|Home | About | Journals | Submit | Contact Us | Français|
Embryonic left-right (LR) patterning is a fascinating aspect of embryogenesis. The field currently faces important questions about the origin of LR asymmetry, the mechanisms by which consistent asymmetry is imposed on the scale of the whole embryo, and the degree of conservation of early phases of LR patterning among model systems. Recent progress on planar cell polarity and cellular asymmetry in a variety of tissues and species provides a new perspective on the early phases of LR patterning. Despite the huge diversity in body-plans over which consistent LR asymmetry is imposed, and the apparent divergence in molecular pathways that underlie laterality, the data reveal conservation of physiological modules among phyla and a basic scheme of cellular chirality amplified by a planar cell polarity-like pathway over large cell fields.
The bilaterally-symmetric external body plan of vertebrates and many invertebrates conceals consistent left-right (LR) asymmetry in the shape and placement of internal organs. The striking conservation in the direction of asymmetry (e.g., consistency of the side on which the heart is located) is a remarkable aspect of embryonic development in a world in which no macroscopic force distinguishes left from right. LR patterning involves many of the key themes that fascinate developmental biologists: evolutionary mapping of a morphogenetic problem and molecular mechanisms upon different body-plans, a patterning event that cuts across multiple scales of organization, and a linkage among epigenetic, biophysical, and transcriptionally-mediated mechanisms (Levin 2006, Speder et al 2007b). Furthermore, errors of laterality have a significant clinical impact as birth defects (Ramsdell 2005).
Establishing consistent asymmetry requires three logical steps (Fig. 1). In the first phase of LR patterning, a mechanism must orient the LR axis with respect to the other two axes (Brown & Wolpert 1990), making one side different from the other. Defects in this process result in loss of asymmetry (midline heart, polysplenia or asplenia, etc.). The orientation of the LR axis must occur reliably with respect to the anterior-posterior and dorso-ventral axes so that all individuals are asymmetric in a consistent direction. Defects in this step result in a mixed population of offspring with mirror-image asymmetries (R- and L-handed forms). Downstream of this initial symmetry-breaking event lie asymmetric gene expression cascades; many aspects of this phase appear to be conserved in vertebrates including the asymmetric expression of members of the Nodal and Lefty families of TGF-β molecules and the transcription factor Pitx2 (Burdine & Schier 2000). The information needs to be amplified and transmitted to multiple organ systems, and midline structures must keep L-sided signals from affecting the R side and vice-versa. The final steps of organogenesis comprise a range of differential migration, proliferation, tension, and adhesion, as individual primordia utilize molecular cues to carry out distinct morphogenesis programs on either side of the midline. Lack of coordination causes a loss of coherence among organs’ interpretation of LR information, resulting in heterotaxia (independently random placement of asymmetric structures among offspring).
It is now known that morphological asymmetry is determined by asymmetric gene expression on the L and R sides after gastrulation. However, despite exciting work on molecular mechanisms upstream of asymmetric transcription, the field still faces fundamental puzzles (Aw & Levin 2008, Tabin 2005). What is the first event that computes LR orientation? How are direction and position with respect to the midline communicated from putative LR organizers and imposed over distant cell fields? How conserved are the relevant molecular mechanisms throughout phyla? Here, we discuss several models based on novel recent findings (Aw et al 2008, Danilchik et al 2006, Takano et al 2007, Xu et al 2007).
The classic treatment of asymmetry was put forth prior to the avalanche of molecular data produced more recently (Brown & Wolpert 1990). This lucid analysis postulated a chiral intracellular component (such as a chemical structure occurring as only one enantiomer in living cells) that was tethered or oriented with respect to the other two axes. This “F-molecule” could then nucleate directed intracellular transport in one consistent direction along the LR axis. This model is broadly compatible with many physical mechanisms and the basic scheme has held up remarkably well for the last 18 years.
