Mitochondrial Group I Introns in Hexacorals Are Regulatory Genetic Elements

Hexacoral mitochondrial genomes are highly economically organized and vertebrate-like in size, structure, and gene content. A hallmark, however, is the presence of group I introns interrupting essential oxidative phosphorylation (OxPhos) genes. Two genes, encoding NADH dehydrogenase subunit 5 (ND5) and cytochrome c oxidase subunit I (COI), are interrupted with introns. The ND5 intron, located at position 717, is obligatory in all hexacoral specimens investigated. The ND5-717 intron is a giant-sized intron that carries several canonical OxPhos genes. Different modes of splicing appear to apply for the ND5-717 intron, including conventional cis -splicing, backsplicing, and trans -splicing. Three distinct versions of hexacoral COI introns are noted at genic positions 884, 867, and 720. The COI introns are of the mobile-type, carrying homing endonuclease genes (HEGs). Some COI-884 intron HEGs are highly expressed as in-frame COI exon fusions, while the expression of COI-867 intron HEGs appear repressed. We discuss biological roles of hexacoral mitochondrial ND5 and COI introns and suggest that the ND5-717 intron has gained new regulatory functions beyond self-splicing.

Hexacorals have a global marine distribution pattern typically recognized in tropical seas at shallow waters living in close relationships with endosymbiotic photosynthetic alga. However, coral reefs and sea anemones in deep offshore waters have more recently been investigated [3][4][5][6]. These cold-water hexacorals occur in low temperatures at high latitudes or great depths. Among the approximately 1500 stony coral species known, 50% are located in cold-water habitats [7,8]. A common feature among cold-water deep-sea hexacorals is that they are non-endosymbiotic in respect to the photosynthetic alga.
Group I introns are intervening sequences interrupting functional genes in eukaryotic (mitochondrial, chloroplast, nuclear, viral) and prokaryotic (eubacterial, archaeal, phage) genomes [21]. Like other mobile genetic elements, horizontal transfer of a group I intron can affect the host by altering the function of surrounding genes, potentially interrupting vital processes but also creating diversity and beneficial alterations. Mitochondrial group I introns in metazoans are rare and restricted to some orders within the basal phyla of Placozoa, Porifera, and Cnidaria [11,22]. Unlike spliceosomal introns, which are abundant in the nuclear genome of eukaryotes, group I introns encode catalytic RNAs (ribozymes) with the unique  [23]. The nine conserved secondary structure paired segments of the catalytic core (P1-P9) are shown, and the three tertiary domains (scaffold, substrate, catalytic) are indicated by blue, yellow, and green boxes, respectively. Essential nucleotide positions in P1 (U, G), P7 (G, C), and P9 (G) are indicated in red. 5 0 , upstream exon sequence; 3 0 , downstream exon sequence. ability to self-splice as naked RNA. These introns sometimes even code for homing endonucleases, giving additional mobility to the ribozymes. The intron RNA processing reaction is catalyzed by the ribozyme, which folds into at least nine conserved paired segments (P1-P9), further organized into hallmark helical stacks named the catalytic domain, the substrate domain, and the scaffold domain ( Figure 1B) [23][24][25]. Group I intron sequences are removed from precursor transcripts in a guanosine-dependent two-step transesterification reaction, leading to exon ligation and intron excision [21].
This chapter reviews recent developments in the characterization of hexacoral mitochondrial genomes with a focus on gene organization and rearrangements, complex obligatory group I introns in the NADH dehydrogenase subunit 5 (ND5) gene, and mobile-type group I introns in the cytochrome c oxidase subunit I (COI) gene.

Mitochondrial gene organization and expression in hexacorals
Five common features in the gene organization can be drawn from the 200 available mitochondrial genome sequences representing all five hexacoral orders (Appendix Table 1). (1) The 13 annotated OxPhos genes encode the same set of proteins as in vertebrate mtDNA [26], representing Complex I (ND1, 2, 3, 4, 4 L, 6, and 6), Complex III (CytB), Complex IV (COI, II, and III), and Complex V (ATPases 6 and 8). The additional approximately 70 OxPhos proteins are nuclear encoded [27]. (2) All canonical mitochondrial genes (OxPhos genes, rRNA, and tRNA genes) are encoded by the same DNA strand. (3) The tRNA gene repertoire is highly reduced, corresponding to tRNA fMet and tRNA Trp in sea anemones, stony corals, mushroom corals, and black corals, and only tRNA fMet in colonial anemones [14,28,29]. This indicates extensive tRNA import into mitochondria [20]. (4) The ND5 gene is split into two exons at nucleotide position 717 (human ND5 gene numbering [19]) by a group I intron found in all hexacorals studied so far (see Section 3 below). (5) The mitochondrial gene synteny appears highly conserved within, but not between, different hexacoral orders.

