159
Continuing Evolution of Xenopus Genome
As mentioned before, several hundreds of long scaffolds
had been assigned to 18 chromosomes by FISH analysis using
BACs located in the scaffolds as probes. The homeologous
relationships of the scaffolds could be determined by their
chromosomal locations as well as the presence of homeologous
genes that the scaffolds contained. However, to distinguish the
subgenomes consisting of nine chromosomes each, derived
from the two diploid ancestor species, we needed to f nd some
“marks” specifc to either of the parental chromosome sets.
For this purpose, we searched for “fossil” transposons which
were horizontally transferred into either of the parental species
and expanded but became inactivated before allotetraploidization (Figure 10.1C, D). After identifying and analyzing a
large number of fossil transposon sequences in the scaffolds,
we found that two PIF/harbinger-type and one Tc1/marinertype DNA transposons were distributed unevenly in either of
each homeologous pair of scaffolds, suggesting that they were
subgenome-specifc fossil transposons. The biased distribution of fossil transposons on either of actual homeologous
chromosomes was confrmed by FISH using the Tc1/marinertype transposon as a probe (Figure 10.1E) (Session et al.,
2016). This result indicated that the set of nine chromosomes
belonging to each of the extinct ancestor species remained
largely intact as a subgenome in X. laevis without massive
reorganization.
Once we successfully identifed the subgenomes, we
needed to name them. Naming them “A” and “B” would
probably have been the frst choice in general, but the suff xes
“a” and “b” were already used randomly for homeologs, so
we had to f nd another pair of suff xes. In parallel with this,
we were analyzing gene syntenies between homeologous
scaffolds and fnding cases that either of the homeologs was
absent, being a single-copy gene or “singleton.” Furthermore,
we realized that singleton genes and tandemly repeated genes
(gene expansions; see subsequently) were biased toward the
Tc1/mariner-negative subgenome. Meanwhile, we noticed
from karyotypes that the Tc1/mariner-positive chromosomes
all appeared shorter than their counterparts (Figure 10.1E).
Therefore, we precisely measured the relative lengths of all
chromosomes, and found that this observation was correct
(see Table 1 in Matsuda et al., 2015). The biased deletion of
genes from the Tc1/mariner-positive chromosomes matches
with those chromosomes being shorter.
If the chromosomes tend to have fewer gene deletions/
more gene expansions, they will become relatively longer
than those with more gene deletions/less gene expansions.
Based on this, we proposed to name the homeologous chromosomes “L” for “longer” and “S” for “shorter,” as well as
the corresponding subgenomes “L” and “S.” The difference
in length between homeologous chromosomes is consistent
with the characteristics of the subgenomes.
10.2.5. ASYMMETRICAL EVOLUTION OF
THE L AND S SUBGENOMES
By distinguishing the L and S subgenomes, we compared
the divergence of protein-coding gene sequences and estimated that the two ancestor species arose about 34 Mya
(Figure 10.1A) (Session et al., 2016). A draft genome sequence
of X. borealis, which is related to X. laevis and shares a common allotetraploid ancestor, allowed us to estimate their speciation about 17 Mya. Together with estimation of expansion
and cessation periods of the L- and S-specif c transposons,
the allotetraploidization event occurred about 17–18 Mya
(Figure 10.1A) (Session et al., 2016).
Based on the assembly sequence ver. 9.1 and gene annotation ver. 1.8, the number of protein-coding genes in the
genome of X. laevis was estimated to be 45,099 in total,
about twice that of the diploid X. tropicalis (Session et al.,
2016). According to the allotetraploid nature of X. laevis, a
single gene in X. tropicalis is supposed to correspond to a
pair of its orthologous homeologs in X. laevis : for example,
a gene on XTR1 is present as copies on both XLA1L and
XLA1S. If a pair of homeologs became a singleton in X. laevis, the relationship would be one to one. Therefore, genes
of X. tropicalis corresponding to those of X. laevis in either
a one-to-two or one-to-one relationship were counted. As a
result, of the 15,613 protein-coding genes of X. tropicalis,
8806 corresponded to homeolog pairs of X. laevis, whereas
the remaining 6807 genes corresponded to singleton genes.
