Zebrafish Research at UCL

Because of its outstanding suitability for imaging and transgenesis approaches, the developing zebrafish is set to become a leading vertebrate model for studies of brain circuitry, synaptic plasticity and behaviour. An essential prerequisite for studies of circuit connectivity and behaviour is a clear understanding of the neuroanatomy of the brain. However, there is a lack of detailed neuroanatomical information for this vertebrate model - this lack of knowledge presents a major bottleneck in the field. Although we have a reasonable understanding of the general principles of how neurons are generated and acquire identities, our knowledge of the circuitry subsequently established by the neurons is very fragmentary.

Furthermore, research to date has tended to focus on a small number of specific brain regions but even within relatively simple regions of the fish CNS, such as the spinal cord, there is still much to learn about the detailed neuroanatomy of the circuits that drive behaviour. If we do not know the normal connectivity of a class of neurons, we will not be able to address how the circuits in which they integrate function.

Figure 1 - 4 day old zebrafish embryo labelled with SV2 and acetylated tubulin antibodies showing axon tracts(green) and neuropil(red)viewed from lateral(top) and dorsal(bottom) orientations.

The paucity of detailed neuroanatomical data is surprising given the outstanding optical properties of the zebrafish which permit imaging of the entire intact brain at high resolution by confocal microscopy. This property, coupled with the availability of large numbers of transgenic lines in which subsets of neurons are labelled by fluorescent reporters make the zebrafish a fantastically powerful model for neuroanatomical studies. Transgenic animals are an excellent and novel way of resolving neuroanatomy. Traditional neuroanatomical tracing techniques, such as lipophilic dye labelling of axons, suffer from the problem that it is hard and time consuming to restrict labelling to unique subsets of neurons. Genetic approaches get around this problem by enabling the cell-type specific expression of transgenes encoding proteins that label the entire morphologies of the expressing cells. This approach can be used to label entire classes of neurons or indeed individual neurons.

Figure 2 - Triptych of a 5dpf zebrafish forebrain viewed from dorsal(anterior to the left), frontal and lateral(anterior to the right) aspects. The embryo has been labelled with SV2 and acetylated-tubulin antibodies showing axon tracts(blue) and neuropil(pink). In these images we can clearly see the different components of the zebrafish forebrain; the olfactory bulb is situated most anteriorly in front of the telencephalic lobes. We can also see the habenulae from dorsal and lateral view points in the dorsal diencephalon.

Despite the exceptional properties of the zebrafish embryo as a model system, neuroanatomical resources are scarce and most detailed neuroanatomical data has to be divined from histological atlases of larval and adult brains. Although these paper atlases are undoubtedly an accurate and indispensable resource, they are not easy to use unless one is an expert in fish neuroanatomy. Traditional atlases standardly use annotated images from serially cut histological sections and so to fully understand the information presented, the user must already have a good understanding of the anatomical relationships and connectivity of the brain as this is difficult to interpret from 2D sections.

Figure 3 - A time-series of tubulin immunostaining through development from 36hpf to 5 day old larva. Note the initial simplicity of the network of tracts and commissures which rapidly increases in complexity through development. The network provides convenient and reproducible landmarks as points of reference for comparisons between specimens.

For the past four years, in collaboration with Jon Clarke's, lab, we have been developing an online, high-resolution atlas of the neuroanatomy of the developing zebrafish brain. Funding for a pilot version of the atlas has been provided in part through an FP6 EU-funded integrated project (ZF-Models) that aimed to generate tools, resources and datasets to facilitate the development of the zebrafish as a model for human diseases. The source data for the atlas have been obtained through confocal imaging of intact brains in which individual neurons or subsets of neurons are labelled using antibodies and/or transgenes. The online atlas will be quite unlike traditional neuroanatomy atlases, because we aim to use a mix of media types to show the data, including confocal sections and reconstructions, 3D and 4D movies and schematic models. Information will be presented in whatever format is most helpful, not being constrained by standardised orientations for viewing. Indeed, interactive 3D models will allow the user to see structures in any and all possible orientations.

Figure 4 - A-F Identification of early embryonic precursors of neuronal structures. A-C. High resolution confocal projections of Tg(tbr:YFP) transgenics at multiple stages. Continuous expression of GFP through development permits the tracing back of structures which can be morphologically delineated at later stages to place their early origins; eg. the olfactory bulb (OB) which originates in the dorsal early telencephalon (36hpf; A) but moves to the anterior pole of the brain by 4 days. (C) D-F. Tubulin immunohistochemistry (red) aids orientation of GFP expression in transgenics in this case Tg(tbr1:YFP) (Mione at al. 2008, Dev Neurosci;30:65-81)

We envisage a typical inexperienced user may have found a phenotype affecting one region of the brain or have identified a gene with localised CNS expression and may come to the atlas wanting to know information about that particular part of the brain. Without knowing anything about neuroanatomy, the user will pick a brain model of appropriate age and navigate around the brain to identify the region of interest. He/she will then be directed to a tutorial about the structure with 2D images and interactive 3D reconstructions to explain the topography and connectivity of the structure. A more experienced user may enter a specific neuroanatomical term and will be taken directly to images, movies and text about the structure, its composition, cytoarchitecture and connections, listing transgenic lines, antibodies, genetic markers and other resources relevant to the structure and other structures linked by, for instance, connectivity or neurotransmitter type. The atlas will be searchable by reference to neuroanatomical structure or region, neuronal sub-type, gene expression, neurotransmitter expression and in some cases gene/transgene expression.

