Morphogenesis underlying the development of the everted teleost telencephalon
© Folgueira et al.; licensee BioMed Central Ltd. 2012
Received: 29 June 2012
Accepted: 5 September 2012
Published: 18 September 2012
Skip to main content
© Folgueira et al.; licensee BioMed Central Ltd. 2012
Received: 29 June 2012
Accepted: 5 September 2012
Published: 18 September 2012
Although the mechanisms underlying brain patterning and regionalization are very much conserved, the morphology of different brain regions is extraordinarily variable across vertebrate phylogeny. This is especially manifest in the telencephalon, where the most dramatic variation is seen between ray-finned fish, which have an everted telencephalon, and all other vertebrates, which have an evaginated telencephalon. The mechanisms that generate these distinct morphologies are not well understood.
Here we study the morphogenesis of the zebrafish telencephalon from 12 hours post fertilization (hpf) to 5 days post fertilization (dpf) by analyzing forebrain ventricle formation, evolving patterns of gene and transgene expression, neuronal organization, and fate mapping. Our results highlight two key events in telencephalon morphogenesis. First, the formation of a deep ventricular recess between telencephalon and diencephalon, the anterior intraencephalic sulcus (AIS), effectively creates a posterior ventricular wall to the telencephalic lobes. This process displaces the most posterior neuroepithelial territory of the telencephalon laterally. Second, as telencephalic growth and neurogenesis proceed between days 2 and 5 of development, the pallial region of the posterior ventricular wall of the telencephalon bulges into the dorsal aspect of the AIS. This brings the ventricular zone (VZ) into close apposition with the roof of the AIS to generate a narrow ventricular space and the thin tela choroidea (tc). As the pallial VZ expands, the tc also expands over the upper surface of the telencephalon. During this period, the major axis of growth and extension of the pallial VZ is along the anteroposterior axis. This second step effectively generates an everted telencephalon by 5 dpf.
Our description of telencephalic morphogenesis challenges the conventional model that eversion is simply due to a laterally directed outfolding of the telencephalic neuroepithelium. This may have significant bearing on understanding the eventual organization of the adult fish telencephalon.
The mechanisms underlying CNS regionalization and specification of neuronal identity are remarkably well conserved across all vertebrates [1–3], yet there is huge diversity in the mature morphology of brains of different species [4, 5]. In addition, the mechanisms underlying differential morphogenesis and growth regulation of homologous CNS domains between species are not understood; indeed this topic has barely been studied. The region of the brain that shows the greatest variation in its morphology is the telencephalon, and this region is probably least well understood in terms of homology between vertebrate classes [6–11]. This is particularly true of the brains of ray-finned fish, which are markedly different from those of other vertebrates [11–13]. Since the zebrafish brain has become an important experimental model for vertebrate brain development, a deeper understanding of its morphogenesis is required, to better define its homologies and functional organization.
The current widely held view of how eversion occurs is based largely on extrapolation from the morphology of the telencephalon in adult and larval fishes [13, 15, 17–22], but no detailed experimental analyses of telencephalic morphogenesis have been performed. According to the current view, starting from a simple hollow tube in early development, eversion is thought to consist of an outward bending of the lateral walls of the telencephalon, resulting in a lateral outfolding of the dorsal region of the neural tube, such that the dorsal telencephalon (pallium) folds laterally over the ventral telencephalon (subpallium) [5, 11, 23] (Figure 1). By this process, the initially narrow telencephalic ‘roof plate’ expands laterally over both left and right telencephalic lobes to become a thin tela choroidea covering the everted telencephalic surface and enclosing the ventricular space (Figure 1). However, the positions of the pallial areas in the adult telencephalon, especially at the caudal pole of the pallium, do not fit with some of the predictions based on the hypothesis of simple mediolateral outfolding [9–11, 19, 24]. Thus, modifications of the classical model have been proposed, to explain the location of posterior pallial areas in the adult brain [9, 10, 19, 24]. Yamamoto and collaborators , for example, suggest that there must be major differences in the morphogenetic movements of eversion along the anteroposterior axis, and they imply that mediolateral folding at the posterior pole is more complex than at the anterior pole. These new hypotheses for eversion have yet to be tested in the embryo.
