Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

16.1 Introduction

The study of the origin and evolution of vertebrates, the subphylum to which we belong, has stood at the crossroad between genome evolution and molecular developmental biology since the late 1960s, when Susumo Ohno published his famous work on Evolution by Gene Duplication and proposed his hypothesis about the pivotal role of genome duplication in the origin of vertebrates and their diversification (Ohno et al. 1968; Ohno 1970). Vertebrates comprise all animals that have a backbone and include mammals, birds, reptiles, amphibians, fishes, and agnathans—the jawless lampreys and hagfishes. Vertebrates together with urochordates (tunicates) form the Olfactores, which together with cephalochordates (amphioxus or lancelets) constitute the chordates (Fig. 16.1). All chordates share a common basic body plan at least during the larval stage of their life cycle, consisting of a notochord running through a post-anal tail, with a dorsal hollow nerve cord, longitudinal blocks of muscle along the notochord, and ciliated pharyngeal gill slits (Brusca and Brusca 2002). Recent phylogenomic analyses have dethroned cephalochordates from the long-assumed position as sister group of the vertebrates; this position is now occupied by urochordates, which include ascidians, larvaceans and thaliaceans (Wada et al. 2006; Oda et al. 2002; Bourlat et al. 2006; Delsuc et al. 2006; Putnam et al. 2008) (Fig. 16.1).

Fig. 16.1
figure 1

Evolutionary tree of chordates representing the three possible scenarios for the timing of the 2R-WGD and the origins for vertebrate innovations. Evidence suggests that the 2R-WGD might have had a significant impact on the diversification (black star) of vertebrate innovations, including structures derived from neural crest cells and placodes, as well as the development of a big complex brain in the stem vertebrate. But whether the 2R-WGD was crucial for the evolutionary origin of these structures remains unclear, and that the hypothesis that the origin of these innovations dated back to at least stem olfactores (gray star) cannot be dismissed. (Lamprey picture courtesy of Juan Pascual-Anaya)

Full size image

Ohno’s 2R-hypothesis was based on comparative analyses of genome sizes and isozyme complexity among chordate taxa. Ohno found that basally divergent chordate subphyla had smaller genomes and less isozyme complexity than vertebrate lineages. This observation led him to suggest that the combination of tandem gene duplication and in particular an octoploidization event involving two rounds of whole-genome duplication were key to the invertebrate–vertebrate transition, and for the subsequent successful vertebrate diversification (Ohno et al. 1968).

Ohno was one of the pioneers in conceiving the evolutionary significance of whole-genome duplication (see also Chap. 1, this volume). Ohno emphasized the importance of gene duplication as probably the main source of raw genetic material for the evolution of new gene functions (reviewed in Taylor and Raes 2004). In Ohno’s classical model, one of the duplicated genes retains the original function whereas its duplicate either disappears by accumulation of detrimental mutations (called pseudogenization or nonfunctionalization) or it is preserved after gaining advantageous mutations that confer positively selected novel functions (neofunctionalization) (Ohno 1970; Nowak et al. 1997; Force et al. 1999). The duplication, degeneration, complementation hypothesis (or DDC model) suggests a third possibility for duplicate gene preservation: subfunctionalization, the complementary partitioning of ancestral structural and/or regulatory subfunctions between two duplicate genes, so that the sum of their functions provides at least that of the original pre-duplication gene (Force et al. 1999). The DDC model predicts that subfunctionalized genes will have lower pleiotropy than the original pre-duplicated gene and lower evolutionary constraints, and thereby will be more permissive to the accumulation of mutations that might confer novel functions. The acquisition of new functions is favored if the duplication affects the entire genome at once, as opposed to multiple individual gene duplications, because when entire gene networks are duplicated, gene stoichiometry is maintained, and therefore deleterious gene dosage effects can be counteracted (Birchler and Veitia 2007, 2010; Van de Peer et al. 2009; Makino and McLysaght 2010; see also Chap. 2, this volume). Genes that originated by gene duplication are called paralogs. Genes that have been duplicated via genome duplication, however, are a special type of paralogs referred as ohnologs, a term suggested by Wolfe (2000) in honor of Ohno’s contribution. This term is useful because of the special properties that ohnologs possess at their birth compared to duplications that arise by other local mechanisms such as unequal crossing-over, tandem gene duplication, or retrotransposition.

The sudden creation and evolution of ohnologs by the 2R-WGD that occurred in the stem vertebrate lineage has been suggested as one of the potential key events underlying the increase of morphological complexity, facilitating the acquisition of genetic and developmental innovations of vertebrates (Shimeld and Holland 2000; Aburomia et al. 2003). Genome duplication doubles the number of genes, many of which have the chance of evolving ‘novel’ functions that might provide new selectable advantages promoting species diversification (Lynch et al. 2001; Van de Peer et al. 2009). Some vertebrate species that have undergone recent polyploidization, such as the frog Xenopus laevis that experienced tetraploidization ~40 million years ago (Hellsten et al. 2007), show a higher adaptability to a variety of different environments, such as drought, salt, cold, and disease resistance, than closely related diploid species, such as Silurana tropicalis (for further details on frog polyploidization see also Chap. 18, this volume). Interestingly, only a limited number of retained ohnologs present evidence of neofunctionalization or subfunctionalization in X. laevis, suggesting that additional selective mechanisms might act on preserving gene duplicates that could promote species diversification (Chain and Evans 2006; Semon and Wolfe 2008).

