Right across the tree of life: The evolution of left–right asymmetry in the Bilateria

Tunicates

The Tunicata are the invertebrate lineage most closely related to the vertebrates (Fig. 2). Tunicate embryos develop into a tadpole larva with dorsal, hollow neural tube and underlying notochord. Larvae typically show asymmetry in the sensory vesicle sited in the anterior brain (Boorman and Shimeld, 2002a). The larvae of ascidians, which comprise the majority of tunicates, metamorphose into a sessile filter feeding adult, which also shows asymmetry, most obviously in the positioning and coiling of the gut and other internal organs (Boorman and Shimeld, 2002a).

The regulation of asymmetry has been primarily studied in two ascidian species, Ciona intestinalis and Halocynthia roretzi. These species belong to widely divergent ascidian lineages; however, their early development is remarkably similar with each showing essentially the same stereotypical pattern of cleavage and cell lineage (Lemaire, 2009). In both species, the nodal–Pitx signaling system is deployed on the left-hand side of the body, as it is in vertebrates (Boorman and Shimeld, 2002b; Morokuma et al., 2002). However, the primary tissue differs, with both H. roretzi and C. intestinalis genes being broadly expressed in the left-side ectoderm at the neurula and tailbud stages, as opposed to around the node and in lateral plate mesoderm in vertebrates (Collignon et al., 1996; Lowe et al., 1996). One possible explanation for this difference is that tunicates are thought to have lost the node-based body axis patterning used by vertebrates (and amphioxus; see below), replacing it with lineage and local signaling-based mechanisms (Kourakis and Smith, 2005; Lemaire, 2009). Conservation of the process extends to the regulation of Pitx by nodal (Morokuma et al., 2002; Yoshida and Saiga, 2008) and to the deployment of lefty: although the precise regulatory role of lefty is yet to be dissected, overexpression experiments produce similar phenotypes to treatment with the nodal receptor antagonist SB431542, suggesting that it also functions to inhibit nodal signaling (Mita and Fujiwara, 2007).

A small number of studies have also begun to investigate the mechanism underlying how left-sided expression of these genes is initially triggered, although with sometimes conflicting results. Shimeld and Levin (2006) investigated whether manipulation of ion channels or pumps by pharmacology affected asymmetry in C. intestinalis, finding that compounds which in vertebrates interfered with H⁺K⁺ ATPase activity could affect asymmetry as revealed by ectopic right-sided expression of Pitx in treated embryos. With respect to the potential for cilia to be involved in symmetry-breaking in tunicates, classical descriptions of ascidian development have failed to identify cilia prior to the larval stage, when asymmetry is already well established (for review, see Satoh, 1994). However, this question has been recently revisited using fixation protocols designed to preserve primary cilia (Thompson et al., 2012). This study identified small (1–2 µm long) transient cilia on all ectoderm cells (other than those at the midline) at the late neurula and early tailbud stages, coincidental with the activation by these cells of nodal expression. The cilia are posteriorly localized on each cell; however, analysis of their structure by TEM found collapsed axonemes and a lack of dynein arms, consistent with primary cilia structure but inconsistent with motility. This led the authors to postulate that the cilia were not involved in a nodal flow-like mechanism in ascidians but were more likely to be sensory in nature. However, a parallel study in H. roretzi showed that during this period of development, embryos rotated gently inside their chorions and always came to rest on their left side (Nishide et al., 2012). Furthermore, they experimentally demonstrated that contact between the side of the embryo and chorion wall was sufficient to induce nodal gene expression. These authors also presented evidence that other ascidian embryos rotated inside their chorions, including those of C. intestinalis, although the degree of rotation varied and embryos did not always end up lying in the same orientation. Cilia, although not proven to be involved, are obviously likely candidates for driving such rotation.

These conflicting data have yet to be reconciled. One possibility is that ascidian embryos develop different types of cilia at different places and/or times such that TEM analysis missed those responsible for embryo motility. Alternatively other forces may drive embryo rotation, for example, the C. intestinalis chorion is lined internally with a dense layer of maternally supplied test cells that are only a few micrometers away from the embryo, in contrast to the large space inside a H. roretzi chorion. Nonmotile cilia might also be involved in asymmetry via a sensory function, as contact between embryo and chorion substratum appears necessary for symmetry breaking in H. roretzi. Mechanistic diversity between ascidians is also possible; as discussed above, although morphologically similar, C. intestinalis and H. roretzi are genetically very divergent. Furthermore, removal of the chorion has different effects on the two species, leading to bilateral expression of left-sided genes in C. intestinalis (Shimeld and Levin, 2006; Yoshida and Saiga, 2008), but loss of left-sided expression of nodal in H. roretzi (Nishide et al., 2012). This suggests that if an embryo–substrate contact mechanism is deployed in these species to break symmetry, it is working in opposite ways in the two species (repressing left-sidedness in C. intestinalis, activating it in H. roretzi).

