Structural characteristics of the melon mitochondria genome

Liverwort (Marchantia polymorpha) was the first complete mitochondrial genome sequenced^[1]. In Cucurbitaceae, sequencing of the mitochondrial genome of watermelon, zucchini^[2] and cucumber^[3] have been completed. Ward et al.^[4] suggested that the mitochondrial genome of melon was about 2.9 Mb. In 2011, the mitochondrial genome sequence of melon was reported, mainly composed of five scaffolds and four contigs^[5]. In 2020, we reported the completed mitochondrial genome of melon including a main loop and two small loops^[6]. Melon mitochondrial genome can provide additional genetic information, such as cytoplasmic male sterility and mitochondrial RNA editing. Therefore, further research focusing on the mitochondrial genome of melon is expected.

Plant mitochondrial genomes have more complex structural characteristics. The size of plant mitochondrial genomes varies greatly between plants, ranging from 187 kb in liverwort^[1] to 11.3 Mb in flycatcher^[7], but there is no linear relationship between the size of the mitochondrial genome and the number of genes it encodes, flycatcher has 32 coding genes, but ground money has 74 coding genes. The GC content of the mitochondrial genome is about 43%−45%. The number of protein-coding genes in plant mitochondrial genomes is typically 30−50, mainly including complex I-V genes, cytochrome C biosynthetic genes, ribosomal protein genes, matR genes, mttB genes and ORFs of unknown function. The non-protein-coding genome includes rRNA and tRNA genes. rRNA includes 5s rRNA, 18s rRNA and 28s rRNA; tRNA includes three different types of tRNAs, including its own intrinsic tRNA and tRNAs transferred from the chloroplast and nuclear genomes, but with the exception of liverwort, the mitochondrial genomes of all other plants do not cover tRNA genes encoding all 21 amino acids.

RNA editing refers to a predetermined codon modification caused by nucleotide changes at the RNA level^[8]. This molecular process mainly exists in chloroplasts and mitochondria of plants, where it maintains the normal biological functions of these organelles. Mitochondrial RNA editing mostly occurs along protein coding regions. In this case, RNA editing can increase the conservation of the encoded protein product, with regards to its primary structure, between different species^[9]. The number of RNA editing sites can vary greatly among different species. There are only 11 RNA editing sites in the mitochondrial genome of the moss Physcomitrella patens^[10], in contrast to 456 and 692 RNA editing sites identified, respectively, in the mitochondria of Arabidopsis thaliana^[11] and Gossypium spp^[12]. Lu et al.^[13] studied the process of RNA editing in eight species from four families of gymnosperms and found a substantial difference in the number of RNA editing sites as well as their positions along the DNA of distinct families and genera of gymnosperms. Mitochondrial RNA editing can affect many important traits in plants. RNA editing of the cotton mitochondrial Ghatp1 gene, at the C1292 and C1415 loci, affects ATPase production and promotes epidermal hair and fiber elongation^[12]. In tomato, a decreased RNA editing of nad3 and sdh4 genes can disrupt the biological function of mitochondria, therefore, reduce the respiratory efficiency of the fruit, which can ultimately inhibit its ripening^[14]. Inadequate and deviated mitochondrial RNA editing may be associated with certain biological conditions, such as cytoplasmic male sterility^[15]. The structural analysis of the mitochondrial genome will provide a theoretical basis for an in-depth study of the genetic characteristics of melon mitochondria.

Mitochondrial genomes were sequenced and assembled using dark-treated MR-1 yellowing seedlings as plant material^[6].

Genome annotation

The annotation of protein-coding genes in the mitochondrial genome was carried out using BLASTN and BLASTX to search the nucleotide and protein libraries of the NCBI/GenBank database. rRNA and tRNA annotations were carried out using RNAmmer (rnammer -S euk-m lsu,ssu,tsu -xml melon.xml -gff melon.gff -hmelon.hmmreport < melon.fsa) and tRNAScan-SE (tRNAscan-SE -o tRNA.out -f rRNA.ss -m tRNA.stats /home/gjs/fasta/h.fa).