The retention rate of homeolog pairs is calculated to be 56%.
Comparison of homeologous gene loci between L and S
shows that the L chromosomes are more similar to X. tropicalis, while the S chromosomes have several large inversions (Figure 10.1B), suggesting that the S chromosomes
are more prone to accumulating changes (Session et al.,
2016). Furthermore, focusing on the genes with one-to-one
correspondence between X. tropicalis and X. laevis (singletons), we found that more genes were lost from the S subgenome (8.3% of the genes in L and 31.5% in S were lost).
Differences in expression among homeologs were examined
by RNA-seq analysis of 14 developmental stages and 14
adult tissues. The results showed that zygotic expression levels of L genes were, on average, approximately 25% higher
than those of the corresponding S genes. In addition, genes
with lower expression levels tended to accumulate more
mutations, suggesting that they are pseudogenizing or perhaps subfunctionalizing.
Another typical example for asymmetry of the L and S
subgenomes is the rDNA (45S pre-ribosomal RNA gene) cluster. The number of rDNA genes in the cluster was originally
estimated to be about 450 (Brown and Dawid, 1968) and, more
recently, about 750 (Michalak et al., 2015). The nucleolus is
formed at the rDNA cluster, and therefore the rDNA region is
also called the nucleolar organizing region (NOR). The NOR
in X. laevis was known to be on XLA3L (the same as the
previously called chromosome 12) (Schmid and Steinlein,
1991). FISH analysis using rDNA as a probe conf rmed
that the rDNA gene cluster was detected only on XLA3L,
not on XLA3S (see Extended Data Figure 5a in Session
et al., 2016). In general, of the two NORs inherited from the
parental species of a hybrid, only one is actively transcribed,
the phenomenon called nucleolar dominance (reviewed in
McStay, 2006; Preuss and Pikaard, 2007). This suggests that
the parental L species of X. laevis had nucleolar dominance,
leading to asymmetric evolution of the rDNA cluster.
Continuing Evolution of Xenopus Genome
As mentioned before, several hundreds of long scaffolds
had been assigned to 18 chromosomes by FISH analysis using
BACs located in the scaffolds as probes. The homeologous
relationships of the scaffolds could be determined by their
chromosomal locations as well as the presence of homeologous
genes that the scaffolds contained. However, to distinguish the
subgenomes consisting of nine chromosomes each, derived
from the two diploid ancestor species, we needed to f nd some
“marks” specifc to either of the parental chromosome sets.
For this purpose, we searched for “fossil” transposons which
were horizontally transferred into either of the parental species
and expanded but became inactivated before allotetraploidization (Figure 10.1C, D). After identifying and analyzing a
large number of fossil transposon sequences in the scaffolds,
we found that two PIF/harbinger-type and one Tc1/marinertype DNA transposons were distributed unevenly in either of
each homeologous pair of scaffolds, suggesting that they were
subgenome-specifc fossil transposons. The biased distribution of fossil transposons on either of actual homeologous
chromosomes was confrmed by FISH using the Tc1/marinertype transposon as a probe (Figure 10.1E) (Session et al.,
2016). This result indicated that the set of nine chromosomes
belonging to each of the extinct ancestor species remained
largely intact as a subgenome in X. laevis without massive
reorganization.
Once we successfully identifed the subgenomes, we
needed to name them. Naming them “A” and “B” would
probably have been the frst choice in general, but the suff xes
“a” and “b” were already used randomly for homeologs, so
we had to f nd another pair of suff xes. In parallel with this,
we were analyzing gene syntenies between homeologous
scaffolds and fnding cases that either of the homeologs was
absent, being a single-copy gene or “singleton.” Furthermore,
we realized that singleton genes and tandemly repeated genes
(gene expansions; see subsequently) were biased toward the
Tc1/mariner-negative subgenome. Meanwhile, we noticed
from karyotypes that the Tc1/mariner-positive chromosomes
all appeared shorter than their counterparts (Figure 10.1E).