Left-right asymmetry is a fundamental property of the Universe, revealed in parity violations during quantum decay (Wu et al 1957). Recent data indicate that chirality is derived within individual cells and may likewise be universal. One of the most significant findings for the field of LR asymmetry was recently made in the investigation of polarity in single neutrophil-like, unciliated HL60 cells that can orient the LR axis in culture (Xu et al 2007). Eighty eight percent of HL60 cells extended pseudopods to the left of a line connecting the nucleus and centrosome. Remarkably, these cells recapitulate all of the important phenotypes observed after manipulation of embryonic LR patterning, exhibiting isomerism, heterotaxia, and situs inversus after alteration of Par6, dynein, and GSK3β signals, respectively.
What subcellular component is responsible for the crucial orientation event that defines “leftward”? One likely possibility is that the coordination of the 3 axes is performed by a cytoskeletal organizing center such as the centriole or basal body. The early Drosophila embryo uses cytoskeletal nucleating centers to set up major body axes (Steinhauer & Kalderon 2006). The microtubule organizing center (MTOC) is an ideal and versatile candidate for an ancient, highly-conserved mechanism from which asymmetry can be leveraged. The MTOC nucleates microtubule assembly (Kellogg et al 1994) and can position itself in the exact center of even irregularly-shaped cells (McNiven & Porter 1988), suggesting it could act as the center of geometric computation to integrate the 3D morphology of the whole cell. In ciliates, basal bodies have a LR asymmetry that is linked to the overall chirality of the cell (Beisson & Jerka-Dziadosz 1999, Bell et al 2008). Consistently, in human Bardet-Biedl patients, basal body dysfunction induces LR randomization (Ansley et al 2003). Recent work in a number of systems showed that cytoskeletal organization (Adams et al 2006, Qiu et al 2005, Thitamadee et al 2002) and the intracellular motors that move cargo molecules along those cytoskeletal tracks (Hozumi et al 2006, Nonaka et al 1998, Supp et al 1997) are all crucial for LR patterning. Indeed, in the cleaving frog egg, the cytoskeleton has a consistent bias that enables the asymmetric localization of physiological determinants of asymmetry (Aw et al 2008).
A recent study (Danilchik et al 2006) showed that the Xenopus egg cortex contains a maternally-derived cytoskeleton that has biased directionality. When the actin organization of the egg is disrupted pharmacologically, the animal pole rotates counter-clockwise with respect to the animal-vegetal axis, revealing the existence of consistent chirality in the egg. This suggests a model for how a cytoskeletal structure can derive a consistent direction along the LR axis from the dorso-ventral and anterior-posterior axes during fertilization and cleavage. The biased cytoskeleton, in an embryo with an inherent “North-South” axis (animal-vegetal, created by the maternal distribution of egg mRNAs and proteins) provides an “East-West” or counter-clockwise circumferential vector. Because consistent LR asymmetry (a true linear axis perpendicular to the other two) is set by the 2nd cleavage (Adams et al 2006), a mechanism is needed to convert the counter-clockwise cytoskeletal chirality of the egg into an orthogonal LR directionality in the embryo. The sperm entry point selects a unique place on the egg’s circumference at which the counter-clockwise structure of the actin cytoskeleton provides an orthogonal LR arrow positioned perpendicular to the other two axes (Fig. 2A–A″).
Most discussions of asymmetry assume the existence of the midline. Behind this term are concealed several issues we do not understand. During organogenesis and later in development, the midline is defined as an imaginary line running from head to toe, across which the internal organs are normally placed asymmetrically. Earlier in development, i.e. during the period of asymmetric gene expression, the midline acts a barrier to prevent leak-over from side-specific signals (Bisgrove et al 1999). At this stage, the midline is what differentiates between directional information that is identical in all cells (which way is left-ward?), and positional information (am I on the R or L side of the midline?), which is different on the left and right sides of the embryo. The latter is required for side-specific gene expression. The midline also has considerable biomedical relevance (Martinez-Frias 1995); embryos that lack intact midline structures have laterality defects (Izraeli et al 1999, Przemeck et al 2003).