In addition to the evolutionary significance of novel fates of duplicate genes on species biology, another mechanism that might contribute to species diversification after genome duplication is reciprocal ohnolog loss between different populations, a concept known as ‘divergent resolution’ that can lead to reproductive isolation (Lynch and Force 2000). Reciprocal ohnolog loss is likely to occur in the period of relaxed selection that duplicate genes experience while they are functionally redundant (Werth and Windham 1991; Lynch and Conery 2000; Lynch and Force 2000; Scannell et al. 2006; Taylor et al. 2001; Semon and Wolfe 2007). The divergent resolution of gene redundancies, such that one population loses one ohnolog copy while the second population loses the other ohnolog copy, leads to chromosomal restructuring such that gametes produced by hybrid individuals can be completely lacking in functional genes for a duplicate pair. In addition to the isolation due to reciprocal gene losses, this model can be further expanded to isolation due to independent processes of gene duplicate subfunctionalization between two populations, in which hybrids will lack one or more subfunctions (Force et al. 1999; Lynch and Force 2000). Hence, large-scale reciprocal ohnolog loss and independent subfunctionalization of ohnologs can be the cause of reproductive isolation of two populations after polyploidization, favoring genetic divergence of these newly incipient future species. This hypothesis is supported by the analyses of both fish and angiosperm lineages that have undergone polyploidization and include more species diversity (e.g. salmonids, catostomids, eudicots, grasses) than their sister groups that did not go through polyploidization and include a lower number of species (Nelson 1994; Ferris et al. 1979; Soltis et al. 2009). Recent integrated approaches of comparative genomics and gene expression analyses in teleosts, however, provide limited evidence supporting the significance of differential ohnolog loss in reproductive isolation and diversification (Kassahn et al. 2009) (see also Chap. 17, this volume).

Many studies have tackled the central question of whether or not the 2R-WGD had a significant impact on the origin of vertebrate innovations and their subsequent diversification. This chapter first reviews evidence supporting the 2R-hypothesis and information regarding the timing and potential mechanisms underlying the 2R-WGD in vertebrates. Second, this chapter examines the impact that the 2R-WGD may have had on the evolution of vertebrate genome structure, number of genes, and the fate of retained ohnologs in comparison with non-vertebrate chordate paralogs. Finally, this chapter discusses how the 2R-WGD might have affected the origin and evolution of vertebrate innovations, with special emphasis on the vertebrate big complex brain, and structures derived from neural crest cells and placodes. Recent data, however, suggest that neural crest cells and placodes could already have been present in stem chordates. Hence, the impact of the 2R-WGD cannot be related to the origin of neural crest cells and placodes, but it could be related to their subsequent diversification and development of a wide variety of complex structures.

16.2 Supporting Evidence for the 2R-Hypothesis

If two rounds of genome duplication (2R-WGD) have occurred, we would expect the presence of many paralogs (ohnologs) in conserved, syntenic genomic regions, which are known as paralogons (Coulier et al. 2000) (or ohnologons (Gout et al. 2009)). Paralogons, therefore, consist of series of linked (but frequently functionally and phylogenetically unrelated) genes on one chromosome region, many of which have linked paralogs on at least one other chromosome region. The discovery of paralogy groups made of four paralogons in the genome of human and mouse was interpreted as remnants of the two events of tetraploidization that occurred early during vertebrate evolution and therefore provided the earliest strong evidence supporting Ohno’s 2R-hypothesis (Lundin 1979, 1993; Pebusque et al. 1998). Possibly one of the best and first examples of a paralogy group supporting the 2R-hypothesis is the case of the four Hox-bearing regions on human chromosomes Hsa2, Hsa7, Hsa12, and Hsa17 (Fig. 16.2) (Kappen et al. 1989; Bailey et al. 1997; Larhammar et al. 2002; Lundin et al. 2003). In contrast, only a single Hox cluster is present in the cephalochordate amphioxus (Garcia-Fernàndez and Holland 1994). This 4:1 ratio is consistent with Ohno’s hypothesis of two tetraploidization events after the cephalochordate–vertebrate split (Holland et al. 1994; Sidow 1996; Garcia-Fernandez 2005). In addition to the Hox paralogy group, several other similar examples have been identified (e.g. MHC (Katsanis et al. 1996), Tbx (Ruvinsky and Silver 1997), G-protein-coupled receptors (Fredriksson et al. 2003), ParaHox clusters (Ferrier et al. 2005), linked receptor tyrosine kinases (Siegel et al. 2007), endothelin ligands and receptors (Braasch et al. 2009), Fox cluster (Wotton and Shimeld 2006), and the EGF ligand paralogons (Laisney et al. 2010)).

Fig. 16.2
figure 2

Conserved synteny in the vertebrate genome generated by the 2R-WGD. a Simplified representation of a genomic region that has been amplified by the 2R-WGD (R1 and R2) showing conserved genes (colored boxes) in four syntenic regions (a–d), which have suffered genomic rearrangements and gene loss (white boxes) and different degrees of conservation (green and red lines label ohnologs preserved in two or more than two regions, respectively). b Representation of the four human paralogons containing Hox A-D clusters in chromosomes Hsa2, Hsa7, Hsa12, and Hsa17, displaying high amounts of conserved synteny between ohnologs in two (green lines) or at least three (red lines) different paralogons. This representation has been generated using a 100-gene sliding window in the Synteny Database (Catchen et al. 2009)

Full size image

Several databases of conserved syntenic chromosomal regions between different species are available, and these provide additional evidence of two rounds of genome duplication. Popovici et al. (2001), for instance, identified 14 paralogons containing more than 1600 genes assembled in a human genome paralogy map (http://u119.marseille.inserm.fr/Db/paralogy.html). The ParaDB (http://abi.marseille.inserm.fr/paradb) predicted that the human genome far exceeds 1000 paralogons that contain more than three pairs of duplicated genes (Leveugle et al. 2003). The Genomicus database v60.01 (http://www.dyogen.ens.fr/genomicus) predicted that 18,228 ancestral vertebrate genes were grouped in 2,642 conserved ancestral synteny blocks with a median N50 size of 5 genes (Muffato et al. 2010). The Synteny Database (http://teleost.cs.uoregon.edu/synteny_db) predicted 231 paralogy clusters with more than 5 genes, and 102 paralogy clusters with more than 10 genes, in a more rigorous count using a 100-gene sliding window and taking amphioxus genes as the outgroup for paralogy assignment in human (Catchen et al. 2009, 2011).

