MICROSCOPY RESEARCH AND TECHNIQUE 48:47–54 (2000) New Ways to Look at Axons in Caenorhabditis elegans HARALD HUTTER* Max-Planck-Institut für medizinische Forschung, Jahnstr. 29, 69120 Heidelberg, Germany KEY WORDS nervous system; GFP; confocal microscopy ABSTRACT In the nematode Caenorhabditis elegans, a well-established model organism for the analysis of nervous system development and function, nerve processes can be labelled in the intact animal with markers based on the ‘‘Green Fluorescent Protein’’ (GFP). The generation of GFP variants with improved brightness and modified emission spectra potentiated the use of this marker for in vivo labelling of subcellular structures. This made it possible to label different groups of neurons and their axons in the same animal with GFP variants of different spectral characteristics. Here I show with double labelling experiments that spatial relationships of axons in small axon bundles can now be resolved at the light microscopic level. In the future this will largely circumvent the need for time-consuming electron microscopic reconstructions to detect local defects in axon outgrowth. Furthermore, I demonstrate that neuronal processes can now be traced even in the head ganglia, an area of the nervous system that was previously almost inaccessible for analysis due to the compact arrangement of cell bodies and axons. Microsc. Res. Tech. 48:47–54, 2000. r 2000 Wiley-Liss, Inc. INTRODUCTION Some 30 years ago Sydney Brenner and his colleagues started to cultivate a small free-living soil nematode called Caenorhabditis elegans in their lab. A major goal was to use this animal to study development and function of the nervous system, specifically how these processes are controlled genetically. All developmental stages of C. elegans are transparent. This enabled Sulston and co-workers to follow cell divisions and migrations in the intact developing animal. They described the complete developmental history and final position of every cell in the animal (Sulston et al., 1983; Sulston and Horvitz, 1977). These positions are highly invariant from animal to animal and can be used by the experienced researcher to identify a particular cell. This knowledge allowed White and co-workers to initiate an even more ambitious project, the complete electron microscopic (EM) reconstruction of the nervous system including the connections among neurons. In serial sections they traced all neuronal processes back to their cell bodies in order to determine their identity. Chemical and electrical synapses were scored and revealed how the nervous system is wired (White et al., 1986). Almost all of the neurons have a very simple morphology with only one or two processes that typically are unbranched. Synaptic connections were mainly found in process bundles, where an axon develops a local swelling and establishes a synapse with a neighbouring dendrite. Therefore, a particular process is always found in a defined position within a process bundle surrounded by its synaptic targets. Consequently, every bundle is a highly ordered structure where the relative position of axons reflects the connectivity (Fig. 1). Outgrowing processes in C. elegans find their targets one by one on their way as they grow out, and they have to navigate very precisely over the entire distance in order not to miss any of them. In their initial attempt to identify and analyse the genes that control neuronal development and function, r 2000 WILEY-LISS, INC. Brenner colleagues isolated a number of neurological mutants (Brenner, 1974). These were first identified as having characteristic behavioural defects, e.g., defects in the coordination of their movement. EM reconstructions revealed specific defects in axonal outgrowth for some of them that could be interpreted in terms of specific guidance signals affected or cell identities misspecified. It is important to note that the extremely high degree of invariance in the anatomy of C. elegans allowed researchers already early on to analyse in detail the consequences of experimental manipulation without the need for molecular markers that became available only later. However, the number of neurological mutants that were isolated soon exceeded the number that could be analysed by EM reconstructions. Histochemical markers were discovered that allowed visualisation of groups of neurons with a common neurotransmitter (McIntire et al., 1993) or sensory neurons with openings to the outside that could be filled with fluorescent dyes simply by soaking the animal in dye solution (Hedgecock et al., 1985). These staining methods were used to rapidly screen the existing mutant collection in order to identify those mutants that had axonal outgrowth defects (Hedgecock et al., 1985; McIntire et al., 1992; Siddiqui and Culotti, 1991). One disadvantage of histochemical markers compared to EM reconstruction is the limited spatial resolution. The need to kill and fix the animal before the staining procedure frequently reduces the resolution even further compared to observations of the intact tissue in a living animal, so that it becomes difficult or even impossible to discern certain details of axon trajectories. The introduction of the ‘‘Green Fluorescent Protein’’ (GFP) as a marker for in vivo labelling (Chalfie et al., 1994) overcame some of these limitations. Cur*Correspondence to: Harald Hutter, Max-Planck-Institut für medizinische Forschung, Jahnstr. 