Evidence for hybridization-driven heteroplasmy maintained across generations in a ricefish endemic to a Wallacean ancient lake

Abstract

Heteroplasmy, the presence of multiple mitochondrial DNA (mtDNA) haplotypes within cells of an individual, is caused by mutation or paternal leakage. However, heteroplasmy is usually resolved to homoplasmy within a few generations because of germ-line bottlenecks; therefore, instances of heteroplasmy are limited in nature. Here, we report heteroplasmy in the ricefish species Oryzias matanensis, endemic to Lake Matano, an ancient lake in Sulawesi Island, in which one individual was known to have many heterozygous sites in the mitochondrial NADH dehydrogenase subunit 2 (ND2) gene. In this study, we cloned the ND2 gene for some additional individuals with heterozygous sites and demonstrated that they are truly heteroplasmic. Phylogenetic analysis revealed that the extra haplotype within the heteroplasmic O. matanensis individuals clustered with haplotypes of O. marmoratus, a congeneric species inhabiting adjacent lakes. This indicated that the heteroplasmy originated from paternal leakage due to interspecific hybridization. The extra haplotype was unique and contained two non-synonymous substitutions. These findings demonstrate that this hybridization-driven heteroplasmy was maintained across generations for a long time to the extent that the extra mitochondria evolved within the new host.

1. Introduction

Mitochondria independently multiply. Heteroplasmy occurs when multiple mtDNA haplotypes arise within cells of an individual, which can result from a de novo mutation [1–3]. However, heteroplasmy is difficult to be maintained across generations because mtDNA haplotypes rapidly segregate by germ-line bottlenecks, i.e. bottlenecks due to reduction in the number of mitochondria and/or copied mitogenomes during oogenesis, which resolves heteroplasmy to homoplasmy [4–12]. This is considered a reason why homoplasmy is prevalent in nature [13,14].

Another source of heteroplasmy is paternal mtDNA transmission to fertilized eggs. In animals, sperm mitochondria are destroyed during gametogenesis or after fertilization, which ensures maternal inheritance of mtDNA [2,3,15,16]. However, interspecific hybridization is known to promote paternal mtDNA transmission [2,3]. This paternal leakage is thought to occur because the efficiency of egg–sperm recognition mechanisms decreases when genetic divergence between parents increases [3,17]. However, heteroplasmy resulting from paternal leakage is also difficult to maintain across generations because of the germ-line bottlenecks.

Here, we report heteroplasmy in the ricefish species Oryzias matanensis, which is endemic to Lake Matano, an ancient lake in Sulawesi Island (figure 1a,b). We noticed that one O. matanensis individual (O. matanensis F04) in previous field collections [18] had many heterozygous sites in the mitochondrial NADH dehydrogenase subunit 2 (ND2) gene (figure 1c) [18]. Therefore, in this study, we cloned the ND2 gene and demonstrated that some O. matanensis individuals were heteroplasmic. We then discuss the possibility that the heteroplasmy of this species is maintained across generations.

Figure 1.

(a) Malili Lake system and the collection site location of Oryzias matanensis in Lake Matano. The map was modified from [18]. The original map was provided by Thomas von Rintelen. (b) A male O. matanensis individual. (c) A part of the sequencing electropherogram of ND2 in an O. matanensis individual (O. matanensis F04) in [18].

2. Material and methods

(a) . Field collections

Lakes Matano belongs to the Malili Lake system in central Sulawesi. This lake system comprises five tectonic lakes, i.e. Lakes Matano, Mahalona, Towuti, Lantoa and Masapi, which share a common drainage (figure 1a). Lake Matano is located in the Matano Fault [19], and the Malili Lakes are postulated to have formed by the movement of this fault in the last 1.5–2 Myr [20,21]. Five Oryzias species, O. matanensis, O. hadiatyae, O. marmoratus, O. profundicola and O. loxolepis, have been reported from this lake system [22]. Oryzias matanensis is endemic to Lake Matano (figure 1a).

Previous research examined five male and five female O. matanensis [18]. In this study, we examined 10 additional males and 10 additional females which were collected along with the previous 10 individuals from a single locality (Pantai Pontada) [18] (electronic supplementary material, table S1). After euthanization with MS-222, the right pectoral fin was taken from each individual on-site and preserved in 99.9% ethanol. In addition, five male and five female O. marmoratus were collected from Lake Towuti (figure 1a) because this species was not included in [18]; ‘O. marmoratus’ from Lake Towuti in [18] was later described as a different species, ‘O. loxolepis’, by [22] (electronic supplementary material, table S1). After euthanization, bodies were preserved in ethanol without the head.

