Worm Breeder's Gazette 16(1): 24 (October 1, 1999)

These abstracts should not be cited in bibliographies. Material contained herein should be treated as personal communication and should be cited as such only with the consent of the author.

Neuroanatomy: A second look with GFP reporters and some comments

Oliver Hobert1, David H.Hall2

1 Columbia University, College of Physicians & Surgeons, Department of Biochemistry, New York, NY
2 Albert Einstein College of Medicine, Department of Neuroscience, Bronx, NY

For everyone who deals with the characterization of expression patterns in the nervous system, the truly impressive paper of White et al., 1986 ("Mind of the Worm") serves as the ultimate source of knowledge. While the largely invariant neural cell body positions described by Sulston et al. are an essential tool in the identification of neurons, the axon morphologies described by White et al. greatly facilitate the identification of a given neuron.

In the seventies, White et al. used electron microscopy to delineate the axonal trajectories of each neuron in the nervous system. Now, GFP reporter gene technology presents an alternative method for describing axonal anatomy. In the vast majority of cases, these GFP reporter gene studies have nicely confirmed the EM analysis. We describe here how GFP reporter gene studies helped to resolve ambiguous results from the EM analysis, confirmed presumptive assignments or, in some cases, points to differences to the EM analysis by White and colleagues.

However, GFP reporter axonal studies are clearly not perfect. In at least one case (mec-7::gfp), several laboratories have observed GFP reporter gene inducing defects in the axonal branching patterns of given neurons. In the mec-7::gfp case, reporter gene-induced defects were obvious given the comparison to other staining methods of the touch neurons, and also because the defects were highly variable. As such, reporter gene studies need to be taken with caution. GFP data should simply help to point to ambiguities, which researchers should use to their own judgement.

We have assembled below some neuroanatomical data that we and others have obtained using GFP methods and compare them to the White et al. data.


GFP reporter studies suggest that the ventral cord axon of this tail motor neuron does not go into nerve ring as shown by White et al., 1986. Instead, it appears to terminate shortly before the vulva. We made this observation with an unc-47::gfp (McIntire et al. 1997) in an unc-30 mutant background (to eliminate D-type motor neuron staining)(Hobert et al., 1999) and by examining in wildtype animals an unc-47::gfp deletion derivative that is expressed in DVB, but not D-type motor neurons (provided by Yishi Jin). The DVB-homologous motor neuron in Ascaris also appears to terminate in the posterior body region (C.Bargmann, pers.comm.). This observation differs from the DVB-ascribed axon in the nerve ring described by White et al.; we discuss this further below in the section about PVT.


White et al. noted that the axon of the PVT tail neuron presumably terminates somewhere in the posterior part of the ventral nerve cord. Given technical circumstances, they pointed to the tentative nature of this assignment. Using a zig-2::gfp reporter gene fusion, we found that the axon of this tail neuron does not appear to stop in the anterior half of the animal; instead it migrates into nerve ring, where it branches (Aurelio and Hobert, unpublished). This morphology is highly reproducible. Hall and Russell (1991) also cite data saying that PVT reaches the nerve ring and branches there. This data was supplied by John White prior to publication of "Mind of the Worm." It is possible that the final axonal shape reported in White et al. is different because of a later switch in assignment. Anterior processes of DVB and PVT could have been mistakenly switched due to their neighboring positions in the posterior ventral cord. In tracing across many serial sections, the possibility of a switch is unwelcome, but not impossible. This is especially true for cell types which lack a bilateral homologue to provide an internal reference for comparison. This switch is even more plausible with the knowledge that the projection of these two processes within the posterior ventral cord was only traced once (animal "N2Y"; see Figure 1A of the Appendix). No one else has thoroughly reconstructed that portion of the posterior ventral cord for comparison on PVT axonal shape. This switch still fails to explain why the switched anterior process (shown as "DVB") lacks any branching. This may be due to some variability in branchiness among the three nerve rings in animals "N2T", "N2U" and "JSH." An unbranched version may have been chosen for the final paper. (Hall and Russell, 1991 tail reconstructions benefited from multiple rounds of comparison with data from Lois Edgar, John Sulston and John White, to resolve confusions of this type.)