Until recently, however, available evidence did not permit us to discard the possibility that these groups of paralogous genes originated by multiple independent block duplications, rather than two duplications of the entire genome (Skrabanek and Wolfe 1998; Wolfe 2001; Larhammar et al. 2002). An initial hypothesis, based on extensive phylogenetic analysis and dating of the duplications that produced hundreds of vertebrate gene families, proposed a ‘big-bang mode’ of sudden large-scale gene origin resulting from two waves of gene duplications, rather than the alternative hypothesis of a constant generation by small-scale duplications (Gu et al. 2002). Wave-I was suggested to consist of tandem or segmental duplications that occurred after the mammalian radiation, and wave-II was interpreted as a rapid increase of paralogs in the early stage of vertebrate evolution after their split from non-vertebrate chordates, consistent with one round of whole-genome duplication (Gu et al. 2002). The first analyses of the human genome draft led to the conclusion that the most parsimonious explanation of the current structure of the human paralogons was a ‘big-bang’ expansion event by a paleopolyploidy that included the whole genome or substantial sections of it. However, no specific evidence was found for two rounds of polyploidy as opposed to one (Venter et al. 2001; McLysaght et al. 2002; Panopoulou et al. 2003).

Recently, however, several analyses have provided definitive support for the 2R-hypothesis (reviewed in Kasahara 2007). Dehal and Boore (2005) developed an elegant, compelling approach to test the 2R-hypothesis by plotting the genomic map position of only those genes that were duplicated prior to the fish–tetrapod split, which rendered a clear global physical pattern of four-way paralogon organization covering most of the human genome. Dehal and Boore’s work therefore provided unmistakable evidence of two distinct rounds of genome duplication during early vertebrate evolution.

Furthermore, the recent sequencing of the whole genome of the chordate amphioxus Branchiostoma floridae (sister to all other chordates) provided even more indisputable evidence supporting the 2R-hypothesis (Putnam et al. 2008). Despite the fact that small-region comparison between human, chicken, teleost fish, and amphioxus genomes revealed low gene-order conservation at the local level (microsynteny), striking extensive gene linkage conservation was observed when entire chromosomes were considered (macrosynteny). Syntenic analysis reconstructed 17 chordate linkage groups (CLG) that might represent the proto-chromosomes of the last common chordate ancestor (Putnam et al. 2008). Exhaustive evaluation of the 17 CLGs revealed that most of the human genome (112 segments spanning 2.68 Gb, which is the equivalent of 95 % of the euchromatic genome) was affected by large-scale duplication events that occurred on the stem vertebrate lineage before the teleost/tetrapod split. Analysis of the distribution of the human segments among the 17 CLGs showed that nearly all ancient chordate chromosomes were quadruplicated (Putnam et al. 2008) (Fig. 16.3). This result robustly demonstrated the occurrence of two rounds of genome duplication, corroborating previous lines of evidence based on analysis of specific regions of interest, such as the Hox-bearing regions (Garcia-Fernàndez and Holland 1994; Larhammar et al. 2002) and the major histocompatibility regions (Vienne et al. 2003; Danchin and Pontarotti 2004). Spring (1997) proposed the term “tetralogs” to refer groups of quadruplicated vertebrate genes at four different chromosomal locations formed by the 2R-WGD corresponding to a single invertebrate gene, with all four more similar to each other than to members of the other tetralogy group.

Fig. 16.3
figure 3

Quadruplicated conserved syntenic pattern between the amphioxus and the human genome as a result of the 2R-WGD. Dot-plots display the distribution throughout human chromosomes (y-axes) of human orthologs (blue dots) of amphioxus genes (black dots) located in two arbitrarily selected genomic regions of approximately 1 Mb (a) and 10 Mb (b) (x-axes). The dot-plots reveal four major human chromosomes (yellow shadow) of conserved synteny as the product of the two rounds of whole-genome duplication. In panel (a), the four paralogons coincide with the Hox-cluster bearing chromosomes 2, 7, 12, and 17, whereas in panel B the four paralogons coincide with the endothelin receptors and ParaHox-cluster bearing chromosomes 2/5, 4, 13, and X. The dot-plots were generated as described in Canestro et al. (2009) using the Synteny Database (Catchen et al. 2009)

Full size image

16.2.1 Timing of the Vertebrate 2R-WGD

Analyses of the completely sequenced genome of the cephalochordate amphioxus (Putnam et al. 2008; Holland et al. 2008) and the genomes from various urochordates (Dehal et al. 2002; Small et al. 2007; Denoeud et al. 2010) validated Ohno’s hypothesized lower-bound timing for the 2R-WGD as after the split between vertebrates and non-vertebrate chordates. Regarding the upper-bound timing, extensive analysis of gene duplicates (Robinson-Rechavi et al. 2004) and the identification of the four clusters in the genome of the elephant shark suggested that the 2R-WGD took place before the cartilaginous/bony vertebrate split (Venkatesh et al. 2007; Ravi et al. 2009).

Within this time window, the most prevalent hypothesis suggests a scenario in which the first round (R1) occurred before the split between gnathostome and jawless vertebrates, and the second (R2) occurred in the stem jawed vertebrates after their divergence from jawless vertebrates (Fig. 16.1). However, a second scenario proposes that both rounds (R1 + R2) of genome duplication took place before the split between gnathostome and jawless vertebrates (pan-vertebrate quadruplication (PV4) hypothesis (Kuraku et al. 2009)) (Fig. 16.1). Comparative analysis of 55 gene families revealed a common expansion in both jawless and jawed vertebrates, which has been interpreted as evidence supporting this second scenario (Kuraku et al. 2009). Available information from sea lampreys and hagfish does not permit us to discern between these two hypothetical scenarios, because these organisms also appear to have suffered lineage-specific duplications and reciprocal gene losses compared to vertebrates, which together obscure the assessments of orthology/paralogy (reviewed in Kuraku 2008, 2010). For instance, multiple Hox gene surveys in different species of sea lampreys and hagfish suggested that extensive independent duplications of Hox genes might have occurred during the evolution of jawless vertebrates (Pendleton et al. 1993; Sharman and Holland 1998; Takio et al. 2004; Force et al. 2002; Irvine et al. 2002; Fried et al. 2003; Stadler et al. 2004; Kuraku et al. 2009). Some of the jawless Hox clusters might have disintegrated, casting doubt as to the usefulness of Hox genes as reliable markers to trace duplications during genome evolution in stem vertebrates (Kuraku 2011). Finally, recent phylogenetic analysis of the degenerated ParaHox cluster in hagfish has opened the possibility of a third scenario, in which both rounds (R1 + R2) occurred in stem jawed vertebrates after their divergence from jawless vertebrates (Furlong et al. 2007) (Fig. 16.1). The validation of this third scenario could have a significant impact on our understanding of vertebrate evolution, because it would imply that the 2R-WGD would have not been important for the origin of vertebrate innovations (i.e. big brain, neural crest cells, and placodes, which clearly exist in jawless vertebrates (Kuratani and Ota 2008; Kuratani 2009)). According to this third scenario, however, the 2R-WGD would have been important for the radiation of gnathostomes into cartilaginous fish, bony fishes, and tetrapods. A solid answer about the timing of the 2R-WGD may have to wait until larger-scale comparisons of the whole-genome organization of hagfish and lampreys are available.

