163
Continuing Evolution of Xenopus Genome
fugu or by losing one of the two HoxD clusters in the lineage
including zebrafsh (reviewed in Kuraku and Meyer, 2009;
Pascual-Anaya et al., 2013).
The third round of WGD in the teleost lineage happened
about 306 Mya (Inoue et al., 2015), which is a much more
ancient event than the WGD in X. laevis about 18 Mya. As
expected, the X. laevis genome has eight Hox clusters consisting of pairs of HoxA, B, C, and D clusters on L and S chromosomes, exactly doubling the four Hox clusters present in X.
tropicalis (Figure 10.2B) (Session et al., 2016; Kondo et al.,
2017). The 76 hox genes in X. laevis included hoxb2p.L,
which appeared to be pseudogenized, since its complete coding sequence could not be identifed and indels were found
when compared to its homolog hoxb2.S. Although the Hox
clusters are highly conserved, comprehensive gene expression analyses of hox genes revealed differences in expression
patterns between L and S hox genes during development or
in adult tissues, which suggest subfunctionalization. In addition, the hoxb.L genes in the HoxB.L cluster appear to be
rapidly diverging compared to hoxb.S genes (Kondo et al.,
2017). This trend is the opposite of other homeologous
gene sets, in which generally S genes tend to be subjected
to mutations, pseudogenization, or partial or total deletions
(Session et al., 2016).
Collinearity is the idea that the order of genes in the
Hox cluster is correlated to spatial and temporal sequential
gene expression during development. Spatial collinearity
has been demonstrated in various organisms in which the
anterior hox genes (located to the 3’ end of the cluster) are
expressed anterior to posterior genes (located to the 5’ end).
However, the defnition of “temporal collinearity” is ambiguous, that is, it is not clear whether “gene expression” refers
to the accumulation of mRNA or activation of gene transcription. Moreover, due to this ambiguity, the evidence supporting temporal collinearity has been weak; as a result, the
“collinearity” of a complete set of genes within a cluster has
never been proven. We analyzed the developmental expression patterns of all hox genes of X. laevis by RNA-seq and
grouped them according to their profles by clustering analysis, that is, when mRNA accumulation starts and reaches the
maximum amount. We realized that temporal collinearity
hypothesis could not be supported (Kondo et al., 2017).
To further examine the temporal collinearity hypothesis, since genomic and transcriptome data of X. tropicalis are available and there are no homeologs to distinguish
in this species, we used X. tropicalis to investigate two
aspects of “gene expression” to test the temporal collinearity hypothesis. First, with the help of high-resolution transcriptome analysis in X. tropicalis (Owens et al., 2016),
we examined the order of genes whose transcript level
reaches a certain threshold. Second, the timing of the start
of de novo transcription was determined using RT-qPCR
to detect pre-spliced transcripts for all genes in the HoxA
cluster and some from other clusters (Kondo et al., 2019).
As a result, we were able to demonstrate that the temporal
collinearity theory is not experimentally supported. These
analyses were possible because of the recent enrichment of
genome and transcriptome information of the two species.
Methodologically, the detection of de novo transcripts was
based on designing qPCR primers at introns or exon/intron
junctions using genome sequence information.
10.3.3. DETAILED GENE ORGANIZATION IN THE
XENOPUS LAEVIS SUBGENOMES
From the late 1980s to the 1990s, many interesting genes
were identifed in X. laevis, but some of them were not found
in mice or humans as orthologs, and some other genes had
multiple paralogs, probably by local gene duplication or
expansion. To clarify orthologous, paralogous, and homeologous relationships, elucidation of chromosomal localization and syntenic gene organizations of these genes was
inevitable. Our gene annotation group in the international
consortium of the X. laevis genome project examined such
well-studied genes encoding transcription factors (Watanabe
et al., 2017; Haramoto et al., 2017), peptide growth factors
( Michiue et al., 2017; Suzuki et al., 2017a), signal transduction components (Suzuki et al., 2017b), cell cycle regulators
(Tanaka et al., 2017), and others. In addition, genomic organization was further clarifed for large gene clusters, such
as the olfactory receptor gene clusters (see Extended Data
Figure 5a in Session et al., 2016), and the type I and II keratin gene clusters (Suzuki et al., 2017). In the following, we
introduce some remarkable fndings from these analyses.