Order-specific gene organization
Each hexacoral order harbors a closely related primary mitochondrial gene organization (Figure 2A). This is an interesting notion since the orders have been separated from each other for 100 million years or more [28]. Stony corals and mushroom corals share some mtDNA synteny [28,30], and similarly, some segments of synteny appear conserved between sea anemones, colonial anemones, and black corals [30]. The only mitochondrial gene synteny common to all species in all five orders is the upstream proximity of the tRNA fMet gene to the large-subunit (LSU) rRNA gene (Figure 2). This suggests co-expression similar to that of tRNA Val and LSU rRNA genes in vertebrate mitochondria [26]. Recent studies in human and rat conclude that the mitochondrial encoded tRNA Val has replaced the 5S rRNA and become an integrated component as a structural rRNA of the mitochondrial ribosome [31]. Thus, tRNA fMet is considered as an interesting candidate for a similar dual function in hexacorals.
Deviations from the primary order arrangements have been reported in some sea anemones, stony corals, and mushroom corals and apparently confined to nonendosymbiotic deep-water species. Among the stony corals ( Figure 2B), Madrepora has a rearrangement in the COII and COIII gene order, and Lophelia and Solenosmilia have a more dramatic rearrangement involving three genes (CytB, ND2, and ND6) [19,32]. The latter example involves a dramatic shift in the size of the ND5-717 intron from approximately 10 kb (primary arrangement) to 6 kb (see Section 3.1 about transfers of OxPhos genes into the intron). Two different deviations were noted in the mushroom corals Corallimorphus and Corynactis [30]. These rearrangements appear complex and involve a drastic size reduction of the ND5-717 intron from approximately 18 kb (primary arrangement) to 12 kb and 10 kb, respectively ( Figure 2B). The most dramatic mitochondrial genome rearrangement is seen in the deepwater sea anemone Protanthea [16]. Here, the 21 kb mtDNA is arranged along two circular mitochondrial chromosomes, MCh-I and MCh-II ( Figure 2C). The mitochondrial gene order is heavily scrambled compared to the primary sea anemone arrangement. Different from all other hexacorals, genes at MCh-I are coded on both DNA strands. The ND5-717 intron size was increased from approximately 2 kb (primary sea anemone arrangement) to 15 kb in Protanthea (Appendix Table 1). Interestingly, the smaller MCh-II encodes the mitochondrial COII and one allele of the small subunit (SSU) rRNA. Phylogenetic analysis indicates that MCh-II is horizontally transferred into Protanthea from a distantly related sea anemone [16]. Not all deep-water hexacorals have mtDNA rearrangements. The Relicanthus sea anemone, sampled at a depth of 2500 m, [4] harbors the primary arrangement [33]. Similarly, Bolocera specimen samples at 40 m (Atlantic Ocean) [12] and at 1100 m (Pacific Ocean) [5] contain the same primary sea anemone arrangement.

Mitochondrial RNA in hexacorals
Mitochondrial RNAs have been investigated in a few hexacoral species representing sea anemones, colonial anemones, and mushroom corals [6,12,[14][15][16]. RNAseq data were obtained from 454 pyrosequencing and Ion Torrent PGM sequencing. Several general features are noted: (1) ribosomal RNA constituted more than 90% of the reads and is found to be at least 10-20 times more abundant than most OxPhos gene transcripts; (2) all the conventional genes were transcribed, and the Complex IV OxPhos genes appeared most expressed; (3) group I introns were perfectly spliced out from ND5 and COI mRNA precursors; (4) COI-884 intron splicing appeared more efficient than that of the ND5-717 intron, suggesting intron retention of ND5 mRNA [16]; and (5) noncanonical mitochondrial genes, such as the intron-encoded HEG and non-annotated open reading frames (ORFs), were clearly expressed. One of these ORFs, corresponding to a 306-amino-acid unknown protein in the mushroom coral Amplexidiscus, was highly expressed and located at the opposite strand compared to canonical OxPhos genes [16].

