«Two Molecular Models of Initial Left-Right Asymmetry Generation Michael Levin1, Ph.D. and Nanette Nascone, M.S. Cell Biology dept. Harvard Medical ...»
Two Molecular Models of Initial Left-Right Asymmetry Generation
Michael Levin1, Ph.D. and Nanette Nascone, M.S.
Cell Biology dept.
Harvard Medical School
200 Longwood Ave.
Boston, MA 02115
1 Author for correspondence:
fax: (617) 432-7758
keywords: embryo, morphogenesis, axis, left-right, asymmetry, connexin-43,
Left-Right (LR) asymmetry is a fascinating problem in embryonic
morphogenesis. Recently, a pathway of genes has been identified which is involved in LR patterning in vertebrates (1,2). Although, this work characterizes the interactions of several asymmetrically-expressed genes, it is still entirely unclear how such asymmetric expression is set up in the first place. There are two promising molecular candidates which may play a role in such a process: the motor protein dynein, and the gap junction protein connexin-43 (Cx43). We present two models, significantly supported by previous findings, which hypothesize that (1) dynein asymmetrically localizes LR determinants in individual cells to establish cell-autonomous LR biasing, and (2) asymmetric activity of Cx43 gap junctions within key cells sets up electric potentials in multi-cellular fields, thus establishing large-scale LR asymmetry.
Introduction Symmetry in morphogenesis Animal body plans occur in a wide variety of symmetries: spherical (volvox), radial (starfish), chiral (snails, ciliates), bilateral (drosophila) and pseudo-bilateral (man). Most vertebrates have a generally bilaterally-symmetrical body plan, but this symmetry is broken further into a pseudo-symmetry by the consistently asymmetric placement of various internal organs such as the heart, liver, spleen, and gut, or an asymmetric development of paired organs (such as brain hemispheres or lungs).
Symmetries are often broken in development. For example, the radial symmetry of the early chick blastoderm is broken into a bilateral symmetry by the appearance of Köhler’s sickle and then the primitive streak (3). This is further broken into a pseudo-symmetry by the right-sided looping of the heart tube. In contrast, the early sea-urchin larva has bilateral (and then, pseudo-bilateral) symmetry. The adult, however, has a five-fold radial symmetry. Such axial patterning is the most fundamental process in embryogenesis because it lays a foundation and provides a context for all subsequent morphogenetic events.
Left-Right Asymmetry Asymmetry along the left-right (LR) axis (defined as an invariant, among normal individuals, difference between the left and right sides of an animal’s morphology) is fundamentally different from asymmetries in the other two axes. First, there is no feature of the macroscopic world which differentiates right from left.
While gravity is a ubiquitous feature of the world which can be used to define the dorso-ventral axis, and any chosen direction of motion automatically picks out an anterior end (since that is the end which is best used for sensory and processing organs), there is no independent way to pick out the left (or right) direction.
Second, all normal members of a given species are asymmetrical in the same direction. However, animals with complete mirror reversal of internal organs are otherwise phenotypically unimpaired (4,5). Thus, while it is possible to come up with plausible evolutionary reasons for why organisms might be asymmetric in the first place (optimal packing of viscera, etc.), there is no obvious reason for why they should all be asymmetric to the same direction. It is, after all, easier to imagine a developmental mechanism for generating bilateral asymmetry (such as positive-feedback and amplification of stochastic biochemical differences) than for biasing it to a given direction. The left-right axis is thus unique, and especially interesting, among the three axes.
Many kinds of situs anomalies have been reported in the human teratology literature, associated with such syndromes as Kartagener’s and Ivemark’s (5).
These include dextrocardia, situs inversus (a complete mirror-image reversal of the sidedness of asymmetrically positioned organs and asymmetric paired organs), heterotaxia (where each organ makes an independent decision as to its situs), and right or left isomerism (where the organism is completely symmetrical, leading to polysplenia or asplenia). Of these, only the complete (and rare) situs inversus is not associated with physiological difficulties. The rest, especially heterotaxia, often result in serious health problems for the patient. Laterality defects can arise in a single individual but are especially associated with twinning (4, 6, 7). These syndromes are paralleled to various degrees by mouse mutants such as: iv (8) which results in roughly 50% of the offspring being phenotypically situs inversus, and inv (9) which have 100% of the offspring showing mirror image inversions of the internal organs.
The molecular mechanisms underlying antero-posterior and dorso-ventral asymmetry have been studied in detail (10). However, the basis for LR asymmetry is much less well understood. Neville (11) presents an extensive and fascinating survey of various asymmetries, including the well-known asymmetric organs such as the heart, as well as flatfish which consistently settle on and undergo eye migration to one side, and even a species of parasite (the arthropod Bopyrus) which lives only on one side of prawn and shrimp. There has been little information, however, shedding light on the mechanisms which determine the sidedness of such asymmetries. Previously, information on the molecular basis of LR asymmetry centered around three lines of inquiry: the genetics of chirality in snails, a list of drugs which cause alterations in LR patterning, and several mammalian mutants which have phenotypes associated with LR asymmetry. Recently, a pathway of genes has been described which are asymmetrically expressed in the chick embryo and control the situs of the heart and other organs (1, 12).
The initial steps of LR determination remain unknown
LR patterning can be conceptually divided into three phases: (1) cell(s) in the very early embryo must ascertain their own right vs. left sides, presumably by a model like that involving a tethered chiral molecule (13), and (through lineage relationships, migration, and cell-cell inductive interactions) this cell-autonomous LR information is converted into asymmetrical multi-cellular domains of expression; (2) these asymmetrically-expressed genes regulate each other in sequential (and perhaps branched) pathways to establish and maintain asymmetric gene expression domains; and finally, (3) the various organ primordia read this information and determine their situs.
