From genes to circuits and behaviour
Structural and functional asymmetries in the nervous system are found throughout the animal kingdom. In humans, for instance, aspects of language processing occur predominantly in the left hemisphere. This brain lateralization is thought to increase cognitive performance, whereby specialization of one hemisphere leaves the other free to perform different tasks. Compromised brain asymmetries have been linked to several neuropathologies including schizophrenia, autism, and neuronal degenerative diseases. Yet despite the prevalence and importance of nervous system asymmetries, our knowledge of the mechanisms that underlie the development and functional consequences of asymmetry is far from complete.
The best-described neuroanatomical asymmetries in vertebrates are found in the epithalamus, a major division of the diencephalon, and consist of differences in size, neuronal organisation, neurochemistry and connectivity (see Concha and Wilson, 2001 and Bianco and Wilson, 2009 for reviews from our group). Over the past ten years, we and others have established the zebrafish as a model to study the genetic and developmental mechanisms underlying the establishment of asymmetry, to analyse how asymmetry is encoded within circuitry and to address the behavioural consequences of asymmetric circuitry. In zebrafish, the epithalamus is composed of paired habenular nuclei and the adjacent pineal complex, which comprises two photoreceptive nuclei, the symmetrically positioned pineal and the left-sided parapineal (Figure 1A).
The habenular nuclei are part of a highly conserved conduction system linking the limbic forebrain to the interpeduncular nucleus (IPN) and Raphe in the ventral midbrain and hindbrain (Figure 1B,C). The habenulae display striking neuroanatomical and molecular asymmetries, including differences in gene expression, subnuclear regionalization, timing of neuronal differentiation, neuropil organization and connectivity. Left-right asymmetries in habenular projection neurons are converted into a dorsoventral asymmetry in the targeting of the habenular axons to the midbrain IPN, with left-sided habenular axons predominantly innervating the dorsal IPN (dIPN) and right-sided axons projecting to the ventral IPN (vIPN, Figure 1C,F).
A) Dorsal view of the brain showing the pineal (blue) at the midline, and left-sided parapineal (green) innervating the left habenular nucleus (red)
B) Lateral view of the brain showing telencephalic (t) and diencephalic (d) projections to the left habenula via the stria medullaris. The medial (*) and lateral (**) subdivisions of the habenula are evident and the olfactory bulb (ob) is indicated
C) Schematic showing connectivity of the habenular complex in larval zebrafish. Afferent inputs derive from the eminentia thalami (EmT), posterior tuberculum and a subset of pallial neurons (Pa) that are a source of asymmetric innervation, selectively terminating in a small medial domain of the right habenula (purple). In the epithalamus, the left-sided parapineal (pp) exclusively innervates the left habenula (lHb). Habenular neurons project efferent axons that course in the fasciculus retroflexus (FR) to the interpeduncular nucleus (IPN). A smaller contingent of habenular axons terminates caudal to the IPN in the serotonergic raphe (R). For full details and references see Bianco and Wilson (2009).
D) Image of an intact fish brain in which a single electroporated habenular neuron is expressing GFP revealing morphology from its soma on the dorsal surface of the brain to its terminals on the ventral surface
E) shows a higher resolution image of two electroporated neurons and their spiraling terminals
F) Three-dimensional confocal reconstruction showing habenular axon terminals in the ventral midbrain. Left-sided axons were labelled with DiD (blue) and right-sided axons with DiI (red), and the oculomotor nucleus is labeled with a transgene (green). The dorsal IPN is almost exclusively innervated by left-sided axons, whereas the ventral IPN receives a majority of right-sided inputs
A, C, Modified from Bianco et al. 2008; D,E from Bianco and Wilson 2009. Panel F is modified from Aizawa et al. 2005, a collaborative study with Hitoshi Okamoto's group at the Riken Institute in Tokyo.
We have taken a systematic approach to studying brain asymmetry in fish, embarking on research encompassing the early events that establish asymmetry, the precise neuroanatomy of the asymmetries and the behaviours of fish with disrupted brain lateralization. Below we outline briefly some of our past and ongoing projects.
