The Posterior Nervous System of the Nematode Caenorhabditis elegans Citation Hall, David Howard (1977) The Posterior Nervous System of the Nematode Caenorhabditis elegans. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/sx21-x768. https://resolver.caltech.edu/CaltechTHESIS:12092025-220135147 Abstract Two young adult C. elegans have been serially sectioned and reconstructed from the tail tip forward through the anterior end of the pre-anal ganglion. Thirty-nine neurons can be identified in the tail, twelve cells in each lumbar ganglion, twelve cells in the pre-anal ganglion, and three cells in the dorso-rectal ganglion. Each cell in the tail can be reproducibly identified on the basis of a set of morphological features, including cell body position, fiber projections, fiber size, and cytoplasmic appearance. Eleven neurons in each lumbar ganglion are bilaterally homologous. Many lumbar cells have sensory dendrites in the tail. Two pairs of lumbar cells which lack sensory dendrites are prominent interneurons in the synaptic interactions of the tail. Virtually all synaptic contacts in the tail are found in the pre-anal. ganglion. Most synapses involve lumbar fibers and fibers from cells whose cell bodies lie anterior to the reconstructed region. Pre-anal ganglion cells themselves are relatively minor participants in these synaptic interactions . A complete connectivity matrix has been constructed for both animals, involving about 150 synapses in each case. Certain cells make repeated contacts with one another (up to thirteen contacts) in both animals. Other instances of non-reproducible synapses are found, usually involving one contact in one animal and none in the other. No self-synapses are observed, but sensory cells frequently synapse onto their bilateral homologues. Homologously paired cells make similar sets of synaptic contacts. One class of reciprocal synapse formation is found. Eighty per cent of the contacts are dyadic, with one pre-synaptic cell and two post-synaptic ones. Ten per cent of the contacts are triadic; the remaining ten per cent are apparently conventional synapses with a single post-synaptic element. Each dyadic synapse generally involves three different types of neurons - none homologous to another - such that A → B C . Each type of pre-synaptic neuron (A) contacts only a few preferred pairs of fibers (B,C). Most dyadic contacts are involved in multiple routes of information flow, such that A → B C and, elsewhere, B → C. The formation of dyadic synapses appears to follow strict rules which may reflect important factors in the development of the nervous system. Most synaptic interactions can be included in a simple wiring diagram by which information flows from sensory cells through multiple routes to converge on a pair of interneurons which project forward into the ventral cord. Positional information is used to identify three pairs of interneurons which are important both in ventral cord synaptic patterns and in the synaptic interactions of the pre-anal ganglion (White et al., 1976). The nematode's behavioral responses to sensory stimulation of the tail or the head are analysed combining known circuitry of the ventral cord, the tail, and the head. Many neurons in C. elegans are derived after hatch from stereotyped sets of cell divisions (Sulston, 1976). Cell body positions and detailed morphologies are used to identify the cells of the lumbar ganglia which are derived from post-hatch cell divisions (in collaboration with Lois Edgar, John White, and John Sulston). Many of these cells are involved in sensory transduction in the adult tail; some as supporting cells of the phasmids, others as ciliated sensory neurons with synaptic output in the pre-anal ganglion. Six cells in the pre-anal ganglion are derived from the post-hatch cell divisions of the most posterior pair of precursor cells of the ventral cord (Sulston, 1976). Daughter cells of homologous precursor cells which enter the anterior ventral cord are known to form particular classes of motoneurons (Sulston, 1976; White et al., 1976). Of the six pre-anal ganglion lineage cells (identified by their cell body positions a la Sulston, 1976), some do become motoneurons, possibly of the proper classes for their lineage. However, two of these cells definitely do not become motoneurons in the adult nematode, violating the general pattern for their lineage. Seven sensory neurons are identified in the tail. Several different modes of sensory transduction appear to be utilized . Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: (Biochemistry) Degree Grantor: California Institute of Technology Division: Biology Major Option: Biochemistry Thesis Availability: Public (worldwide access) Research Advisor(s): Russell, Richard L. Thesis Committee: Unknown, Unknown Defense Date: 9 July 1976 Record Number: CaltechTHESIS:12092025-220135147 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:12092025-220135147 DOI: 10.7907/sx21-x768 ORCID: Author ORCID Hall, David Howard 0000-0001-8459-9820 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 17790 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 10 Dec 2025 21:30 Last Modified: 10 Dec 2025 22:43 Thesis Files PDF - Final Version See Usage Policy. 50MB Repository Staff Only: item control page