Understanding the development of the two fundamentally different telencephalic morphologies - everted and evaginated - will be useful not only for deciphering common and divergent mechanisms in CNS morphogenesis, but also for understanding and assigning homologies between the neuronal groups and regions in fish and other vertebrates. Therefore, we have performed a detailed analysis of early telencephalon morphogenesis in the zebrafish, a teleost with an everted telencephalon. Our data show two key steps in telencephalon morphogenesis. The first step involves the generation of a deep ventricular sulcus between telencephalon and diencephalon (the anterior intraencephalic sulcus or AIS), which creates a posterior telencephalic wall and shifts the most posterior pole of the dorsal telencephalon to a more lateral position. This is followed by a large expansion of the pallial territory, effectively elongating the dorsal telencephalon along the anteroposterior axis so that the ventricular surface of the pallium bulges into the dorsal aspect of the AIS. During this phase, the roof of the AIS expands forward over the telencephalon to form the tela choroidea. Our data do not support the widely held model that the everted phenotype results from an outward bending of the lateral walls of the telencephalon.
Our time-lapse data show that during formation of the AIS, the most posterior neuroepithelial cells of the prospective pallial telencephalon are displaced laterally (Figure 2A,B). To test this idea further, we analyzed expression of transgenes expressed in neural cells in anterior and posterior pallial telencephalon (see below and Methods for details of the transgenes). At 30 hpf we find that pallial neurons that express Et(gata2:GFP)bi105 are displaced laterally compared with the more anterior olfactory bulb neurons expressing Tg(−10lhx2a:EGFP)zf176. Both transgenes are expressed just in front of the dorsal rim of the AIS, with the more topologically anterior olfactory bulb (OB) expression located medial to the more topologically posterior pallial expression (Figure 2C,D,E).
By the end of the first day of development, each of the two telencephalic lobes is a roughly ellipsoidal solid, with flat medial surfaces adjacent to the narrow midline ventricle, and bounded posteriorly and ventrally by the curved ventricular surface of the AIS. So at this stage, both telencephalic lobes have a medial ventricular surface sandwiched between the two lobes and a posterior ventricular surface, which we will refer to as the posterior telencephalic wall. The AIS has a thin diamond-shaped roof dorsally.
Additional file 2: Figure S1. Olfactory bulb markers and fate mapping suggest that origin of OB is close to the dorsal rim of the AIS. A-D. Whole-mount in situ hybridizations for lhx1a at different stages from 1 dpf to 5 dpf, showing that the distance between the olfactory bulb domain (arrowhead) and the AIS (dotted line) increases between days 1 and 5. All lateral views with anterior to the left. E and F. YFP expression in the olfactory bulb of the transgenic Et(CLG-YFP)smb8 embryos also becomes increasingly distant from the AIS (white dotted line) between 2 and 5 dpf. Neuropil is counterstained with an antibody recognizing acetylated tubulin (red). All images are lateral views, anterior to the left. G to J. Photoconversion of Kaede-expressing cells close to the dorsal rim of the AIS at 1 dpf reveals that descendants from these cells populate the olfactory bulb at 5 dpf. G. Drawing illustrating the three regions of the telencephalon targeted by Kaede photoconversion for fate mapping at 1 dpf (T1-3, lateral view). H. Summary distribution in lateral views at 5 dpf of cells photoconverted in the regions illustrated in G. All data derived from confocal z-stacks of the whole telencephalon. I-J. Example of photoconversion and fate from T1. I is a lateral non-confocal view of the photoconverted cells (blue arrow) at 1 dpf in region T1. J is a single confocal section taken at 5 dpf at the level illustrated by the red line in H. Photoconverted cells are red and non-converted cells remain green. Scale bars 50 μm. AC: anterior commissure; E: epithalamus; Ha: habenula; Hy: hypothalamus; OB: olfactory bulb; SM: stria medullaris; SOT: supraoptic tract; Tel: telencephalon; V: ventricle. (MOV 8 MB)
To assess the overall growth and form of the telencephalon between 2 dpf and 5 dpf, we used the Tg(−8.0cldnb:lynEGFP)zf106, which fortuitously strongly labels the membranes of all telencephalic cells (Figure 4G and H). Observations from the dorsal and lateral aspect show rather little change in mediolateral and dorso-ventral dimensions but a significant increase along the anteroposterior (AP) axis (Figure 4I and J).
