INTRODUCTION

Genetic introgression between closely related fish species is widely recognized (Smith, 1992; Mukai, 2001 and references therein). The process of gene introgression has been represented by production of fertile hybrid and back-cross gradually incorporating genes into recipient populations upon genetic recombination. In this report we show an unusual, probably non-recombinant, and leaping mode of mitochondrial gene introgression which is mediated by a unisexual hybridogenetic system.

Some loaches (family Cobitidae, Osteichthyes) contain diploid-polyploid complexes (Kim and Lee 2000; Saitoh et al., 2000 and references therein; Zhang and Arai, 1999). Occurrence of unisexual (all-female) populations of hybrid origin in some of these complexes is emphasized to be a source of establishment of gonochoric tetraploid population (Vasil'ev et al., 1989), but no one except Kim and Lee (2000) recognized that normal diploid individual can be born from unisexual hybrids. Establishment of tetraploids via uni-sexual hybrids in loaches has been thought to be a one-way process.

A diploid-triploid hybrid complex occurs (Cobitis hankugensis-longicorpus [Cobitis sinensis-longicorpus] complex) in Nakdong River, Korea (Kim and Lee, 1990; Kim et al., 2003). The hybrid complex contains both diploid and triploid populations with few male occurrences. The diploid hybrid contains haploid genomes from C. hankugensis and C. longicorpus each, and the triploid does two haploid genomes from C. hankugensis and one from C. longicorpus (Kim and Lee, 1990). Artificial crossing experiment showed the diploid hybrid produces unreduced diploid hybrid ova, and the trip-loid does ova of C. hankugensis haploid genome eliminating C. longicorpus genome and reducing (Fig. 1) (Kim and Lee, 2000). Then, normal diploid C. hankugensis regenerates from the hybridogenetic triploid, crossing with male C. hankugensis. The process of establishment of polyploid populations thus may not be a one-way process. If so, this hybrid complex can mediate mitochondrial gene introgression from C. longicorpus to C. hankugensis.

Figure 1

Reproductive mode of Cobitis hankugensis-longicorpus complex. Single letters stand for C. hankugensis (H) or C. longicorpus (L) haploid genome. Solid arrows indicate experimental hybridization (Kim and Lee 2000), while dotted arrows denote presumed pathways. Large circles indicate eggs or oocytes. Haploid genomes being eliminated or released as polar bodies are set in small circles.

MATERIALS AND METHODS

We have sequenced a portion of mitochondrial genome (3329–3330 bp) from two female C. hankugensis, three diploid hybrids (one male and two females), three triploid hybrids (one male and two females), and two female C. longicorpus individuals. Species and ploidy diagnosis followed Kim and Lee (1990) employing morphological and chromosomal examination. These loaches came from Inwol-myon, Namwon-gun, Chollabuk-do, Korea (127°35′E, 35°27′N) except one C. hankugensis. The other C. hankugensis being examined as a comparative material was from Seangchomyon, Sanchong-gun, Gyeongsangnam-do, Korea (127°50′E, 35°27′N). Both collecting sites are in the Nakdong River basin. The sequenced region encompasses from upstream NADH dehydrogenase subunit-6 (ND6) to small subunit ribosomal DNA, corresponding to nucleotide positions from 14260 to 1017 of C. striata mitochondrial genome (Saitoh et al., 2003). We employed the two step PCR direct sequencing technique (Miya and Nishida, 1999; Kawaguchi et al., 2001). About 7 kb region was first amplified from genomic DNA with long-PCR primer pair. The long-PCR products then worked as templates for short PCRs with combination of 27 primers (Table 1) for direct sequencing using a commercial kit (Amersham, Bucks, UK) and an ABI373S automated DNA sequencer (ABI, Norwalk, USA).

Table 1

PCR and sequencing primers used in this study

RESULTS

Nine individuals of C. hankugensis, C. longicorpus, and hybrids from Inwol-myon turned out to be very close or identical (none to one nucleotide gap, none to three transitions, and no transversion) in their mitochondrial DNA sequences regardless of sex (Table 2). On the other hand one C. hankugensis individual from a different collecting site carried a heterogenic mitochondrial genome with 28–30 transitions, five transversions, and two amino-acid substitutions being observed between this individual and other nine individuals. Number of estimated nucleotide substitutions per site (Kimura, 1980) (transition/transversion ratio=5.71) between the two C. hankugensis individuals from different localities was 0.01. On the other hand, the values between individuals from the same locality were 0–0.0009 (average=0.0003) regardless of their biotypes.

Table 2

Sequence variable sites (corresponding to C. striata [Saitoh et al., 2003]) in Cobitis hankugensis-longicorpus complex

We could not find any sequence differences between one C. hankugensis, one C. longicorpus, two diploid hybrids and one triploid hybrid individual. Similarly, one C. longicorpus and one diploid hybrid individual were identical in mitochondrial DNA sequences. A maximum parsimony tree with C. striata sequence as an outgroup showed a nested distribution of the three biotypes in the tree (Fig. 2).

Figure 2

A maximum parsimony tree of two equally parsimonious topologies obtained by an exhaustive search without character weighting on PAUP* ver. 4.0b (Swofford 1998). Gaps are treated as the fifth character. See Fig. 1 for genome composition of three bio-types. Letters in parentheses indicate male (M) or female (F). A C. hankugensis individual with an asterisk was collected at a different collecting site (Seangcho-myon) from other nine individuals.

DISCUSSION

Sequence divergence between Cobitis species is so far reported to range between 4.6 to 19.2% (Kim et al., 2000; Perdices and Doadrio, 2001; Kitagawa et al., 2001). Sequence divergence between C. hankugensis individuals from different localities actually was 1% indicating loach populations are localized and prone to diverge even within a single basin. From an empirical view taking these reports and our result into account, it is unusual that two morphologically and cytologically (Kim and Lee, 1990) distinct species carry mitochondrial genomes with little sequence divergence. Lineage sorting is unlikely over a geological time-scale.

One possible explanation is the diploid-triploid hybrid complex as a vehicle of mitochondrial genome between two parental diploid species. If the mother of the initial diploid hybrid was C. longicorpus, it transferred the C. longicorpus mitochondrial genome to triploid hybrids of the next generation (Fig. 1) (Kim and Lee, 2000). The triploid hybrids produce ova with C. longicorpus mitochondrial genome and C. hankugensis haploid chromosome set which sometimes presumably accept C. hankugensis sperm in the natural habitat. Since hybridogenetic complexes show no genetic recombination between heterospecific genomes (Graf and Pelaz, 1989; Schmidt, 1996, but see Mateos and Vrijenhoek, 2002), next generation individuals would be nucleo-cytoplasmic hybrids between C. hankugensis and C. longicorpus. This pathway can accomplish hereby the unusual, probably non-recombinant, and leaping mode of mitochondrial gene introgression. Ecological study is necessary focusing on a mate recognition system between the hybrid complex and their parental diploid species.

Our study sheds light over unusual mitochondrial grouping among some diploid fish species. Carmona et al. (1997) observed unusual mitochondrial clustering of minnows and postulated a hybrid origin or ancient lineage sorting. Kitagawa et al. (2001) postulated mitochondrial genome exchange between two Cobitis lineages. Introgression events at the diploid level (hybridization and backcrossing) may be responsible for such mitochondrial clustering or genome exchange, but also hybridogenesis can mediate gene introgression. We should especially consider the latter possibilities in minnows and loaches, because unisexual reproduction and polyploidy of hybrid origin occur frequently in these fish groups.