Detection of heteroplasmy and nuclear mitochondrial pseudogenes in the Japanese spiny lobster Panulirus japonicus

Direct nucleotide sequencing

Readable electropherograms were obtained from both direction in COI fragments of all three individuals of the Japanese spiny lobster. COI sequences determined by direct nucleotide sequencing ranged from 807 to 864 bp and have been deposited in International Nucleotide Sequence Database Collection (INSDC) under accession numbers of LC571524‒LC571526. No stop codon was observed in these sequences (designated by PJK1-direct, PJK2-direct, and PJK3-direct). No indel was observed between these sequences. All nucleotide substitutions at 19 variable sites observed between these sequences were transition at the 3rd position of a codon, and all substitutions were synonymous. The mean Kimura two parameter (K2P) distance between these three haplotypes was 1.510 ± 0.352% SE and that between these sequences and a reference sequence of P. japonicus (NC_004251) was 1.087 ± 0.270%, which were all well within the range reported for Japanese spiny lobster samples collected in Japan and Taiwan^9,10.

Electropherograms obtained by forward primer for 12S fragments were not readable, while those by reverse primer were readable in all individuals. 12S sequences determined by direct nucleotide sequencing using reverse primer alone ranged from 551 to 570 bp and have been deposited in INSDC under accession numbers of LC605705‒LC605707. Of nine variable sites, eight were transition and one was indel. The mean K2P distance between these three haplotypes (designated by PJK1-12Sdirect, PJK2-12Sdirect, and PJK3-12Sdirect) was 0.970 ± 0.338%, and that between these sequences and a reference sequence of P. japonicus was 0.835 ± 0.282%.

Electropherograms obtained by both primers for Dloop fragments were readable only in one individual (PJK2). This Dloop sequence determined by direct nucleotide sequencing was 762 bp and deposited in INSDC under accession number of LC605749. K2P distance between this haplotype (designated by PJK2-Dloopdirect) and a reference sequence of P. japonicus was 3.666%. No indel was observed between the two sequences, and 25 of 27 variable sites were transition.

Phylogenetic analysis of clones, heteroplasmy and NUMTs

Among the 36–42 positive COI clones examined per individual, sequences (809–892 bp) of 22–31 clones per individual (75 clones in total) were successfully determined. After alignment, both ends of all sequences were trimmed to fit the shortest sequence obtained by direct nucleotide sequencing, yielding 774–810 bp sequences. Eleven clones of PJK1 were identical to PJK1-direct, as well as seven of PJK2 to PJK2-direct and three of PJK3 to PJK3-direct. These dominant haplotypes (807 bp) were determined to be genuine COI haplotypes of each individual, and representative sequences of these three genuine haplotypes were deposited in INSDC (LC 571527, LC571533 and LC571538). Nucleotide sequences of the remaining 54 clones were all different one another, in which 20 haplotypes were observed in PJK1, 14 in PJK2, and 20 in PJK3 (LC571541–LC571577, OK429332–OK429343, LC654683-LC654687).

Phylogenetic tree constructed using three genuine COI haplotypes, 57 unique haplotypes and eight sequences of reference lobster species is shown in Fig. 1. Haplotypes detected from P. japonicus were segregated into four groups (designated by A, B, C and D). Among the outgroup species used, Australian rock lobster (P. cygnus) that morphologically and genetically belongs to the P. japonicus group^11,12, appeared to be the closest kin to all haplotypes detected from P. japonicus. All haplotypes in group A were of the same length (807 bp), and no indel was observed. Three distinct clades (designated by c-I to c-III) were observed in group A, in which 14 haplotypes from PJK1, 11 from PJK2 and 11 from PJK3 were cohesively clustered together with their corresponding genuine haplotypes (bold italic). PJK1-C25 was outlier, having 10 nucleotide differences from the genuine COI sequence. The numbers of variable nucleotide sites between haplotypes within c-I, c-II and c-III were 20, 15 and 26, respectively, of which nonsynonymous nucleotide substitutions were observed at 11, 13 and 10 sites. Stop codon was observed only in one haplotype (PJK3-C1). The mean K2P distance between different haplotypes within these clades ranged from 0.320 ± 0.075 to 0.561 ± 0.103%. The mean K2P distances between three clades ranged from 1.343 ± 0.339 to 2.178 ± 0.464%. Although group A must be composed of sequences containing those caused by Taq polymerase error or true heteroplasmic sequences as well as genuine haplotypes, it is difficult to determine the former two categories. All of the non-genuine haplotypes in group A had singleton difference one another, supporting the occurrence of Taq polymerase error. We determined haplotypes (marked with dagger in Fig. 1) differed by less than two substitutions from the genuine haplotype to be due to Taq polymerase error. This criterion may be reasonable, since Taq polymerase-mediated errors were estimated to occur approximately at a frequency of 7.2 × 10⁻⁵ per bp per cycle¹³ to one mutation per 10,000 nucleotides per cycle¹⁴. When Taq polymerase error is taken into account, these K2P distances within and between clades and number of haplotypes are likely to be somewhat overestimated. PJK1-C25, two (PJK1-C5 and PJK1-C60) in c-I clade, one (PJK2-C26) in c-II, and five (PJK3-C1, PJK3-C5, PJK3-C26, PJK3-C31, PJK3-C34) in c-III differed by 3 to 10 nucleotides from their genuine haplotypes, which were determined to be heteroplasmic haplotypes.