Taken together, these data show that the OB initially differentiates close to the dorsal rim of the AIS and is separated by only approximately 20 μm from the dorsal diencephalon. Between 2 dpf and 5 dpf, telencephalic domains posterior to the OB grow considerably, bulging into the AIS and increasing the distance between the OB and the dorsal diencephalon.
Additional file 3: Movie 2. Three-dimensional reconstruction from confocal z-stack of the telencephalon of a 2 dpf embryo stained for ZO-1 (red) to reveal ventricular surface and GFAP (green) to show depth of neuroepithelium. As it rotates, note that the ventricular surface is only present on the medial and posterior surfaces of the telencephalic lobes, but not on the upper telencephalic surface at this time. Anterior is to the left, initial view is from dorsal aspect and this rotates to give first a posterior and then lateral view. Roof of the AIS has been dissected away to allow better resolution of the ventricular surface of the telencephalon. (MOV 8 MB)
The expansion of proliferative cells over the upper pallial surface from 2 dpf should be accompanied by a thin epithelial tela choroidea covering these superficial ventricular cells. To assess this, we followed tela choroidea formation using the epithelial junction protein ZO1 to reveal these flattened epithelial cells. At 1 dpf and 2dpf, the prospective tela forms a roughly diamond-shaped roof over the dorsal extremity of the AIS. From 2 dpf through to 5 dpf, the area of the roof enlarges greatly along its AP dimension but relatively little along its transverse (mediolateral) dimension (Figure 5D-F), so that by 5 dpf almost all the upper surface of the telencephalon is covered by tela choroidea (Figure 5F).
The OBs do not become covered by tela; rather, the tela is attached just caudal to their posterior limit (Figure 5F). This close anatomical relation of OB to the attachment of the tela choroidea is maintained into adulthood and has been documented in other teleost brains (Additional file 4: Figure S2 and [12, 30, 31]).
The AIS is a prominent fold in the early forebrain ventricular surface that lies at the interface of the telencephalic and diencephalic territories, and was described at least 100 years ago . Despite this early documentation, its relevance to the distinctive everted morphology of the telencephalon of ray-finned fish has been largely ignored [9, 11, 13, 20, 22]. Our work confirms the generation of the AIS as a primary feature of telencephalic development in the zebrafish, and our data suggest that it may have significant implications in understanding subsequent organization of regional identity within the telencephalon. The potential implications for regional organization stem from our demonstration that AIS formation folds the most posterior territory of the dorsal telencephalic neuroepithelium to a more lateral location (Figure 7A). This folding process generates a posterior ventricular wall to the telencephalon, resulting in anteroposterior positional values becoming realigned along the mediolateral axis (Figure 7A). If specification of pallial regional identity is established before the AIS is initiated, then the neuroepithelium that is located at the most posterior region of the dorsal telencephalic anlage will become placed first more laterally, and this is supported by our analysis of OB and pallial markers.
Following formation of the AIS and posterior telencephalic wall, the next significant morphogenetic process is the expansion of the pallium domains between days 2 and 5 (Figure 7B-D). During this phase, fate-mapping data and analysis of tela choroidea expansion show that the major axis of neuroepithelial growth is aligned along the embryo’s AP axis. Our data are reminiscent of those of Droogleever Fortuyn , who suggested that eversion of the stickleback dorsal telencephalon includes a phase of forward-directed growth (see his Figures 10 and 11). We believe that our data are best interpreted as the pallial expansion being accompanied by a bulging convexity of the posterior telencephalic wall pushing the dorsal aspects of the posterior telencephalic wall up against the roof of the AIS. We believe that this is the key aspect of telencephalic morphogenesis that initiates the everted phenotype by pushing the ventricular surface of the posterior telencephalic wall up towards the dorsal surface. This generates the narrow ventricular space and tela choroidea that begin to enclose the upper surface of the telencephalon. Our fate mapping suggests that there is very little mediolateral reorganization of the ventricular surface during this period. Our data, therefore, do not support the widely held model that the everted phenotype results from an outward bending of the lateral walls of the telencephalon.