In summary, although it is clear that ascidians share left-sided nodal and Pitx expression with vertebrates and that the regulation of Pitx by nodal is also conserved, it is less clear whether upstream mechanisms of symmetry breaking are also shared ancestrally. An intriguing possibility is that a node cilia-based symmetry breaking mechanism was deployed by the common ancestor and that ascidians have adapted this to the use of cilia (be they sensory, motile, or both) over the whole ectoderm as a corollary of losing the node. Understanding symmetry breaking and nodal regulation in amphioxus would help resolve this.

Cephalochordates

Amphioxus and allies form the Cephalochordata, the earliest diverging living chordate lineage (Fig. 1). The amphioxus larva is one of the most asymmetric bilaterians described. It possesses a canonical chordate body plan, with dorsal neural tube, notochord, and somite muscle blocks (Conklin, 1932). However, the somites are asymmetrically aligned, the mouth opens on the left side of the head, the gill slits extend up to the right side of the head, and the anterior head cavities undergo very different morphogenesis on left and right sides (reviewed by Boorman and Shimeld, 2002a). Some symmetry is restored at morphogenesis; however, asymmetry clearly persists in brain, gut, and associated organs.

Both nodal and Pitx are left sided in amphioxus (Boorman and Shimeld, 2002b; Yasui et al., 2000; Yu et al., 2002). Their expression more closely resembles that of vertebrates than ascidians, despite the latter two being more closely related; for example, nodal expression initiates symmetrically before resolving into asymmetry, and both genes are mesendodermally expressed (Yasui et al., 2000; Yu et al., 2002). This may be a corollary of organizer-based patterning in amphioxus and of the involvement of nodal in mesoderm induction (Onai et al., 2010; Yasui et al., 2000).

To date, the mechanism that breaks symmetry in amphioxus and leads to asymmetric nodal expression is unknown. Early amphioxus embryos are heavily ciliated and rotate inside their chorion prior to hatching, powered by these cilia (Conklin, 1932). Gastrulation internalizes ciliated cells such that the archenteron of the early neurula is lined with cells with long cilia, although whether these remain actively beating has not been described. Although this raises the possibility that cilia flow could account for symmetry breaking in amphioxus, hard evidence is lacking, and further experiments are needed to investigate this, for example, direct imaging of cilia activity, measurement of fluid flow within the amphioxus archenteron, and/or inhibition of cilia activity.

Echinoderms and Hemichordates

Echinoderms and hemichordates form a monophyletic grouping termed the Ambulacraria (Fig. 1). Most hemichordates are long, thin, and worm-shaped animals, although one hemichordate lineage, the pterobranchs, are tube dwellers (Sato and Holland, 2008). At least some adult hemichordates have subtle asymmetries, for example, in the location of an anterior pore (Kaul-Strehlow and Stach, 2013). Whether pterobranch hemichordates are asymmetric is less clear; classical descriptions of these animals as having an asymmetric gonad have recently been shown to be antisymmetric rather than directionally asymmetric (Sato and Holland, 2008). As of yet, the regulation of hemichordate asymmetry has not been described.

Echinoderms are a diverse Phylum with a variety of adult body plans including sea urchins, starfish, brittle stars, sea cucumbers, and sea lilies, all united by underlying pentaradial symmetry in the adult form. However, larvae are bilaterally symmetrical, and in sea urchins, the adult forms from a rudiment that develops on the left side of the larva (Fox et al., 2004). The positioning of this rudiment is under the control of nodal–Pitx signaling across the LR axis, demonstrating mechanistic conservation with chordates. The regulatory interactions of these genes are quite well dissected but will not be discussed further here; instead, the reader is referred to these recent studies (Duboc and Lepage, 2008; Luo and Su, 2012; Warner et al., 2012). In contrast to chordates, however, nodal and Pitx are deployed on the right-hand side of the embryo. The implication of this difference will be discussed below in detail, when we consider lophotrochozoan asymmetries.