29, 69120 Heidelberg, Germany. E-mail: hutter@ mpimf-heidelberg.mpg.de Received 31 May 1999; accepted in revised form10 August 1999 48 H. HUTTER Fig. 1. The nervous system of C. elegans. A: A schematic drawing of the head region of C. elegans illustrating the location of various parts of the nervous system. Interneuron cell bodies are located in head ganglia close to the nerve ring, whereas motorneuron cell bodies are positioned along the ventral midline. Note that interneuron axons use different routes into the nerve ring and run in defined positions within the ring. B: Cross-section through the ventral cord after White et al. (1986). The ventral cord consists of two longitudinal axon tracts separated by a hypodermal ridge. Also positioned between the axon tracts are cell bodies of motorneurons. Most of the axons run in the right tract. The spatial relationships of axons within the bundle are highly invariant from animal to animal. Interneuron axons (AVB, AVD, AVE , AVA, PVC) and motorneuron axons form distinct subbundles and maintain a particular order even within the subbundle. C: Synaptic connections between interneurons and motorneurons (type A and type B) within the ventral cord. A large number of synaptic connections is formed among the interneurons and between interneurons and motorneurons. Arrows point from the presynaptic to the postsynaptic process. Since synapses are formed only between processes that contact each other, the connectivity is reflected in the relative positioning of neuronal processes. rently several GFP variants with improved brightness and different colours are available (Heim and Tsien, 1996; Ormö et al., 1996). Here I demonstrate that the use of GFP in combination with advanced microscopy and imaging techniques allows a new view of the nervous system of C. elegans. Double labelling experiments reveal that the spatial resolution in an intact animal is sufficient to visualise relative positions of axons even within small axon bundles. In the brain region, which was virtually inaccessible at the light microscopic level due to its complexity of cell bodies and neuronal processes, GFP labelling now allows the detailed reconstruction of axon trajectories. MATERIALS AND METHODS GFP Constructs The following plasmids were used as a starting point for cloning experiments. KP#6 kindly provided by J. Kaplan contains the glr-1 promoter for expression of VISUALISING AXONS IN C. ELEGANS reporter genes in a subset of interneurons (Hart et al., 1995). pPD95.75 kindly provided by A. Fire is a promoterless vector containing the S65C variant of GFP (green GFP). pPD115.46 also provided by A. Fire is a promoterless vector containing the Y66W N146I M154T V163A variant of GFP (cyan variant or CFP). The 5.3 kb PstI-BamHI fragment from KP#6 containing the glr-1 promoter was cloned into pPD95.75 resulting in pG75#KP6 and also into pPD115.46 resulting in pG46#KP6. Transgenic Strains Transgenic strains were made using standard techniques (Mello and Fire, 1995). Briefly, plasmid DNA (pG75#KP6 or pG46#KP6) was injected into C. elegans hermaphrodites. dpy-20(⫹) DNA was used as coinjection marker; injections were made into a dpy20(e1282ts) strain. Stables lines were generated by gamma-irradiation followed by outcrossing. Animals where motorneurons are labelled with GFP and interneurons with CFP were generated by crossing a strain carrying pG46#KP6 with one expressing GFP in DA and DB motorneurons (kindly provided by J. Culotti). Microscopy and Imaging Images for Figure 2 were taken on a Leica SP confocal system equipped with an Argon laser using a Plan Apo 100x, 1.4NA oil objective. CFP was excited at 458 nm, GFP at 488 nm. Emitted light was collected between 468 and 482 nm for CFP using a 465-nm dichroic filter, and between 500 and 560 nm for GFP using a 488-nm dichroic filter. Sequential scans were used to collect images in the two channels. Images for Figure 3 were taken on a Leica NT confocal microscope equipped with an Argon-Krypton laser using a Plan Apo 63x, 1.4NA oil objective. GFP was excited at 488 nm, a TK500 dichroic filter was used together with a BP525/50 barrier filter. Images were further processed with the Huygens2 (Bitplane AG, Switzerland) deconvolution software and displayed using the Imaris program. Within Huygens2 the Maximum Likelihood Estimation method (MLE) algorithm (Snyder et al., 1992) was used for deconvolution with