Sustained Heterozygosity Across a Self-Incompatibility Locus in an Inbred Ascidian

Abstract

Because self-incompatibility loci are maintained heterozygous and recombination within self-incompatibility loci would be disadvantageous, self-incompatibility loci are thought to contribute to structural and functional differentiation of chromosomes. Although the hermaphrodite chordate, Ciona intestinalis, has two self-incompatibility genes, this incompatibility system is incomplete and self-fertilization occurs under laboratory conditions. Here, we established an inbred strain of C. intestinalis by repeated self-fertilization. Decoding genome sequences of sibling animals of this strain identified a 2.4-Mbheterozygous region on chromosome 7. A self-incompatibility gene, Themis-B, was encoded within this region. This observation implied that this self-incompatibility locus and the linkage disequilibrium of its flanking region contribute to the formation of the 2.4-Mb heterozygous region, probably through recombination suppression. We showed that different individuals in natural populations had different numbers and different combinations of Themis-B variants, and that the rate of self-fertilization varied among these animals. Our result explains why self-fertilization occurs under laboratory conditions. It also supports the concept that the Themis-B locus is preferentially retained heterozygous in the inbred line and contributes to the formation of the 2.4-Mb heterozygous region. High structural variations might suppress recombination, and this long heterozygous region might represent a preliminary stage of structural differentiation of chromosomes.

Ciona intestinalis, self-incompatibility, inbred line, chromosome evolution

Introduction

Balancing selection maintains diversity of self-incompatibility loci (Wright 1939), and self-incompatibility loci are maintained heterozygous, because they prevent self-fertilization. Hence, self-incompatibility loci could cause linkage disequilibrium of flanking regions, and recombination suppression (Uyenoyama 1997), and might eventually cause structural differentiation of chromosomes, as a sex determining gene does in sex chromosomes (Eichler and Sankoff 2003; Fraser and Heitman 2004; Charlesworth et al. 2005; Uyenoyama 2005; Bachtrog 2013). Although this prediction has partially been tested in plants (Kamau and Charlesworth 2005; Llaurens et al. 2009), it could also be tested by inbreeding more directly. First, genomic regions near self-incompatibility loci will be retained heterozygous by linkage disequilibrium. Second, if recombination is suppressed, large regions flanking to self-incompatibility loci will be retained heterozygous.

The genome of Ciona intestinalis, which is a hermaphrodite chordate belonging to a sister group of vertebrates (Delsuc et al. 2006), provides a unique opportunity to address this problem. It encodes two self-incompatibility loci, Themis-A and Themis-B (Harada et al. 2008). These loci are encoded on different chromosomes and are expected to be preferentially maintained heterozygous, because fertilization rarely occurs between sperm and eggs bearing self-incompatibility proteins that are encoded by the same alleles in both loci. Therefore, theoretically, either of the self-incompatibility loci is always heterozygous, and outcrossing is the natural reproductive mode of C. intestinalis (Morgan 1944). However, the self-incompatibility system of this animal is however incomplete, and self-fertilization occurs in laboratory conditions.

Each of these incompatibility loci encodes two genes (Harada et al. 2008; Saito et al. 2012); one is expressed in sperm and the other is expressed in eggs. The Themis-B locus encodes the genes, s-Themis-B and v-Themis-B. s-Themis-B is expressed from a haploid genome of sperm, and v-Themis-B is expressed from a diploid genome of oocytes, and their protein products interact with each other to reject fertilization. These two genes contain a hypervariable region, which enables a specific interaction between s-Themis-B and v-Themis-B proteins encoded by the same allele. The Themis-A locus has a structure similar to the Themis-B locus. An incompatible reaction occurs, only if both Themis-A and Themis-B alleles are matched between sperm and eggs.