The Cytoskeleton and LR Asymmetry in Molluscs

Mollusc embryos primitively undergo spiral cleavage (Fig. 2). This is apparent from the four- to eight-cell stage, when at the third cleavage, the tier of four animal micromeres is slightly offset from the four vegetal macromeres, creating a spiral form that either twists sinistrally or dextrally. Successive cleavages follow this offset pattern through to gastrulation. While in larval development dextrally cleaving embryos show dextral shell coiling, sinistrally cleaving embryos show sinistral shell coiling. Spiral cleavage is shared by many other lophotrochozoan Phyla and may be ancestral for the Lophotrochozoa, although this is somewhat contentious (Giribet, 2008; Hejnol, 2010). Although not described outside the Lophotrochozoa, spiral cleavage may hold clues as to the ways in which asymmetry is established in other clades.

There is excellent experimental evidence that the angle of spindle orientation during the third cleavage is both coincident with and determines later asymmetries (Freeman and Lundelius, 1982; Shibazaki et al., 2004). A particularly elegant study by Kuroda et al. (2009) showed that the direction of spiral rotation during third cleavage is directly connected to the orientation of shell coiling in later development. This was shown by physically manipulating blastomeres in dextral and sinistral snails, changing the arrangement of spindles and hence forcing spiral cleavage in the opposite orientation. This resulted in the concordant alteration of chirality in later development, with shell coiling corresponding to the direction of blastomere manipulation, regardless of the genetic background of the embryo. Subsequent generations reverted back to their genetically inherited direction of cleavage and coiling.

Early genetic studies indicated that the chirality of at least some mollusc embryos depends on a maternally inherited nuclear locus, termed sinistral (Asami et al., 2008; Boycott et al., 1930; Diver et al., 1925; Sturtevant, 1923). An experimental embryological study has further connected this genetic basis to a molecular mechanism, with the transfer of cytoplasm of a dextral egg into a sinistral egg causing subsequent dextralization (Freeman and Lundelius, 1982). This shows that there is a specific factor in the cytoplasm that communicates dextrality; however, the nature of this factor is still unknown. The pond snails L. stagnalis and Lymnaea peregra have historically been used for such studies, as they can be naturally found and cultured in both chiral forms (for review, see Schilthuizen and Davison, 2005). Mapping of the locus responsible has been undertaken (Liu et al., 2013), and although the nature of the gene or genes is not yet clear, this can only be a matter of time now. A likely scenario is that the locus will encode a maternally supplied protein or proteins that organize dextral orientation of the spindle at the third cleavage.

Interestingly (and somewhat counterintuitively), sinistral and dextral snails, although mirror images at the gross level of adult morphology and shell coiling, are not mirror images of each other at the cellular level during cleavage. Instead, they show morphologically and temporally distinct mechanisms of orientated cell division involving the cytoskeleton, at least for the small number of species studied to date (Shibazaki et al., 2004). Future studies are needed to better understand how these mechanisms operate, how phylogenetically widespread they are, and how they could be related to the maternally inherited sinistral locus.

These studies, while promising clear insight in the near future into the nature of sinistral versus dextral development in molluscs, do not yet tell us how symmetry is initially broken in mollusc embryos. It is worth noting that there is evidence for a cytoskeletal role in establishing asymmetry across other axes in the development of other mollusc species. Subcellular- and blastomere-specific RNA localization has been noted in the eggs and early cleavage stages of the marine gastropod mollusc Ilyanassa (Kingsley et al., 2007; Lambert and Nagy, 2002; Rabinowitz and Lambert, 2010), with maternally deposited RNA playing a role in specifying the DV axis via formation of the D-quadrant organizer, a specific vegetal cell that directs subsequent DV development (Lambert, 2010). Whether this extends to LR asymmetry of maternally deposited mRNA is not clear; however, we speculate that somewhere between the first cell cleavage and the specification of the D-quadrant organizer, asymmetric cues could be communicated to specify the LR axis. Asymmetric RNA localization has also been identified in the gastropod Crepidula fornicata (Henry et al., 2010) and would benefit from studies in other mollusc lineages and other lophotrochozoan Phyla to understand whether this is an ancestrally conserved mechanism or one that was convergently evolved with a shared cleavage pattern.