a calculated point spread function. RESULTS Determining the Relative Positions of Axons Within an Axon Bundle In C. elegans, the relative position of a particular axon within an axon bundle is well conserved. Determining these positions so far was only possible in timeconsuming EM reconstructions. Using GFP as a marker in living animals, I attempted to visualise these spatial relationships at the light microscopic level. To this end, I generated transgenic animals expressing two variants of GFP in different groups of axons. A cyan variant (CFP) was used to label interneurons of the motor circuit whereas a green variant was used to label a subset of motorneurons. Both groups of axons are found in the major longitudinal axon tract, the ventral cord. From EM reconstructions we know that interneurons and motorneurons form distinct subbundles within the ventral cord with the motorneuron bundle lying ventrally adjacent to the interneuron bundle (White et al., 1976; Fig. 1B). This relative position is maintained over 49 a long distance and reflects the local synaptic connections within the ventral cord (Fig. 1C). The entire ventral cord in cross-section is about 2 µm wide with individual axons about 0.2 µm in diameter. A single axon, therefore, is at the resolution limits of a light microscope and with current histochemical markers the relative positions of axons within an axon bundle cannot be resolved. How far can the resolution limits be pushed using modern microscopy and imaging techniques together with in vivo markers that allow imaging of an intact animal? 3D image stacks of the ventral cord of double-labelled animals were recorded with a confocal microscope. Figure 2A shows a central focal plane with the interneuron bundle in green and the motorneuron bundle in red. The relative positions of these axons clearly can be determined. The motorneuron axons are positioned ventrally adjacent to the interneuron subbundle (Fig. 2B). Both groups of axons are tightly bundled together. To further improve the spatial resolution and quality of the image a deconvolution algorithm was applied (Maximum Likelihood Estimation method (MLE) algorithm (Snyder et al., 1992)). The resulting pictures (Fig. 2C, D) demonstrate that this indeed leads to a sharper image with less blur around the edges of axons, which makes it easier to determine the borders of the subbundles. CFP and GFP have overlapping excitation and emission spectra and it is impossible to completely separate the two variants in standard fluorescence microscopes equipped with a mercury lamp for illumination. On a Leica SP confocal microscope, however, the emission window can be adjusted freely, a feature that allows one to determine for each sample individually the windows that minimise bleed-through (in this example CFP was detected between 465 and 482 nm, GFP was detected between 500 and 560 nm). Laser strength for illumination and photomultiplier settings for detection can be adjusted so that bleed-through is essentially undetectable (Fig. 2E–H). This allows a clean separation of CFP and GFP signals. Interneuron and motorneuron axons maintain their relative positions over the entire length of the ventral cord. Figure 2I shows a larger region of the ventral cord between two neighbouring motor neuron cell bodies. Again the motorneuron axons can be seen running adjacent to the interneuron axons. A single axon that clearly runs apart from the interneuron bundle on the dorsal side of the ventral cord (small arrow) belongs to the AVG neuron. This is the first axon to grow into the ventral cord on the basement membrane that separates muscle and epidermis. Figure 2J–M show crosssections at 4 different positions indicated with arrowheads in Figure 2I. Note that the motorneurons are always found in the same position with respect to the interneurons. Note also that the single AVG axon which runs only about two axon diameters away from the interneuron bundle can easily be resolved in the crosssections (arrows in Fig. 2J,K). Axon Trajectories in the Brain Region The major neuropil in C. elegans is called the nerve ring. This is a large bundle of about a hundred axons that loop around the pharynx (Fig.1A). The majority of synapses is found here. The nerve ring is surrounded by Fig. 2. VISUALISING AXONS IN C. ELEGANS a number of ganglia containing cell bodies of neurons projecting their axons into the ring. There are a number of different entry and exit points and a major nerve leaving the ring on the ventral side, which eventually becomes the ventral cord (Fig. 1). The large number of cell bodies, the short distances between cell bodies and nerve ring, and the compact architecture of the ring itself made it almost impossible to discern individual features with histochemical markers. The use of GFP again changes this situation and allows one to trace axon trajectories in the brain area even in a complex staining pattern with