THE POS'rERIOR NERVOUS SYSTE~1 OF THE NEMATODE CAENORHABDITIS ELEGANS Thesis by David Howard Hall In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy California In s titute of Technolo g y Pasadena, California 1977 (sub1aitted July 9, 1976) ii ACKNOWLEDGEMENT Working with the Russell group has been a rewarding experience. My fe llow 'Tades offered friendship an4 support which helped to carry me through my many dubious ventures until this project finally began to take form. I wish to thank Dick Russell, who provided me the opportunity to explore the biology of the nematode and gave me the freedom to search for the approach which best suited my psyche. Randle Ware was instrumental in designing the methods and equipment for doing serial section reconstructions, and he gave me valuable assistance in learning the ropes. Dave Clark, Lauren Hollen and Margo t Szalay shared the many day to day drudgeries required to keep my experiments going. I also wish to thank Merri l Bernstein for typing this manuscript . Loi s Edgar and John White were very kind to share their data with me . My collaboration with Lois was very enjoyable , despite the frantic pace which we adopted. Many of my happiest hours were devoted to collaborative efforts to isolate and conquer the ultimate mutant . Among the 'T ades who shared these times were Randy Ca ssada , Carl Johnson , Ed Hedgecock, Jim Rand, and Joe Culotti. Our group effects were especially pleasing, and it is g ratifying to see them begin to pay off . I would a l so li ke to thank Joe and Mimi Culotti for iii introducing Nancy and me to the world of birding, which helped us both to retain our sanity during the writing of this thesis. Finally, I would like to thank my wife, Nancy, for supporting me through my many traumas. She has put up with the strange working hours and all of the moods which resulted. She has shared my successes and my failures, and she has joined me in my searches for new experiments, new games, and new places to explore. Nancy has given me the encouragement to make this thesis possible, and I gratefully dedicate this work to her. j_v ABSTRAC'I' Two young adult f~ elegans have been serially sectioned and reconstru c t ed from the tail tip forward through the anteri or end of the pre-anal gangl ion. can be Thirty-nine neurons identified in the tail, twelve cells in each lumbar ganglion, twelve cells in the pre-anal ganglion, and three cells in the dorso-rectal ganglion. Each cell in the tail can be reproducibly identified on the basis of a set of morphological features, including cell body position, fiber projections, fiber size, and cytoplasmic appearance. Eleven neurons in each lumbar gang lion are bilaterally homologous. th e tail. Many lumb ar cells have sensory dendrites in Two pai rs of lumb ar cells which lack sensory dendrites ar e prominent interneurons in the synaptic interactions of the ta:L l. Virtually all synapti,c contacts in the tail are found in the pre-anal. ganglion. Most synapses involve lumbar fibers and fib e r:3 from cells whose cell bodie s lie anterior to the reconstructed r egion. Pre-anal ganglion cell s themselves are relatively minor participants in the se synaptic interactions . A complete connectivity matrix has been constructed for both animals, involving about 150 synapses in each case. Certain cells make repeated contacts with one another (up to thirteen contacts) in both anima ls. Other instances of non-repro ducible synapses are found, usually involving one V contact in one animal and none in the other. No self- synapses are observed, but sensory cells frequently synapse onto their bilateral homologues. Homologously paired cells make similar sets of synaptic contacts. One class of reciprocal synapse formation is found. Eighty per cent of the contacts are dyadic, with one pre-synaptic cell and two post-synaptic ones. Ten per cent of ~he contacts are triadic; the _ remaining ten per cent are apparently conventional synapses with a single post-synaptic element. Each dyadic synapse generally involves three different types of neurons - none homologous to another such that A~ B c· Each type of pre-synaptic neuron (A) contacts only a few preferred pairs of fibers (B,C). Most dyadic contacts are involved in multiple routes of information flow, such that A+~ and, elsewhere, B + C. The formation of dyadic synapses appears to follow strict rules which may reflect important factors in the development of the nervous system. Most synaptic interactions can be included in a simple wiring diagram by which information flows from sensory cells through multiple routes to converge on a pair of interneurons which project forward into the ventral cord. Positional information is used to identify three pairs of interneurons which are important both in ventral cord synaptic patterns and in the synaptic interactions of the vi pre-anal ganglion (White et al., 1976). The n ematode I s behavioral responses to sensory stimulation of the tail or the head are analysed combining known circuitry of the ventral cord, the tail, and the head. Many neurons in C. elegans are derived after hatch from stereotyped sets of cell divisions (Sulston, 1976). Cell body positions and detailed morphologies are used to identify the cells of the lumbar ganglia which are derived from post-hatch cell divisions (in collaboration with Lois Edgar, John Wlfi. te, and John Sulston). Many of these cells are involved in sensory transduction in the adult tail; some as supporting cells of the phasmids, others as ciliated sensory neurons with synaptic output in the pre-anal ganglion. Six cells in the pre-anal ganglion are derived from the post-hatch cell divisions of the most posterior pair of precursor cells of the ventral cord (Sulston, 1976). Daughter cells of homologous precursor cells which enter the anterior ventral cord are known to form particular classes of motoneurons ( Suls ton, 19 76; White et al. , 19 76) . Of the six pre-anal ganglion lineage cells (identified by their cell body positions a la Sulston, 1976), some do become motoneurons, possibly of the proper classes for their line age . However, two of these cells definitely do not become motoneurons in the adult nematode, violating the vii genera l pattern for their lineage . Seven sensory neurons are i dentified in the t ail . Several d i f ferent modes of sensory transduct i on appear to be ut il ized . viii TABLE OF CONTENTS Ackn ow l edgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii . . . . . . . . . . . . . . . . . . ........................... . iv Abstract Tab le of Contents ........ . .......................... . . viii List of Figures X I. Introduction 1 II. Materials and Methods Nematodes ........................... . ........... • 13 Fixation and Embedding .......................... 13 Reconstruction .................................. 14 Catalogue of Reconstructed Animals .............. 15 Mutant Selection 15 Feulgen Staining 17 III. Observations: GeneraJ. Anatomy of the Adult Hermaphrodite Hypodermi s /Cuticle 18 Dige s tiv e System 19 Body Musculature 24 Nervous System of the Tail . . . . . . . . . . . . . . . . . . . . . . 27 Sensory Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 IV. Observation s : Detailed Nervous Anatomy of the Adult Hermaphrodite Tail Cell Identification: Neurona l Morpho l og ie s and Position s ........ ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Lumbar Ganglion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 ix Pre-Anal Gan g lion .............. . ..... ...... 53 Dor sore ct al Ganglion ................ .... ·~. 58 Ant er ior Ce ll s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 Synaptic Organization of the Pre-Anal Ganglion . ........................... ... . . .... .. . 66 Further Indentification of Tail Neurons by Fiber Position and Lineag e ........................... 95 Interneurons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Pre -An a l Ganglion Lin eage ...... .. .......... 101 Lumbar Lineage ....... . ..... . ...... ... ...... 11 5 V. Di s cuss i on Compar ative Anatomy ............................. 118 Chemosensory Transdu ction in t he Tail .... . .... .. 119 Mechanosensory Transduction in the Ta il .. ....... 122 Deve lo pment of th e Tail : Change s After Hatch .... 125 Deve l opment o f the Tail : Rules of Fiber Growt h and Synaps e Formation 1 28 Functional Analy sis of the Pre-Anal Ganglion Circui try 140 Su.,'Timary . . ......... .. . .......... ·. ............. .. ..... . . 157 References 160 Appe'."ldix I 168 X FIGURES 1. Nematode Tail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2. Feulgen/ Schemat ic tail/ Serial sections ......... 29 3. Synapse distribution .. . ......... . ................ 34 Lt. Synapses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 5. Phasmid ........... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 6. Lvmbar panels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.19 7. P/\.G panels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 6 8. DRG panel~-; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 9. Cell body positions ... .. .............. .. ......... 64 10. Lwnbar/ P/1.G cornrn issurc s .............. . ........ .. .. 68 11. Lumbar/PAO commissur es ........ . . . . . . . . . . . . . . . . . . . 70 12. PAG cross-section s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 13. Synapse chart ... 47 x 47 ............... . ........ 75 IL!. Gap junc tions ........................ . ........... 79 15. Simplified synapse chart ... 23 x 23 ............. 82 16 . Synaps e ch 2rt / Homologous cells combined. . 8 x 8 . 85 17 . Preferred dyads ... 8 x 8 x 8 .................... 89 18. Spatial config uration of synapses on alpha fibers. 19 . Fiber positions in ventral cord . . . . . . . . . . . . . . . . . . 99 20 . PAG cell body positions - Lineage ............... 104 21. Motoneuron types - Morphologies . . . . . . . . . . . . . . . . . 106 22. Lineage cell divisions . . . . . . . . . . . . . . . . . . . . . . . . . . 108 23. \Yiring d:Lagrarns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1Lt8 94 1 I. INTRODUCTION In the complex behavioral array of a high er organi sm it seems inevitable that each element must be the con se quenc e of a definite pattern of neuron a l interactions . However , in mo s t cases relatively little is known about the underlying pattern, and indeed the very complexity of higher nervous systems most often precludes a detailed analysis of this pattern, either in functional terms or in term s of the development of it s component synaptic contacts. In consequence, many neurobiolo g ists have turned in stead to simpler invertebrate nervous systems , where individual neurons can often b e r e liably identified and where the behavioral roles and d e velopmental ori gins of each can be reproducibly studied. Th e very variety of systems which have been chos en for these purposes testifies to the notion that no one of them is ideally suited to answer al l of the types of behavioral and developmental questions which arise . Indeed, i t appears at the moment as thou gh th e chosen system s divid e themselves into two broad class es , 1) a group with relatively larse , imp alable cells, in which isolated ganglia contain cell s whi ch are few enough in nu2be r and sufficiently identifiable to permit an exhaustive electrophysiological study of the beh avior rol e of each cell, and 2) a group which are smal l, rapidly reproducing, and ge netically we ll suited for mutational analysis of neural developme nt . The k in ds of informat ion currently 2 obtainable from these two classes of systems are complementary, although it i s certainly to be hoped that greater overlap will deve lop in time. Among systems of the first class, one of the earliest to be studied was the crayfish, in which Wiersma first reco gnized the occurrence of individual command interneuron s (Wier sma , 1952; Wiersma and Ikeda, 1964). Eventually Wiersma and his colleagues were able to catalogue over one hundred types of crayfish interneurons, each with a characteristic behavioral role (Wiersma, Ripley and Chri stensen , 1955; Wiersma and Hughes, 1961; Wiersma and Bush, 1963). • These ear ly studies were particularly important