To maintain the incompatibility system of Themis-A and Themis-B, recombination needs to be suppressed within these loci, because such recombination would destroy their function by rendering gametes carrying different alleles incompatible. However, there are no indication that structural differentiation occurs around the Themis-A and Themis-B loci, because the heterozygosity of regions close to these two loci is not prominently high compared with other regions (Satou et al. 2012). Hence, if these two self-incompatibility loci do evoke structural differentiation of chromosomes, the process remains at a preliminary stage.

In this study, we found that a long heterozygous region in the genome of an inbred line of C. intestinalis contained the self-incompatibility Themis-B locus. We examined a possibility that this locus contributes to the formation of this long heterozygous region.

Results

Sequencing of Genomes of Two Sibling Animals of an Inbred Line

We established an inbred strain of the tunicate, C. intestinalis, from an individual that was caught in Onagawa-Bay, Miyagi Prefecture, Japan. Taking advantage of the incomplete self-incompatibility of this animal (Harada et al. 2008), we repeated self-fertilization. Using sperm, we analyzed genomes of two mature siblings (specimens A and B) obtained after 11 self-fertilization (F11 generation). Using illumina and 454 sequencers, we obtained approximately 1.6 × 10⁸ sequence tags from each of these two sibling genomes (table 1). Over 1 × 10¹⁰ bases were mapped onto the nonrepeated region of the reference genome (Dehal et al. 2002; Satou et al. 2008), which corresponded to over 100× sequence coverage. Then, we called genotypes of the 89,770,299 and 89,531,196 nucleotide positions, which correspond to approximately 80% of the reference sequence (table 1).

Table 1.

Statistics of Sequencing Data and Mapping Data.

	Specimen A	Specimen B
Number of Illumina sequence tags	158,996,466	162,332,248
Total length of Illumina sequence tags	15,693,529,667	16,034,418,242
Number of 454 sequence tags	953,649	860,633
Total length of 454 sequence tags	130,066,159	102,565,614
Total nucleotide length	15,823,595,826	16,136,983,856
Total number of nucleotides mapped onto the reference genome sequence (112,162,187 bases)	11,393,757,651	11,895,122,867
Mean sequence depth	101.5x	106.1x
Total number of nucleotides mapped onto the repeat-masked genome sequence (97,042,271 bases)	10,032,892,695	10,576,941,089
Nucleotide positions where genotypes are called	89,770,299	89,531,196
Heterozygous positions	38,072	39,177
Homozygous positions	89,732,207	82,492,000
Homozygous positions that are different from the reference sequence	991,115	992,810
Ambiguous positions	20	19

We found that 0.04% of nucleotide positions were heterozygous in each sibling genome (table 1). Genotypes could not be determined for 20 and 19 positions in the genomes of specimens A and B, respectively, which could be due to misalignments of sequence tags to the reference genome. The observed frequencies of heterozygous sites of specimens A and B were much lower than heterozygosity rates in natural populations (1.1–1.2%) (Dehal et al. 2002; Satou et al. 2012), but higher than the rate expected after 11 generations of self-fertilization assuming neutrality (1.1–1.2% × (1/2)¹¹ =0.00054–0.00059%).

Among homozygous sites, 1.1% of genomic positions of the inbred animals were different from the reference (table 1). Thus, each of the four haplotypes of the two siblings, we analyzed differed from the reference by approximately 1.12% (=1.1% + 0.04%/2). This estimate showed great agreement with the mean nucleotide diversity of C. intestinalis (Dehal et al. 2002; Satou et al. 2012), confirming the accuracy of genotype calling.

There Are Few Neutral Heterozygous Sites That Would Become Homozygous by Further Inbreeding

To test whether any neutral heterozygous sites or regions were remaining in the genome of the inbred strain, we compared genome sequences between the two sibling specimens. If neutral heterozygous sites or regions remained in the genome of the inbred strain, a quarter of these sites would become homozygous in both specimens, another quarter would remain heterozygous in both specimens, and the rest would become heterozygous in one specimen and homozygous in the other (Hom/Het sites or regions). At 89,068,420 positions where genotypes were determined commonly in both siblings, 2,962 positions were homozygous in specimen A and heterozygous in specimen B, and another 2,820 positions were heterozygous in specimen A and homozygous in specimen B (5,782 Hom/Het sites in total; table 2).