One of the key issues that we wanted to address was how the left-sided parapineal formed. Our two hypotheses were that either (1) the equivalent group of cells on the right ultimately had a different fate (such as death or contributing to a different structure) or (2) the left-sided parapineal started off symmetric with precursors on both left and right sides of the midline and later migrated to the left. To resolve this, we traced the origin of the parapineal cells and followed them as they generated a left-sided nucleus (Figure 2; Concha et al. 2003). These experiments revealed that parapineal precursor cells are located on either side of the dorsal midline of the brain and that to form a left-sided nucleus, these cells coalesce and migrate as a coherent primordium to the left side of the brain.
So why does the parapineal nearly always migrate to the left side of the brain? We discovered that this consistent direction to the asymmetry ("laterality" or "handedness") is dependent on left-sided activity of the Nodal signalling pathway. At early stages, Nodal pathway genes are asymmetrically expressed on the left side of the epithalamus (Figure 3). Crucially, if Nodal signalling occurs on both sides of the brain or is absent, brain asymmetries still develop, but are randomised, such that normal brain laterality and reversed brain laterality are equally likely outcomes (Concha et al, 2000). These results indicate that Nodal signalling is not required for the development of neuroanatomical asymmetry per se but is required to determine the laterality of asymmetry by biasing an otherwise stochastic laterality decision to the left side of the epithalamus.
If the left habenula is partially ablated using a laser to kill cells, the parapineal sometimes migrates to the right side of the brain suggesting that signals from the left habenula influence left-sided parapineal migration (Concha et al. 2003). This means that Nodal signals could act directly on parapineal cells to attract them leftwards, indirectly by promoting secondary cues coming from the left habenula or by both routes. Although we have not completely resolved this issue, we recently discovered that Nodal does appear to have some direct effects on habenular precursor cells.
When we studied the earliest stages of neurogenesis, we found that progenitors/neurons (marked by cxcr4b expression) appear earlier in the left habenula than in the right (Roussigné et al, 2009). The temporal left-right difference in neurogenesis occurs prior to the leftward migration of the parapineal and is still detected if the parapineal is ablated. In contrast, removing the left-right bias in Nodal signalling renders habenular neurogenesis symmetric (Figure 4). This tells us that Nodal does affect the left habenula by introducing a bias in neurogenesis, which could subsequently affect parapineal migration to the left. Habenular asymmetry does still develop when Nodal is removed, mostly due to the signals coming from the parapineal (see below); however, when the parapineal is removed subtle differences remain between left and right habenulae - perhaps these are a consequence of early asymmetric Nodal signalling.
So left-sided Nodal signalling is important for introducing a left-right bias in neurogenesis and for determining the direction of parapineal migration. But how Nodal comes to be on the left in the first place? From various genetic experiments, we proposed a model in which left-sided Nodal signalling was required to overcome repression of Nodal expression on the left side of the brain (Concha et al 2000). Fitting with this idea, Micheal Rebagliati's group showed that Nodal signalling mediated by Southpaw from the left lateral plate mesoderm is required to activate Nodal signalling, mediated by Cyclops, in the brain (Long et al. 2003). However, we think that the role for Southpaw is remove repression of cyclops expression rather than to directly activate transcription.
Evidence from parallel studies from our group (Carl et al. 2007) and from Lila Solnica-Krezel's group (Inbal et al. 2007) suggests that Nodal signalling in the prospective brain is repressed at late gastrulation stage and that the role for Southpaw is to alleviate this repression on the left. Adi Inbal in Lila's group found that if Six3 activity is compromised, Nodal is activated on both sides of the brain; we found the same phenotype when Wnt signalling is enhanced on both sides of the brain. The bilateral activation of Nodal signalling in the brains of the mutants does not require the presence of Southpaw. This implies that if repression is not imposed in the first place, then Southpaw is not needed – confirmation that Southpaw's role is to remove repression. Increased Wnt signalling is likely to suppress six3 expression and so that's probably why the phenotypes seen by Lila's group and ours are similar, though we haven't tested this hypothesis directly yet. The mechanism of repression is not yet clear. It might involve Six3, but repression is established many hours before cyclops is activated in the brain and how the cells "remember" this is not known.
As is often the case, we also found that the Wnt pathway is used reiteratively at other steps during the development of body and brain asymmetry (Figure 5 and see below).
If Nodals are not absolutely required for the development of neuroanatomical asymmetry, what is? In order to elucidate the genetic mechanisms underlying the Nodal-independent breaking of brain symmetry, we screened lines of fish for mutants lacking brain asymmetry and found that the fgf8 mutant, ace, shows symmetric development of the epithalamus (Regan et al. 2009; Figure 6).