With the identification of a cascade of asymmetrically expressed genes which regulate each others’ expression (1), a significant part of phase 2 has been uncovered. Likewise, it has become clear that this pathway controls the LR patterning of many aspects of laterality (12), and experiments are currently under-way to determine the mechanisms by which organs such as the heart respond to this information (Sylvia Pagan, personal communication). Thus, significant progress is on the horizon for phase 3. However, what is conspicuously missing are clues to the most interesting part of this problem: how the LR axis is oriented with respect to the AP and DV axes in the first place, and how this orientation at the single-cell level is converted into asymmetry on the scale of the whole embryonic field.
The existence of isomerisms (where the body is symmetrical, consisting of either two normally-left sides or conversely, two right sides) and randomized asymmetry (i.e., 50% incidence of complete situs inversus) as two distinct genetic conditions suggests that normal LR asymmetry is accomplished in two dissociable steps: a random asymmetry is generated, which is then biased in the correct direction with respect to the other two axes (13). The pathway identified by Levin et al. most likely directs the second step, since right-sided misexpression of leftdetermining genes such as Shh or nodal results in heterotaxia (1,12), not true isomerism. Thus, there is still no molecular data on how (random) asymmetry is generated in the first place.
We would like to sketch out some ideas regarding two of the only promising molecular candidates for the primary steps in LR patterning, dynein and connexin-43. While the models described in this paper are quite speculative, they illustrate the types of mechanisms which are very likely to play a role in these early events. The dynein model is designed to show how a chiral molecule could differentiate L from R within a single cell. The Cx43 model shows how, once a cell becomes LR asymmetric, this information could become transduced into multi-cellular domains of asymmetric gene expression. It should be noted that the two models are independent of each other and describe complementary phases in LR patterning.
A Specific chiral molecule model Microtubule motors The first model, based on the ideas of Brown and Wolpert (13), hypothesizes a cytoskeletal component, such as a centriole, which is chiral. It is oriented with respect to the AP and DV axes of the egg by means of other cytoskeletal filaments, and serves as a nucleation center for filaments or microtubules which run along the LR axis. Consistent with this model, tubulin was identified as one of the proteins modified in iv homozygotes relative to w.t. mice (14). The head-tail attractive feature of microtubule assembly (15) ensures that the chiral nature of the nucleating center is passed on as a directionality of the LR tracks. Interestingly, the mouse egg has no centriole (one forms anew after several cell divisions), so that defects in the origin of chirality would show up as zygotic (as in the mouse LR mutants, such as iv); in contrast, the maternal mode of inheritance of chirality in snails (16) may be explained by the fact that the snail egg’s cytoskeletal components are formed by the mother.
The next step results in a microtubule motor, such as dynein, riding the LR tracks carrying mRNA or protein determinants, which become transported to one side of the cell. These determinants could become localized with cell division (which is possible in molluscs or even frogs), or this process could happen anew in each cell during various phases (which is most likely in the chick and mouse) followed by the kind of process discussed below for generating domains of LR gene expression.
Evidence for a Dynein model
An excellent candidate for such a mechanism is dynein (17), a motor protein which serves to actively translocate sub-cellular cargo (18-20). There appear to be at least 13 axonemal (used in cilia and flagella), and 2 cytoplasmic (presumably involved in axonal transport, mitosis, etc.) dynein genes (reviewed in (21)).
Despite the involvement of dynein proteins in many disparate events (such as ciliary function, vesicle transport in axons, mitosis, etc.), it is clear that certain dynein genes have very specific expression. For example, Dhc64c is expressed only in ovaries, testes, and very early embryos in Drosophila; furthermore Dhc64c is asymmetrically (though not LR) localized in the Drosophila oocyte (22-23). Thus, it may be expected that specific lesions in one of the many dynein genes can affect LR patterning without lethal effects on mitosis or organelle transport. There is compelling evidence that early AP and DV embryonic patterning is controlled in part by the cytoskeleton (24-25), and most importantly, disruption of the microtubule array in Xenopus by UV light causes 25% situs inversus (26).
Predictions of this model Under this model, the heterotaxia phenotype could result from a broken dynein motor which is unable to perform localization of determinants. This would allow the factor to homogeneously accumulate in both halves of a cell, resulting in double-R or double-L (depending on the nature of the determinant) compartments. As shown by Levin et al. (12), this leads to independent randomization of organ situs. Interestingly, human patients with heterotaxia as part of Kartagener’s syndrome do show defects in dynein (27-29).
The inv mutant could result from a nucleation center that is either the opposite enantiomer of one with the proper chirality, or simply becomes oriented incorrectly. The former possibility is much less likely (since the complete reversal of such a complex structure would require several coordinated mutations); it is most likely that whatever binding site is used to tether it with respect to the DV and AP axes is altered. This would result in embryos which are normal except for the consistently incorrect situs (as is observed in inv mice, where nodal is expressed always on the incorrect side only (2)).
The iv mutant may represent a nucleation structure that was not tethered at all.
Thus, it would face in different directions (randomly) in different cells. Depending on stochastic events, this would result in a mosaic of domains of cells which were oriented properly, adjacent to cells which were not. This would be magnified by cell proliferation and lineage relationships and could thus easily account for the full spectrum of nodal expression patterns observed in the mutant mice (2), corresponding to normal situs, reversed situs, or double L or R sides.
An alternative explanation for why dynein defects are associated with laterality disturbances has been proposed: that cilia directly influence the situs of the gut (27-30). This is unlikely because it has been shown that several kind of asymmetries are present long before gut looping (1), and because some patients with heterotaxia do have normal cilia function (31-32). This may be a consequence of the fact that only cytoplasmic, not ciliary, dynein is important for this process.