Table 2.

Comparisons of Heterozygous and Homozygous Sites between the Two Sibling Specimens.

	Number of Sites
Genomic positions whose genotypes were called in both siblings	89,068,420
Homozygous sites that are identical between the siblings	89,028,708
Homozygous sites that are not identical between the siblings	1
Heterozygous sites that are identical between the siblings	33,903
Heterozygous sites that are not identical between the siblings	0
Sites that are homozygous in the specimen A and heterozygous in the specimen B	2,962 (68a)
Sites that are homozygous in the specimen B and heterozygous in the specimen A	2,820 (39a)
Ambiguous sites	26

^aCandidates for bona fide Hom/Het sites (see the main text and Materials and Methods).

Deep inspection of raw data revealed that at 5,675 of these 5,782 Hom/Het sites, two heterozygous bases were found in aligned tags in both specimens, although in one animal, frequencies of the secondarily common bases were below the threshold (see Materials and Methods). Thus, these 5,675 (2,894 and 2,781 in specimens A and B, respectively) sites could have been miscalled in either specimen (not bona fide Hom/Het sites). Even if all of these sites were indeed heterozygous, the above estimated frequencies of heterozygous sites are not significantly increased.

The remaining 107 (=5,782–5,675) Hom/Het sites were considered candidates for bona fide Hom/Het sites. Because of genetic linkage, most Hom/Het sites are expected to make clusters on the genome. Of the 107 Hom/Het sites, 56 heterozygous sites were indeed found in a 5,224-bp region of chromosome 5 of specimen B (fig. 1A–C), and therefore these sites likely represent bona fide Hom/Het sites, or multiple copies of this region in one haplotype of the genome of specimen B. In the case of multiple copies, the above 56 sites might not be actual Hom/Het sites, but the corresponding region should be considered as a Hom/Het “region.” Indeed, the mean sequence coverage of this region in specimen B (160×) was 1.6 times as deep as that in specimen A (101×). An additional 31 potential Hom/Het sites were found in seven small genomic regions (length = 2, 7, 11, 29, 461, 1,662, and 2,612 bp), although the remaining 20 (=107-56-31) sites were dispersed over the genome. Statistically, half of the neutral heterozygous sites/regions are expected to become homozygous in one sibling and heterozygous in the other. Hence, even if the above 107 sites or 87 (=107-20) sites in eight regions actually represented neutral Hom/Het sites, most heterozygous regions in the parental F10 animal could hardly become homozygous by further inbreeding.

Fig. 1.

The number of sequence tags aligned to the region from the 303,101-th to 317,100-th positions on chromosome 5. At each position, the number of the nucleotide that was most frequently found is shown by black dots, and the number of secondarily most frequently identified nucleotide is shown by red dots. Regions where no sequence tags are aligned have no dots. Two siblings are shown in (A) and (B), respectively. Two genes encoded in this region are indicated in (C). Exons are shown by black boxes and introns are shown by thin lines.

As described earlier, a quarter of neutral heterozygous regions in the parental genome are expected to become homozygous in both siblings, and half of them are expected to be different from each other. There was only one such site at nucleotide position 5,250 of scaffold KHL81 (“G” in specimen A and “C” in specimen B). Direct sequencing of the amplicon of this region of the ancestral F9 genome revealed a homozygous C, but showed no indication of G (fig. 2). Therefore, the G nucleotide most likely occurred by mutation between the F9 and F11 generations. This observation again supports the concept that there are few neutral heterozygous sites that could become homozygous by further inbreeding, and that most heterozygous sites found in the inbred strain will be maintained by further inbreeding.

Fig. 2.

A chromatogram showing homozygosity of the 5,250-th nucleotide position of scaffold KHL81 of the F9 animal. Although the deep-sequencing data indicated that this position is homozygous G in specimen A and homozygous C in specimen B, direct sequencing of a genomic fragments amplified from the ancestral F9 genome revealed that this position is actually C (an arrow).