16.2.2 Mechanisms Underlying the Vertebrate 2R-WGD

A question that still remains is how did the stem vertebrate genome become octoploid by two rounds of tetraploidization. There are two main mechanisms of tetraploidization observed in many species of plants and animals (Van de Peer et al. 2009). The first mechanism is allotetraploidy, which occurs when two related but not identical genomes are combined by hybridization of closely related species and associated (often subsequent) genome duplication. In the case of allotetraploidy, the pairs of distinct ‘homologous’ chromosomes that are sufficiently different due to their separated origin are called homeologs. The second mechanism is autotetraploidy, which occurs when the genomes are not sufficiently diverged into homeologous sets; autotetraploidy therefore ranges from the combination of genomes of two conspecific individuals (perhaps from different populations) to the combination of identical genomes from a single individual. The genetic attributes of allo- and autotetraploids differ and may have substantial effects at individual, population, and species levels (see also Chap. 2, this volume). Both allotetraploidy and autotetraploidy could be generated by several processes such as: (i) an abnormal non-disjunction of sister chromatids at meiosis; (ii) the uncoupling of mitotic DNA replication and cell division during early development of the germ line (this process, for instance, occurs normally during the endoreplication of megakaryocytic bone-marrow precursors of blood platelets, or during the development of the oikoblastic epithelia that secrete the house in basal urochordate larvaceans); (iii) potential cell fusion during early embryo development or in germ-line precursors in syncytial gametogenesis (cell fusion is observed naturally, for instance, in skeletal muscle cells and placenta) (reviewed in Storchova and Pellman 2004; Shemer and Podbilewicz 2000).

In the case of allotetraploids, each pair of homologous chromosomes should segregate normally during meiosis, and genetic interchange between homeologous chromosomes is rare. If two consecutive events of allotetraploidization occurred in stem vertebrates, we would predict that in the ideal situation in the absence of gene losses, a phylogenetic tree of homeologs will render a symmetrical (A,B) (C,D) topology (Furlong and Holland 2002).

In autopolyploids, however, meiotic pairing might occur between any of the four identical chromosomes at meiosis I, facilitating genetic interchanges freely among the four alleles, and leading to ‘tetrasomic inheritance’. Eventually the alleles, and chromosomes, might diverge, starting a process of diploidization that reestablishes diploidy. Randomly one of the chromosomes will diverge first and no longer form homologous structures, while the other three will keep pairing until another further divergence. Hence, if two consecutive events of autotetraploidy occurred in quick succession (pseudo-octoploidy) in stem vertebrates, we would predict that in the absence of gene losses, gene family phylogenetic trees will likely render asymmetrical (((A,B),C),D) topologies (Furlong and Holland 2002). Because many vertebrate gene families do render asymmetrical tree topologies (Friedman and Hughes 2001; Hughes 1999; Hughes and Friedman 2003), two quick consecutive events of autotetraploidy have been considered a likely mechanism for the 2R-WGD in stem vertebrates (Furlong and Holland 2002; Lynch and Wagner 2009).

16.3 Consequences of the 2R-WGD on the Evolution of Vertebrate Genome Structure

It has been suggested that polyplodization events, at least in plants, can trigger genomic stress associated with major genomic rearrangements, in many cases mediated by a burst of mobilization of transposable elements (Matzke and Matzke 1998; Comai 2000). Transposable elements can be substrates for unequal and illegitimate recombination and can be responsible for a variety of genome reorganizations associated with the transposition, including chromosomal insertions, deletions, inversions, translocations, and duplications. Lineage-specific genome rearrangements mediated by transposable elements might facilitate rapid evolution, reproductive isolation of different populations, and consequently species diversification (Parisod et al. 2010).

Contrary to possible genome reorganization after polyploidization, as noted above for plants, in stem vertebrates, recent work based on proximate gene pair methods and measurement of syntenic clustering conservation found that the 2R-WGD in vertebrates were not followed by an increase of genome rearrangement (Hufton et al. 2008). Unexpectedly, this work measured massive genome rearrangements prior to the 2R-WGD, which has been interpreted as a pre-existing ‘disposition’ toward genomic structural change (Hufton et al. 2008). Interestingly, in contrast to the archetypal condition that has been described in the organization of particular genomic regions (e.g. Hox-cluster region (Garcia-Fernàndez and Holland 1994)), the amphioxus genome structure is not exceptionally well conserved, evolving its own particular type of repetitive elements (e.g. ‘mirage’ minisatellites (Cañestro et al. 2002b; Ebner et al. 2010)), undergoing extensive local tandem gene duplications (see section below), and experiencing a moderate rate of synteny loss similar to that of sea urchin or sea anemone (Hufton et al. 2008). Therefore, the amphioxus genome structure cannot be considered a fossil genome representing the pre-duplication condition, at least in terms of genome structure (Garcia-Fernàndez et al. 2001; Hufton et al. 2008), although it is far less divergent from the vertebrate genome structure than is any known urochordate genome (Dehal et al. 2002; Denoeud et al. 2010; Louis et al. 2012).

There have been several attempts to infer the karyotype and genome structure from common chordate ancestors and to reconstruct the evolutionary history leading to present chromosome structures. The first comparisons of conserved syntenic associations in different vertebrate karyotypes, using an in silico chromosome painting approach, allowed reconstructions of the ancestral vertebrate genome containing 10–13 ancestral proto-chromosomes (Kohn et al. 2006; Nakatani et al. 2007). Recently, the sequencing of the amphioxus genome has allowed researchers to reconstruct the ancestral chordate genome as consisting of 17 conserved syntenic blocks, which might represent the ancient chordate proto-chromosomes (Putnam et al. 2008).