(b) . Mitochondrial sequencing and cloning

Total DNA was extracted from the pectoral fins or the muscles using a DNeasy Blood & Tissue Kit (Qiagen, Germany). The ND2 gene was amplified for each individual by PCR and Sanger-sequenced using the methods and primers described by [23]. Because phylogenetic trees estimated from the ND2 sequences and those estimated from the whole mitochondrial sequences were consistent [24,25], we consider that the ND2 gene sufficiently represents the whole mitochondrial genome. We found that seven O. matanensis individuals were suspected of heteroplasmy, i.e. the sequencing electropherogram of these individuals showed many heterozygous sites (see Results). We therefore performed cloning and subsequent sequencing for each of them as follows.

After PCR, DNA fragments were extracted from agarose gels. The extracted DNA was purified using the Wizard SV Gel and PCR Clean-Up System (Promega, USA). Purified DNA was then inserted into the pGEM vector using pGEM-T Easy Vector Systems (Promega, USA) and transformed into an E. coli strain (JM109). After transformation, we isolated colonies by blue/white selection and performed colony PCR using the T7 and Reverse universal primers (5′-AATACGACTCACTATAG-3′ and 5′-AACAGCTATGACCATG-3′, respectively). The sequencing was performed using the 3500 Genetic Analyzer with the BigDye Terminator v. 3.1 Cycle Sequencing Kit (Applied Biosystems, USA).

We separately determined sequences of four clones for each of the seven possible heteroplasmic individuals above. When multiple identical sequences were obtained, we treated that sequence as a haplotype. We obtained two haplotypes for each of five out of the seven individuals, and only one haplotype was obtained for each of the remaining two individuals.

(c) . Phylogenetic analysis

The ND2 sequences obtained from direct sequencing and cloning were used for phylogenetic analysis along with the ND2 sequences used in [18], which included 10 individuals each of O. matanensis, O. hadiatyae, O. profundicola and O. marmoratus from Lake Lantoa; O. marmoratus from Lake Mahalona; and O. dopingdopingensis from the Doping-doping River, a river that shares an estuarine region with the drainage of the Malili Lake system (Larona River).

All sequences were aligned using MUSCLE in MEGAX version 10.1.8 [26]. We obtained an alignment of the full-length ND2 gene (1053 bp). A maximum-likelihood (ML) phylogeny was estimated with RAxML-NG version 1.2.0 [27] using the codon-specific GTRGAMMAI model. We assigned O. dopingdopingensis as the outgroup, and bootstrap support values were calculated by a transfer bootstrap expectation analysis of 10 000 replicates.

(d) . Detection of positive selection on the extra haplotype

To detect a signature of positive selection at specific sites on the extra haplotype found in five O. matanensis individuals (see Results), we used EasyCodeML version 1.41 [28] and implemented a branch model in the codeml program in PAML [29], in which the two-ratio model (H₁) was compared with the one-ratio model (H₀) using a likelihood ratio test. In the test, we used a topology obtained by clipping only the O. marmoratus haplotypes (10 haplotypes) and the O. matanensis extra haplotype from another ML tree estimated by the same method as above but based on only 52 unique haplotypes extracted using DnaSP version 6.12.03 [30] (see electronic supplementary material, figure S1).

3. Results

Electropherograms of ND2 direct sequencing revealed that seven out of the 20 O. matanensis individuals had many heterozygous sites (figure 1c). We successfully obtained two haplotypes from each of five individuals by cloning.

The ML tree revealed that one of the two haplotypes (Hap A) clustered with haplotypes of other O. matanensis individuals, whereas the other haplotype (Hap B) clustered with haplotypes of O. marmoratus, a congeneric species distributed in adjacent lakes (Lakes Mahalona, Lantoa and Towuti) (figure 2). Only Hap A was obtained from the remaining two individuals with heterozygous sites. No individual having only Hap B was found.

Figure 2.

A maximum-likelihood phylogeny of Oryzias species endemic to the Malili Lake system based on the 1053-bp mitochondrial ND2 sequences. Numbers on branches are bootstrap values.

Four variants were found in Hap A among the seven possible heteroplasmic individuals. By contrast, the Hap B sequence was unique and the same in all individuals. No reading frame was broken in either Hap A or Hap B. Two non-synonymous substitutions were found in Hap B (I236T and R315C). The branch model also detected significant positive selection on the branch of the extra haplotype (p = 0.0088).

