Structural analysis of leucine, lysine and tryptophan mitochondrial tRNA of nesting turtles Caretta caretta (Testudines: Chelonioidea) in the Colombian Caribbean

Biological samples

Peripheral blood samples were collected from 11 healthy captive loggerhead turtle of undetermined sex and maintained at ambient temperature (average 30 °C, minimum 27 °C) in an outdoor seawater pool at the CEINER Oceanarium in San Martin de Pajares island, Cartagena, and two from Don Diego beach (11°16′N y 73°45′O) in PNN Tayrona—Santa Marta characterized by a semi-arid climate (identified in this study as CC1, CC2, CC3, CC4, CC5, CSM-1, CSM-2, CSM-3, CR-2, 2C y 3C according to molecular biology laboratory nomenclature). The blood was obtained from the dorsal cervical sinus in accordance with Dutton (1996). The samples were placed in sterilized tubes with Tris-EDTA buffer 0.1 M (GreinerBio-one^®, Kremsmünster, Austria) solution and were transported at 4 °C to the Molecular Biology lab of the Universidad Jorge Tadeo Lozano, Bogota campus. The samples were collected following the ethical standards established by the legislation and the study obtained permission from the Ministry of Environment and Territorial Development (No 24 of June 22, 2012) and the Genetic Resources Access contract (No 64 of April 2013)

Extraction of DNA from the blood tissue

The genomic DNA was obtained from eleven blood samples using the GF1GF-1 Tissue DNA kit (VIVANTIS-Malasia). The DNA obtained was visualized by electrophoresis in agarose gel at 1% p/v with Safeview stain (two µg/ml) (Fig. S1). The DNA was quantified using the Nanodrop 1000 Spectrophotometer equipment and was analyzed with the ND-1000 V3.7.1, (ThermoScientific, Maltham, MA, USA) program.

PCR amplification and sequencing

The selection of 3 of the 22 tRNA (tRNA^Trp, tRNA^Lys and tRNA^Leu) was sustained in the prior results obtained by Otálora & Hernández-Fernández (2018) who identified point mutations in these three genes. For the PCR amplification of the tRNA genes, three primer pairs were used; cc7, cc11 and cc16 (Beltrán et al., 2013), that amplified three fragments of 800 bp that contain the coding sequences of tRNA^Trp, tRNA^Lys and tRNA^Leu respectively (Fig. S2). The mixture of PCR reaction contained 1 unit of MyTaq DNA Polymerases (high reliability) (Bioline Inc., California, USA, EE.UU.), 2 mM of MgCl2, 1 mM of primer, 1X PCR tampon (50 mM of KCl and 10 mM of Tris-HCl pH 8,3), and 0,2 mM of deoxyribonucleotide (DNTP´s), in a final volume of 25 µL. The thermocycling program consisted in one step of initial denaturing of five minutes at 95 °C, followed by 35 cycles of 95 °C for 1 min, 40 and 44 °C (depending on the primer) for one minute and 72 °C for another minute, with a final extension of 10 min at 72 °C. The PCR was conducted with an automatic thermocycler (Labocon Systems Ltd, Hampshire, UK). The PCR products were purified using the GF-1 Gel DNA Recovery kit (Vivantis, MALASIA) and were sequenced in both directions (5′–3′ and 3′–5′) using the tagDyeDeoxy Terminator Cycle-sequencing method, in an 3730XL sequencer (Applied Biosystems, Foster City, CA, USA) at SSIGMOL (Universidad Nacional de Colombia: http://www.ssigmol.unal.edu.co/).

Data analysis

The tRNA sequences were assembled using the Geneious R6^® program (Biomatters, Ltd., New Zealand). With the online tool BLAST (Altschul et al., 1990) (http://blast.ncbi.nlm.nih.gov/) basic local alignment was conducted to determine the similarity percentage of the sequences obtained with the sequences of C. caretta previously described. All sequences obtained in this study and the sequences coded for the same tRNA in the seven sea turtle species previously reported in GenBank were aligned using the algorithm ClustalW (Thompson, Higgins & Gibson, 1994). The inference of the 2D structures of the tRNA was conducted with the software ARWEN (Laslett & Canback, 2008) (http://130.235.46.10/ARWEN/). Parameters: default metazoan mitochondrial code. Output structures were manually curated. The prediction of the 3D structures was conducted using the online program RNA composer 3D (Popenda et al., 2012) (http://rnacomposer.cs.put.poznan.pl/). Modeling was performed with ‘batch mode’ with secondary structures as the input, and was visualized through the program Geneious R6^® (pdb format), were the distance between the anticodon and unpaired nucleotide in the 3′ end of each structure was measured. The potential interaction networks and the non-canonical base pairs were identified for the tertiary structures obtained. The nucleotide sequences and inferred structures were identified with the same name of the individual to which they belonged. The positions of the nucleotides were numbered according to the conventional standards (Sprinzl et al., 1998), and definition of canonical tRNA cloverleaf and numbering of nucleotides (Salinas-Giegé, Giegé & Giegé, 2015).