many axons labelled. GFP expressed under the control of a glutamate receptor promoter (glr-1 promoter) was used to visualise axons in the brain region. The cells expressing glr-1 have been described (Hart et al., 1995) and their axon trajectories are known from EM reconstructions (White et al., 1986). In Figure 3 a number of axon trajectories into the nerve ring are illustrated. Figure 3A shows a side view of the brain area. The picture is a projection of several focal planes. Starting from the cell body, the AVJL axon can be followed as it projects anteriorly towards the nerve ring. Upon entering the nerve ring subdorsally, it abruptly turns and runs in a posterior position within the ring towards the dorsal midline where it crosses over to the contralateral side. Individual focal planes illustrating different parts of the trajectory are shown in Figure 3B–D. A dorsal aspect of the nerve ring illustrating the same trajectory is displayed in Figure 3E. Another neuron, RIML, whose cell body is close by, uses an entirely different route into the nerve ring. It first projects ventrally through the amphid commisure into the ventral cord, where it turns anteriorly to enter the nerve ring from the ventral side (Fig. 3F). Even for neurons that have their bodies immediately adjacent to the nerve ring, it is possible to follow their axons into the ring. In Figure 3G, the RMDL axon is shown entering the nerve ring laterally and extending into the anterior part of the nerve ring. Upon entering a process bundle, axons typically maintain their relative positions for long distances (see Fig. 2), i.e., they travel in a constant neighbourhood surrounded by the same axons, which frequently are synaptic partners. Many processes, however, make abrupt changes of neighbourhood at certain points. These points frequently are found where process bundles meet or split up (White et al., 1986). Within the nerve ring, a prominent point of change is the dorsal midline. Certain processes terminate here whereas others make abrupt changes in direction as they cross the midline. In Figure 3H, axons can be seen as they switch from a posterior position Fig. 2. Double labelling of axons in the ventral cord. A: False colour image of interneuron (CFP-labelled) and motorneuron axons (GFPlabelled) in the ventral cord. Interneurons axons are in green, motorneuron axon in red. Side view. Animals are oriented with anterior to the left, dorsal is up. Scale bar ⫽ 2 µm. B: Cross-section through the ventral cord at the position marked by the arrowhead in A. C,D: Same images after deconvolution with Huygens2. E–H: Single channel images of C,D. I: A larger region of the ventral cord with cell bodies of motorneurons visible. The single AVG axon which is the dorsalmost axon in the ventral cord can be seen running separate from the interneuron bundle (small arrow). J–M: Four cross-sections through the ventral cord at positions indicated by arrowheads in I. 51 within the ring to a more anterior one after crossing the dorsal midline. The high spatial resolution achieved by the use of GFP markers in living animals in combination with confocal microscopy allows the analysis of axonal outgrowth in areas that were almost inaccessible before. DISCUSSION Understanding the underlying logic of a complex process like the development of the nervous system during embryogenesis first of all requires a thorough description of the fully developed structure and the intermediates stages through which it develops. To this end, appropriate methods of visualisation are essential, for the developing nervous system most importantly a method to visualise nerve processes at high resolution. For the nematode C. elegans, a uniquely detailed description of the adult nervous system is available in the form of EM reconstructions (White et al., 1986). This information serves as a basis to interpret the results of various experimental manipulations, most notably the inactivation of individual genes that play a role in various aspects of neuronal development. EM reconstructions were used to analyse the defects in the first neurological mutants isolated by Brenner and colleagues (Brenner, 1974). However, since they are very time-consuming and require extraordinary experience, they were the method of choice only for a small number of researchers and a few selected experiments. Many researchers had to limit themselves to an analysis at the light microscopic level with its reduced spatial resolution. This imposed certain constraints on the analysis especially in particular regions of the nervous system like the brain with its complex and compact architecture. As a consequence, certain aspects of neuronal development like guidance of axons in local environments or synaptic target recognition were difficult to analyse. The introduction of the Green Fluorescent Protein (GFP) as in vivo marker was a technical revolution for cell biologists. In