The problem is the lack of a molecular definition of the midline early in development, i.e. prior to asymmetric Nodal signaling. Thus, the early midline is currently just an imaginary line drawn through the middle of an organizer (expression of Siamois in frog, or base of the primitive streak in chick for example, Fig. 2B). It is unknown how the boundary between the L and R compartments of an embryonic field is determined. Data in Xenopus have provided one of the only glimpses into the mechanistic, molecular definition of the L:R boundary. In the 16-cell frog embryo, the L and R ventral cells are gap-junctionally isolated (Guthrie 1984); this zone of isolation splits an otherwise gap junction-coupled field and demarcates the midline of the early embryo. Targeted disruption of the ventral isolation zone via misexpression of constitutively open gap junctions results in specific randomization of asymmetry (Levin & Mercola 1998), confirming functionally the significance of the midline established by lack of gap junctional communication between cells separated at the very first cleavage.
This mechanism relies on the fact that in Xenopus, the cleavage plane of the first large blastomeres normally coincides with the prospective midline of the whole embryo such that the left side of each cell also corresponds with the left side of the embryo (Fig. 2B′). What happens in embryos that establish the LR axis when there are thousands of cells present that do not span the whole embryo and when intracellular transport in individual cells cannot distribute components across the entire embryo? Pharmacological or molecular disruption of gap junctional communication in chick (where the midline is not apparent until there are thousands of cells) also randomizes asymmetric gene expression (Levin & Mercola 1999). However, it is not known whether the observed pattern of junctional isolation throughout the streak (Levin & Mercola 1999) is involved in the generation of a sharp midline within the streak field.
When does the midline really appear? It is thought that the radial symmetry of the avian blastoderm is broken by a streak-inducing center and that the midline, and thus LR asymmetry, is elaborated during gastrulation. While an axis can be experimentally induced at later stages in mice and birds, there is the possibility that the midline, and perhaps asymmetry, is set up in these organisms much earlier.
Gynandromorphs display both male and female characteristics due to non-disjunction of sex chromosomes during early cell divisions. Bilateral gynandromorphs provide a striking visible demarcation of the midline in organisms where the male and female are sexually dimorphic, typically because of coloring differences (Fig. 2C–C″). Gynandromorphism occurs in a wide variety of species including insects (most commonly reported in butterflies), crabs, chickens (reviewed in (Levin 2006)), and even human patients with X-linked diseases such as CHILD syndrome (Happle 2006). The boundary is often exactly down the middle, including in the brain (Agate et al 2003), and in roosters, is respected even by the comb on top of the head of the male half.
The sharp midline separation suggests that the first cell cleavage may produce L and R halves that inherit differential chiral information. Is this plausible in mammals? The main axes of rabbit embryos are known to be established prior to the appearance of the streak (Viebahn et al 1995) and this has been proposed for mouse development (Gardner 2001, Piotrowska & Zernicka-Goetz 2001). Even in humans, unilateral X-inactivation required to explain the harlequin pattern observed in CHILD syndrome (Happle 2002) requires that the midline already exist at very early stages. This would be consistent with the puzzling “mirroring” (opposite sidedness of unilateral traits) often observed in human monozygotic twins (reviewed in (Levin 1999)).
Ultimately, early pathways must initiate differential gene expression cascades on the L and R sides. There are at least two types of biophysical mechanisms functioning upstream of asymmetric transcription.
Elegant studies examining movement of cilia show that their beating generates a biased fluid flow (Kramer-Zucker et al 2005, Nonaka et al 2002, Schweickert et al 2007) that could result in asymmetric extracellular signals. While cilia are commonly viewed as an initiator of asymmetry, we have argued elsewhere that no data indicate that cilia initiate asymmetry; much of the genetic data are also consistent with intracellular transport roles for ciliary proteins (Levin 2003). We view the cytoskeleton-dependent events described above as more fundamental and broadly relevant to numerous species where asymmetry is established without cilia or long prior to the appearance of cilia (Levin & Palmer 2007, Speder et al 2007a). Additionally, ciliary function is not likely a general LR initiating step because their function is not upstream of some LR pathways such as asymmetric Notch signals (Krebs et al 2003, Raya et al 2003), and because human patients with primary ciliary diskinesia display normal handedness and brain laterality (McManus et al 2004). Ciliary beating may instead be a middle step that amplifies or converts earlier signals.