Although the characteristics of an everted telencephalon are clear [11, 12] (see Figure 1), the process that generates such morphology (eversion) has been difficult to define. Based mainly on the morphology of the adult telencephalon, eversion has been thought to consist of a lateral outfolding of the dorsal region of the neural tube, such that the dorsal telencephalon (pallium) folds laterally over the ventral telencephalon and the roof plate expands mediolaterally [11, 15, 17–23]. Implicit in this model is the assumption that the anterior to posterior regional identity of the ventricular tissue undergoing eversion should be unaffected. However, there have been no previous experimental studies that directly address tissue and cell morphogenesis in living embryos. Based on our current work, we can now ask how the zebrafish data compares with the conventional view.
Our data show that early telencephalic morphogenesis does include a lateral outfolding of the neuroepithelium but that this folding occurs at the telencephalic-diencephalic interface to generate the AIS and folds the posterior telencephalon to a more lateral position (Figure 7A). Unlike the folding event in the conventional model of eversion, this folding process does not bring the medial ventricular surface of the telencephalon to the upper surface but instead generates a posterior ventricular wall to the telencephalon. Our data suggest that from approximately 48 hpf the posterior ventricular wall bulges up and into the AIS. This brings the posterior ventricular wall of the telencephalon into close proximity with the roof of the AIS, and this is the first step that causes a small area of ventricle to lie over the upper surface of the telencephalon. The dorsal telencephalon then grows along the AP axis, increasing the amount of ventricle on the upper surface of the telencephalon and stretching the roof of the AIS forward to become the tela choroidea (Figure 7B-D). These data between days 1 and 5 are not consistent with the conventional view that eversion is driven by an outward bending of the lateral walls of the telencephalon. However, after 5 dpf, the everted morphology could well be refined to acquire the final morphology observed in the adult by a variety of processes that could include differential growth, further folding of the neuroepithelium, and cell migration.
Our data are not the first to suggest that the commonly held view of eversion needs modification. Previous studies of the organization of telencephalic areas and their connections in adult goldfish found that this did not fit the predictions of the classic model for telencephalic eversion [9, 10]. Yamamoto and colleagues  proposed a new eversion model based only on their regional analyses in the adult. A major component of their hypothesis is a ‘caudolateral’ folding of the embryonic pallial neuroepithelium that could be consistent with the movements we describe generating the AIS. Our analysis demonstrates that this outfolding precedes the phase of eversion that brings the ventricular cells and their covering of tela choroidea onto the superficial surface of the telencephalon, whereas their hypothesis implies that it follows this event (see their Figure 4A). Interestingly, their study suggests, and our data demonstrate, that the mediolateral outfolding is more complex at the posterior pole of the telencephalon than the anterior pole. In contrast to the rearrangements at the posterior pole of the telencephalon, our data do not support the Yamamoto et al. view that a relatively simple lateral outfolding of the anterior pallium occurs. However, it is, of course, possible that such folding could occur beyond day 5 of development.
Our results also show that there is more than one way to obtain the characteristic organization of radial glia often used as one criterion in support of the conventional model of eversion (e.g. ). Coronal sections of adult teleost telencephalon show radial glia with cell bodies adjacent to the upper everted surface and their radial processes curving laterally to the pial surface. Our analysis shows that a similar organization of cell shapes is generated in the neuroepithelium during formation of the AIS, but at this time the fold in the neuroepithelium is in the horizontal plane, not the coronal plane. However, the upwards and backwards bulging expansion of the posterior wall of the telencephalon towards the tela choroidea of the AIS between 2 and 5 days would reorient this cellular arrangement from the horizontal plane to the coronal plane.
The generation of the AIS must be driven by local regulation of neuroepithelial cytoskeleton to control cell shape changes required to fold the epithelium. Cells destined to lie at the lateral extremes of the AIS must shorten and constrict their apical profile. The second step of our model involves the expansion of pallial territory along the AP axis to increase the distance between the OB primordia and the dorsal diencephalon. This appears to be coupled to an increase in the convexity of the dorsal aspects of the posterior telencephalic wall. The posterior wall appears to bulge back and up towards the roof of the AIS. The AP vector of ventricular expansion could either be driven by oriented divisions along the AP axis or contain an element of cell intercalation that expands clones along a particular orientation.
