.de/) (Bernt et al., 2013). The genetic code was set to vertebrate, and other settings followed the default parameters of MITOS. The annotation results were further verified after manual correction and using MitoFish (Iwasaki et al., 2013) (http://mitofish.aori.u-tokyo.ac.jp/), online tools, and tRNAscan-SE v1.3.1 (Lowe and Eddy, 1997). Circular mitochondrial genome maps were generated in MitoFish (Iwasaki et al., 2013).

The two newly obtained mitochondrial genomes were combined with 12 published mitochondrial genomes from the genus Corydoras for genome analysis. Base compositions and genetic distances were determined using MEGA v7.0 (Kumar et al., 1994). PhyloSuite v1.2.2 (Zhang et al., 2020) was used to calculate the number of codons, AT and GC content, AT skew [AT skew = (A – T)/(A + T)], and GC skew [GC skew = (G – C)/(G + C)] in the mitochondrial genome (Perna and Kocher, 1995). The Ka/Ks ratio and nucleotide diversity for the 14 Corydoras species were calculated using DNAsp v5.1 (Librado and Rozas, 2009). These results were plotted in origin v2018 (Moberly et al., 2018).

For the analysis, we created a dataset of 14 Callichthyidae mitochondrial genomes (12 Corydoras, one Brochis, and one Hoplosternum available in the NCBI database) and the two newly sequenced genomes; Hyphessobrycon amandae MT484069 (Sun et al., 2021) in the Characidae family was selected as an outgroup (Table I). IQ-TREE v1.6.8 (Nguyen et al., 2015) integrated in PhyloSuite was used to build a maximum likelihood (ML) phylogenetic tree based on 13 PCGs and two rRNAs. The best partition model was screened using ModelFinder (Kalyaanamoorthy et al., 2017). Branch confidence was assessed by 200,000 ultrafast bootstrap replicates (Minh et al., 2013) and the Shimodaira–Hasegawa-like approximate likelihood-ratio test (Guindon et al., 2010). Mbayes v3.2.6 (Huelsenbeck and Ronquist, 2001) was used to construct a Bayesian inference (BI) phylogenetic tree under the partition model (two parallel runs of 20,000,000 generations each), in which the initial 25% of sampled data was discarded as burn-in. The optimal partition models for ML and BI are listed in Table II. iTOL (Letunic and Bork, 2016) was used to visualize the resulting phylogenetic trees.

The total lengths of the mitochondrial genomes of C. hastatus and C. cruziensis were 16,518 bp (GenBank accession number: OP562095) and 16,531 bp (GenBank accession number: OP562096), respectively. Complete mitochondrial genomes are double-chained rings consisting of a heavy chain (J strand) and a light chain (N strand) (Fig. 1). Both mitochondrial genomes contained 37 genes (13 PCGs, 22 tRNAs, and two rRNAs) and one D-loop

non-coding control region (Table III). There were nine genes on the N chain, namely tRNA-Gln, tRNA-Ala, tRNA-Asn, tRNA-Cys, tRNA-Tyr, tRNA-Ser, tRNA-Glu, tRNA-Pro, and ND6, and the remaining 28 genes were on the J chain. The length of most of the gene sequences and the interval repeats were identical between C. hastatus and C. cruziensis (Table III). There were 11 spacer regions in the mitochondrial genomes of C. hastatus and C. cruziensis, with spacer lengths of 69 and 70 bp, respectively. The largest spacers were between tRNA-Asn and tRNA-Asn, comprising 31 and 30 bp in the genomes of C. hastatus and C. cruziensis, respectively. Eight adjacent genes in the two genomes overlapped. The maximum overlap of 13 bp was between COI and tRNA-Ser, and the minimum overlap of 1 bp was between tRNA-Gln and tRNA-Met and between

tRNA-Cys and tRNA-Tyr. The start and stop codons of the PCGs in C. hastatus and C. cruziensis were identical and similar to other Corydoras species. The start codons were ATG and GTG, and the stop codons were TAG, TAA, and the incomplete stop codon T-.