Sequence size of haplotypes in groups B to D ranged from 774 to 810 bp. K2P distance between haplotypes of groups A and B ranged from 7.169 to 8.177% with a mean of 7.754 ± 0.973%, that between A and C ranged from 12.073 to 17.392% with a mean of 14.521 ± 1.151%, and that between A and D ranged from 17.472 to 23.880% with a mean of 21.042 ± 1.600%. Multiple stop codons were observed in a haplotype of group B, in five of eight haplotypes of group C, and all haplotypes of group D. Three haplotypes in group C had no stop codon but differed in four to 10 deduced amino acids from the genuine haplotypes. BLAST homology search revealed no identical sequence for haplotypes in groups B to D but indicated that the closest species were P. japonicus or P. cygnus with moderate similarity (83–89% homology). Therefore, all haplotypes of groups B to D (LC571565–LC571570, LC571572–LC571577, LC654683-LC654687) were determined to be NUMTs.

Among the 30–35 positive 12S clones examined per individual, sequences (772–806 bp) of 25–27 clones per individual (77 clones in total) were successfully determined. After alignment, primer sequences were trimmed, yielding 731–765 bp sequences. Thirteen clones of PJK1 were identical one another, as well as 12 of PJK2 and three of PJK3, and these were identical to PJK1-12Sdirect, PJK2-12Sdirect and PJK3-12Sdirect, respectively. These dominant haplotypes ranging from 761 to 762 bp in size were determined to be genuine 12S haplotypes of the individual, and representative sequences of these three genuine haplotypes were deposited in INSDC (LC605708‒LC605710). Nucleotide sequences of the remaining 49 clones were all different one another, in which 12 haplotypes were observed in PJK1, 23 in PJK2, and 14 in PJK3 (LC605711‒LC605748, OK429126–OK429131, LC654678-LC654682).

Since incorporation of all eight Panulirus species sequences made sequence alignment ambiguous because of multiple indels, reference sequences of P. japonicus and of closely related P. cygnus were used for constructing phylogenetic tree (Fig. 2). Haplotypes detected from P. japonicus were segregated into three groups (designated by A to C). Sequence size of haplotypes in group A ranged from 760 to 762 bp. Three distinct clades (s-I to s-III) were observed in group A, in which 12 haplotypes each from PJK1, PJK2 and PJK3 were cohesively clustered together with their corresponding genuine haplotypes (bold italic). The numbers of variable nucleotide sites between haplotypes within s-I, s-II and s-III were 24, 17 and 16, respectively. Of these variable sites, transversion was observed at five, one and three sites, and indel was observed at one, zero and one sites, respectively. The mean K2P distances between different haplotypes within these clades ranged from 0.345 ± 0.081 to 0.519 ± 0.101%. The mean K2P distances between three clades ranged from 0.936 ± 0.275 to 1.371 ± 0.359%. Haplotypes differed by less than two substitutions (including indel) from the genuine haplotypes are marked with dagger. Five haplotypes in s-I clade and two haplotypes in s-III clade differed by three to six nucleotides from their genuine haplotypes, which were determined to be heteroplasmic copies.