A 2.4-Mb Region Containing a Self-Incompatibility Locus Is Retained Heterozygous

Although the above analysis showed that most neutral sites became homozygous in the F11 genome, sequence coverage and distribution of heterozygous sites of specimens A and B in nonoverlapping 100-kb chromosomal windows showed that an approximately 2.4-Mb region on chromosome 7 is highly heterozygous (fig. 3A and supplementary fig. S1A, Supplementary Material online). This region contained over 40% of the heterozygous sites we identified, and encoded more than 300 genes. Because the result in the preceding section predicts that this region will not easily become homozygous by further inbreeding, and because the recombination rate in C. intestinalis is estimated to be 25–49 kb/cM (Kano et al. 2006), this heterozygous region likely lacks recombination.

Fig. 3.

A chromosomal distribution for heterozygosity in specimen A. (A–C) Sequence coverage (blue lines) and frequencies of heterozygous sites (red lines) (A) in nonoverlapping 100-kb chromosomal windows across the 14 chromosomes, which comprise 68% of the current assembly, (B) in nonoverlapping 10-kb windows across the highly heterozygous region on chromosome 7, and (C) in nonoverlapping 1-kb windows across the region encoding the Themis-B locus. Positions of Themis-A and Themis-B are shown by arrows in (A) and (B). In (C), the constant regions of three copies of Themis-B are indicated by with black arrows. Arrows shows directions of s-Themis-B on the genome. The hypervariable region of the first (left) copy of Themis-B is shown with a red line.

We confirmed the high heterogeneity of this 2.4-Mb region on chromosome 7 by sequencing an amplicon of a 783-bp long genomic region of the F4–F9 genomes. These genomes all contained 19 heterozygous sites (supplementary table S1, Supplementary Material online).

The detailed view of the 2.4-Mb region of specimen A in nonoverlapping 10-kb windows showed that there were two peaks of heterozygosity at the 1,840,001–1,850,000-th and 1,920,001–1,930,000-th positions (fig. 3B). This and a still more detailed view of the regions containing these two peaks in nonoverlapping 1-kb windows (fig. 3C) showed that the peak regions encoded a self-incompatibility locus, Themis-B (in the reference sequence, there are three Themis-B genes). Figure 3C also shows that the observed frequency of heterozygous sites of regions around the Themis-B locus, particularly the intervening region, is higher than the mean heterozygosity rate in natural populations (1.1–1.2%; Dehal et al. 2002; Satou et al. 2012) and in more distant regions.

A hypervariable region is contained in the N-terminal end of s-Themis-B (Harada et al. 2008). Although no sequence tags were mapped onto the hypervariable regions of the second and third copies of Themis-B genes (middle and right genes in fig. 3C), a considerable number of sequence tags were mapped onto the hypervariable region of the first copy (left gene in fig. 3C). This Themis-B allele appears to be encoded in either haplotype of the inbred strain, because no heterozygous nucleotides were found and the sequence coverage was almost half the average (fig. 3C). Careful manual inspection of Roche 454 tags succeeded in identifying four different alleles of Themis-B, which we will further confirm in the following section. In contrast, a substantial number of sequence tags were mapped onto the constant regions of the three Themis-B genes. Therefore, the inbred strain likely had multiple copies of Themis-B, and sequence tags for their hypervariable regions were not mapped because of their variability.

In contrast, the genomic region containing the other self-incompatibility locus, Themis-A, did not show high heterogeneity (supplementary fig. S1B, Supplementary Material online). In addition, deep inspection of Roche 454 tags succeeded in identifying one allele but not multiple alleles. To confirm this observation, we amplified a genomic fragment containing the hypervariable region of Themis-A using a primer for a constant region of the s-Themis-A gene with a primer for the upstream region of Themis-A from the F9 genome. We obtained only one allele, which was the same as that identified by deep inspection of Roche 454 tags (supplementary fig. S2, Supplementary Material online). Although we could not determine whether the Themis-A locus in the F0 genome was homozygous, the above results strongly indicate that the Themis-A locus is homozygous in the inbred strain.