4. Discussion

We found an extra ND2 haplotype (Hap B) in some O. matanensis individuals. Because no reading frame was broken in the extra haplotype, it is very unlikely that this haplotype resulted from nuclear mitochondrial DNA segments inserted into the nuclear genome (e.g. [31,32]). However, to corroborate if this extra haplotype is mitochondrial, it is necessary to determine the whole mitochondrial genome sequence for each of the two mitochondria within a single individual.

We also found that the extra haplotype in O. matanensis resembled haplotypes of O. marmoratus, a congeneric species endemic to the Malili Lake system [22]. This indicates that the extra haplotype originated not from a de novo mutation but from hybridization. Heteroplasmy via interspecific hybridization and resultant paternal leakage has been reported in many animals (e.g. [33–35]). However, most of the observations were made on laboratory crosses of different species (e.g. [36–39]); there are not as many examples reported of hybridization-driven heteroplasmy in nature (e.g. [40–46]). This indicates that heteroplasmy is resolved to homoplasmy within a few generations due to the germ-line bottlenecks [4,6,8,9,32]. Indeed, in many cases, heteroplasmy in nature has been observed when hybridization is ongoing (e.g. [35,45]).

Therefore, the most straightforward interpretation would be that the heteroplasmy observed in O. matanensis also reflects ongoing or recent hybridization with O. marmoratus. However, there is no evidence for this based on population structure and phylogenetic network analyses using genome-wide single-nucleotide polymorphisms (SNPs) obtained from double-digest restriction-site-associated DNA sequencing. Mandagi et al. [18] demonstrated that O. matanensis individuals, including the possible heteroplasmic individual (O. matanensis F04), have been long isolated from the other congeners in the Malili Lake system (electronic supplementary material, figure S2). Moreover, the extra haplotype was not completely identical to O. marmoratus haplotypes, and even two non-synonymous substitutions were found. These findings demonstrate that this hybridization-driven heteroplasmy is maintained across generations for a long time to the extent that the extra mitochondria have evolved within the new host. Indeed, Oryzias species in this lake system are considered to have repeatedly undergone ancient hybridization [18]. Demographic inference also revealed that O. matanensis may have experienced admixture with O. marmoratus in Lake Mahalona, a lake adjacent to Lake Matano, about 30 000 years ago [18], although it remains unknown whether that was the origin of their heteroplasmy.

Theoretically, it is possible that O. marmoratus exists somewhere in Lake Matano, and ongoing hybridization with O. matanensis might be occurring. However, there have been no reports of O. marmoratus from Lake Matano despite several field surveys [47–49]. Of course, because an absence of evidence is not evidence of the species' absence, that possibility cannot be ruled out. However, no data have so far been obtained that positively supports the presence of O. marmoratus in Lake Matano, including there is no trace of hybridization on the genome of O. matanensis, as mentioned above.

If the heteroplasmy in O. matanensis originates from ancient hybridization, it remains a mystery how the heteroplasmy managed to avoid germ-line bottlenecks. One possibility is that segregation of mitochondria by germ-line bottlenecks might occur in every generation, but heteroplasmic oocytes or individuals might have higher fitness than homoplasmic ones. Another possibility is that the extra mitochondria might be selfishly transmitted to oocytes, leading to segregation distortion between the two mitochondria as in [50–52]. In any case, we hypothesize that the maintenance of heteroplasmy involves the non-synonymous substitutions and positive selection detected in the extra mitochondria.

Ethics

Field collections were conducted under the research permits issued from the Ministry of Research, Technology, and Higher Education, Republic of Indonesia (research permit no. 394/SIP/FRP/SM/XI/2014, 106/SIP/FRP/E5/Dit.KI/IV/2018, and 20/E5/E5.4/SIP.EXT/2019). We followed the Regulations for Animal Experiments at the University of the Ryukyus for handling the fish, and all experiments were approved by the Animal Care Committee of the University of the Ryukyus (approval nos. 2018099 and 2019084).

Data accessibility

Data are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.vdncjsz27 [53].

Supplementary material is available online [54].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors' contributions

H.N.: formal analysis, investigation, writing—original draft, writing—review and editing; I.F.M.: investigation, writing—review and editing; K.W.A.M.: investigation, writing—review and editing; J.K.: formal analysis, writing—original draft, writing—review and editing; N.I.: conceptualization, investigation, writing—original draft, writing—review and editing; K.Y.: conceptualization, formal analysis, funding acquisition, investigation, project administration, writing—original draft, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This study was supported by JSPS KAKENHI (grant nos 17H01675 and 22K18370) and JST CREST (grant no. JPMJCR20S2).