Sequence size of haplotypes in group B varied from 731 to 762 bp. K2P distance between groups A and B ranged from 1.336 to 7.445% with a mean of 3.449 ± 0.398%, and those between a reference sequence of P. japonicus and groups A and B were 0.864 ± 0.236% and 3.189 ± 0.410%, respectively. Sequence size of haplotypes in group C varied from 744 to 765 bp. K2P distance between groups A and C ranged from 3.104 to 22.434% with a mean of 12.049 ± 0.901%, and those between a reference sequence of P. japonicus and group C ranged from 3.951 to 21.287% with a mean of 11.764 ± 0.901%. BLAST homology search indicated that the closest species for haplotypes in groups B and C was P. japonicus or P. cygnus with moderate to high similarity (84–98% homology). Therefore, all 13 haplotypes (LC605741‒LC605748, LC654678-LC654682) in groups B and C were determined to be NUMTs.

Among the 36–49 positive Dloop clones examined per individual, sequences (777–893 bp) of 26–38 clones per individual (92 clones in total) were successfully determined. After alignment, primer sequences were trimmed, yielding 736–853 bp sequences. Three clones (821 bp) of PJK1 were identical one another and determined to be genuine haplotype of this individual. Nine clones (813 bp) of PJK2 were identical to PJK2-Dloopdirect and determined to be genuine haplotype of this individual. Three clones (821 bp) of PJK3 were identical one another and determined to be genuine haplotype of this individual. Representative sequences of these three genuine haplotypes were deposited in INSDC (LC605750‒LC605752). Nucleotide sequences of the remaining 78 clones were all different one another, in which 25 haplotypes were observed in PJK1, 17 in PJK2, and 36 in PJK3 (LC605753‒LC605815, LC654419-LC654430, LC654675-LC654677).

Incorporation of all eight Panulirus species sequences made sequence alignment considerably unreliable because of multiple indels, reference sequences of P. japonicus and of closely related P. cygnus were used for constructing phylogenetic tree (Fig. 3). Haplotypes detected from P. japonicus were segregated into four groups (designated by A to D). Sequence size of haplotypes in group A ranged from 812 to 822 bp. Three distinct clades (d-I to d-III) were observed in group A, in which 17 haplotypes from PJK1, 13 from PJK2 and 15 from PJK3 were cohesively clustered together with their corresponding genuine haplotypes (bold italic). The numbers of variable nucleotide sites between haplotypes within d-I, d-II and d-III were 27, 61 and 28, respectively, of which indels were observed at five, two and four sites and transversion was observed at 0, six and six sites. The mean K2P distance between different haplotypes within these clades ranged from 0.340 ± 0.067 to 1.097 ± 0.139%. The mean K2P distance between these three clades ranged from 7.577 ± 0.951 to 8.770 ± 0.984%. Haplotypes differed by less than two substitutions (including indel) from the genuine haplotypes are marked with dagger. Eight haplotypes in d-I clade, three in d-II clade, and four in d-III clade differed by three to five nucleotides from the genuine haplotype were determined to be heteroplasmic copies.

Sequence size of haplotypes in groups B to D largely varied from 736 to 853 bp. K2P distances between group A and others ranged from 14.748 ± 1.030% (A vs B) to 61.619 ± 3.045% (A vs D), whereas that between haplotypes of group A and a reference sequence of P. japonicus was much smaller (6.333 ± 0.663%). BLAST homology search revealed no identical sequence for haplotypes in groups B to D and indicated that the closest species for haplotypes in groups B and C was P. japonicus with low to moderate similarity (74–88% homology). On the other hand, no significantly similar sequence was found for haplotypes in group D. Therefore, all 31 haplotypes (LC605788‒LC605815, LC654675-LC654677) in groups B to D were determined to be NUMTs.

Impact of heteroplasmy and NUMTs for direct nucleotide sequencing

Partial electropherogram obtained by direct nucleotide sequencing for COI amplicon of PJK3 is shown in Fig. 4 (top). Peak signals of this electropherogram are readable, but there are a number of sites where two (asterisk) or three (dagger) signals overlap. Alignment of a genuin haplotype (PJK-C7) and nine NUMTs sequences, corresponding to this partial electropherogram, is shown in Fig. 4 (bottom). At the sites where plural peaks overlap, different NUMT haplotypes were observed to share the same nucleotide different from the PJK3-direct. Heteroplasmic copies in COI determined in this study may have little negative impact on direct nucleotide sequencing, since nucleotides different from the genuine haplotypes were all unique to each heteroplasmic haplotype. Thus, the plural peaks at a site were composed of signals from genuine plus NUMT haplotypes, and the intensity of each peak was positively related to the copy numbers of these haplotypes. Frequent failure to obtain readable electropherograms in 12S and Dloop regions by direct sequencing may be due to extensive indels observed in the NUMT haplotypes.

Source: Ecology - nature.com