Structural Variations of the Self-Incompatibility Locus

The above observation raised the possibility that the Themis-B locus, which is thought to be maintained heterozygous, is responsible for the long heterozygous region. Although a previous study speculated that multiple copies of Themis-B are artifacts caused by misassembly of the genome (Harada et al. 2008), the above observation implied that multiple copies were indeed encoded in each haplotype of the inbred line.

To confirm that the inbred line has multiple copies of the Themis-B locus, we amplified a genomic fragment containing the hypervariable regions between alleles of s-Themis-B and v-Themis-B using a primer for a constant region of the s-Themis-B gene with a primer for the v-Themis-B gene. We obtained four different genomic fragments from the F9 genome and determined the nucleotide sequences of them individually (alleles A, B, C, and D in supplementary figs. S3–S5, Supplementary Material online). Allele B was identical to the allele encoded in the first copy of the reference genome sequence. Thus, the inbred animals had multiple copies of the Themis-B locus per haplotype.

To further confirm this result, we determined the copy number of s-Themis-B in the genome of the F9 animal using quantitative polymerase chain reaction (qPCR). Under the assumption that two copies of a control Macho-1 gene are encoded per diploid, the qPCR assay indicated that six copies of s-Themis-B, and two copies of another control gene, FoxA-a, are encoded in the F9 diploid genome (fig. 4A). Thus, our assay found six copies of four different Themis-B alleles in the F9 genome.

Fig. 4.

Allele variants, copy number variation, and variant number variation of the Themis-B locus may affect the rate of self-fertilization. (A) Copy numbers of FoxA-a (gray bars) and s-Themis-B (red bars) in eight animals including the F9 inbred animal. Error bars indicate standard errors among triplicates. (B) Alleles found in the present study are indicated by red boxes. (C) In addition to the copy numbers of FoxA-a (gray bars) and s-Themis-B (red bars) in nine animals, numbers of Themis-B variants identified by PCR cloning (blue bars) and the rate of self-fertilization (a black line) are shown. Error bars on gray and red bars indicate standard errors among triplicates.

Next, we examined the genomes of seven wild animals (wt1–wt7) and found that these animals carried two to ten copies of the s-Themis-B gene (fig. 4A). By genotyping of the Themis-B locus in the wt1 and wt2 genomes, which have five and seven copies of Themis-B, respectively, we identified three (alleles D–F) and two (A and C) alleles of Themis-B (fig. 4B and supplementary figs. S3–S5, Supplementary Material online). Therefore, multiple copies of multiple Themis-B variants are likely encoded in these genomes, and are not specific to the inbred strain.

The self-Fertilization Rate in Association with Structural Variations of the Self-Incompatibility Locus

Multiple copies of Themis-B imply that the self-incompatibility system is not as simple as previously proposed. Indeed, 79% of eggs obtained from the F9 animal (884 of 1,117) were self-fertilized. We reasoned that structural variations of this locus should be related to the self-fertilization rate, if the Themis-B locus is responsible for formation of the 2.4-Mb heterozygous region. For this purpose, we determined the rate of self-fertilization in association with copy numbers and nucleotide sequences of alleles of Themis-B in nine different wild type animals (wt8–wt16) (fig. 4B and C and supplementary figs. S3–S5, Supplementary Material online). We identified only one allele (allele A) in wt16, the eggs of which were almost completely self-fertilized. This observation indicates that the pair of s-Themis-B and v-Themis-B encoded in this allele does not effectively block self-fertilization. On the other hand, s-Themis-B and v-Themis-B proteins encoded in allele H, which was the only allele found in wt10, apparently reject self-fertilization efficiently (fig. 4B and C). This allele H is most similar to allele F; s-Themis-B and v-Themis-B proteins encoded in these two alleles were 90% and 81% identical, respectively (supplementary figs. S4 and Supplementary Data, Supplementary Material online). Two individuals with allele F (wt8 and wt9) showed the lowest self-fertilization rates (fig. 4B and C). Thus, different variants likely have different reactivity, and self-fertility and genotypes of this locus appear correlated.