C. elegans, with its small size and transparent body, GFP tagging can be used to follow developmental processes in the intact animal with high temporal and spatial resolution. Microscopy and imaging techniques have also been improved significantly over the last years. The introduction of confocal microscopy techniques together with improved image processing algorithms enable us now to push the resolution limits of the light microscope. New Avenues for the Analysis of Neuronal Development in C. elegans Distances within the nervous system of C. elegans are small enough, so that axons do not have to branch extensively in their target area to contact different synaptic partners. Axons develop local swellings and establish synapses with neighbouring processes within process bundles (White et al., 1986). As a consequence, these bundles are highly ordered structures where individual axons are found in fairly invariant positions within the bundle. The local signals that are used by axons to navigate within an axon bundle as well as the signals that initiate synapse formation with particular neighbours are essentially unknown. Identifying these signals was difficult, since the only way to determine 52 H. HUTTER Fig. 3. VISUALISING AXONS IN C. ELEGANS the position of a particular axon within a process bundle were EM reconstructions. The results presented here illustrate that it is now feasible to determine the relative position of axons within a process bundle at the light microscopic level, facilitating future studies of local guidance signals in C. elegans. A few signals that provide long-range orientation cues for axons have been identified already. One important signal for guidance in dorsoventral directions is encoded by the gene unc-6 (Ishii et al., 1992). The unc-6 gene product, netrin, is thought to form a gradient along the along the dorsoventral axis with its highest concentration on the ventral side (Wadsworth et al., 1996). Certain axons are attracted towards the ventral midline by netrin while others are repelled (Hedgecock et al., 1990). Analysis of the expression pattern revealed that netrin is also expressed by cells bordering the nerve ring (Wadsworth et al., 1996). This implies that unc-6 signalling might also be important in the developing nerve ring. Axon trajectories, however, are difficult to determine in this region of the nervous system with histochemical markers due to the compact architecture of the head ganglia and process bundles. GFP markers in combination with confocal microscopy and imaging techniques now provide access to these regions of the nervous system of C. elegans. An evaluation of outgrowth defects of axons in this area is possible now and will lead to a deeper understanding of how the wiring of the nervous system is controlled. Potential and Limitations of GFP Markers An individual GFP-labelled axon can be followed until it joins a bundle containing equally labelled axons. Beyond this point, it is impossible to unambiguously determine its further trajectory without additional information. In combination with the EM reconstruction data, however, sometimes further conclusions can be drawn. An example, where a large number of Fig. 3. Axon trajectories into the nerve ring. Interneuron cell bodies and axons labelled with GFP under the control of the glr-1 promoter. Pictures in multiple focal planes were recorded with a confocal microscope and deconvolved with Huygens2. Arrowheads point to the axons in question. A–D: Head region of an animal showing the AVJL axon projecting into the nerve ring. Side view, anterior is to the left. The AVJL axon first projects anteriorly from the cell body, enters the nerve ring subdorsally and immediately turns dorsally (arrowheads). A: Maximum intensity projection of 14 focal planes taken at 120-nm intervals illustrating the entire axon trajectory. Scale bar ⫽ 5 µm (valid for all panels except panel G). B–D: Single focal plane images. E: Dorsal view of a different animal illustrating the AVJL and AVJR axon trajectories from the dorsal side. Despite the large number of labelled axons in the ring the AVJ axons can be clearly followed as they enter the ring and stay near the posterior end of the ring while growing dorsally (arrowheads). F: The RIML neuron, whose cell body is located close to the AVJL cell body, sends its axon into the nerve ring by a rather different route. It first grows ventrally towards the ventral cord and enters the nerve ring from the ventral side. G: The RMDL cell body is located close the nerve ring. Its axon also extends into the ring but in contrast to the AVJL axon grows through the ring past the axons that the AVJL axon joins after turning dorsally (small arrow) Scale bar ⫽ 2 µm. H: Closeup of the dorsalmost part of the nerve ring, dorsal view, anterior to the left. One pair of axons growing dorsally near the posterior end of the nerve ring abruptly change directions at the dorsal midline and continue to grow on the contralateral side near the anterior end of the ring. These axons most probably are the AVEL and AVER axons, since those are the only ones among the GFP-labelled ones that are known for this kind of behaviour from EM reconstructions. 