An obligatory group I intron in the ND5 gene
All hexacoral mitochondrial genomes harbor ND5-717 introns (Appendix Table 1), making this group I intron an obligatory feature. Evolutionary analyses of ND5-717 introns have previously been performed and show a strict vertical inheritance pattern and a fungal origin [19]. Homologous group I introns at the ND5 insertion site 717 are frequently noted in the fungi Ascomycota, Basidiomycota, and Zygomycota [34], which include mobile-type versions with HEGs [35,36]. This supports an ancient transfer with a subsequent progression into an obligatory strict vertical inherited intron. Interestingly, HEG-containing ND5-717 was also reported in the mitochondrial genome of choanoflagellates, species considered as the animal ancestors [37].

The ND5-717 intron is a giant group I intron
Phylogenetic analysis supports the early version of hexacoral ND5-717 introns to harbor two OxPhos genes (ND1 and ND3) in P8 [19]. This ancient organization is represented by sea anemones, colonial anemones, and black corals (Figure 2A). Insertions of ORFs into loop regions are a common feature in group I introns, and engulfing these compulsory genes might be a strategy for the intron in becoming essential to the host genome. RNA secondary structure folding of the ND5-717 ribozyme reveals that the catalytically important ωG (last nucleotide of the intron) is replaced by ωA (Figure 3). This replacement is likely to have a dramatic effect on intron biology, leading to host-factor dependent splicing and inhibition of 3 0 hydrolysis-dependent intron RNA circularization [38].
In some hexacoral orders, mitochondrial genome rearrangements resulted in additional transfers of canonical genes into the P8 segment. In stony corals two versions of 6 and 11 genes are intron-located (Figure 2A and B). Furthermore, it was noted that robust-clade species have developed a highly compact ribozyme core compared to complex-clade species (and all other hexacorals) [19]. The most complex ND5-717 introns are found in mushroom corals and in the Protanthea sea anemone [6,16,28,30]. Whereas three versions of 9, 11, and 15 intron-located genes are noted in mushroom corals, 14 genes are present in P8 of Protanthea ( Figure 2B and C). The ND5-717 intron in mushroom corals represent the largest group I intron known to date with an approximate size of 19 kb.

Unconventional splicing of ND5-717 introns
Mitochondrial RNA sequencing reveals perfectly ligated ND5 mRNA exons in sea anemones [12,15], colonial anemones [14], and mushroom corals [16], which support a biological splicing activity of ND5-717 introns. In the mushroom corals Ricordea and Amplexodiscus, the splicing efficiency of the ND5-717 intron was reported to be about 10% of that of the COI-884 intron located in the same mitochondrial genome [16]. The complex ND5-717 intron contains 2-15 mitochondrial genes within P8 that challenges its mode of splicing. The shortest forms of ND5-717 introns (approximately 1.6-2.4 kb) detected in sea anemones, colonial anemones, and black corals are likely to be excised by conventional group I intron cis-splicing from one single precursor RNA ( Figure 4A). The longest forms of ND5-717 introns (approximately 15-19 kb), present in mushroom corals [28,30] and the deep-water Protanthea sea anemone [6], contain almost the entire mitochondrial genome within P8. Recently, experimental support of intron removal by backsplicing in mushroom corals was reported [16]. It was found that the primary ND5 transcript contains a permuted exon arrangement where exon 2 is followed by exon 1 (Figure 4B). Correct ND5 exon ligation was achieved by involving a circular exon-containing RNA intermediate, which is a hallmark of intron backsplicing [16]. This is the first example of a natural group I intron removed by backsplicing and may explain why some hexacorals tolerate giant ND5-717 group I introns.
How the ND5-717 introns in stony corals are removed from their precursors by splicing is currently not known. These introns (sizes from approximately 6-12 kb) [19,39] may be too large and complex to be removed by conventional cis-splicing, and the ND5 exons may be too distant apart for backsplicing. Thus, a more plausible alternative is trans-splicing that generates a ligated ND5 mRNA from two separate precursor RNAs (Figure 4C). An interesting notion is that group I intron transsplicing has been reported in mitochondrial transcripts of placozoan animals [40].

Mobile-type group I introns in the COI gene
The gene encoding COI is a frequent host of group I introns in hexacoral mitochondrial genomes. Of the total 133 species inspected (Appendix Table 1), about 50% harbor an intron insertion. COI introns are present in all five hexacoral orders, but at different distribution patterns.