Figure 5 - A-C Other staining methods which can be used for orientation. A SV2 immunohistochemistry (red), here combined with tubulin (green) shows areas of neuropil that can act as reference points. B & C. Nuclear dyes (grey (B); blue (C)) show areas of somal coalescence (brain nuclei) and also expose morphological boundaries between brain areas (B). Neuropil is not marked and is revealed as dark spaces (B). Thus nuclear staining is complementary to neuropil markers like SV2 or tract tracing methods (red and green; C).

Aside from the development of methods to present neuroanatomical data, we have also been taking advantage of the excellent imaging properties of the zebrafish embryo to generate very high quality neuroanatomical data for presentation in the atlas. Our approach has been to collect high-resolution confocal stacks from many different transgenic lines at multiple developmental stages (eg. figure 4 A-F) and reconstruct images into flattened projections, 3D (and occasionally 4D) models and/or movies. Anti-acetylated tubulin labelling of axonal tracts provides a scaffold on which to anchor the locations of expression of the transgenes (figure 3; figure 4 A-F; figure 6 A-D) and other antibodies (for instance, against synaptic vesicles figures 1&2, or neurotransmitters, figure 7) or histological staining methods (figure 5 A-C) help us to further characterise expression.

Concurrent with this, we have developed a single cell focal electroporation technique that enables us to express transgenes in individual neurons in the intact brain, providing exceptionally high-resolution information on the morphologies of specific neuronal sub-types (figure 8 A-C). These techniques allow us to visualise different neuronal subgroups within the CNS and characterize their connectivity. The fact that we work with the intact brain hopefully means that our neuroanatomical data will be more easily interpreted by users from the zebrafish community who, for the most part, work with wholemount preparations. In addition to providing a useful neuroanatomical reference resource, detailed analysis of neuronal connectivity within the embryonic zebrafish CNS will become more and more essential as the zebrafish community moves into novel areas of research, particularly circuit physiology and behaviour.

Figure 6 - A & B Distinct populations of neurons are GFP positive (green) in Tg(brn3a-hsp70:GFP)(Okomoto et al. 2008, Dev Growth Differ. 2008 Jun;50) (A & B)and Tg(1.4dlx5a-dlx6a:GFP)transgenics(Mione et al. 2008, Dev Neurosci ;30:65-81 (C) counterstained with anti-tubulin immunostaining (red). D Calretinin immunohistochemistry (green) of neurons at the fore/midbrain boundary

The atlas is conceived as an open access tool for the scientific community, to be used not only by scientists from the zebrafish field but also by researchers from other fields or model systems. As part of this open philosophy, we would also like to get feedback from the community itself, from experts on different regions of the brain and from general users. As the atlas becomes more established, we plan to include high quality neuroanatomical data from other researchers who themselves may be expert in particular brain regions.

Figure 7 - GFP expression in the ET11 enhancer-trap transgenic line (Choo et al. BMC Dev Biol. 2006 Feb 14;6-5). The box highlights a population of neurons in the dorsomedial tectum some of which are GABAergic.

Aside from finding funding to support the project, one of our biggest challenges in the future is the annotation of 3D models. All of the high-resolution confocal stacks we have collected of the various transgenic lines and immunolabelling have the potential to be rendered in 3D as interactive models. Interactive 3D models will allow the user to see structures in any and all orientations. If we are able to properly annotate and integrate these models in a user friendly manner into the atlas, they would enhance it greatly. Traditional atlases relying on a single media type usually present images in strictly transverse, sagittal or horizontal views that are not so easily interpretable. By using a mixture of media types including confocal sections and reconstructions, 3D and 4D movies and schematic models, we can present information in whatever format is most helpful and we are not constrained by standardised orientations for viewing. Our intention is to present data from whole brain reconstructions down to high-resolution reconstructions of typical morphologies of individual neurons of specific classes (figure 8). Alongside the data repository of annotated media, we will present tutorials on specific CNS nuclei, pathways and structures (example, figure 9). Tutorials will provide the context for understanding the images and movies within the atlas. They also facilitate browsing within the atlas, allowing the user studying one structure to quickly find information on linked nuclei or pathways. To develop these tutorials, input, feedback and corrections from experts in different brain regions will be crucial to make them a fully useful and accurate resource.We hope that in addition to being a useful and intuitive tool for the community, the atlas will provide a showcase for the beautiful images and movies being produced by the zebrafish community using cutting edge imaging techniques and rendering software.

Figure 8 - Single cell labelling. A. Focal electroporation of GFP in a single habenular neuron; the entire morphology is revealed from the cell body on the dorsal surface of the brain to the ventral terminals in the interpeduncular nucleus (IPN). B. Exquisite detail of the morphology of a pair of habenular neurons. C. The somal and dendritic morphology of a large reticulospinal neuron revealed by anti-calretinin antibody. Images in A and B adapted from Bianco et al. 2008, Neural Developent 3, 8.

We intend to make a β-version of the atlas available for testing by a few selected members of the community early in 2009 and make it fully available once any major glitches and problems have been ironed out. To date, we have only received sufficient funding to generate a pilot version of the database with data from a limited number of transgenic lines and a focus on several specific regions of the CNS. We are now trying to obtain funding to continue and expand this project. We hope that with community support, enthusiastic reviewers and generous funding sources, we will be able to move forward and provide an outstanding resource to the community.

Figure 9 - An overview of a typical region tutorial from the database

We thank many colleagues in the community, including Tom Becker, Vladimir Korzh, Hitoshi Okamoto and Marina Mione, for providing fish lines and other resources to help with this project.

Support for this project has been provided by ZF-Models, an FP6 EU-funded integrated project, and a Wellcome Trust Programme Grant to SW. Future funding opportunities are available to any Funding Bodies, Charities, Philanthropists or others with ambitions that dovetail with those of the project!