DNA extraction, amplification and sequencing

The genomic DNA obtained from the blood sample of 11 individuals of (C. caretta) loggerhead turtle were of good quality, obtaining concentrations of ±50 ng/µL and purity in a range of 1.8–2.0. Lastly, 11 sequences of the tRNA^Trp and tRNA^Leu(CUN) mitochondrial genes respectively and nine sequences for the tRNA^Lys genes were obtained; two of the sequences did not present clear chromatograms, therefore, they were not taken into account. On some occasions there may be problems in the sequencing reaction that prevent having a good sequence, related to: The primer could be linked in several positions to the template DNA, it is also possible that more than two PCR products have been amplified and sequenced in parallel presenting wrong chromatograms, in addition the primers of the original PCR may not have been removed. These sequences obtained were deposited in the GenBank database (accession numbers KX063643–KX063673). The BLAST analysis established a similarity percentage of 98–100% between the sequences obtained and the mt-tRNA sequences deposited in GenBank.

Multiple alignment of sequences

The molecular analysis of tRNA^Trp was conducted using 78 nucleotide positions, of which, 62 positions were identical for the seven species, with a conservation of 79.4% of this gene (Fig. S3). C. caretta was differentiated from the other taxa because the tRNA^Trp gene revealed an additional thymine in position 26 and a C→T transition in position 48 that represent interspecific variations. A mutation in position 14 (transition C→T) was identified in this same gene, shared by nine of the eleven turtles of this study, but absent in the previously described sequences. This result is in agreement with the one described by Otálora & Hernandez (2015), Otálora & Hernández-Fernández (2018). Apparently, this mutation would be fixed in the nesting colony of the Colombian Caribbean, constituting an endemic haplotype. This hypothesis is supported by the genetic differences identified in populations of different geographic areas (Bowen, Nelson & Avise, 1993; Encalada et al., 1996) as a result of reproductive isolation and low genetic flow that could cause the philopatry (Cuadros-Arasa, 2013).

The analysis of the tRNA^Lys gene sequences was based in 73 nucleotide positions, of which, 66 were identical for all the species (90.4% of conservation for this gene). Seven mutations between positions 52 and 62 were identified (Fig. S4). C. caretta was uniform in its sequences, with a C→T transition in position 62 that was not shared with any other species. Only the Colombian sequence described by Otálora & Hernandez (2015) presented a transversion in the nucleotide position 18 (mutation at the level of the individual).

The tRNA^Leu gene presented 72 nucleotide positions in 66 identical places in all the sequences (98.1% of conservation). Six mutations were identified, four of which were contributed by D. coreacea, evidencing the divergence of this species (Eckert et al., 1999). The majority of sequences of C. caretta presented an adenine at position 43 as described by Otálora & Hernández-Fernández (2018); nevertheless, two of the sequences obtained in this study and two of the sequences previously described presented a A→G transition in the same position (Fig. S5).

Secondary structures

The secondary structures inferred from the tRNA^Trp sequences presented just one shared motif by ten of the eleven sequences obtained in this study (Fig. 1A). This motif was characterized by having a T-loop composed of 7 nucleotides, anticodon loop with 7 nucleotides and a larger one, D-loop, composed of 13 nucleotides (Table 1). The large size of this loop is because this gene presented a size of 77 bp, 4 more than the 73 bp generally described for the same gene in other species (e. a. the hawksbill turtles, Eretmochelys imbricata) (Jung et al., 2006). The D-stem is formed by 4 pairings including one non-canonical pair, adenine–adenine, that could be recognized as a secondary interaction conserved in the tRNA^Trp structure, a characteristic presented by the seven species of sea turtles (Fig. 1C). A second 2D structural motif for the tRNA^Trp gene was identified from the sequence corresponding to the individual CC2 (Fig. 1B) that was differentiated by the presence of a non-canonical pairing A₂₈–C₄₂ in the anticodon stem as a result of the T→C transition identified in the multiple alignment in position 48. In addition two new mutations in a non-complementary A₇–A₆₆ and a non-canonical A₆–G₆₇ pairs in the acceptor stem (because of C→A and T→A transversion in positions 6 and 7) its creating a risk of interrupting the postranscription processes, such as aminoacylation (Blakely et al., 2013). When comparing the motifs obtained for tRNA^Trp of C. caretta with the consensus structure of tRNA^Trp of sea turtles (Fig. 1C), it was observed that the variable region is small, with only 4 nucleotides, and that the D loop contains the majority of variant positions, suggesting that this region presents greater freedom to mutate within the tRNA^Trp molecule when remaining conserved in the variable region.