10.3.3.1. nodal5, nodal3, and vg1 Clusters
Extreme asymmetry between subgenomes L and S was
found in the nodal5 and nodal3 clusters (Session et al., 2016;
Suzuki et al., 2017a), as well as the vg1 cluster (Suzuki et al.,
2017a) (Figure 10.3A, B, C ). These are examples of clusters in
which all functional genes were lost from the S subgenome.
nodal5 and nodal3 are TGFß family members, and the numbers of the genes are expanded (Figure 10.3A, B). Probably
more than fve copies of nodal5 genes exist on XLA3L (personal communications from Dr. Shuji Takahashi and Dr.
Yoshikazu Haramoto). Other TGFß family members, vg1
and derriere, are duplicates of a common ancestor gene and
became subfunctionalized paralogs; vg1 is a maternal factor
in the egg, whereas derriere is zygotically expressed. vg1 is
present as a single copy gene in X. tropicalis, whereas in X.
laevis, the vg1.L gene has expanded, forming a gene cluster,
but vg1.S was pseudogenized (Figure 10.3C ). Notably, there
are two types of the expanded vg1 genes, a functional Ser20
type (S20) and a non-functional Pro20 type (P20) as paralogs (Suzuki et al., 2017a), in which vg1 (P20) was f rst identifed as a maternal mRNA (Rebagliati et al., 1985). These
three clusters have common features: (1) the numbers of
expanded genes, including pseudogenes and mutated genes,
in the subgenome L of X. laevis are more or less the same or
more than those in X. tropicalis, whereas the corresponding
genes are missing or pseudogenized in subgenome S and (2)
the expanded genes appeared to be functionally equivalent
(not subfunctionalized), implying that these gene expansions
Continuing Evolution of Xenopus Genome
fugu or by losing one of the two HoxD clusters in the lineage
including zebrafsh (reviewed in Kuraku and Meyer, 2009;
Pascual-Anaya et al., 2013).
The third round of WGD in the teleost lineage happened
about 306 Mya (Inoue et al., 2015), which is a much more
ancient event than the WGD in X. laevis about 18 Mya. As
expected, the X. laevis genome has eight Hox clusters consisting of pairs of HoxA, B, C, and D clusters on L and S chromosomes, exactly doubling the four Hox clusters present in X.
tropicalis (Figure 10.2B) (Session et al., 2016; Kondo et al.,
2017). The 76 hox genes in X. laevis included hoxb2p.L,
which appeared to be pseudogenized, since its complete coding sequence could not be identifed and indels were found
when compared to its homolog hoxb2.S. Although the Hox
clusters are highly conserved, comprehensive gene expression analyses of hox genes revealed differences in expression
patterns between L and S hox genes during development or
in adult tissues, which suggest subfunctionalization. In addition, the hoxb.L genes in the HoxB.L cluster appear to be
rapidly diverging compared to hoxb.S genes (Kondo et al.,
2017). This trend is the opposite of other homeologous
gene sets, in which generally S genes tend to be subjected
to mutations, pseudogenization, or partial or total deletions
(Session et al., 2016).
Collinearity is the idea that the order of genes in the
Hox cluster is correlated to spatial and temporal sequential
gene expression during development. Spatial collinearity
has been demonstrated in various organisms in which the
anterior hox genes (located to the 3’ end of the cluster) are
expressed anterior to posterior genes (located to the 5’ end).
However, the defnition of “temporal collinearity” is ambiguous, that is, it is not clear whether “gene expression” refers
to the accumulation of mRNA or activation of gene transcription. Moreover, due to this ambiguity, the evidence supporting temporal collinearity has been weak; as a result, the
“collinearity” of a complete set of genes within a cluster has
never been proven. We analyzed the developmental expression patterns of all hox genes of X. laevis by RNA-seq and
grouped them according to their profles by clustering analysis, that is, when mRNA accumulation starts and reaches the
maximum amount. We realized that temporal collinearity
hypothesis could not be supported (Kondo et al., 2017).