After the 2R-WGD, under the naive assumption of absence of loss or fusions of chromosomes, we would expect 68 (17 × 4) proto-vertebrate segments, but parsimonious reconstruction of chromosome history revealed that numerous chromosomal fusions and translocations have occurred. These reconstructions predict at least 20 fusions that led to 37–49 chromosomes in the bony vertebrate ancestor, which became 12 chromosomes in the stem teleost ancestor due to many additional fusions, and 33–45 chromosomes in the stem tetrapod ancestor due to at least 4 fusions shared between human and chicken genomes (Putnam et al. 2008; Naruse et al. 2004; Nakatani et al. 2007). An excellent example of chromosomal rearrangement after the 2R-WGD has been recently provided by a phylogenetic analysis of members of the four Hox paralogons that resulted in a (B(A(C,D))) topology. These results suggest that two chromosomal rearrangements between protochromosomes 11 and 4, and 7 and 5 occurred after the clusters duplicated but before the diversification of extant vertebrates 450 million years ago (Lynch and Wagner 2009). These chromosomal rearrangements resolve conflicting data regarding the order of linked genes and support the hypothesis that the 2R-WGD occurred by two consecutive events of autotetraploidy, and thereby the ancestral vertebrate might have been “pseudo-octoploid”. Interestingly, the asymmetrical (B(A(C,D))) topology of the vertebrate Hox cluster (Lynch and Wagner 2009) contrasts with the symmetrical (A,B) (C,D) topology inferred from the cartilaginous elephant shark using the amphioxus Hox cluster as the outgroup (Ravi et al. 2009). Further extensive analyses including HoxA-D clusters from a broader representation of cartilaginous and bony vertebrates will be required to resolve these conflicting topologies, which could suggest that the Hox-cluster rearrangement took place after the cartilaginous/bony vertebrate split and not immediately subsequent to the 2R-WGD.

16.4 Consequences of the 2R-WGD on the Evolution of Vertebrate Gene Fate

16.4.1 Function of Gene Duplicates After 2R-WGD

After polyploidization, a period of transilience may follow in which genes might enjoy extra ‘degrees of freedom’ to mutate without selective penalty (reviewed in Soltis and Soltis 1999; Otto 2007). Understanding the processes by which genome duplication might influence the fate of duplicated genes is crucial to evaluate how the 2R-WGD might have impacted the evolution of vertebrate innovations. Neofunctionalization and subfunctionalization are the two main processes driving the functional fate of newly generated ohnologs after the 2R-WGD and have been extensively discussed in the literature (Hughes 1994; Force et al. 1999; Lynch and Conery 2000; Durand 2003; Postlethwait et al. 2004; Hoekstra and Coyne 2007; Conant and Wolfe 2008; Semon and Wolfe 2007, 2008; Jimenez-Delgado et al. 2009). A prominent example of neofunctionalization related to the 2R-WGD occurred during the expansion of the vertebrate retinoic acid receptor (RAR) family, which acquired new functions in both their expression domains and in their structural protein activities (Escriva et al. 2006). There are also examples, however, in which neofunctionalization and subfunctionalization are related to both 2R-WGD and local tandem duplications (e.g. the expansion of the vertebrate globin superfamily, which promotes the vertebrate innovation related to oxygen transport and storage (Hoffmann et al. 2012)), or merely related to local tandem duplications, and not the 2R-WGD, such as the expansion of the vertebrate Alcohol Dehydrogenase (Adh) family, which promotes the acquisition of new enzymes for the synthesis of retinoic acid (Cañestro et al. 2000, 2002a, 2003b). Therefore, not all vertebrate innovations can be exclusively attributed to the 2R-WGD, and the global weight of the impact of the 2R-WGD on the evolution of vertebrate gene functions remains unknown.

16.4.2 Gene Network Rewiring by Tranposons After 2R-WGD

In many cases, neofunctionalization and subfunctionalization can be due to alterations in cis-regulatory elements that might lead to adaptative changes in duplicated genes (Force et al. 1999). Many of the cis-regulatory elements appear to be embedded in distinct repeat families, especially in transposable elements (TE) (Thornburg et al. 2006; Polak and Domany 2006; Bourque et al. 2008). Analysis of the distribution of 10,000 TEs in the human genome, for instance, revealed that most TEs are concentrated under strong purifying selection near regulatory and developmental genes (Lowe et al. 2007). Most of the described examples of TE mobilization and rewiring of gene regulatory networks have been associated with relatively recent events of TE mobilizations. For instance, TE-mediated rewiring for neofunctionalization after gene duplication has been recently described for the sex-determining gene dmrt1bY in medaka fish, in which a novel regulatory element driving a negative feedback on dmrt1bY has been acquired due to the insertion of an Izanagi transposon (Herpin et al. 2010), or for the origin of a novel gene regulatory network dedicated to pregnancy in placental mammals, which was due to a transposition of the MER20 TE (Lynch et al. 2011).

A massive expansion of TEs appears, therefore, as a powerful mechanism that could boost a vast redeployment of cis-regulatory elements into new gene regulatory networks (Feschotte 2008), promoting large-scale events of neofunctionalization and subfunctionalization (van de Lagemaat et al. 2003; Bennetzen 2005; Bejerano et al. 2006). Polyploidization can trigger the mobilization of transposable elements (Matzke and Matzke 1998; Parisod et al. 2010), because recently duplicated genomes contain many redundant genes and substantial repetitive DNA, which serve as buffer against TE insertional mutagenesis (Matzke et al. 2000). According to this expectation, bursts of TE mobilization have been described after polyploidization in different organisms (Matzke and Matzke 1998; Comai 2000; SanMiguel et al. 1996, 1998).