For the tRNA^Lys gene a 2D structural motif was identified (Fig. 1D) in which the T-loop with nine nucleotides is larger than the D with 7 (Table 1). In the D loop, two residues of invariant guanine were identified (G₁₈ and G₁₉), identical to the canonical tRNA (Westhof & Auffinger, 2001). The distinctive characteristic of this structure is the acceptor stem conformed by 8 intracatenary pairings and not by 7 as presented in the canonical tRNA (Westhof & Auffinger, 2001; Giegé et al., 2012). In addition, two of these pairings are of the non-Watson–Crick type. This characteristic, typical of tRNA^Lys in sea turtles, is shared by the other six species (Fig. 1E). On the other hand, at position 62, thymine was observed to characterize the sequences of C. caretta. This generates a canonical pairing A₅₂–U₆₂ in the T-stem (Fig. 1D), absent in the other turtle species.

The tRNA^Leu(CUN) analysis produced a secondary structure (Fig. 1F). A mutation at position 44 was identified, which did not generate changes in the secondary structure. In this structure it was observed that the D/T loops and anticodon are formed by 7 bases (Table 1). The variable region presented 5 nucleotides (one more than in the tRNA^Trp and tRNA^Lys). A non-canonical pairing at the first base pair of the acceptor stem was identified, a characteristic shared by the secondary consensus structure of sea turtles.

Tertiary structure

Three tertiary structures from the eleven sequences analyzed of the tRNA^Trp gene were obtained (Fig. 2). The CSM-3 and CC5 individuals presented a tertiary structure that differs from (Fig. 2B) the dominant motif for this gene (Fig. 2A), because they had a cytosine mutation for a thymine at position 14. This transition generated a greater folding of the D-loop producing a larger bend in the nucleus of the molecule, shortening the distance between the anticodon and the 3′ end. This change shows the susceptibility of the tRNA to present distortions in its architecture for a point mutation (Laslett & Canback, 2008).

A tRNA is aminoacylated if it has a distance between the anticodon and the CCA termination at the 3′ end between 7.5–8.0 nm (Westhof & Auffinger, 2001; Suzuki, Nagao & Suzuki, 2011). The predominant tertiary motif of tRNA^Trp and the motif found for individuals CSM-3 and CC5 presented a distance of 7.218 and 6.872 nm respectively; at this distance there is still a need to add the length of the CCA sequence, which is added at the tRNA in a postranscriptional manner, a process that could compensate the remaining distance (Giegé et al., 2012). The tertiary motif obtained for the tRNA^Trp from the sequence belonging to the individual CC2 (Fig. 2C) showed a length within the functional range (7.716 nm). This length for this motif is attributed to the presence of the non-canonical pairing and the non-complementary pair identified in the secondary structure that distorted the helix that forms the acceptor stem. According to Fritsch & Westhof (2005) only a small number of non-Watson–Crick base pairs may be incorporated in the stems without interrupting the helical structure, but they anyhow cause distortions that perturb normal function. In this case in which the helix is only 7 base pairs long, the presence of two non-canonical stocked pairs produces this notable distortion, because definitely pathogenic mutations are mainly produced in the stems (Blakely et al., 2013).

From the inferred 2D structure for the tRNA^Lys, a 3D structure was obtained (Fig. 2D) which showed a close disposition to the typical “L” folding, maintaining opposites in the 3′-end and the anticodon. The T-loop presented a greater torsion than the one commonly presented in a typical tRNA, caused perhaps by the larger size of this loop producing a folding over it. The presence of two non-Watson–Crick pairs evidenced in the 2D structure generated a distortion of the helix of the acceptor stem where the major groove is wider and the minor groove is more closed. In the anticodon loop an atypical distortion was observed that could cause problems in the codon-anticodon alignment; nevertheless, it is known that modified nucleosides are inserted in the anticodon loop and are necessary for its stability (Suzuki, Nagao & Suzuki, 2011). The distance between the anticodon and the 3′-end of 7.78 nm is within the functional range (Westhof & Auffinger, 2001; Suzuki, Nagao & Suzuki, 2011).