To further examine the temporal collinearity hypothesis, since genomic and transcriptome data of X. tropicalis are available and there are no homeologs to distinguish
in this species, we used X. tropicalis to investigate two
aspects of “gene expression” to test the temporal collinearity hypothesis. First, with the help of high-resolution transcriptome analysis in X. tropicalis (Owens et al., 2016),
we examined the order of genes whose transcript level
reaches a certain threshold. Second, the timing of the start
of de novo transcription was determined using RT-qPCR
to detect pre-spliced transcripts for all genes in the HoxA
cluster and some from other clusters (Kondo et al., 2019).
As a result, we were able to demonstrate that the temporal
collinearity theory is not experimentally supported. These
analyses were possible because of the recent enrichment of
genome and transcriptome information of the two species.
Methodologically, the detection of de novo transcripts was
based on designing qPCR primers at introns or exon/intron
junctions using genome sequence information.
10.3.3. DETAILED GENE ORGANIZATION IN THE
XENOPUS LAEVIS SUBGENOMES
From the late 1980s to the 1990s, many interesting genes
were identifed in X. laevis, but some of them were not found
in mice or humans as orthologs, and some other genes had
multiple paralogs, probably by local gene duplication or
expansion. To clarify orthologous, paralogous, and homeologous relationships, elucidation of chromosomal localization and syntenic gene organizations of these genes was
inevitable. Our gene annotation group in the international
consortium of the X. laevis genome project examined such
well-studied genes encoding transcription factors (Watanabe
et al., 2017; Haramoto et al., 2017), peptide growth factors
( Michiue et al., 2017; Suzuki et al., 2017a), signal transduction components (Suzuki et al., 2017b), cell cycle regulators
(Tanaka et al., 2017), and others. In addition, genomic organization was further clarifed for large gene clusters, such
as the olfactory receptor gene clusters (see Extended Data
Figure 5a in Session et al., 2016), and the type I and II keratin gene clusters (Suzuki et al., 2017). In the following, we
introduce some remarkable fndings from these analyses.
10.3.3.1. nodal5, nodal3, and vg1 Clusters
Extreme asymmetry between subgenomes L and S was
found in the nodal5 and nodal3 clusters (Session et al., 2016;
Suzuki et al., 2017a), as well as the vg1 cluster (Suzuki et al.,
2017a) (Figure 10.3A, B, C ). These are examples of clusters in
which all functional genes were lost from the S subgenome.
nodal5 and nodal3 are TGFß family members, and the numbers of the genes are expanded (Figure 10.3A, B). Probably
more than fve copies of nodal5 genes exist on XLA3L (personal communications from Dr. Shuji Takahashi and Dr.
Yoshikazu Haramoto). Other TGFß family members, vg1
and derriere, are duplicates of a common ancestor gene and
became subfunctionalized paralogs; vg1 is a maternal factor
in the egg, whereas derriere is zygotically expressed. vg1 is
present as a single copy gene in X. tropicalis, whereas in X.
laevis, the vg1.L gene has expanded, forming a gene cluster,
but vg1.S was pseudogenized (Figure 10.3C ). Notably, there
are two types of the expanded vg1 genes, a functional Ser20
type (S20) and a non-functional Pro20 type (P20) as paralogs (Suzuki et al., 2017a), in which vg1 (P20) was f rst identifed as a maternal mRNA (Rebagliati et al., 1985). These
three clusters have common features: (1) the numbers of
expanded genes, including pseudogenes and mutated genes,
in the subgenome L of X. laevis are more or less the same or
more than those in X. tropicalis, whereas the corresponding
genes are missing or pseudogenized in subgenome S and (2)
the expanded genes appeared to be functionally equivalent
(not subfunctionalized), implying that these gene expansions