A question that remains unclear is whether there was or was not a massive TE mobilization after the 2R-WGD that could have favored a significant redeployment of cis-regulatory elements into new gene regulatory networks in the stem vertebrate lineage. Recent comparison of the diversity and content of TEs between vertebrates and amphioxus has provided some insights that might help to answer this question (Canestro and Albalat 2012). The dynamics of the TE content within a genome follows a competition model in which the expansion of a particular TE might cause the reduction of other types of TEs, consequently reducing the TE diversity, until a new equilibrium that preserves the functionality of the genome is reached (Abrusán and Krambeck 2006). According to this model, if a massive expansion of TEs occurred after the 2R-WGD, we expect that the diversity of TEs shared among vertebrates should be smaller than in cephalochordates. Consistent with this prediction, a recent comparative study reveals that the shared TE diversity of vertebrates (14 superfamilies in lampreys, 28 in ray-finned fishes, 20 in amphibians, 14 in reptiles, 10 in birds, and 15 in mammals) is lower than the TE diversity in amphioxus (33 superfamilies), which makes plausible the hypothesis that a TE burst could have occurred after the 2R-WGD in the stem vertebrate lineage (Canestro and Albalat 2012). Further comparative genomic analysis between different vertebrates and cephalochordates will be required to test this hypothetical burst of TEs, and especially to evaluate its putative impact on the evolution of gene functions after the 2R-WGD.

16.4.3 Ohnologs Gone Missing After 2R-WGD and Impact on Surviving Ohnologs

While several studies focus on the functional fate of retained gene duplicates, less attention has been paid to how losses of paralogs or ohnologs might impact the evolution of the functions of other genes (reviewed in Cañestro et al. 2007). Loss of one copy of two fully redundant gene duplicates should not usually have significant impact, but loss of one of the paralogs after functional divergence likely has evolutionary consequences. Recent analyses of gene losses by comparative genomics have led to the unexpected finding that significant components of the developmental toolkit might be lost without major changes to the body plan (Cañestro and Postlethwait 2007; Holland 2007), which suggests the presence of compensatory mechanisms or the acquisition of innovations that have preserved unaltered the ancestral condition (Cañestro et al. 2007). Tracing the evolution of gene families throughout ancestral proto-chromosomes using blocs of conserved synteny has become a powerful tool to clarify uncertain phylogenies, to detect ohnologs gone missing (OGM) (Postlethwait 2007; Catchen et al. 2009, 2011), to provide robust assessments of orthology and paralogy between different species, and to discern evolutionary innovations from losses of ancestral features in sister lineages (Canestro et al. 2009).

There are cases in which different ohnologs in different species acquire the same expression pattern, which has been called function shuffling (McClintock et al. 2001) and synfunctionalization (Gitelman 2007), and in some cases the convergence of expression patterns between paralogs can be related to OGM (Postlethwait 2007; Canestro et al. 2009). The evolution of the vertebrate retinaldehyde dehydrogenease Aldh1a family provides a paradigmatic example of how uncovering the evolution of gene family members through the 2R-WGD has been fundamental to illuminating how gene functions evolve among newly generated paralogs after genome duplications in the face of loss of ohnologs (Canestro et al. 2009). For instance, analysis of conserved synteny revealed that the presence of Aldh1a1 in tetrapods and its absence in teleost fish was not due to a tetrapod innovation, but to an OGM in the teleost stem lineage, which was accompanied by a re-acquisition of ancestral functions by surviving paralogs (Canestro et al. 2009). Medaka provides a more radical example in which aldh1a2, the only survivor of the aldh1a family in this species, recapitulates the expression pattern of all other aldh1a paralogs that have been lost in medaka. This result is in agreement with a model of functional evolution in which surviving genes re-acquire ancestral gene family roles in the face of loss of ohnologs. Other examples that illustrate the importance of identifying OGMs ohnologs are shown in the endothelin and agouti systems, in which the exclusive presence of endothelin 4 (edn4) and the agouti-signaling protein 2 genes (asip2a/b) in teleost fish was not due to a fish innovation related to the teleost-specific whole-genome duplication, but instead to a loss of ohnologs that originated in the 2R-WGD in the tetrapod lineage (Braasch et al. 2009; Braasch and Postlethwait 2011). To understand acquisition of functions of vertebrate ohnologs that were generated by the 2R-WGD, both the impact of the retention of neo- or subfunctionalized ohnologs, as well as the impact of OGM, on the functions of other survivor gene family members should be studied.

16.5 Consequences of the 2R-WGD on Vertebrate Gene Number and Functional Evolution

How many of the genes that were part of the original fourfold increase in genes generated by 2R-WGD in the stem vertebrate have actually survived nonfunctionalization? And importantly, how significant have the functional consequences of those retained genes been for promoting the origin and evolution of vertebrate innovations? Estimates on gene retention in other organisms that have experienced a WGD have reported ~13 % retention over ~100 million years (MY) in yeast (Wolfe and Shields 1997), ~72 % in maize over ~11 MY (Ahn and Tanksley 1993; Gaut and Doebley 1997), and ~77 % in Xenopus over ~40 MY (Hellsten et al. 2007). In vertebrates, a ~33 % retention of divergent functional genes after the 2R-WGD over ~500 MY was inferred initially based on theoretical models applied to 270 gene families of the human genome (Nadeau and Sankoff 1997). More recent and broader analyses based in the complete catalog of human ohnologs estimated a rate of retention between 20 and 30 % (Putnam et al. 2008; Huminiecki and Heldin 2010; Makino and McLysaght 2010).

But how can we assess the impact of the 2R-WGD on the origin of vertebrate complex features? A naive approach to estimating this impact could be to perform a comparison of the total number of retained paralogs and their distribution among functional categories in vertebrates and non-vertebrate chordates that did not undergo any WGD since their split from our last common chordate ancestor. Comparison of the gene catalog of the three chordate subphyla (i.e. cephalochordates, urochordates, and vertebrates) has allowed us to identify a lower bound of 8,437 gene families with members that descend from a single gene in the last common chordate ancestor (Putnam et al. 2008). Through subsequent genome or local duplication, these families account for 13,610 amphioxus genes, 13,401 human genes, and 7,216 ascidian genes, the latter being a significantly lower number due to the extensive gene losses that have occurred in urochordate lineages (Dehal et al. 2002; Cañestro et al. 2003a; Edvardsen et al. 2005; Denoeud et al. 2010). Although it is frequently true that the multiple ohnologs of a vertebrate gene family are represented by a single gene in amphioxus, the total number of paralogs derived from a single-copy gene in the last common ancestor is surprisingly similar between amphioxus (13,610) and human (13,401) (Putnam et al. 2008). Therefore the mere total numbers of retained genes after the 2R-WGD duplicates might not be the key to explain the gain of complexity during the evolution of the vertebrate lineage in comparison with amphioxus.