Other Lophotrochozoans

The Lophotrochozoa remain under-investigated in the developmental literature, and outside the mollusc examples discussed above, there have been no studies indicating that similar mechanisms for establishing asymmetry are at work in other Phyla within this clade. Spiral cleavage, however, is widespread, being found in at least annelids, platyhelminthes, nemerteans, entoprocts, molluscs, and phoronids (Lambert, 2010; Merkel et al., 2012; Pennerstorfer and Scholtz, 2012). This suggests that underlying mechanisms of symmetry breaking and cleavage orientation may be equally as ancient, although this idea needs to be treated with some caution as there are some differences between molluscs and other spirally cleaving embryos; for example, although sinistral mollusc strains and species are naturally (if rather rarely) found, the majority of other lineages seem to lack such sinistral forms. Furthermore, there is still the possibility that despite a conserved method of cleavage, similar mechanisms of asymmetry establishment may have evolved convergently in spiraling cleaving animals. Recent advances in genome sequencing show that the nodal gene is broadly conserved in lophotrochozoans (Kenny et al., manuscript submitted), although whether it is involved in LR asymmetry remains to be determined. Given the modifications of spiralian cleavage that can be found within this Superphylum, from equal cleavage to unequal cleavage and polar body formation, this provides interesting material to help us understand the relationship between cleavage patterns and molecular mechanisms in the establishment of LR asymmetry.

Summary of the Lophotrochozoa: An Emerging Picture of Cytoskeletal Architecture Linking to Asymmetric Nodal Expression

In summary, investigation into genetic, molecular, and cellular mechanisms underlying LR asymmetry in the Lophotrochozoa indicate that one aspect of this process—the role of the Nodal pathway in directing asymmetric morphogenesis—is conserved with deuterostomes. This would make it ancestral for these two lineages, and hence a character of the bilaterian common ancestor (if we exclude the acoelomorphs; Fig. 1). Axis establishment in the Lophotrochozoa is closely linked to the chirality of early cell cleavage, and although the exact mechanism for establishing polarity remains unclear, this is likely via cytoskeletal organization, certainly through spindle orientation, and possibly through asymmetric RNA localization (although direct evidence here is lacking). This model makes some sense when we consider deuterostomes and ecdysozoans (as discussed below). However, it should be recognized that it is effectively the tip of the iceberg drawn from a small number of studies in a few species primarily within the Gastropoda, that is, just one Class of one Phylum, the Mollusca. We still do not know whether these mechanisms are conserved in other molluscs, let alone other lophotrochozoan Phyla.

A further consideration that influences our understanding of the establishment of LR asymmetry in these taxa is that spiral cleavage is visible at the third cleavage of spiralian embryos. This is before DV axis specification, which occurs after the third micromere quartet forms and is mediated by the D-quadrant macromere (for review, see Lambert, 2010). Thus, early spirality correlates with LR asymmetry but precedes DV specification. As left and right sides cannot in theory be defined until both AP and DV are established, spiral cleavage does not break LR symmetry per se, although may provide underlying chirality that can be “read” into LR asymmetry after the DV axis is set. Further studies are needed that probe the interaction between the spiral third cleavage, the D-quadrant macromere, and LR asymmetry. Much remains to be determined, and given the diversity of the Lophotrochozoa, additional surprises are possible.

Cytoskeletal Cues in Ecdysozoan LR Asymmetry

Asymmetric cell divisions play a key role in differentially segregating cellular determinants in a wide variety of species (e.g., Betschinger and Knoblich, 2004) and have the potential to initiate the development of morphological asymmetries, although hard evidence linking such cell level processes directly to organismal LR asymmetry is generally lacking. In C. elegans, early cell divisions are asymmetric, and pioneering work by Wood (1991) showed that physical reversal of cleavage asymmetry by micromanipulation led to reversal of subsequent LR asymmetry of the cell lineage. This experiment is analogous to that performed by Kuroda et al. (2009) as described above, in which lophotrochozoan spiral cleavage was physically reversed with similar results. In C. elegans a chiral actin cytoskeleton is present in the egg (Pohl and Bao, 2010). The daughter cells that result from this polarity have markedly different cytoplasmic constitution and size. It is hypothesized in this species that asymmetries may be formed by differential forces in spindle elongation due to chiral cortical and cytoskeletal components, which would generate asymmetric cell divisions (Pohl, 2011; Pohl and Bao, 2010). This hypothesis needs to be tested in other Ecdysozoa with asymmetric cell divisions in the early embryo, such as the tardigrades, to understand whether this could be an ancestrally shared mechanism within the Ecdysozoa. A critical question remains as to what subsequent mechanism then generates morphological asymmetry, although one recent study has shown that in C. elegans, this may in part be based on the regulation of cortical contractility and cleavage furrow formation (Pohl and Bao, 2010).