for establishing the extent to which interneurons could be specifically involved in some behaviors but not in others. Another system of the first class is the sea hare Aplysia, and more specifically its pari eto visceral ganglion, where Hughes and Tauc (1 962) , Kandel et al. (1 96 7) and Strumwasser (1967) have determined the constancy of cell number and position and have shown each morphologically identifiable cell to have a defined set of electrophysiological properties. Of special importance in this case has been the demonstration of cells with endogenous rhythmic properties appropriate for timing behavior (Strumwasser, 1963, 1968 ). Still another system wi t h large, impalablc cells is 3 the segmental ganglion of the leech Hirudo, of which each leech contains twenty-one nearly identical copi es with about 350 cells each. Each ganglion contains a complement of fourteen pairs of excitatory motoneurons and three pairs of inhibitory motoneurons . Each motoneuron innervates a territory of muscle fibers which is consistent in size and location from segment to segment (Stuart, 1970). Each ganglion also contains seven pairs of sensory cells, and each sensory cell has a receptive field along the body . surface which is consistent from segment to segment. Three pairs of sensory cells are receptive to light touch, two pair s are receptive to maintained pressure and two pairs are receptive to noxious stimuli. These fourteen cells comprise all of the rapidly conducting sensory cells of the segmental ganglion (Nicholls and Baylor, 1968). Intra- cellular recordings of combinations of these cells indicate that the sensory and motor cells of the segmental ganglion are synaptically interconnected in a stereotyped pattern (Nicholls and Purves , 1970). Stereotyped synaptic inter- connections are also found between cells in adjacent segmental ganglia. If the connectives between two segmental ganglia are severed and allowed to regrow, synaptic conn e ctions are reestablished in a consistent pattern but the function of these regenerated connectives is significantly altered from th e function of normal conn ect ives. This altered function is then obs e rved not only in th e re generat e d con n e ctives, but in all of the connectives of the experimenta l animal (Jan s en and Nicholls, 1972) . The stereotyp e d pattern s of synaptic interactions betw e en sensory cells and motoneurons in the l eech have been exploited to analys e some of the behavioral functions of the segmenta l ganglion , notably the shortening reflex ( Nicholls and Pu rves , 1972) and the control of swimming motions ( Kr i sta n, Stent and Ort, 1974ab; Ort, Ktistan and Stent, 1974). Th e l eech is a promising sy stem for exhaustive electrophysiological analy s i s of a given behavior, altho ugh it probably has too many cells to be studied to completion. Th e stomatogastri c gang lion of the lobster, which drives the lob s t e r's gastric mill, is anoth e r intere sting preparation for ne uro bio lo g ical study. Tl1is gang l ion has about thirty ce ll bodies, most of which can be impaled with relative ease, and the synaptic interconnections of these cells are known to be very standardized, as measured by simultaneous intracellular recordings of pairs of cells ( Morris and Mayn ard , 1970; Maynard, 1972; Mullaney and Selverston , 1974ab; Selverston and Mullaney , 1974; Hartline and Maynard, 1975; Maynard and Selverston, 1975) . In view of the relativ e s impljcity of the gastric mill ' s action , attempts are now in progress to make a detailed model of 5 the gang lion' s output, based s ol e l y on the p ropert ies directly me asured for ea ch c el l. If s uccessful, this mod e ll i ng should p r ovide a particu l arly detailed view of the impo rtanc e of each cell and e ach pairwise int e racti on in an ove rall behavioral pattern . In the stomatogastric ganglion , attempts are a l so being made to examine the mo rphological basis of the observe d functional con tacts betwe e n c e ll s. TI1e gangli on is structured in typical inverteb r ate fa sh ion, with a she ll of cell bodies surro undin g a c ore of syn aptic n europi l. There are no sy napti c contact s onto ceJ l bodies; all ce ll bodies and proximal fibe rs are s heat hed in glial cell s . Only the distal fibers in th e fine - textur e d synaptic n e u ropi l h a ve syn aptic in terac tion s (King, 1976a ). Individual neuron s identi fied by intrac e ll u l ar recording are known to have rath er constant cell body sizes and posit io ns , but the proj ecti ons of these cells int o the synaptic n europ il are high l y complex and variable. This vari ab ilit y , plus th e large number of synap ses in th e gangli on , (e stimate d to be one mi ll i on (King;l976a)), has g reatly increas e d the difficul ty of demons trating a morpho l ogical ba sis for t he observed fun ctio nal contacts . ~ Jong systems of the second , genet i cal l y tractable cl ass , the fruit fly, Dros op hila melan6gas ter, i s one of the most attractive ( Benze r, 1 971) . Drosophila is routj_nely 6 maintained isogenically and has an abundance of interesting behaviors. Many behavioral mutations have been obtained, and the use of genetic tricks such as mosaicism has proved especially powerful in the analysis of these mutants. With mosaics, the site of gene function can be localized to specific tissues in the fly (Hatta and Benzer, 1972), and through the use of cell marker mutants, development of the nervous system can be revealed by fate mapping (Kankel and Hall, 1975). Particularly good progress has been made in the analysis of sensory reception (Harris, Stark, and Walker, 1975; Harris, 1976) and motor output (Jan and Jan, 1976), where the number of cells or cell types involved is sufficiently small. However, the study of central pro- cessing in the Drosophila brain remains