BDUL/R are interneurons that send a process anteriorly into the nerve ring (White et al., 1986). However, they do not appear to be monopolar, but bipolar. A posteriorly directed process can be seen with zig-3::gfp (Aurelio and Hobert, unpublished) and punc-53::gfp (N.Pujol, pers.comm.). Way and Chalfie (1988) have also observed a bipolar morphology using Nomarski optics on clr animals.


White et al. stressed the tentative nature of the assignment of the axon PQR tail neuron as terminating in the posterior body region (see Appendix). This assignment was confirmed by Peter Swoboda who used gcy-32::gfp (Yu et al., 1997) to visualize this ciliated sensory neuron in the tail of the animal. The PQR cell body is in the lumbar ganglion. The dendrite goes posterior and ends in the pseudocoelomic space. The axon enters the ventral nerve cord, goes anterior and usually ends just posterior to the vulva (in ~ 2/3 of the cases). In roughly 1/3 of the cases, the axon ends somewhere in the middle between vulva and rectum, and in ~ 2 % of the cases the axon ends anterior to the vulva (P.Swoboda, pers.comm.). These observation were made at young to mid adult stages.


The IL2 axon morphologies seen with DiO labeling differ from those in White al., 1986. Liz Ryder and Ralf Baumeister noted that the dorsal and lateral IL2s branch and look much more like the IL1 morphologies that are shown in White et al., 1986. While it is possible that the morphologies of the IL1 and IL2 neurons might be switched, J.White points out that differences in branching structures were seen from animal to animal; yet the synaptic connectivities were constant (White, pers.comm.). Since the dorsal and ventral IL1s as well as the dorsal and ventral IL2s have similar patterns of synaptic connectivities it indeed appears unlikely that dorsal IL1s and IL2 might have been switched.


Using an odr-2::gfp fusion provided by Cori Bargmann's lab Bruce Wightman examined PVP axonal morphology in adults and observed that both PVP axons form branches at the vulva, similar to those formed by the HSN axons at the same locations. The branches raise the possibility that the PVP axons may form synapses with egg-laying muscles. J.White notes that this observation is puzzling, since all neuronal processes in the vulval region have been followed; while PVP branching might have been missed, it is possible that PVN and PVP processes might have been switched along the ventral nerve cord (White, pers.comm.). White et al. stress in their paper that a complete series from head to tail has never been reconstructed from a single animal; instead processes were carried from one series to another based on relative position within the cord and synaptic behaviour (J.White, pers.comm.), thus making these kind of switches possible.


As noted by White at al. in the Appendix of their paper, the axons of the ALA, CAN and PVD neurons run together alongside the excretory canal. Two of them terminate at the anus, and the third enters the lumbar ganglion and synapses onto PVC. The latter has been tentatively assigned as ALA (White et al., 1986). This tentative assignment was confirmed using GFP and antibody staining. Nathalie Pujol looked at two different strains, pceh-17::gfp and punc-53::gfp to examine ALA axons. Both strains show that the ALA axons stop after the anus, and make a little foot where they stop. Using FMRFamide antibodies, Whightman et al. (1996) also show that the ALA axon terminates posterior to the anus in the lumbar ganglion.


In White et al., the sublateral nerves, lying out of view, far away from the primary nerve cords, were not closely examined over their full lengths at high resolution. They have never been fully reconstructed at the EM level. Currently, Jim Rand, Janet Duerr and Dave Hall have been looking at the sublaterals by several methods for evidence of synapses (see previous WBG articles and posters at the 1997 Int'l meeting and 1998 East Coast meeting). GFP reporters and antibody staining are being used to discover the relative lengths of sublateral projections, and the frequency of their synaptic swellings. If people wish to share any data they might have in this regard, please contact O.Hobert at or38@columbia.edu. The eventual goal is to have all this data integrated into AceDB/CeDB

Acknowledgements: Thanks to many members of the worm community for general comments, John White and Cori Bargmann for specific comments, Eileen German for her technical help and P.Swoboda, B.Wightman, N.Pujol, R.Baumeister and L.Ryder for specific comments and supply of data.


Hall and Russell, 1991, J. Neurosci. 11(1): 1-22; Hobert et al., 1999, Development 126(7): 1547-62; McIntire et al., 1996, Development 122(2): 671-82; White et al., 1986, Philos. Trans. R. Soc. Lond. (Biol.) 314: 1-340; Way and Chalfie, 1988, Cell 54(1): 5-16; Yu et al., 1997, Proc Natl Acad Sci 94(7): 3384-7