Forward repeats in the mitochondrial genome were analyzed using Reputer software^[16] with minimum repeat size set to 20 bp, hamming distance of 0, and similarity greater than 90%. Inverted repeats were analyzed using IRF (Inverted Repeats Finder) software^[17] with parameters set to: match 2, mismatch 3, delta 5, match probability 80, indel probability 10, min-score 40, max-length to report 500,000, max-loop 500,000. Tandem repeat was analyzed using TRF (Tandem Repeat Finder) software with parameters set to: min. align. score 50, max. period size 500. Simple sequence repeat was analyzed using MISA software^[18] with parameters set to 1 base. The parameters were set to 10 and more repetitions of 1 base, five and more repetitions of 2 bases, four and more repetitions of 3 bases, four and more repetitions of 4 bases, four and more repetitions of 5 bases, four and more repetitions of 6 bases, and only those base repetitions meeting the criteria were considered as microsatellite sequences.

Comparative analysis of mitochondrial genomes

Analyses were delineated using Easyfig^[19], and then illustrated with MapChart 2.2^[20]. Comparative analysis of sequences between organelle and nuclear genomes was performed by BLASTN and Tbtools.

Prediction of mitochondrial RNA editing sites

The RNA editing sites of all protein-coding gene sequences, located in the mitochondrial genomes of melon (Cucumis melo), cucumber (Cucumis sativus), watermelon (Citrullus lanatus) and zucchini (Cucurbita pepo), were predicted by PREP-Mt^[21].

Phylogenetic analysis

Mitochondrial genomes were aligned using ClustalX (www.clustal.org/clustal2). A phylogenetic tree was constructed via Neighbor Joining (NJ) using Mega7 software.

The melon mitochondrial genome contains 4,861 pairs of forward repeats, 439 pairs of inverted repeats, 653 tandem repeats and 218 SSR sequences. The total length of these repeats is about 44.2% of those detected in the whole genome. The coding genes accounted for 1.54% of C. melo mitochondrial genome (Table 1). In fact, non-coding gene sequences accounted for 98.46% of the mitochondrial genome, but their function still require more detailed characterization.

Collinearity analysis of mitochondrial genomes

By comparing the mitochondrial genomes of four cucurbit plants, we discovered that sequences, with a consistency of no less than 80% in C. sativus, C. pepo and C. lanatus, accounted for 33%, 40% and 65% of the whole mitochondrial genome length in melon, respectively. The sequence shared by the four species accounts for about 6% of the full length of the melon mitochondrial genome. In addition, both melon and the other three crops have similar gene coding regions, but the non-coding regions are quite distinct (Fig. 1).

Figure 1. 

Collinearity analysis of C. melo mitochondrial genome with other three Cucurbitaceae plants.

The linear relationships of mitochondrial coding genes among C. melo, C. sativus, C. pepo and C. lanatus were compared. Two or more collinear gene groups are called gene clusters, and we get 7 to 13 gene clusters with different gene numbers (Table 2). A higher number of collinear gene clusters was identified in C. melo and C. sativus, contrarily, a fewer number of collinear gene clusters was found in C. pepo and C. lanatus.

Comparison of four mitochondrial genomes

There were no significant differences in mitochondrial genomic GC content among C. melo, C. sativus, C. pepo and C. lanatus (between 42.8 and 45.1%, Table 3). The total number of repeat sequences in the melon mitochondrial genome was the highest among all four species, accounting for 44.2% of the total genome sequence. The number of repeat sequences in the mitochondrial genome of C. sativus, C. pepo and C. lanatus was 44.1%, 24.4% and 9.6% of the total genome, respectively.