relatively difficult, since individual central neurons cannot yet be easily identified. Daphnia magna, a small crustacean, is an isogenic organism which has been studied by serial section reconstruction to examine the distribution and development of normal contacts in the visual system (Macagno, LoPresti and Levinthal, 1973; LoPresti, Macagno and Levinthal, 1973). The eye of Daphnia contains 176 receptor cells organized into t\·renty-two ommatidia, containing eight cells each. receptors each send one fiber to the optic ganglion. The The optic ganglion contains 110 neurons, five for each set of 7 ommatidial fibers. Primary synaptic contacts occur between sets of eight receptor cells and five lamina cells in a stereotyped fashion, which can be reproducibly identified morphologically. The fiber projections cf the receptor cells show a limited degree of variation in branching pattern, but the synaptic connections between receptor c e lls and lamina cells show a very strong pattern which is stable in the four anirr,als examined. There i s some variation in the numbers of specific synapse types within the general pattern. Th u s a limit ed degree of morphological variation does occur in isogenic animals raised under controlled conditi ons , but the pattern of synapse formation is highly conserved ( Macagno , LoPresti and Levinthal, 1973). In the development of the optic lamina in Daphn:La, a single recept or neuron sends a fiber to the developin g lamina cell s wh ich acts as a leader for the other seven receptor fibers. Only the lead fib e r displays a growth cone during development; the other seven fib ers follow the lead fiber by growing out in close apposition to it. The. lead fiber becomes wrapped transiently in a glial fashion by the first two laminar neuroblasts which it contact s . Ti, e laminar neuroblasts undergo differentiation after contact with the lead fiber, in the same order in which they are contacted (LoPresti, Macagno and Leventhal, 1973). These studies seem particularly important because they 8 indicate rules for fiber growth and contact formation th a t may be general. A final system of the second class is the n ematode Caenorhabditis elegans, which i s one of the s:implest organisms to possess a neurally mediated r epe rtoire of behavior. It s entire nervou s system consists of f ewer than 300 c e lls, and b ehavi oral mutations are easily obtained and genetically analyzed ( Brenner , 1974). The nervous system off. elegan s is being studied in parts, u si n g serial reconstruction techniques to detail each region. The bulk of the nervous system is in c epha lic ganglia bordering the nerve ring; the sensory input and the motor output in thi s region have been descri bed, (Ward et al . , 1975; Ware et al . 1975) and the structure of the neuropil in the nerve ring h as also been studied ( Brenner , White and associates, unpublished). The structure of the pharyngeal nervous system and that of the ventral and dorsal nerve cords h2v2 a lso recently been described, including patterns of synaptic int eract io n (Albertson and Thomson, l o 7 or ; llh l. Lve et 2. l . , 1976). _.,1 , t~ J. These ~t~dies have revealed striking constancies in the organization of the n ervous syst em in Q. elegans. Sensory and motor structures are formed by well-defined sets of cells with very stable morphologies (Ward et al., 1975; Ware et al ., 19 75 ), and indi v i dual motoneurons and inter- 9 neurons can be r e producibly identifi e d in th e ph arynx and in the ventral cord ( Albertson and Thomson , 1976; White et al . , 1976) . Furthermore, the fibers in th e ventral cord and in the sensory connectives of the snout are hi ghly organ i zed , and many can be ide ntified solely by their positions relative to one another . In addition , the synaptic interaction s are reproducible , and th e synaptic pattern s of the ventral cord permit some speculations con cerning the roles of th e identified moton e uron classes in behavior ( White, 1975). On the developm ent al side , Sulston (1 976 ) has recently developed a techniq ue for following individual ce ll divi si on s in the living n ematode by Nomarski optics , and has shown th at many motoneurons in the ve ntral cord are n ot present at hatch, but are derived in stead fro m the later cell divisions of a few prec ursor c e lls. Each of a set of thirte en pre curso r c e lls unde rgoes the same pattern of cell divisions to form thirteen similar sets of daughter c211s, a nd within each set of daughter c e lL.3, the same subset of fiv e cells become motoneurons o.f f=Lve dLfferent classes (Sul sto~ , 1976; White et al ., 1976) . '11hus the lin eage of each ce l l in the ventral cord appears to govern its fate in the adult . From th e results obt a ine d so f ar it appears that Q. ele g ans will be particularly useful for obtain in g a complete morphological description of an e n tire nervous 10 system and for examining th e relation sh ip b etween a c e ll ' s lin eage history a nd its eventual f unction. The work to be reported below is an ext e nsion of the anatomi c al ana l ysis of t he C. elegans n e rvou s system into the tai l, where a c omb1nat i on of limited c ell number and favorable geometry has made a comp let e sy n apti c ana l ysis pos s ibl e . Th e tails of many ot her n ema todes have been examined to a limit e d extent as a means of clas s ification . (M any workers use t he presence or absence of th e se n sory structures k n own as phasmids to divide the phy lum Nematoda into two classes ( Phasmidia a nd Aphasrnid i a ). Many of the aphasmidial species have other specialized tail structure s which sec r e te a s ub stance from a post erior spinneret, so th at t he tail serves as a holdfast organ ( Chitwood and Chitwood , 1950). Many of the phasmidial species have mu l tiple sets of phasrnids