Presence of different mtDNA types
Out of 424 individuals of Mytilus galloprovincialis 36 spawned: 17 males and 19 females. Total DNA was isolated from the foot and sperm of these males and from the foot of females. The PCR assay for the C-genome was negative in all samples. All foot samples, from either female or male individuals, were positive for the F-genome, but no sperm sample tested positive for this genome. The M-specific test was positive in two female foot preparations out of 19 (or 10% of cases) and in 11 male foot preparations out of 17 (or 65%). The frequency of occurrence of the M genome in the foot tissue is statistically higher in males (chi-square 11.885, D.F. = 1, P = 0.0006). This result agrees with that by Garrido-Ramos et al. , who observed that the M genome was more frequently found in the foot of males than females of M. edulis. Further comparison of the two studies is not possible because of the different ways of detection of the M genome they used.
To obtain an equal number of M-positive foot samples from both genders we extracted DNA from the foot of females that did not spawn (see Methods). The complete set of 30 sequences we obtained (ten foot sequences from ten females and ten foot and ten sperm sequences from ten males) are given in Additional file 1. The complete sequence of VD1 (see Additional file 1), including the A-rich region that was excluded from the sequence comparison. In contrast, only the variable positions of the 16S-rRNA and the CD fragments are shown. The most striking observation is that in each of the ten males the sequence from the foot and the sequence from the sperm were identical. Among the 30 sequences one was encountered 5 times (f7, m9ft, m9sp, m17ft, m17sp), one 4 times (f80, f81, m16ft, m16sp), one tree times (f1, f2, f11), 7 twice each in a different male and 4 once each in a different female. This provides a rough estimate, the best we can have on the basis of our sample size, of haplotype frequencies under the hypothesis that the foot and sperm haplotype in a male are independently derived. For the hypothesis of egg heteroplasmy to be compatible with our data we would require that the male's father and the male's maternal grand-father happened to have the same sequence. The probability, p, that this may occur for any single male is p = (5/30)2 + (4/30)2 + (3/30)2 + 7(2/30)2 + 4(1/30)2 = 0.091. The probability, P, that this will happen for all ten males is P = (0.091)10 = 4 × 10-11, an extremely small number. We may confidently conclude that the presence of the M genome in the foot of males is due to sperm mtDNA leakage. On the basis of this conclusion, the set of independently derived sequences is reduced to 20. The new probability for obtaining the same sequence in two draws is h = 11(1/20)2 + 3(3/20)2 = 0.095. This is not very different from the one obtained when the two sequences in a male were assumed to be independently derived, and this is due to the high diversity of the VD1 region.
For females we cannot have direct evidence that the M genome found in the foot is that of their father because we do not have an independent way of knowing what the father's M genome was, as we do have in males. This could be done only in females with a known male parent. Even though several studies of DUI have used pair-matings [7, 31, 32], none of these compared the M genome of the male parent with that found in the somatic tissues of females. Our results from females do not provide information about the origin of the M genome in somatic tissues but suggest that the M sequences found among females are no less diverse than those found in males, in agreement with expectation. Among the ten females we encountered one sequence three times (f1, f2, f11), another sequence twice (f80, f81) and five singletons (h = 0.02). Assuming that the two sequences in a male have the same origin, we encountered one sequence twice (m9 and m17) and 8 singletons (h = 0.014).
Our results can be discussed in the context of the studies by Cao et al. , Obata & Kumaru , Cogswell et al.  and Kenchington et al. . In these studies sperm of Mytilus edulis (in the first, third and fourth study) and of Mytilus galloprovincialis (in the second study) was treated with MitoTracker Green FM, a fluorescent stain that binds to the outer surface of mitochondria, and used to fertilize eggs. The sperm mitochondria were observed microscopically in the embryo up to the 4-cell stage (and, occasionally, up to the 8-cell stage). The eggs used in these experiments were derived from mothers that were known from previous studies to produce daughters either exclusively or in very high frequency (daughter-biased mothers) or from mothers that produced sons in high frequency (son-biased mothers). The researchers found that in embryos from daughter-biased mothers the sperm mitochondria dispersed randomly in both blastomeres at the 2-cell stage or in all four blastomeres at the 4-cell stage and that the distribution pattern was random. Thus, in female embryos, the sperm mitochondria appear to be randomly mixed with the egg mitochondria, which vastly outnumber them, and are either entirely lost or become barely detectable minorities in the somatic tissues of adult females. In contrast, in embryos from son-biased mothers the five sperm mitochondria formed a condensed mass that was found in the larger cell resulting from the first or second egg division. The researchers suggest that this is a developmental mechanism through which the sperm mtDNA is delivered in the male's primordial germ cells. Thus in males sperm mitochondria appear to follow a regimented developmental process, the result of which is that the male gonad becomes mainly or exclusively occupied by the sperm mtDNA (M-type), whereas the somatic tissues contain mainly the egg mtDNA (F-type). We say mainly, not wholly, because according to the results of several studies [15–19] the M-type is also found in somatic tissues of some males.
Our study is the first to compare the M genome that is always found in the gonad of a male and the M genome that is occasionally found in the somatic tissues of the same male (but see ) and leads to the conclusion that the M molecule found in the foot of these individuals comes from leakage of sperm mtDNA into somatic tissues. This leakage may occur if one or more of the approximately five sperm mitochondria that enter the egg after fertilization fail to become part of the aggregate that is formed by the sperm mitochondria and, as a result, ends up in cells that will form somatic tissues. Cao et al. , Cogswell et al.  and Kenchington et al.  have indeed observed repeatedly cases of "orphan" sperm mitochondria, i.e., mitochondria that were not part of the aggregate and segregated independently from it. Our results do not exclude the hypothesis that some cases of somatic heteroplasmy for the M genome in species with the DUI system may result from heteroplasmic eggs, i.e., eggs that contain a minority of M mtDNA genomes along with the vast majority of F genomes. This hypothesis is supported by the work of Obata et al. [18, 19], who reported a high incidence of egg heteroplasmy. Further evidence for this comes from the work of Theologidis , who has observed several cases of triplasmy in male and female tissues, i.e., presence of F, M and C genomes. That this situation may have arisen through sperm heteroplasmy (a sperm that contained the M and the C genomes) appears unlikely in view of the results of Venetis et al.  who failed to detect in the sperm of 35 males any other mtDNA except the paternal one. We note that the mechanism of egg heteroplasmy, if it occurs, requires that the female that produces the heteroplasmic egg must have received its M genome from its father or from a more remote ancestral male. This means that sperm mtDNA leakage is the ultimate cause of the presence of the M genome in either male or female somatic tissues.