The two mitochondrial genomes obtained in this study were combined with 12 published mitochondrial genomes for a comprehensive analysis that included the base composition, base bias, paired genetic distance, nucleotide diversity, Ka/Ks ratio, and number of codons. The mitochondrial genomes with the 13 PCGs and two rRNAs of the 14 Corydoras species showed a positive AT skew but a negative GC skew, except for ND6 (Fig. 2), which has also been reported in a mitochondrial genome study on other fish species (Ruan et al., 2020). Among the 13 PCGs and two rRNAs, the AT content of ATP8 and ATP6 was the highest, and that of ND4L was the lowest (Fig. 3). The GT content exhibited the opposite trend. Except for ND4 of C. agassizii and C. schwartzi, the AT content of all other genes was greater than the GC content, which coincided with the fact that the mitochondrial base composition of teleost fish exhibits a preference for A and T (Broughton et al., 2001; Sun et al., 2021).

To estimate the average divergence among the mitochondrial genomes of Corydoras, the overall mean K2P genetic distances were analyzed based on 13 PCGs (Fig. 4). Congruent results showed that both COII (0.073) and COIII (0.097) had the smallest genetic distance, whereas ND4 (0.136) had the largest, thereby representing the most conserved and the most variable genes, respectively. The results of the nucleotide diversity analysis (Fig. 4) were consistent with those of genetic distance.

To evaluate selection pressure (Lemos et al., 2005; Sun et al., 2020) on the mitochondrial genome of Corydoras fish, the Ka/Ks values of 13 PCGs in the mitochondrial genome were estimated, and a histogram of this ratio was constructed (Fig. 5). The Ka/Ks values of the 13 PCGs ranged from 0.006 to 0.072 and were less than one, indicating strong purification selection. The Ka/Ks values of COI (0.006) were the lowest, suggesting that this gene was under the greatest purifying selection pressure during evolution.

The PCGs of the 14 mitochondrial genomes of Corydoras were translated into 3,785–3,797 codons. Ile (307.35 ± 9.60 codons), Thr (312.58 ± 2.45 codons), Ala (312.52 ± 5.80 codons), and Leu1 (470.88 ± 14.61 codons) were the four predominant codon families (Fig. 5) and might be associated with the coding function of the chondriosome (Gu et al., 2022). In contrast, Cys (25.12 ± 0.83 codons) and Ser1 (52.82 ± 2.57 codons) were with the smallest number of codons.

The phylogenetic tree of C. hastatus, C. cruziensis, and 14 species of the family Callichthyidae, based on the tandem sequences of 13 PCGs and two rRNAs in the mitochondrial genome, was constructed using the ML and BI methods (Fig. 6). Consistent with previous studies (Alexandrou et al., 2011; Roxo et al., 2019; Sun et al., 2022), the ML and BI tree topologies were congruent, confirming the monophyly of the genus Corydoras; however, Brochis multiradiatus was clustered within this genus. Corydoras hastatus and C. pygmaeus formed a highly supported clade (BI potential probabilities, PP = 1; ML bootstrap, BS = 89), which is consistent with the results reported by Alexandrou et al. (2011). Corydoras cruziensis clustered with (Corydoras rabauti + Corydoras aeneus) in a highly supported clade (PP = 1; BS = 100). Fourteen species of the genus Corydoras clustered together quite well. Similar to Sun et al. (2022), we believe that the clustering of C. trilineatus and C. sterbai is attributable to identification errors, introgressive hybridization, or that the two names are homonyms.

In this study, the mitochondrial genomes of two Corydoras species were sequenced and assembled. The results showed that the sequenced gene arrangements were consistent with the putative ancestral fish mitochondrial genomes, as understood today. Comprehensive analysis of the two new and 12 published Corydoras mitochondrial genomes showed that the 14 mitochondrial genomes had similar AT and GC content, AT and GC skew, genetic distances, nucleotide diversity, number of codons, and Ka/Ks values.

The nucleotide diversity and genetic distance of PCGs in the Corydoras mitochondrial genomes showed that ND2 and ND4 were the most variable genes, whereas COII was the most conserved gene. An analysis of the selection pressures on each gene showed that COI was associated with the strongest purifying selection.

Phylogenetic analyses based on PCGs and rRNAs from the mitochondrial genomes of 17 species have thus clarified the phylogenetic relationships of Corydoras. The sister-group relationships between C. hastatus and C. pygmaeus and between C. cruziensis and (C. rabauti + C. aeneus) were well supported at the mitogenome level. Our findings also suggest that mitogenome sequences are effective molecular markers to study the phylogenetic relationships within Corydoras and Callichthyidae.

This research was funded by the Doctoral Research Fund of Mudanjiang Normal University [grant number 1002319042].

The animal study protocol was approved by the Ethics Committee of Mudanjiang Normal University.

The original contributions presented in this study are publicly available. This data can be found in the GenBank repository under accession numbers OP562095 and OP562096.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.