and ana l papillae .) Much of the earl y nematode anatomy was done on t wo r e l a t e d phasrnidial species , Asca:c:-Ls lumbr ico:L des and Ascaris rnegalocephala . Thes e lar ge paras itic n ematodes can b e s tudi ed by li ght microscopy and are n ow u seful for neu rophysio lo gica l st udi es as we ll . Hesse ( 1892 ) de scribed the se n sory endings and published a detailed drawin g of the male t ai l of As cari s ~-, but he did not dwe ll on the fema l e tail . Voltzenlogel ( 1902 ) de sc rib ed the tai l s of both species more extensively , with many fin e drawings of the male anat omy and a single 11 drawing of the female anatomy . Goldschmidt (1 903, 1 908 , 1 909 , 1910), in connection with his detailed reconstructions of the anterio r anatomies of both species , also reported briefly on t he se n sory anatomy of th e tail of Asc~ris . A comp l ete picture of the n ervous system of AscAris_has be e n put to gether by Chitwood and Chitwood ( 1950 ), based upon the work of Vo l tzenloge l and Goldschmidt . ( As will be shown below, the overall organizat ion of the Asearis tail i s quite similar to th at reported here for Q_ . e l egans .) Loo ss (1905) des cribed the a natomy of Ahcylostom a , in c ludi n g th e nervous anatomy of the female tail , and was the first to describe innervation of the phasmids by l atera l caudal nerve s . Chitwood and Chitwood (1 950) pub li shed a r eco n struction of the nervou s system of th e f emal e tail of Spironoura affina . Chit wood and Wehr (193 4) reconstructed the n e rvous system of Rhabditis terr:Lcola , ,,rhich is closely re l ated to Q_ . el egan s . Draw in g s of the femal e tail o f both spec ies are shown by Chit wood and Chi t\\Jood ( 1950). (I n both cases , th e layout of the nervou s system is very s imil ar to that of C . elegans ; the gan g lia , commissures and ne rve cords differ only in small details, as shown below ~) The work reported below on the C. el~ g ans tail extend s the previous studies to the electron microscopic level , 1·1h ere unambig uous cell id e ntific a tions c an be made and 12 where , in particul a r, syn a ptic int eraction s can b e c ata lo g ue d . Th e re s ult s co n firm the genera l conclus ions o f th e earlier s tudi e s a nd in additi on sugg est po ss ibiliti es both for th e function of th e s yn apt ic c onn e ction s o bse rved a nd for th e rule s g ove rning t he ir format i on. 13 II. MATERIALS AND ME THODS NEMATODES : Caenorhabditi s elegans ( var . Bristol, is a small free-living soil nematode . strain N2 ) These animals are self- fertilizing hermaphrodites, facilitating the use of genetic cros ses . Mutage nesis and handlir1g techniques fol l owed tho se of Brenner ( 1974) and Dusenbery et a l. (197 5 ). Worms were grown monoxenically in petri plates containin g nematode growth minimal medium (NGMM ) agar pre-s eede d with Esche~ichja coli strain OP50. Mutant stocks were stored froz en in small plastic tubes kept in liquid ni trogen . FIXATION AND EMBEDDING : Worms were grown up to a desired stage in synchronous culture . Synchrony was obtained by collect i ng n e~ly hatched animal s over a period of approximately 45 minutes. Animals were fixed when they reach egg- laying stage , ge nerally 50 hours post-hatch when grown at 2 0° c . Methods of fixation and embedd in g followed those of Ware et al . ( 1975) . Animal s were rinsed off plates and a~aesthetized for 15 minutes in 0 . 5% l-p henoxy-2-propanol in 0 . lM cacodylate, pH 7.4, for 1 to 4 hours. The tail s were then cut off with an x-acto knife before staining for 45 minutes in 1% uranyl acetate, in 0 . 05M maleate , pl! 6 . 1 . The tails were then embedded in Epon-Araldite . After 14 sectioning, th e thin sections were post-stained on carboncoated formvar girds in l ead citrate for 5 minutes (Reynolds , 1963) . RECONS TRUCTI ON: For microtomy , individual embedded tails were cut out of a flat mold an d mounted onto support block s with epoxy. Serial sections were cut with a Sorvall MT2B < u l tramicrotome and collect ed in strips on slot grids . Continuity of these series was very good , and the loss of sections was generally 3 % or less. Sectioning was done ,· perpendicular to the body axis . Most re g ions of the tail were reconstructed from 50-60 nm . sections , which best reveal synapses and commissures . Typical series were 1000- 2000 sections long . Electron microscopy was done with a Zeiss EM9, u si n g a special roll film camera designed by Randle Ware ( Ware et al ., 1975) . With this came ra , it was po ss ible to store 1 00 feet of 70mm . Rekordak microfilm and expose successive 200 fr ame series . The recon s truction of lon g fiber tracts (500 - 800 sections) was done by se ria l section cinematography , usin g methods and equipment describ e d by Levinthal and Ware (1972) and Ware et al . (1975). Both th e image combiner and the modified Vanguard M35 projector used for reconstruction were designed by Randle Ware . The gangli a were reconstructed from hi gh - magn ification prints of every section ; thi s t echn ique was require d to catalog ue every 15 synapse and to trace difficult commissures . CATALOGUE OF RECONSTRUCTED ANIMALS: Eight adult hermaphro- dites have b ee n serially sectioned over major portions of the tail . 