Three different insertion sites in the COI gene
The COI gene is interrupted by group I introns at three genic positions, where each intron site represents a unique evolutionary history [14,41]. The intron insertion sites correspond to positions 720, 867, and 884 (human COI gene numbering [19]). The COI-884 introns are widespread in hexacorals, present in most investigated species of sea anemones, mushroom corals, and black corals, as well as a few stony corals (Appendix Table 1) [12,41,42]. Colonial anemones harbor COI-867 introns [14], and some Indo-Pacific stony coral species contain COI-720 introns [41,43]. It appears that hexacorals are infected at least three times by COI introns or that this mitochondrial gene is subjected to recurrent group I intron invasion and extinction.
COI introns at different insertion sites are distinct in their ribozyme secondary structure, exemplified by the Urticina sea anemone and Zoanthus colonial anemone introns COI-884 and COI-867, respectively (Figure 5A and B). A common feature, however, is the large insertion within helical segment P8 harboring a HEG that codes for a homing endonuclease of the LAGLIDADG family. These HEGs extend beyond P8 and into the ribozyme domains [12,14,15,43]. Thus, COI-720, COI-867, and COI-884 intron sequences possess dual coding potentials of catalytic RNAs and homing endonucleases. This integration of the endonuclease into the ribozyme core structure ties the two elements closer together, making the endonuclease less prone to degradation.

Expression of intron-encoded homing endonucleases
Mobile-type introns, like the hexacoral mitochondrial COI introns, promote homing into cognate intron-less alleles by gene conversion [44,45]. Intron homing is initiated by a DNA double-strand break catalyzed by the intron-encoded homing endonuclease. Expression of HEGs has been studied in COI introns of sea anemones, colonial anemones, and mushroom corals [12,[14][15][16]. Two main versions were noted, leading to either highly expressed or repressed HEGs ( Figure 5C).
(1) The most successful mode of expression is the in-frame COI-HEG fusion strategy. The HEG, which covers most of the intron sequences (including the ribozyme encoded parts), is fused in-frame with the 5 0 COI exon. Highly expressed in-frame HEGs are observed in the sea anemones Urticina and Bolocera [12], and similar in-frame organizations appear common in other sea anemones such as Isosicyonis, Phymanthus, Actinia, and Stichodactyla ( [15,46,47]; our unpublished results). A COI fusion strategy for intron HEG expression in mitochondria, however, is not unique to sea anemones since several fungi are using this approach [44,48]. (2) Truncated in-frame fusions or freestanding intron HEGs result in

Concluding remarks
A hallmark of hexacoral mitochondrial genomes is the presence of self-catalytic group I introns. What is the biological role of these mitochondrial introns-are they purely selfish genetic elements, or could they have gained new regulatory functions beyond self-splicing? Current knowledge suggests a fungal origin of the hexacoral introns [19,34,49]. The group I introns in the COI gene encode LAGLIDADG-type homing endonucleases, consistent with intron mobility between cognate intron-less alleles [12,45]. The hexacoral COI introns appear gained and lost in multiple cycles during the last 0.5 billion years [42], which supports a selfish intron behavior.
The ND5-717 intron is apparently obligatory in hexacoral mitochondrial genomes, making this genetic element an interesting candidate in gene regulation. Similar obligatory group I introns have been noted in the chloroplast tRNA Leu gene of all green plants and in the nuclear LSU rRNA gene of all Physarales myxomycetes [50,51]. These obligatory mitochondrial, chloroplast, and nuclear introns are considered domesticated group I introns that may have gained new host-specific functions beyond self-splicing [21,25]. The mitochondrial ND5 mRNA stability has a key role in respiratory control in higher animals; it is tightly regulated and contains m 1 A base modification [52][53][54]. Intron retention of ND5 mRNA was recently reported in mushroom corals [16], suggesting possible host regulatory functions in hexacorals. Thus, further investigations on hexacoral mitochondrial intron functions and biological roles are needed and highly welcome. Size of mitochondrial genome. C, completely sequenced; P, partial/almost completely sequenced. 2 Size of ND5-717 group I intron. 3 Size of COI group I intron. COIÀ, no COI intron present; 720, 867, or 884 introns indicated. 4 The sea anemone Aiptasia pulcella may also be annotated as Exaiptasia pallida. 5 Information from our unpublished complete mitochondtial genome sequence of Stichodactyla helianthus. 6 The black coral Cirrhipathes lutkeni may also be annotated as Strichpates lutkeni. 7 The stony coral Lophelia pertusa may also be annotated as Desmophyllum pertusum. 8 The stony coral Sclerophyllia maxima may also be annotated as Acanthastrea maxima.

Species
Appendix Table 1.