53 axons in the nerve ring are GFP-labelled, is shown in Figure 3. At the dorsal midline, a pair of axons abruptly changes its direction of growth (Fig. 3H). It is impossible to trace these axons back to their cell bodies since other labelled axons run close to them in the nerve ring. However, from the EM reconstructions we know that among the labelled axons only the two AVE axons (AVEL and AVER) exhibit such a prominent change of direction at the dorsal midline. These axons, therefore, are likely to be the AVER and AVEL axons. The observations made with the GFP reporter can serve as a valuable basis for a more detailed analysis. Doublelabelling experiments would be necessary to validate and extend the observations made with the single label. However, EM reconstructions still are the method of choice for a complete reconstruction of many different axon trajectories within a single animal in C. elegans. Although it is possible to separate GFP and CFP under optimal conditions, this is not the ideal combination for double labeling where clean separation of the signals is required. A combination of the cyan variant of GFP together with a yellow shifted variant (10Cvariant; Ormö et al., 1996) allows an easier separation with a simple microscope setup (see Ellenberg et al., 1998, 1999 for a summary of GFP variants used in double color imaging). This combination has recently been used successfully to label different subcellular structures in C. elegans (Miller et al., 1999). The overlapping exciation and emission spectra of CFP and GFP, which makes it difficult to separate them, are advantageous when two colors are needed in rapid time lapse recording experiments, since CFP and GFP can be excited and recorded simultaneously on a confocal microscope (Zimmermann and Siegert, 1998). GFP markers are also used successfully in other organisms to improve the resolution in axonal imaging and as a label for particular neurons in living tissue (Rodriguez et al., 1999). The spacial and temporal resolution that is achived with GFP enables researchers now to study in vivo the changes in the dendritic architecture in response to individual factors (Horch et al., 1999) or even the dynamics at a single synapse as it grows (Zito et al., 1999). These and many other studies will soon lead to a deeper understanding of developmental processes in the brain. Towards the Isolation of Genes Important for the Generation of Neuronal Circuits In their initial screens, Brenner and co-workers (Brenner, 1974) isolated a large number of mutants that were characterised by specific behavioural defects. The most prominent behavioural defects they scored were movement defects. The corresponding genes were called unc-genes, a reference to the ‘‘uncoordinated’’ movement of the affected animals. Since then a number of researchers have collected many mutants in genes affecting all aspects of behaviour in C. elegans with the aim of understanding various aspects of neuronal development. The collection of unc-genes now consists of 130 genes, many of which have been studied in detail. Among those, about one-third is important for various aspects of muscle development (Moerman and Fire, 1997), whereas the majority affects neuronal development and function (Riddle et al., 1997). Among the neurological 54 H. HUTTER mutants many play a role in synaptic transmission and encode ion-channels, enzymes involved in neurotransmitter metabolism, or proteins important for neurotransmitter release. A smaller number of neurological mutants have defects in various aspects of neuronal development most notably neuronal cell fate specification and axonal guidance. The fraction of unc-genes involved in this latter aspect is less than 20%, which illustrates that a purely behavioural screen is not the most effective way to identify genes important for axonal guidance. Evaluating the movement of an animal is a rather indirect way of judging the developmental state of its nervous system. With GFP-markers more direct screens for neurological mutants, even such with only subtle defects in axonal outgrowth are now possible. The ability to perform double-labelling experiments in the living animal and the high spatial resolution achieved with modern microscopy techniques will allow us to isolate and analyse genes with important roles in aspects of neuronal development that were not easily accessible so far. In particular, local guidance signals that are used only by a subset of neurons and signals that are used for synaptic target recognition can now be identified through genetic screens. Our laboratory has started to screen for mutants with defects in