Further analysis of mutants such as the zebrafish Seahorse (Kishimoto et al 2008) is likely to shed light on this issue. Seahorse genetically interacts with the ciliary gene Inversin, is highly enriched in heavily ciliated tissues, and is required for normal asymmetry and for preventing kidney cyst formation. However, the LR randomization in Seahorse mutants occurs in the absence of detectable ciliary motility or structure defects. This illustrates the dissociation between associated kidney/laterality syndromes, usually thought to be due to ciliary defects, from ciliary motility per se and supports models of non-ciliary roles for “ciliary” proteins in LR patterning. Indeed, Seahorse contains nuclear import/export sequences; its precise function remains to be characterized.
Another epigenetic system functioning upstream of asymmetric gene expression is bioelectrical. The LR-differential function of specific ion transporters, and their resulting voltage and pH gradients, are required for normal asymmetry at early stages of sea urchin, Ciona, chick, frog, and zebrafish embryonic development (reviewed in (Levin 2006)). The details are clearest in Xenopus, where intracellular localization via a biased cytoskeleton during the first cleavages result in LR-asymmetric distribution of maternal ion transporter proteins. Just as in the mammalian gut (Kaufhold et al 2008), the module consists of the Kir4.1 and KCNQ1 channels and the H+/K+-ATPase exchanger (Aw et al 2008, Levin et al 2002). Together with the V-ATPase H+ pump (Adams et al 2006), this results in the right side of the early embryo being more strongly polarized than the left – a state that is absolutely required for normal subsequent asymmetric gene expression. In the chick, the proximal sensor is Notch, which is activated via a Ca++-dependent pathway (Raya et al 2004).
It is likely that LR information originates intracellularly. This may occur in all cells (if the orientation mechanism is extremely ancient and fundamental, as suggested by data from ciliates and HL60 cells) or perhaps only in a LR “organizer” (Hyatt & Yost 1998, Nascone & Mercola 1997). In either case, intracellular information must be amplified and coherently imposed across large cell fields; this process must interact with the conversion of directional to positional information along the LR axis.
One possibility is the creation of a morphogen gradient by ciliary flow (Tanaka et al 2005). Another is the electrophoretic movement of a LR determinant. In chick and frog, the neurotransmitter serotonin plays a key role in LR patterning upstream of asymmetric gene expression and long before the appearance of neurons and cilia (Fukumoto et al 2005). While a closely related pathway also functions in the LR patterning of plants (Pekker et al 2005), the details of serotonin’s movement in chick and other species are unclear. In Xenopus however, the system has been worked out in considerable detail (Levin et al 2006); maternal serotonin appears to accumulate in a right-ward gradient through a path of gap junction-coupled cells under the electrophoretic force of the asymmetric voltage gradient. Because the relevant parameters are now known quantitatively, a predictive mechanistic model has been formulated of this process (Esser et al 2006). By creating a voltage gradient, a subcellular asymmetric localization of ion transporters is transduced into a multicellular morphogen gradient that provides LR positional information across the entire embryonic field.