In vertebrates, analysis of the functional categories of the gene families that have expanded after the 2R-WGD revealed that cell signalers and transcriptional regulators of developmental pathways are generally retained as multiple ohnologs (Roux and Robinson-Rechavi 2008; Putnam et al. 2008; Hufton et al. 2008; Huminiecki and Heldin 2010). Genes associated with basic cellular functions (i.e. translation, replication, splicing, and recombination, with the important exception of cell cycle), however, have been less successfully retained after the 2R-WGD (Huminiecki and Heldin 2010) (although see Gout et al. (2009) for different results in other organisms that have also undergone WGD). Analysis of the human genome reveals that dosage-balance constraints act on the retention of ohnologs, resulting in an enrichment of dosage-balanced genes, an observation predicted following WGD (Birchler and Veitia 2007, 2010) and also reported for other vertebrates, plants, and yeast (e.g., Paterson et al. 2006). Interestingly, many of these retained ohnologs in humans are refractory to copy number variation, have rarely experienced subsequent small-scale duplication, and are frequently associated with diseases related to dosage-imbalance such as down syndrome (Makino and McLysaght 2010). Analysis of retained genes that have originated in vertebrates by local duplications revealed a strong underrepresentation of genes related to cell communication, cell cycle, and embryo development (Huminiecki and Heldin 2010).

In amphioxus, although a thorough analysis of amphioxus-specific gene family expansions has not been performed, Table 16.1 shows an extensive list of amphioxus-specific duplicated genes reported in the literature (this list is probably not complete, and may be biased toward the research with which I am most familiar). This list shows numerous retained duplicates from a broad array of functional categories, including metabolic enzymes, members of transduction and signaling cascades, members of the immunity system, as well as pivotal transcription factors of developmental pathways. Awaiting a more exhaustive analysis, including different amphioxus species to infer the ancestral cephalochordate condition, the list in Table 16.1 shows no obvious bias toward any particular functional category, although it is noticeable that duplicated developmental transcription factors do not account for more than two paralogs (with the exception of the eight hairy amphioxus paralogs (Minguillon et al. 2003)).

Table 16.1 List of paralogs originated independently in the amphioxus lineage
Full size table

Remarkably, the main difference between the newly acquired paralogs in amphioxus and ohnologs in vertebrates is the mechanism of duplication. While approximately 25 % of the ancestral chordate gene families have two or more ancient vertebrate ohnologs generated by the 2R-WGD, there is strong evidence that most amphioxus paralogs originated by local tandem duplications rather than large-scale chromosomal duplications (Table 16.1). Therefore, considering the functional bias of retention of genes duplicated by WGD or local duplication, it is reasonable to speculate that the key influence of the 2R-WGD promoting the successful diversification of vertebrate features resides in the fact that whole networks were duplicated, in contrast to local duplications such as those that occurred in amphioxus, an organism that seems to have maintained morphological and genetic stasis during the last 200 million years (Garcia-Fernàndez and Holland 1994; Cañestro et al. 2002a; Somorjai et al. 2008; Canestro and Albalat 2012; Paps et al. 2012). Duplication of whole gene networks is dosage-balanced and increases the evolvability to generate novel functions, which in the case of the vertebrate 2R-WGD could have led to an increase in complexity of the signaling and developmental regulatory networks that facilitated the acquisition of innovations.

In addition to the evolutionary role of coding genes in the acquisition of innovations, microRNAs (miRNAs) also play crucial roles during development and have been postulated as important players for the evolution of organismal complexity (Lee et al. 2007; Sempere et al. 2006). Analysis of miRNAs in chordate species showed that the 2R-WGD has increased the diversity of the inventory of miRNAs in vertebrates, which correlated with the increase of complex patterns of tissue specificity of miRNAs (Heimberg et al. 2008; Campo-Paysaa et al. 2011). However, the finding of 41 vertebrate-specific miRNA families, absent in non-vertebrate chordates, suggests that their origin must have occurred in stem vertebrates after their separation from urochordates and is not explained by the 2R-WGD (Heimberg et al. 2008). The appearance of these 41 vertebrate-specific miRNA families has been proposed as a potential key evolutionary force lying behind the dramatic increase of vertebrate complexity (Heimberg et al. 2008). Future exhaustive analysis of the expression patterns of the members of these 41 vertebrate-specific families, and an understanding of their roles, will allow a reevaluation of the importance that this innovation could have had on the origin of vertebrate features.

16.6 Consequences of the 2R-WGD on the Innovation of Vertebrate Features

A major question not resolved yet is the precise impact of the 2R-WGD on the innovation of particular vertebrate features. Three vertebrate features are perhaps the most prominent innovations: derivatives from neural crest cells, sensory organs concentrated in the head derived from ectodermal placodes, and a big complex brain. When taken together, these features probably allowed the transition from ancestral, peaceful, filter-feeding, non-vertebrate chordates to active, voracious, vertebrate predators (Northcutt and Gans 1983; Gans and Northcutt 1983); reviewed in Yu et al. 2008; Holland 2009) (Fig. 16.1).