The tRNA^Leu(CUN) presented a structural motif (Fig. 2E), even when a mutation in the variable region was found. The A→G transition in position 43 did not produce changes in the 3D structure, showing that the variable region is capable of accomodating variations in the number of nucleotides to try to maintain the general architecture of the tertiary structure (Rich & RajBhandary, 1976).

Tertiary interactions

To obtain the typical 3D structure in “L” form, require tertiary interactions between the loops and the helices (Suzuki, Nagao & Suzuki, 2011). In this study, the network of tertiary interactions canonical was validated (Giegé et al., 2012) for each one of the tRNA analyzed. Table 2 shows the proposed tertiary interactions for mitochondrial tRNA, tRNA^Trp, tRNA^Lys and tRNA^Leu(CUN) of C. caretta, based on the similarity and coincidence of residues with the interactions described for canonical tRNA from the visual examination of the secondary structures.

In tRNA^Lys, the identity of some bases changed with respect to canonical interactions fundamentally due to the presence of additional mating in the acceptor arm, and in tRNA^{Leu (CUN)}, by a larger variable loop. In the tRNA^Trp the nucleotide at position 14 intervenes in a tertiary interaction, fact that could explain why the mutation at this position generated changes in the 3D structure that was identified as a different motif. On the other hand, the participation in a tertiary interaction of the non-Watson–Crick A₂₃–A₁₂ bond identified in the D-arm of the tRNA^Trp would indicate why this non-canonical pair is a characteristic conserved in C. caretta and in all sea turtles, because the residues involved in tertiary interactions are under a strong selective pressure (Widmann et al., 2010). The bases that form the tertiary interactions in this study (Table 2) are not strictly equivalent to the bases reported by Giegé et al. (2012), nevertheless, the nucleotides that form the tertiary interactions proposed were identified as invariant residues, which provides support, given that the tertiary interactions occur mainly between conserved and semiconserved nucleotides (Widmann et al., 2010).

Figure 3 shows the verification of the tertiary interactions previously referred. Nucleotides with the same color integrate one interaction and are indicated in Table 2 as follows: interaction 1-red, 2-green, 3-blue, 4-cyan, 5-magenta, 6-yellow, 7-orange. Direct verification on the tertiary structures showed that interactions 1, 2 and 3 are not present in tRNA^Trp and tRNA^Leu(CUN). Nevertheless, it was observed that interaction 1 in the tRNA^Trp (Fig. 3A), known as Levitt pair (Giegé et al., 2012), could be stabilized through the participation of the U_20B nucleotide forming the C₄₈–U_20B–A₁₅ interaction. In the 3D structure obtained for the CC5 and CSM-3 individuals (Fig. 3B) it was observed that the Levitt pair could be replaced by the U₂₀–C₄₈ and A₁₅–U_20B pairs. Interaction 3 in tRNA^Trp and tRNA^Leu(CUN) (Fig. 3D) does not present pairing of positions 46–22, as described Giegé et al. (2012) for tRNA of class II. In the putative structure of tRNA^Lys, the invariant base C₄₈ was not conserved, nevertheless, the pair A₄₈–A₁₅ imitates the canonical Levitt pair (Fig. 3C). It looks like interaction 2 is not present in this tRNA.

Non-Watson–Crick pairs

The comparison of the non-canonical pairs identified in the putative 2D structures obtained in this study with proposed models from studies with X ray crystallography in the RNA of bacteria and yeast (Leontis, Stombaugh & Westhof, 2002) allowed to identify the possible type of non-Watson–Crick pairing, the number of hydrogen bonds that would be forming and the atoms that intervene in such interactions. At the end, it could be determined that all non-canonical pairings may be classified within the same geometrical family, taking into account the border of the nucleotide that participates in the interaction and the orientation of the glycosidic bond according to the axis of interaction given by the hydrogen bonds (Leontis, Stombaugh & Westhof, 2002) within the Cis Watson–Crick/Watson–Crick family (Fig. S6). Only pair A₇–A₆₆ was identified as a possible non-complementary pair. According to Leontis, Stombaugh & Westhof (2002), for this case, no example has been identified and there is yet no reasonable model that indicates how the hydrogen bonds would be formed, and thus are classified as non-complementary.