The high self-fertilization rate of wt15 indicates that alleles D, I, and J do not react very efficiently (fig. 4B and C); if either of these three alleles reacted efficiently, the self-incompatibility reaction would take place (Saito et al. 2012), and no self-fertilization would occur. Wt11 had only alleles I and J, and showed a lower self-fertilization rate than wt15 (fig. 4B and C). Wt11 had a smaller number of Themis-B copies and a smaller number of variants, which might increase the likelihood that a specific type of s-Themis-B expressed in a sperm cell would encounter its counterpart (v-Themis-B) on the egg’s vitelline coat. This hypothesis is also supported by the following observation. The self-fertilization rate of wt12, which had three variants including alleles I and J, was higher than that of wt11, but lower than that of wt15 (fig. 4B and C), whereas the copy number was larger than that of wt11 and smaller than that of wt15. Although we cannot rule out the possibility that some alleles of the Themis-A self-incompatibility locus did also not react efficiently in wt8–wt16, the above observations strongly indicate that the self-fertilization efficiency is affected by different reactivity among variants, copy number, and allele number of Themis-B.

Discussion

Establishment of Inbred Lines

Because of the simplicity and compactness of the genome, as well as simple development, Ciona is widely used for experimental biology including developmental and evolutionary studies. Nevertheless, no inbred Ciona strain has yet been made available, and animals from natural populations are typically used for experiments. Because this animal has a large gene pool (Satou et al. 2012), genetic variations might often have affected reproduction of results obtained from different animals. Deep genomic sequencing of the inbred strain established here has shown that this strain possesses only 0.04% of heterozygous nucleotides, which is much smaller than the nucleotide diversity rate in natural populations (1.1–1.2%). In addition, decoding of the genomes of two siblings predicted that there are few neutral sites that could potentially become homozygous by further inbreeding. Thus, this strain can now be used as a new biological resource.

The Self-Incompatibility Locus Themis-B

We identified 12 Themis-B alleles in the genomes of the F9 and eleven wild animals, which were derived from animals caught in Onagawa Bay. Eight of these alleles were shared among multiple individuals. In addition, the B allele is found in the reference genome, which is derived from a specimen caught on West Coast of the United States (Dehal et al. 2002). These observations imply that this animal has a limited number of variants of the self-incompatibility gene in natural populations.

The number of variants appears much smaller than the theoretical expectation based on population sizes (Wright 1939; Kimura and Crow 1964); 60–90 variants are expected for even a small population of 10,000 individuals. However, our results strongly indicated that different reactivity among variants, copy number variation, and variant number variation all influence self-fertilization efficiency, and a combination of these factors probably increases the functional complexity of this self-incompatibility locus. At the same time, this system makes self-incompatibility between eggs and sperm of Ciona incomplete, such that outcrossing occurs more frequently under natural conditions, where self-sperm and nonself-sperm are available, and self-fertilization occurs in laboratory conditions.

A half of offspring of our inbred line are expected to be homozygous for Themis-B. The observation that this locus is maintained heterozygous indicates that the fertility of animals homozygous for Themis-B is lower than the fertility of heterozygous animals, and animals homozygous for Themis-B rarely generate offspring. In this sense, this incomplete self-incompatibility locus has strong heterozygous advantage in the inbred strain, and likely causes inbreeding depression, which is widely observed in offspring of related individuals (Charlesworth and Willis 2009).