Vertebrate neural crest cells are a transient population of developmental cells that delaminate at the border of the neural plate through an epithelial–mesenchymal transition, migrate, and differentiate at their final destination into a variety of structures such as sensory neurons, glial cells, peripheral nervous system, pigment cells, smooth muscle cells, connective tissue, cranio-facial cartilage, skeletal bones, and teeth (Weston 1970). Vertebrate crest development depends on four crucial sets of genes that form what is called the neural crest gene regulatory network (NC-GRN): (1) patterning signal genes establish the expression of (2) neural plate border specifier genes, which activate (3) crest specifier genes, which turn on (4) neural crest effector genes that provide differentiated products (Meulemans and Bronner-Fraser 2004, 2005; Ota and Kuratani 2007; Sauka-Spengler et al. 2007). Analysis of the amphioxus genome has revealed the presence of cephalochordate orthologs from all of these four sets of genes, including (1): Fgf, Wnt, Bmp, Notch Dlx, AP2, SoxB, Zic, and islet; (2): Pax3/7, Msx, Dlx5, and Zic; (3): Snail, SoxE, AP2, Twist, Id, FoxD, and Myc; and (4): Rho, cRet, Erbb3, Mitf, tyrosinase, and tyrosinase-related genes, with the remarkable exception of the tyrosine kinase c-Kit essential for migration and survival of crest cells, and the gene for myelin protein P0, consistent with the notion that the glial myelin sheath is a vertebrate innovation (Meulemans and Bronner-Fraser 2007; Holland et al. 2008; Holland and Short 2008; Nikitina et al. 2009).

The fact that most of the specifier genes are present as single copy in amphioxus, but multiple paralogs in vertebrates, presumably due to the 2R-WGD, has led to the hypothesis that neofunctionalization and subfunctionalization of paralogs may have facilitated the co-option of ancestral genes into the NC-GRN (Sauka-Spengler et al. 2007; Meulemans and Bronner-Fraser 2007; Holland et al. 2008; Holland and Short 2008). Gene ontology (GO) analysis estimates that 91 % of the neural crest genes in vertebrates have been co-opted from genes already present in basal metazoans, while the remaining 9 % of the neural crest genes are vertebrate innovations (Martinez-Morales et al. 2007), including the assembly of new signaling pathways like the endothelin system (Braasch et al. 2009). The evolution of the vertebrate NC-GRN, therefore, appears as the result of a combination of ancestral gene co-option, newly evolving genes, and amplification of these components by the 2R-WGDs (Braasch et al. 2009).

Interestingly, however, the recent description in urochordates of neural crest-like cells that express typical vertebrate crest marker orthologs, migrate, and differentiate into pigments challenges the idea that neural crest cells are a vertebrate innovation (Jeffery et al. 2004, 2008; Jeffery 2007). Thus, it cannot be discounted that some types of neural crest cells might have been present in the last common ancestor of olfactores (urochordates + vertebrates), followed by losses during the significant morphological and genetic simplification suffered by urochordate lineages (Seo et al. 2004; Cañestro et al. 2005; Cañestro and Postlethwait 2007; Holland 2007). Therefore, it seems plausible to consider that the 2R-WGD might have not been crucial for the origin of the neural crest cells, but the 2R-WGD might have been important for increasing the evolvability of the NC-GRN and the diversification of derivative structures.

Similar conclusions have been reached through studies of the gene regulatory network underlying placode and brain development. Analysis of placode-marking genes (e.g. Eye, Pitx, Six, and Pax) in ascidian and larvacean urochordates suggested that the last common olfactore ancestor already presented multiple placode derivatives, such as olfactory and adenohypophyseal. Additional and independent proliferation and loss of a variety of placodes probably occurred in both urochordate and vertebrate lineages (Bassham and Postlethwait 2005; Mazet et al. 2005), in some cases recruiting paralogs that had been independently duplicated in both urochordates and vertebrates (Bassham et al. 2008).

Despite the fact that non-vertebrate chordates have a simple brain lacking a midbrain and a midbrain–hindbrain organizer (MHB), most brain-making gene orthologs are present in non-vertebrate chordates, suggesting that vertebrate brain features were built on a foundation already present in the ancestral chordate probably facilitated by the new ohnologs created by the vertebrate 2R-WGD (reviewed in Holland 2009). Recent analysis of developmental genes in the ascidian brain revealed that the expression of Fgf8 can reorganize the expression of other brain genes and transform hindbrain structures into an expanded mesencephelon, recapitulating the organizing activity of the vertebrate MHB and therefore suggesting that the MHB was already present at least in the last common ancestor of olfactores (Imai et al. 2009). Analysis of urochordate genomes revealed that important genes (i.e. Gbx) for the positioning of the MHB have been lost in stem urochordates (Cañestro et al. 2005), as has the retinoic acid dependent anterior–posterior axial patterning of the central nervous system (Cañestro et al. 2006), making plausible the hypothesis that the absence of midbrain in urochordates is not due to a vertebrate innovation of a midbrain, but a simplification in urochordates of an ancestral tripartite brain structure (Cañestro and Postlethwait 2007; Cañestro et al. 2007). Evolutionary analysis of the origin of the complex Nova-regulated splice variants of the vertebrate brain genes revealed that many of these variants were already present in the last common olfactore ancestor (Irimia et al. 2011). It is possible that the 2R-WGD promoted the increase of the complexity of Nova-dependent splice variants in the vertebrate brain, although a simplification of this system during urochordate evolution cannot be discarded.

In conclusion, it is likely that the origin of vertebrate features such as neural crest cells, placodes, and a complex tripartite brain are not related to the 2R-WGD, but that these features were already present to some extent in stem non-vertebrate chordates (Fig. 16.1) (reviewed in Donoghue et al. 2008). However, it is likely that the subsequent evolution of these three features has been strongly influenced by the new ohnologs that originated after the 2R-WGD, due to processes of neofunctionalization, subfunction partitioning and subsequent refinement, recruitment of cis-regulatory elements driven by genome rearrangement and transposable element activity, inventions of novel miRNA families, and evolution of novel splice variants, which overall increased the complexity of duplicated developmental gene regulatory networks after the 2R-WGD. Future integrative analysis of comparative genomics, functional evo-devo, and examinations of gene regulatory networks in a wide variety of non-vertebrate chordates as well as basally divergent jawless vertebrates will help to narrow down the precise timing of the 2R-WGD and evaluate its actual impact on the origin and evolution of vertebrate features.

Probably the new ‘2R, or not 2R’ question (Hughes and Friedman 2003) is now to ascertain whether the origins of vertebrate innovations were, or were not, the consequence of the 2R-WGD, and to understand the mechanisms by which the 2R-WGD increased the evolvability of developmental gene regulatory networks that facilitated the diversification of complex vertebrate features.