Therefore, we precisely measured the relative lengths of all
chromosomes, and found that this observation was correct
(see Table 1 in Matsuda et al., 2015). The biased deletion of
genes from the Tc1/mariner-positive chromosomes matches
with those chromosomes being shorter.
If the chromosomes tend to have fewer gene deletions/
more gene expansions, they will become relatively longer
than those with more gene deletions/less gene expansions.
Based on this, we proposed to name the homeologous chromosomes “L” for “longer” and “S” for “shorter,” as well as
the corresponding subgenomes “L” and “S.” The difference
in length between homeologous chromosomes is consistent
with the characteristics of the subgenomes.
10.2.5. ASYMMETRICAL EVOLUTION OF
THE L AND S SUBGENOMES
By distinguishing the L and S subgenomes, we compared
the divergence of protein-coding gene sequences and estimated that the two ancestor species arose about 34 Mya
(Figure 10.1A) (Session et al., 2016). A draft genome sequence
of X. borealis, which is related to X. laevis and shares a common allotetraploid ancestor, allowed us to estimate their speciation about 17 Mya. Together with estimation of expansion
and cessation periods of the L- and S-specif c transposons,
the allotetraploidization event occurred about 17–18 Mya
(Figure 10.1A) (Session et al., 2016).
Based on the assembly sequence ver. 9.1 and gene annotation ver. 1.8, the number of protein-coding genes in the
genome of X. laevis was estimated to be 45,099 in total,
about twice that of the diploid X. tropicalis (Session et al.,
2016). According to the allotetraploid nature of X. laevis, a
single gene in X. tropicalis is supposed to correspond to a
pair of its orthologous homeologs in X. laevis : for example,
a gene on XTR1 is present as copies on both XLA1L and
XLA1S. If a pair of homeologs became a singleton in X. laevis, the relationship would be one to one. Therefore, genes
of X. tropicalis corresponding to those of X. laevis in either
a one-to-two or one-to-one relationship were counted. As a
result, of the 15,613 protein-coding genes of X. tropicalis,
8806 corresponded to homeolog pairs of X. laevis, whereas
the remaining 6807 genes corresponded to singleton genes.
The retention rate of homeolog pairs is calculated to be 56%.
Comparison of homeologous gene loci between L and S
shows that the L chromosomes are more similar to X. tropicalis, while the S chromosomes have several large inversions (Figure 10.1B), suggesting that the S chromosomes
are more prone to accumulating changes (Session et al.,
2016). Furthermore, focusing on the genes with one-to-one
correspondence between X. tropicalis and X. laevis (singletons), we found that more genes were lost from the S subgenome (8.3% of the genes in L and 31.5% in S were lost).
Differences in expression among homeologs were examined
by RNA-seq analysis of 14 developmental stages and 14
adult tissues. The results showed that zygotic expression levels of L genes were, on average, approximately 25% higher
than those of the corresponding S genes. In addition, genes
with lower expression levels tended to accumulate more
mutations, suggesting that they are pseudogenizing or perhaps subfunctionalizing.
Another typical example for asymmetry of the L and S
subgenomes is the rDNA (45S pre-ribosomal RNA gene) cluster. The number of rDNA genes in the cluster was originally
estimated to be about 450 (Brown and Dawid, 1968) and, more
recently, about 750 (Michalak et al., 2015). The nucleolus is
formed at the rDNA cluster, and therefore the rDNA region is
also called the nucleolar organizing region (NOR). The NOR
in X. laevis was known to be on XLA3L (the same as the
previously called chromosome 12) (Schmid and Steinlein,
1991). FISH analysis using rDNA as a probe conf rmed
that the rDNA gene cluster was detected only on XLA3L,
not on XLA3S (see Extended Data Figure 5a in Session
et al., 2016). In general, of the two NORs inherited from the
parental species of a hybrid, only one is actively transcribed,
the phenomenon called nucleolar dominance (reviewed in
McStay, 2006; Preuss and Pikaard, 2007). This suggests that
the parental L species of X. laevis had nucleolar dominance,
leading to asymmetric evolution of the rDNA cluster.