Furthermore, it is known that at the third cell cleavage, spindle orientation by a G protein is crucial for the correct establishment of LR asymmetry (Bergmann et al., 2003). These G proteins have also been shown to have a role in the alignment of spindles in D. melanogaster, by binding to atypical protein kinase C and par proteins (reviewed by Ahringer, 2003). The par genes, originally identified in C. elegans but now known to be broadly conserved, are known to disrupt this asymmetric division process when mutated, affecting spindle localization (reviewed by Nance and Zallen, 2011). Par proteins are themselves asymmetrically distributed by the action of actin and myosin in the cell, which is generally said to follow from the organization of the cytoskeleton around the point of sperm entry (Munro et al., 2004), and have also been identified in an annelid (as suggested by Weisblat, 2007) and a mollusc, with at least one mollusc Par protein able to interact with microtubules and localizing with spindles during early cleavage (Hozumi et al., 2006).

In concert with cytoskeletal components, molecular motors could be the proximate cause of LR asymmetry by asymmetric localization of cargoes such as mRNA. A good example is the type ID unconventional Myosin (MyoID), which has been shown to cause situs inversus when mutated in D. melanogaster (Hozumi et al., 2006; Spéder et al., 2006). Such molecular motors move along F-actin in the cytoskeleton, and any asymmetry in the actin cytoskeleton could then be translated into transcriptional asymmetry. MyoID can also interact with adherens junctions via β-catenin (Petzoldt et al., 2012; Spéder et al., 2006), and these junctions could also act as mediators of LR asymmetry signals. Adherens junctions have been shown to be involved with LR asymmetry in concert with cadherins in D. melanogaster (Petzoldt et al., 2012) and the chick (Burdine and Caspary, 2013). An ancestrally shared role for adherens junctions could also explain the links seen between the Planar Cell Polarity (PCP) pathway (reviewed by Aw and Levin, 2009), inversins (Petzoldt et al., 2012; Watanabe et al., 2003), and LR asymmetry in a diverse phylogenetic grouping of species. However, this should remain a tentative conclusion, as it is easy to imagine that convergently evolved mechanisms of asymmetry involving epithelia would independently co-opt the adherens junction–cadherin machinery typical of such cells.

Conserved Cytoskeletal Mechanisms in Ecdysozoans?

Chirality of the actin microfilament cytoskeleton could be the root of all downstream differences in directionality, with additional cytoskeletal components acting on this, effectively representing the hypothetical “F” molecule, proposed by Brown and Wolpert (1990) as an asymmetric molecule that could integrate AP and DV polarity to specify left from right. Both C. elegans and D. melanogaster seem to use such a system. However, extrapolating this to conclude conservation (and hence assuming it is ancestral for other ecdysozoans) has caveats. First, the timing of action is quite different in C. elegans and D. melanogaster, occurring at early cleavage in the former and much later in development in the latter. This may be a consequence of the unusual mode of early development of D. melanogaster, in which both AP and DV axes are maternally defined, a derived condition for a subgroup of the insects. Without study of other arthropods and other ecdysozoan Phyla, we cannot test this. Second, the actin cytoskeleton is such a fundamental component of eukaryotic cells that convergent evolution is a clear possibility. Identification of additional similarities in underlying mechanisms would help to support the case for conservation rather than convergence.

A Conserved Role for Other Mechanisms Between Ecdysozoa and Other Bilateria?

Some of the genes implicated in modulation of asymmetrical signals in deuterostomes have also been observed playing a similar role in ecdysozoans. Notch signaling and calcium concentration, for example, play a role in establishing asymmetry in vertebrate embryos (Lopes et al., 2010; Raya et al., 2003, 2004). The specification of two neurons in C. elegans—ASE left and ASE right—are driven by the bilaterally asymmetric activation of T-box transcription factors under the control of calcium levels and the Notch pathway (Bertrand et al., 2011; Poole and Hobert, 2006; Schumacher et al., 2012). As the areas where Notch is expressed in vertebrates and C. elegans have no obvious comparison, this is likely to represent co-option of widely used signaling pathways, rather than an ancestrally conserved mechanism. micro-RNAs (miRNAs) have also been shown to be involved in the establishment of asymmetry in C. elegans (Johnston and Hobert, 2003); however, although miRNAs have also been shown to regulate asymmetry via the regulation of Nodal in a range of contexts in vertebrates, there is currently no evidence for homology of this process between the Ecdysozoa and vertebrates (Barroso del Jesus et al., 2011; Choi et al., 2007; Martello et al., 2007).