One other pathway now emerges as a fascinating candidate for a well-conserved role in amplifying LR pattern: planar cell polarity (PCP) - a powerful, elegant, and widely conserved module for imposing a coordinated orientation on a field of cells (Fig. 2D,D′). PCP is the coordinated polarization of cells in the plane of an epithelium, which occurs in the Drosophila wing bristles and the sensory hair cells of the vertebrate ear (Wang & Nathans 2007). PCP must solve the very same issues that confront LR asymmetry (Lewis & Davies 2002), and a logical schematic of PCP maps directly onto the F-molecule model of LR patterning. How is the internal asymmetry of the cell generated? How is the orientation of the axis of asymmetry (the polarity vector) of a given cell coupled to that of its neighbors? And, how is the orientation of a tissue controlled globally by influences from more distant structures, according to its location in the body as a whole? Through a process akin to domineering nonautonomy (Amonlirdviman et al 2005), LR information arising in a small group of cells can thus be amplified across large cell fields in a coordinated manner.
The PCP model requires that embryos at relevant stages propagate LR information through polarized epithelia. The early-cleavage frog embryo is an epithelium (Muller & Hausen 1995) and it was recently shown (Aw et al 2008) that the 4-cell embryo integrates the orthogonal apical-basal (animal-vegetal) and planar (DV and LR) polarities to orient intracellular transport tracks that are biased towards the right-ventral cell. The early chick blastoderm is also an epithelium that is polarized in the plane. The cells have a clear dorso-ventral (apical-basal) polarity, and reveal their knowledge of anterior-posterior polarity by their alignment during streak elongation (Wei & Mikawa 2000), a key process now known to be dependent on the PCP pathway (Voiculescu et al 2007). During stages at which the first known LR asymmetric signal – the left-sided depolarization – appears (Levin et al 2002), the chick blastoderm, like Drosophila, exhibits features dependent on core PCP proteins, including characteristic multicellular rosettes (Blankenship et al 2006, Wagstaff et al 2008), intercalation (Benazeraf & Pourquie 2008), and convergent extension, a process of coordinated, polarized movement of a field of cells (Lawson & Schoenwolf 2001).
Both PCP and LR asymmetry involve conserved, widespread systems for orienting a large-scale third axis in relation to two existing axes. This provides a ready framework to understand the way that subcellular chirality is imposed on a complete body axis and test models in which chiral molecules (Brown & Wolpert 1990) nucleate asymmetry in organisms that pattern the LR axis at later stages, within multicellular fields of different sizes. The origin of LR information in chick is mysterious because asymmetry appears when the blastoderm contains thousand of small cells, which unlike Xenopus blastomeres do not have the opportunity to utilize intracellular transport machinery to localize protein cargo to the L or R sides of the prospective embryonic midline. One of the most attractive features of PCP for understanding LR patterning throughout phyla is its scale invariance. By setting up alignment of polarization on the cellular level, PCP mechanisms allow embryos to achieve large-scale polarization from intracellular events.
We have argued for considerable conservation of basic LR mechanisms (Levin 2006) and suggest widespread reuse of a module consisting of intracellular chirality driven by cytoskeletal organizing centers and amplification of signals through subsequent physiological asymmetries. This proposal is controversial, with many in the field favoring independently-evolved mechanisms (Tabin 2005). A meta-analysis of molecular mechanisms that have been probed in LR patterning (Fig. 3 and (Levin & Palmer 2007)) suggests that while the details differ, many mechanisms are broadly conserved.
More importantly, what may be conserved are the higher-level modules performing functions such as orientation, amplification, and conversion of direction to position with respect to the midline. If the information processing by each of these modules remains intact, the molecular details of each (i.e. the specific proteins involved) can diverge, as can the timing of each step relative to other developmental events. For example, Xenopus uses ion currents to amplify asymmetry at the first few cleavages, while the chick waits until there are ~50K cells. Could this be an example of heterochrony, where different species conserve the same basic steps but change the timing of the LR pathway’s mechanisms to better match their architecture?