In the ace mutant, parapineal cells are present but they fail to migrate from their symmetric midline location. If given a source of exogenous Fgf8, parapineal migration is restored but usually to the left, irrespective of the source of Fgf8. This directionality, as you might suspect from the studies described above, is due to the presence of Nodal signalling on the left. If we remove the lateralised bias in Nodal signalling, then Fgf8 is sufficient to direct the laterality of migration. These results allowed us to propose a model for breaking symmetry in the brain, and suggest that mechanisms to generate asymmetry and direct laterality can be uncoupled and probably evolved sequentially (Figure 7).
These results indicate that Fgf8 is required for parapineal migration and left/right differences in the levels of Fgf8 can bias the direction of migration. Our next goal is to determine how Nodal influences Fgf8-dependent migration. It could do so through promoting a left/right asymmetry in the number of fgf8 expressing habenular precursors and/or by enhancing Fgf pathway activity in habenular and/or parapineal cells. Clearly more work is needed to resolve the interactions between Nodal and Fgf signalling pathways in the epithalamus.
Subsequent to the induction of left-sided migration, work from our group and from Josh Gamse and Marnie Halpern has shown that the parapineal promotes the elaboration of left-sided character in habenular neurons, such that the paired habenular nuclei show left-right asymmetries in gene expression, neuropil organisation and axonal projections (Bianco et al., 2008; Concha et al., 2003; Gamse et al., 2003, 2005). This communication ensures concordance between the two brain asymmetries and indeed, in virtually all asymmetry mutants, there is concordant disruption of laterality of both nuclei. The mechanisms by which the parapineal influences left habenular development are not yet known although the observation that communication between the two nuclei fails to occur in axin1 mutants suggests that modulation of Wnt signalling may be involved (Figure 8; Carl et al., 2007).
To further explore the communication between the parapineal and the habenulae, we are conducting a genetic screen (see the genetic screens page for more details) to identify phenotypes in which a parapineal is present but elaboration of habenular asymmetry is disturbed. From a pilot screen, we have 2 mutations that give phenotypes in which body asymmetry is normal but the brain is symmetric with double left-sided character (Figure 9). We are currently characterising these and other asymmetry mutants and trying to identify the affected genes.
The outcome of these early acting signalling pathways is the development of neurons that show different gene expression profiles on left and right sides of the brain. How then is asymmetry encoded within the circuitry established by these neurons? We presume that the left and right habenulae have distinct computational functions and mediate different responses and so circuitry on left and right must be different.
In collaboration with Hidenori Aizawa in Hitoshi Okamoto's lab, we have examined the pattern of habenular projections and identified a conspicuous left-right asymmetry in habenular efferent circuitry (Aizawa et al. 2005). Left-right asymmetries in habenular neuronal organization are converted into a dorsoventral asymmetry in the targeting of the habenular axons in the midbrain IPN, with left-sided habenular axons predominantly innervating the dIPN and right-sided axons projecting to the vIPN (Figure 1F). In this way, the habenulae relay information to an unpaired midline nucleus in the midbrain whilst maintaining the left-right coding of information. The laterality of this asymmetric connectivity is controlled by the Nodal-signalling pathway and is concordant with neuroanatomical and molecular asymmetries in the epithalamus.
Our next goal was to examine the morphology and connectivity of individual habenular projection neurons to address whether asymmetries in habenular connectivity could be encoded at the level of dendrite, axon morphology and/or at the level of synaptic organization. Together with Jon Clarke's lab, we optimised a single cell focal electroporation technique to do this (Figure 1E).
This approach led to the identification of two projection neuron subtypes that have axon terminal arbours with distinct morphologies and target connectivity (Bianco et al. 2008). Both subtypes are found in both the left and right habenula, but in substantially different ratios. Thus, the vast majority (84%) of left habenular neurons form 'L-typical' axon arbours that are tall and highly branched and localized to the dIPN (Movie 1). Only a very small percentage of the right-sided neurons form L-typical arbours. Instead, over 90 per cent of the right-sided cells elaborate 'R-typical' arbours that are flattened along the DV axis and localized to the vIPN (Movie 2). Because these two arbour subtypes differentially innervate the dorsal and ventral domains of the IPN, the substantial asymmetry in the cell type composition between the left and right habenulae accounts for the laterotopic habenula to IPN connectivity pattern. Finally, we find that signalling from the unilateral, left-sided parapineal nucleus is necessary for both left and right axons to develop arbours with appropriate morphology and targeting. However, following parapineal ablation, left and right habenular neurons continue to elaborate arbours with distinct, lateralized morphologies.