the fasciculation of interneuron axons in the ventral cord. We have collected a first set of mutants with defects in the navigation of these axons within the ventral cord (Hutter and Hedgecock, unpublished data). Some of these mutants show outgrowth defects that do not lead to pronounced behavioural defects and would not have been found in traditional behavioural screens or in screens that rely on histochemical markers. ACKNOWLEDGMENTS I thank J. Kaplan, D. Baillie, and A. Fire for vectors, J. Culotti for an unc-129::GFP strain, S. Liebe and U. Tauer at LEICA Microsystems Heidelberg GmbH for their help with taking images at their LEICA SP confocal microscope, G. Giese for an introduction to the Huygens system, and I. Wacker for critical reading of the manuscript. REFERENCES Brenner S. 1974. The genetics of Caenorhabditis elegans. Genetics 77:71–94. Chalfie M, Tu Y, Euskirchen G, Ward WW, Prasher DC. 1994. Green fluorescent protein as a marker for gene expression. Science 263:802– 805. Ellenberg J, Lippincott-Schwartz J, Presley JF. 1998. Two-color green fluorescent protein time-lapse imaging. Biotechniques 25:838–846. Ellenberg J, Lippincott-Schwartz J, Presley JF. 1999. Dual-colour imaging with GFP variants. Trends Cell Biol 9:52–56. Hart AC, Sims S, Kaplan J. 1995. Synaptic code for sensory modalities revealed by C. elegans glr–1 glutamate receptor. Nature 378:82–85. Hedgecock EM, Culotti JG, Hall DH. 1990. The unc–5, unc–6, and unc–40 genes guide circumferential migrations of pioneer axons and mesodermal cells on the epidermis in C. elegans. Neuron 2:61–85. Hedgecock EM, Culotti JG, Thomson JN, Perkins LA. 1985. Axonal guidance mutants of Caenorhabditis elegans identified by filling sensory neurons with fluorescein dyes. Dev Biol 111:158–170. Heim R, Tsien RY. 1996. Engineering green fluorescent protein for improved brightness, longer wavelength and fluorescence resonance energy transfer. Curr Biol 6:178–182. Horch HW, Kruttgen A, Portbury SD, Katz LC. 1999. Destabilization of cortical dendrites and spines by BDNF. Neuron 23:353–364. Ishii N, Wadsworth WG, Stern BD, Culotti JG, Hedgecock E M. 1992. UNC–6, a laminin-related protein, guides cell and pioneer axon migrations in C. elegans. Neuron 9:873–881. McIntire SL, Garriga G, White J, Jacobson D, Horvitz HR. 1992. Genes necessary for directed axonal elongation or fasciculation in C. elegans. Neuron 8:307–322. McIntire SL, Jorgensen E, Kaplan J, Horvitz HR. 1993. The GABAergic nervous system of Caenorhabditis elegans. Nature 364: 337–341. Mello C, Fire A. 1995. DNA transformation. In: Epstein HF, Shakes DC, editors. Caenorhabditis elegans: modern biological analysis of an organism. San Diego: Academic Press. 48:451–482. Miller DM, Desai NS, Hardin DC, Piston DW, Patterson GH, Fleenor J, Xu S, Fire A. 1999. Two-color GFP expression system for C. elegans. BioTechniques 26:914–921. Moerman DG, Fire A. 1997. Muscle: structure function, and development. In: Riddle DL, Blumenthal T, Meyer BJ, Priess JR, editors. C. elegans II. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press. p 417–470. Ormö M, Cubitt AB, Kallio K, Gross LA, Tsien RY, Remington SJ. 1996. Crystal structure of the Aequorea victoria green fluorescent protein. Science 273:1392–1395. Riddle DL, Blumenthal T, Meyer BJ, Priess JR. 1997. C. elegans II. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press. Rodriguez I, Feinstein P, Mombaerts P. 1999. Variable patterns of axonal projections of sensory neurons in the mouse vomeronasal system. Cell 97:199–208. Siddiqui SS, Culotti JG. 1991. Examination of neurons in wild type and mutants of caenorhabditis elegans using antibodies to horseradish peroxidase. J Neurogenet 7:193–211. Snyder DL, Schulz TJ, O’Sullivan JA. 1992. Deblurring subject to nonnegativity constraints. IEEE Trans Sign Proc 40:1143–1150. Sulston JE, Horvitz HR. 1977. Post-embryonic cell lineages of the nematode Caenorhabditis elegans. Dev Biol 56:110–156. Sulston JE, Schierenberg E, White JG, Thomson JN. 1983. The embryonic cell lineage of the nematode Caenorhabditis elegans. Dev Biol 100:64–119. Wadsworth WG, Bhatt H, Hedgecock EM. 1996. Neuroglia and pioneer neurons express UNC–6 to provide global and local netrin cues for guiding migrations in C. elegans. Neuron 16:35–46. White J, Southgate E, Thomson J, Brenner S. 1986. The structure of the nervous system of the nematode Caenorhabditis elegans. Phil Trans R Soc Lond B. 314:1–340. White JG, Southgate E, Thomson JN, Brenner S. 1976. The structure of the ventral nerve cord of Caenorhabditis elegans. Phil Trans R Soc Lond B 275:327–348. Zimmermann T, Siegert F. 1998. Simultaneous detection of two GFP spectral mutants during in vivo confocal microscopy of migrating Dictyostelium cells. Biotechniques 24:458–461. Zito K, Parnas D, Fetter RD, Isacoff EY, Goodman CS. 1999. Watching a synapse grow: noninvasive confocal imaging of synaptic growth in Drosophila. Neuron 22:719–29.