One of the issues facing the field is that some important and fascinating phenomena are not easy to study in the popular “genetic” model systems. Examples include bilateral gynandromorphs in birds and most insects, and opposite-sided hair whorls in monozygotic human twins. The lack of midline separation of male and female tissues in Drosophila gynandromorphs (unlike moths and butterflies), the random separation of male-female cells in the CHILD syndrome mouse model (unlike the strict midline demarcation in humans), and the lack of data on early, pre-streak physiological LR mechanisms in the highly unusual cylinder of the rodent embryo suggest that atypical features of some model species may lead us astray when looking for general mechanisms. Testing for conservation of LR mechanisms (e.g., gap junctional communication) is further complicated by the importance of genetic background: mice with conditional deletion of the gap junction protein Connexin43 exhibited disruption of neuronal migration ranging from no phenotype in C57B1/6J mice to a strong phenotype in 129SVEV background (Wiencken-Barger et al 2007).
The PCP hypothesis predicts a link between the planar polarity of epithelia and LR patterning. In this model, the known role of planar polarity in determining hair whorl patterns (Guo et al 2004, Wang et al 2006) explains the mysterious relationship between the direction of hair whorls on the scalp and laterality of the brain and body in monozygotic twins (Jansen et al 2007, Levin 1999, Weber et al 2006). Interestingly, genetic deletion of the PCP gene Frizzled6 results in ectopic hair whorls where straight hair patterns normally occur (Guo et al 2004). Future work will reveal whether PCP is necessary to convert a rotary pattern into a straight one (a multi-cellular version of Fig. 2A′) and whether PCP is necessary to mask the chirality inherent in structures, the way that the retinoic acid pathway does in somite patterning (Vermot & Pourquie 2005).
Another major theme for future studies of LR patterning is likely to be epigenetics. For example, the asymmetry of the asexual blastozooid of ascidians appears to not be controlled by the genome (described in (Ishii et al 1994)). Additionally, both ciliary and physiological modules’ signaling does not require changes in transcription. It is crucial to understand the linkage of these pathways to canonical response cascades. We predict that interactions will be detected between canonical planar polarity proteins and H+ pumps and K+ channels, in both vertebrate and invertebrate models of LR asymmetry; testing this hypothesis is an important future direction.
Models in which the chiral computation occurs soon after fertilization imply that twins produced by blastomere separation will exhibit reversals of asymmetry, whereas models where LR asymmetry is initiated by cilia in later embryogenesis suggest that early-split twins should both be normal since the cilia will beat normally in each twin at gastrulation. The conservation of chirality within a pair of cells derived from the same mother cell is beautifully illustrated by the work of Albrecht-Buehler, who showed that after division, daughters of mammalian cells in culture have mirror-image cytoskeletal organization and subsequent migration trajectories (Albrecht-Buehler 1977). The medical literature contains numerous and as yet unexplained examples of such mirroring (see (Levin 1999, Levin 2001)), including opposite hair-whorl direction and unilateral facial defects in monozygotic twins. When separated at the 2-cell stage, Newt embryos exhibit 89% incidence of organ laterality reversal in one of the twins (Takano et al 2007). Thus, the twinning data do not support ciliary models but are a straightforward prediction of asymmetry generation by very early, intracellular chirality mechanisms. Because some bovine strains have hair whorls, future studies using in vitro fertilization could be used to generate monozygotic twins with known timing of cell separation, which can then be examined for hair whorl direction.
Taken together, we believe the data argue for a view of LR patterning as an instance of universal, ancient cellular polarity that originates inside the cell, magnified by functionally-conserved physiological and planar coordination modules. The establishment of new model systems will advance major open directions for the field, including the nature of the midline, the molecular mechanisms underlying organ concordance, and subtle asymmetries such as handedness, hair whorl chirality, and brain laterality. Quantitative modeling at multiple levels of organization, together with molecular investigation of LR signaling components, is likely to reveal fascinating new aspects of developmental and cell biology.
We thank members of the Levin lab and the LR community as a whole for numerous useful discussions, Arthur P. Levin for the photo in Fig. 1C, and C. Schaeberle for assistance with the sketches in Fig. 3. This work was supported by NIH grants R01-GM077425 and R01-GM067227, and American Heart Association Established Investigator Grant #0740088N.