Our expectation is that the post-synaptic neurons of the IPN differentially process and transmit information from left and right sides of the brain. In support of this, individual neurons in the IPN are organised in such a way that this nucleus is likely to have discrete outputs from left and right-sided activation but can probably also integrate inputs from both sides (Bianco et al. 2008). We need to understand more about the neuroanatomy of the epithalamic circuitry and this is a topic that our group, Hitoshi Okamoto's group and maybe other will be actively pursuing in the coming years.
Despite progress in defining the circuitry of the larval epithalamus (Bianco and Wilson, 2009), the functional consequences of these asymmetries are not yet resolved. Larval zebrafish do show behavioural asymmetries dependent upon whether visual cues are presented to left or right eyes but, as yet, there is no data directly linking epithalamic asymmetries to specific behaviours. There is no a priori reason why any such behavioural responses should be asymmetric - for instance, in humans, cortical activity during speech is asymmetric but the motor activity that leads to the production of words is not. We have therefore started to explore a wide repertoire of behaviours that could be influenced by habenular circuitry including the learning of, and response to, fearful visual stimuli, sleep-wake cycles, as well as levels and patterns of motor activity.
Our molecular-genetic and neuroanatomical projects provide excellent resources for behaviour studies. For instance, a line of fish with reversed brain asymmetry allowed us to show that fry preferentially use one or other eye when viewing objects and that this behaviour reverses concordant with reversal of epithalamic asymmetry (Barth et al., 2005). A pilot mutant screen has already generated two lines in which homozygotes have double-left habenulae; both mutants are viable for behaviour testing as fry and at least one is viable to adulthood. As yet, we have no viable mutants with double-right habenulae, but we can create such fish by ablation of the parapineal (Bianco et al., 2008). In addition to genetic mutants, transgene-encoded optogenetic tools for recording and manipulating circuit activity are ideally suited for use in larval zebrafish, given the small size, relative simplicity and optical transparency of the brain. In the future, these tools will enable us to temporarily or permanently remove specific groups of neurons from circuits and/or to drive patterns of activity and assess behavioural consequences. One of the biggest challenges with all such experimental approaches is to define robust assays with which to test behaviour. Our initial approach involved laborious manual tracking of fry but since moving to automated tracking apparatus (Figure 9), we can analyse many more parameters without preconceived notions of what changes we should be looking for.
Our ongoing studies with the fsi line of fish have shown that fry display a stereotypic sequence of switches between left and right eyes over time when viewing conspecifics; that there is a preferred eye for initial viewing and this reverses in fish with reversed brains (Barth et al., 2005); that reversed fry appear to be bolder, showing less avoidance of novel visual stimuli and more exploration of familiar visual stimuli; and, that both normal and reversed fry learn to recognize visual stimuli and to suppress avoidance behaviour (Figure 10). Marnie Halpern's group has shown that reversed fry also show delayed initiation of swimming (Facchin et al., 2009) although we have not observed this in the fsi line of fish that we routinely use for behavioural studies. One current focus for our studies is to define altered/compromised behaviours of fry lacking brain asymmetry as these fish may potentially shed light upon human neuropathologies in which lateralization is reduced. Overall, we aim to establish a repertoire of robust behavioural assays, determine how these behaviours are affected in larvae with disrupted asymmetry phenotypes and, in the future, directly measure habenular circuit activity during these and other behaviours.
If you want to know more about our asymmetry research projects, feel free to contact us. More details on specific publications is available in our research publication summaries and all our papers can be downloaded as pdfs.
Most of our work on this project has been funded by a Wellcome Trust Programme Grant and we have also received some support from a European Communities grant entitled "Evolution and Development of Cognitive, Behavioural and Neural Lateralisation".
Our collaborators include, or have included Alex Schier, Becky Burdine, Richard Andrew, Hitsohi Okamoto, Claire Russell, Miguel Concha and Patrick Blader. In addition to these investigators, other labs working on brain asymmetry in fish include Josh Gamse, Marnie Halpern and Suresh Jesuthesan.