Our data describing telencephalic morphogenesis challenge the conventional model that eversion is simply caused by a laterally directed outfolding of the telencephalic neuroepithelium. This result may have significant bearing on understanding eventual organization of the adult fish telencephalon.
Zebrafish strains (Danio rerio, Cyprinidae, ray-finned fish) were maintained and bred following standard procedures. AB wild type, Tg(elavl3:eGFP)zf8, Et(gata2:GFP)bi105, Tg(1.4dlx5a-dlx6a:GFP)ot1, Tg(−10lhx2a:EGFP)zf176, Et(CLG-YFP)smb8, Tg(β-actin:HRAS-EGFP)vu119, and Tg(−8.0cldnb:lynEGFP)zf106 were used. Embryos were raised at 28.5°C and staged according to standard methods . To prevent pigment formation, 0.003% w/v phenylthiocarbamide (PTU, Sigma) was added to the fish water at 24 hpf (hours post fertilization).
The cDNAs used in this work for RNA synthesis were cloned previously in pCS2 (kaede, ; pard3/asip-egfp, ) and mRNA was generated using a Message Machine RNA synthesis kit (Ambion) according to manufacturer’s instructions. Embryos were injected with 2–3 nl of 0.1 pg/μl mRNA at the one to four cell stage.
Confocal time-lapse imaging was performed using Tg(β-actin:HRAS-EGFP)vu119 line. Once anaesthetized, embryos were embedded in low melting point agarose and imaged from a dorsal view using water immersion objectives. Embryos were maintained at 28.5°C in an environmental chamber and z-stacks were collected at intervals of 5 to 8 min. Imaging was started at around 14 hpf to 16 hpf and usually continued for 10 to 12 hours.
Embryos were transferred to a small dish containing a solution of bromodeoxyuridine (BrdU, 5 mg/ml) diluted in fish water with 15% dimethyl sulfoxide (DMSO, Sigma). The dish was kept on ice for 20 min and then transferred to 28.5°C (20 min). After recovering from treatment (5 min) in fish water, embryos were fixed in 4% paraformaldehyde and immunostained.
For immunostaining, embryos were fixed in 4% paraformaldehyde and stained as whole mounts, following standard procedures . Antibodies and dilutions used were mouse anti-BrdU (1:200, Sigma), rabbit anti-green fluorescent protein (GFP; 1:1000, Torrey Pines Biolabs), rabbit anti-glial fibrillary acid protein (GFAP; 1:400, Dako), zona occludens 1 (ZO1; 1:300, Invitrogen) mouse anti-acetylated tubulin antibody (1:250, Sigma).
For in situ hybridization, antisense RNA probes were generated using digoxigenin RNA labelling kits (Boehringer-Mannheim). Whole-mount in situ hybridization was performed as previously described .
Embryos were anaesthetized and mounted in low melting point agarose. A small number of cells expressing Kaede were photoconverted from green to red  using the 405 nm line on a Leica SP5 confocal microscope. Images of the initial photoconverted region were taken. Embryos were then released from their agarose mount and maintained at 28.5°C until two or three days later, when they were reimaged to assess the fate of the photoconverted cells.
MF has a joint appointment with the Department of Cell and Molecular Biology, University of A Coruña, 15008 A Coruña, Spain.
anterior intraencephalic sulcus
dorsoventral diencephalic tract
posterior telencephalic wall
tract of the anterior commissure
tract of the posterior commissure
tract of the postoptic commissure
PB was funded by a Wellcome Trust PhD studentship. MF was funded through the EU Framework 6 ZF-MODELS programme (to JC and SW), the Wellcome Trust (to SW) and the University of A Coruña. The study was also supported by grants from the BBSRC to JC, TH and SW. We thank Tom Hawkins for advice with transgenic lines. We also thank Corinne Houart, Marina Mione, Ramón Anadón, and all the members of the Clarke and Wilson labs for many helpful discussions on this work. We thank members of UCL Fish Facility for help with fish stock maintenance.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.