The Large Heterozygous Region Might Represent a Preliminary Stage of Structural Differentiation of Chromosomes

The recombination rate in C. intestinalis is estimated to be 25–49 kb/cM (Kano et al. 2006), and three Themis-B genes in the reference genome are encoded within approximately 100 kb. Therefore, the 2.4-Mb heterozygous region in inbred animals strongly suggests that recombination rarely occurs in this region, and it seems likely that this recombination suppression is related to the Themis-B locus. First, Themis-B is retained heterozygous. Previous studies have shown that large genomic regions segregate together with self-incompatibility or mating type loci in other organisms (Boyes et al. 1997; Lahn and Page 1999; Ferris et al. 2002; Lengeler et al. 2002; Liu et al. 2004). Second, any recombination within the hypervariable region of Themis-B across alleles would probably impair their function, because v-Themis and s-Themis encoded in each allele specifically react with each other and do not react with those encoded in other alleles. Third, the high degree of nucleotide variability and copy number variations should physically suppress recombination at this locus. Fourth, it is possible that different haplotypes have different arrangements of multiple Themis-B copies, including inversions. Such variation may be directly related to recombination suppression in the large regions flanking the Themis-B locus, although this hypothesis remains to be tested. Finally, the observation that the frequency of heterozygous sites in the genomic region between the second and third Themis-B genes was higher may suggest that this region has started to diverge.

Intervening homozygous regions within the 2.4-Mb heterozygous region (fig. 3B), however, indicate that recombination is not completely suppressed. There may be several loci with heterozygous advantage in the 2.4-Mb heterozygous region, although we could not find candidate genes with heterozygous advantage other than Themis-B. Linked deleterious mutations might also confer heterozygous advantage. Because the observed frequency of heterozygous sites of regions close to the Themis-B locus is higher than those of more distant regions within the 2.4-Mb heterozygous region, genes or genomic regions with heterozygous advantage in distant regions and modifiers of recombination rates would have been acquired under the linkage disequilibrium caused by the Themis-B locus. Thus, the 2.4-Mb heterozygous region of chromosome 7 might represent a preliminary stage of structural differentiation of chromosomes.

Materials and Methods

Inbreeding and Sequence Data

A single individual originating from a natural population in Onagawa, Miyagi Prefecture, Japan was chosen as the F0 animal. Although C. intestinalis is a hermaphroditic and outcrossing is its normal reproductive mode, self-fertilization occurs under laboratory conditions. Each generation, offspring were obtained by self-crossing. DNA extracted from sperm cells of two F11 animals were individually sequenced with illumina GA2 and Roche 454 sequencers.

Mapping and Genotype Calling

We first trimmed low quality regions (quality values < 25) from sequencing tags with a “TrimmingReads.pl” script (Patel and Jain 2012). Obtained tags were mapped onto the reference genome sequence using ssaha2 (Ning et al. 2001) with default parameters for illumina tags and Roche 454 tags. We used nucleotide positions in nonrepeat regions that were covered by 20 or more tags because of the error-prone nature of sequence tags. Because nucleotide positions that are too deeply covered might represent repeat sequences, we also excluded nucleotide positions that were covered by more than 203 and 212 tags, which were twice as large as the averaged sequence depths of specimens A and B, respectively (table 1).

Before genotype calling, we screened out repeats in the reference genome sequence (KH version) (Satou et al. 2008). We first took 100-bp sliding windows with 50-bp steps and aligned them onto the reference genome using blat (Kent 2002) with the “fastMap” option. Sequence fragments that were more than 50% identical with multiple regions were regarded as repeats. As a result, 15,119,916 nt were found in repeats and were excluded from the subsequent analyses.

Using the mapped sequence tags, genotypes were called using previously published criteria (Harismendy et al. 2009; Nielsen et al. 2011). We called positions with common allele read frequencies over 80% as homozygous, and positions with secondarily common allele read frequencies over 20% as heterozygous, unless ternary common allele read frequencies were over 20%.