A total of 40 protein-encoding genes were annotated in the mitochondrial genome of melon. The number of protein-encoding genes in the mitochondrial genome of C. sativus, C. pepo and C. lanatus was 37, 38 and 39, respectively. Melon mitochondrial genome lost rps19 gene, two copies of atp1 gene, and two more ORF genes (orf1 and orf2). The mitochondrial genome of these four plants contained three different types of ribosomal genes (rrn5S, rrn18S and rrn26S). The highest number of rRNAs were detected in melon, including six copies of rrn5S, and two single copies of rrn18S and rrn26. Six rRNA genes were present in the cucumber mitochondrial genome, and each ribosomal gene presented two repeats. Only three rRNAs were observed in watermelon and zucchini, all of which were single-copy genes (Table 4).

We compared the tRNA use of C. melo, C. sativus, C. pepo and C. lanatus (Table 5). The results showed that the mitochondrial genome sequences of all four species could not encode a complete set of tRNA that could recognize all codons or transport a complete set of 20 amino acids.

Prediction of mitochondrial RNA editing sites

Our results show that C. lanatus had the largest number of RNA editing sites along its mitochondrial genome (509 sites). C. melo and C. sativus contained 507 and 498 RNA editing sites, respectively. C. pepo presented the lowest number of RNA editing sites (total of 486). Nad4, ccmB and ccmFn were the three genes with the highest RNA editing sites (Table 6).

Phylogenetic analysis

In this study, the phylogenetic relationship of C. melo, C. sativus, C. pepo and C. lanatus were analyzed based on 10 conserved coding genes (atp6, nad6, cox3, rps12, atp1, nad4, nad9, nad7, nad4L) that are present in all 14 species. Evolutionary relationships were analyzed (Fig. 2) and it was found that grapes and Cucurbitaceae are closely related, and Cucurbitaceae can cluster well in a clade, but the genetic distance of C. pepo in the phylogenetic tree is closer to C. melo and C. sativus compared to C. lanatus.

Figure 2. 

The construction of phylogenetic tree among 14 species based on mitochondrial conserved genes.

Our study corroborates with previous reports that utilized renaturation kinetics and restriction endonuclease technology to analyze the mitochondrial genome analysis of Cucurbitaceae^[4]. The structure of both C. melo and C. sativus mitochondrial genomes is polycyclic. Likewise, this genome structure has been observed in other plants, including wheat^[22] and rape^[23].

The protein-coding genes in melon mitochondrial genome are like those in the three mitochondrial genomes. Similar conclusions have been reached in studies of other higher plants, where the coding regions of plant mitochondrial genomes are more conserved than the non-coding regions^[24]. Melon mitochondria contain two more ORF genes than the other three reference genomes, and ORFs may encode proteins with important functions. In fact, some mitochondrial ORFs have been associated with cytoplasmic male sterility in many plants.

With the expectation of T-urf13, atp6 of radish and orf256 of wheat, it has been verified that most of the transcripts of protein-coding genes are edited in the mitochondria of higher plants, but the editing degree of transcripts from different genes is variable^{[25, 26]}. The male sterility of plants may be caused by some genes in the mitochondrial genome recombining with ORF to form chimeric genes or inadequate RNA editing^[27]. In the present study, a slightly distinct number of RNA editing sites was detected in the mitochondrial genomes of four Cucurbitaceae crops. The elucidation of these RNA editing sites may provide data support for the research of RNA editing in cucurbits.

The mitochondrial genomes of both melon and three other kinds of Cucurbitaceae plants were linearly analyzed. C. melo and C. sativus showed more collinearity between mitochondrial gene clusters, which further shows that C. melo has a closer relationship with C. sativus. In the four kinds of Cucurbitaceae plants some gene cluster, rps3-rpl16 is widespread in the plant mitochondrial genome typical gene cluster^[28].

This study upon in-depth comparative analysis of the mitochondrial gene structure of C. melo, C. sativus, C. pepo and C. lanatus, revealed that the large number of repetitive and nuclear genome sequences were the potential reasons for the increasing scale and variation of the melon mitochondrial genome. These results provide the basis for the genetic variation of the mitochondrial genome in Cucurbitaceae plants.