PCR Genotyping

We amplified a genomic fragment shown in supplementary table S1, Supplementary Material online, from the F4 to F9 genomes by PCRs with a set of primers; 5′-ACTGCAAACAAGTTTCCGTTGT-3′ and 5′-AGTGTTACTGACTTGAGATTACTA-3′. Amplified sequence fragments were directly sequenced using an ABI3130xl sequencer. Genomic DNA was extracted from sperm cells of the F4–F9 animals. The genomes of F0–F3 and F10 were accidentally lost and were not analyzed.

We also amplified a genomic fragment of the scaffold KHL81 from the F9 genome with the following primers. 5′-AACGAATTGTAAAGTACGCTCACGA-3′, and 5′-AGAAGCTCTCAGCCAATGAGCGT-3′.

The Themis-B locus of the genomes of the F9 animal and wild-caught animals was amplified with the following primers: 5′-CTGGAAAGTTATCCAACCAGTT-3′ and 5′-TTGTTGGGATATTT(A/G)TTGATTTCT-3′. Amplified fragments were cloned and sequenced. For each genome, we sequenced at least 18 clones.

The Themis-A locus of the F9 genome was amplified with the following primers: 5′-ATTTCGATCTCAATAGACACCAA-3′ and 5′-TGTTACATTCAATGTGCCAAGT-3′.

Measurement of Copy Number Variations

DNA extracted from sperm cells of the F9 animal and wild-caught animals were used to determine copy number variation. We performed two distinct multiplex-quantitative-PCRs for each genomic DNA. We amplified a constant region of the s-Themis-B and Macho-1 genes in the first reaction and FoxA-a and Macho-1 genes in the second reaction. We adopted a TaqMan method using a TaqMan Universal PCR Master Mix (Invitrogen). Probes for s-Themis-B, FoxA-a, and Macho-1 are as follows: s-Themis-B, 5′-(FAM)-CAGCGCTATCATTAGAT-(MGB)-3′; FoxA-a, 5′-(FAM)-TCTGCCGTTGAAGTTAGTTCGCCATCC-(TAMARA)-3′; Macho-1, 5′-(VIC)-ACGGTCACTTTAGCACCTCCACCA-(TAMARA)-3′. As calibrators, we made two DNA constructs, which contained amplicon sequences of s-Themis-B and Macho-1, and of FoxA-a and Macho-1. Sequences of these calibrators are shown in supplementary fig. S6, Supplementary Material online. After calibrating the different efficiencies between probe/primer sets with these calibrators, we calculated expected numbers of s-Themis-B and FoxA-a per diploid, assuming that Macho-1 exists in a single copy gene per haploid (two copies per diploid). Standard errors were calculated among triplicates. PCR primers were as follows: s-Themis-B, 5′-TGATGAATGTAAATTGGTTCAAGTCAA-3′ and 5′-TGAGGAACGGTTTCAAACACTTG-3′; FoxA-a, 5′-TTCAACACCACCACACTCAACAG-3′ and 5′-CGTGTTCAATGCCATGTTC-3′; Macho-1, 5′-CCCAGTATGCACCAAATTCAGA-3′, and 5′-TGGTGAGAAAACGGGTGAAAC-3′. The fertilization ratio was determined as previously described (Harada et al. 2008).

Acknowledgments

This research was supported by the National Bio-Resource Project of Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan to Yu.S., T.E., Ya.S., K.I., and N.S., and a Grant-in-Aid from the MEXT, Japan, to Yu.S. (grant number 21671004). Grants-in-Aid from the MEXT to K.I. (grant number 21112004) and Ya.S. (grant number 23681039) also supported this study in part. The authors thank Chikako Imaizumi at Kyoto University and members of Shimoda Marine Research Center, University of Tsukuba for maintaining the inbred Ciona line. Dr Steven D Aird is acknowledged for his help in editing the manuscript. The obtained sequence tags are deposited in the DDBJ Sequence Read Archive (DRA) under the accession number DRA002218.

Author notes

‡Present address: Department of Immunology and Microbiology, Meiji University of Integrative Medicine, Nantan-city, Kyoto, Japan

Associate editor: Naruya Saitou