'I 'he extent of these series is summarized- in Figure 2c. Other animals were sectioned ove r s hort e r region s . The gene ral fe a tures of each re g ion of the tail have been confirmed in four or more animals . The most detail ed reconstructions arc those of B126 and B136 , which featur e complete series of all of th e tail g an g lia and their commissures . I have been very fortunat e to collaborate with Lois Edgar , who has extensively reconstructed th e lumbar regions of anoth e r hermaphrodit e tai l. Many of my morphologica l findin gs have be en confi rmed by her work. Several mo rp ho - logical features were fir s t r ecognized by Loi s Edgar and later confirmed in my own recons truction s . I have also sectioned t he tails of wild-type ma les and some behavioral and morpho lo gi c a l mutants . One extensive seri es of 90 nm . sections has been ass embled to s tudy t he g eneral anatomy of the adu lt ma le t ai l. Attempts to sec- tion juve nil e tails hav e been unsuccessful, due to probl ems in memb ran e fixation. f\'iJ'i'.i-\N'I' SELECTION : Ethylmethanesulfonate was u sed to mutagenize young nematodes, follo wing the methods of 16 Brenner (1974) and Du senb ery e t a l. (197 5) . Mutagenized animals were th e n aJlo we d to self-fertiliz e for two gen e rations at 16°c, a nd rec ess ive mutants were selected from the second ge nerat ion pro gen y by a va r iety of methods . Behavioral mutant s insensitive to tail touch were sought among two populations ; 1) thos e which fail to migrate to th e bacteria on a half-se eded plate and 2) those which migrate to th e bacteria well . Potential mutant s of th e first class were difficult to as ses s, since th ey general ly did not have no rma l mov eme nt capabilities. Seve r a l mutants of the se cond class have been i s olated. Behaviora l mutant s unable t o defecate norm a lly have been selected by feedin g prosp e ctive mutants on a bacterial lawn permeated wi th con go red parti cl es . After 20 hour s growth on thi s s ub s trate, the animals were shift e d to 25°c an d continued to feed . After 4 hours at 25°c th e animal s were rin se d off the stain plat e , moved to a fr esh bacteri a l plate ( without stain), and allowed to f ee d at 25°c . Norma l animals defecate a ll of the con ga red stai~ in a period of 5 - 1 0 minut0s . Pro sp e ct:Lve ( 11 constipatecF ) mut ants are selected from those an i mals holdin g stain after 30 minutes on the destainin g plate. Te mperature sensitive mutants and non - ts mut ant s sho uld both b e among th ese clones. Congo red does not permeate any of the nematode 's tis s ues , no r does it appe a r to a lt er the animal ' s growth or d e velopm e nt . 17 None of our defecation mutants have been ts-conditional, and none of them have shown a complete loss of defecation function. Total lo ss of defecation behavior is presumably a lethal condition. Mutants with morphological defects in the tail, chemotaxis mutants and thermotaxis mutants have also been, isolated by other workers in our laboratory. Their selection methods have previously been described by Dusenbery et al. (1975) and Hedgecock and Russell (1975). ,· FEULGEN STAINING: Nematode cultures were grown on plates to desired stages and than washed off into test tubes with MeOH/HAc fixative (3:1). After a 15 minute fixation the worms were treated with l M HCl at 6o 0 c for 10 minutes, stained with Schiff's rea g ent for 45 minutes, destained in bisulfite for 5 minutes and mounted in xylene on glass slides. 18 III . OBSERVATIONS: GENERAL ANATOMY OF THE ADULT HERMAPHRODITE HYPODEillliIS/CUTICLE: The hypodermis secretes an impermeable cuticle which protects Q_. elegans from its environment and which al l ows the animal to maintain a substantial turgor pressure . Cuticle covers the body and lines the openings of the pharynx and the anus . Smal l openings in the cuticle are found at some sensory structures; the amphids, some cepha l ic papi l lae and the phasmids (Ward et al ., 19 75 ; Ware et al ., 1 975) . Hypoderma l cel l s are restricted to dorsa l, l ateral and ventral ridges along most of the animal ' s length (Chitwood and Chitwood, 1950; White , 1 975 ). An additional central hypodermal ridge is found above the anal region i n the tail ( Figure lb ). Posterior to the anus the anal , lateral and ventral ridges fill most of the available space . At the t i p of the tai l these ridges merge and become i n distinguishable . Similar hypodermal patterns in the nematode tail have been previously reported by Looss (1905) and Chitwood end Chitwood (1950). Hypodermal nuclei are spaced a l ong these ridges ; the hypodermis appears to be syncytial in the tail. For clarity, I have not included the locations of hypodermal nuclei in the figures. However, three posterior hypodermal nuclei are very prominent in the tail tip and they can easily be viewed in feulgen-stained ani mals as the three most posterior nuclei . The outer surface of the hypoderrnal cells in the tail 19 show frequent folded-membrane specializations~ presumably for cuticle secretion. One of these specializations is evident in Figure 12, in the hypodermal tissue adjacent to the pre-anal ganglion. Thin hypodermal sheets connect the hypodermal ridges, lying between the body muscles and the cuticle - thus forming a thin cylinder of hypodermis arouµd the whole body. There are channels along the lateral lines of the animal, where the hypodermal ridges are pulled away from the cuticle. ~hese channels are found along the length of the animal and are conspicuously enlarged in the tail (Figure lb and Figure 5). artifact of fixation. They do not appear to be an There are gland-like cells (cells 7,8) which run laterally along these channels throughout the length of the tail. Similar cell types have been observed along these channels in anterior re gi ons as we ll (White, 1975). White reports that there is a chain of these cells, and he sugges ts that they could be secreting a fluid into the channels. The phasmids are situated at the posterior ends of these channels, where th e phasmids emerge laterally to open to the exterior. The channels could be important in transducing mechanical stimuli to the posterior body regions into pressure waves, which they could direct to the phasmi d s . DIGES'I'IVE SYSTEM: C. ele g ans pumps bacteria into the 20 Figure T. (a ) Side view of hermaphrodite tail ( schematic ). Eight transverse sections are shown in each of four panels ( b) , (c), (d), and (e). The locations of these sections are indicat ed in ( a) . Tail tip is foreshortened ; the true tai l tip (posterior to the phasmids) is considerably longer. G33 , G34,G35 , G36 - Most posterior of the 36 intestinal cell s DC - dorsal cord VM - valve muscle VC - ventr a l cord PM - proctodeal muscles LC - l ateral cord P - ph asmid AC - anal cord LG - lumbar ganglia A - anus V - rectal valve VQ , DQ - ventr al a nd dor sal quadrants of th e longitudinal somatic muscles PAG - pre - anal ganglion DRG - dorsorectal gan g lion IL - in testinal DA - d:Llator ani lumen 21 (a) 3 2 5 4 DRG 8 7 6 ,.,,,,. oc G35 p IL vc PAG 10µ. f-----1 (b) Hypodermi s 3 4 0 0 0 0 0 2 5 (c) Digestive System 0 (d) Muscle 0 (e) Nervous System 0 0 Ftgure l 6 7 8 22 int esti n e through a two-bulb pharynx. The structure of the pharynx has recently been described by Albertson and Thomson (1976). The intestine extends from the posterior pharynx to the tail, occupying roughly two-thirds of the animal 's l ength . by a rectal valve. The intestine is separated from the anus The anus is lined with cuticle, as is the pharyn ge al openin g . The intestine is form ed by successive pairs of cells along its l ength . Seventeen pairs of intestinal cells have been counted by feul ge n staining. A final pair of vestigial inte s tinal cells are observe d in our reconstructed anim a ls. This final pair of cells has a typical intestinal cell cytoplasm, but they are much smaller in size and these cells form intestinal villi for a very short length in the region of the rectal valve ( Figures la,c). The r e ctal valve con sis ts of four dark staining c e lls. Chitwood and Chitwood (1 950) have reported that these cells are modified intestinal c ells in other species of nem atode . The rectal valve seems to restrict the flo w of material to several small c hannel s (approximate ly five sq uar e microns in cross - section) . The po sterio r int est ine is enlarged to roughly 100 square micron s in cro ss -s ection . This region of the inte s tin e has previously been called the proctodeum (C hitwo od and Chitwood, 1950). The anus has an internal cross - section of roughly 35 square microns. 23 Several specialized muscles are present in th e tail which probably have a role in defecation (Fi g ure s la,d). A pair of muscle cells surround the proctodeum, presumably to flatten thi s region during de fecation . A single small muscle c e ll ha s four gro ups of muscle fibers which attach to the four corners of the rectal valve. The two larges t fiber bundles of this muscle cell attach the ventral h a lf of the valve to the ventrolateral body cuticle . The smaller dorsal fib e r bundles attach to the dorsal half of the valve, but th e ir di sta l ends are not well anchored. This muscle cell probably serves to open the passage through the rectal valve during defecation. A similar muscle, found in other species of nematode s , is reported to act a s a sphincter mu s cle (Chitvrnod and Chitwood, 1950) . A larger H- shaped muscle c e ll has musc le fibers which connect the dorsal surface of the anus to the dorsolateral body cuticle. This ce 11 prob ably serve:3 to widen the anal opening durin g dsfecation . This cell has previously been called the depressor ani in other spe c ies of nematode (Voltzenloge l, 1 903 ; Loos s , 1905; Martini , 1 916). Chitwood and Chitwood report that the depressor ani is universally present i n all species of nematode s which have been examined. These four defecation musc le s send muscle arms into a common region above the v e ntral hypodermal rid ge. these musc le arms are evident in Figure 12 . ) (Some of Thes e mus cle 24 arms appear to form gap junctions among thems e lves. (There may also be gap junctions connecting muscle arms of the valve muscle and the de pressor ani with the mu s cle arms of the most posterior set of body muscles.) Neuromuscular junctions are formed by a large nerve fiber along the dorsal surface of the ventral hypodermal ridge, adjacent to the pre-anal ganglion (cell 33 in Figure 12). The association of the muscle arms with this motoneuron process is poorly preserved in my fixed animals, so the identification of these neuromuscular junctions must remain tentative. Figure la indicates the distribution of contractile elements of the valve muscle and the depressor ani . Cross- sections of the depressor ani, the valve muscle and the proctodeal muscles are evident in Figure ld. BODY MUSCULATU~ : Q. elegans moves sinusoidally on an agar plate, lying on one side of its body. Cuticular treads along each lateral line help the animal to maintain traction. Th e sinusoidal motion is generated by longitu- dinal bod J muscles, whidh are organized into four quadrants, two dorsal and t wo ventral (Fi g ure ld). The c ephalic body muscles are under dual control by ventral nerve cord motoneurons and nerve ring motoneurons. The cephalic mu s cl es move the head through comp lex searching motions as well as initiatin g the sinusoid al body motion. Th e innervation of th e cephalic muscles has previously been d es crib ed by 25 Ware e t a.l . (l or7~) . .,1 ..., The b ody muscles for the rest of the animal are controll ed by motoneurons lyin g in the ventral n erve cord (White et al., 1976). Each quadr ant of body musc l es i s served by twenty-four cells, e i g ht of which arc cephalic. The muscle c e ll s are arr ayed sequentially along t he animal's length so that most re gi ons along the animal are served by the sarcomeres of t wo muscle cells per quadr ant . The sarcomeres form a single lay e r against the cuticl e , and are obliquely striated - a type which ha s previously been called meromyarianpl atymyarian ( Martini , 1903, 1906, 1909; Chitwood and Chitwood, 1950; Bird , 1971). Each musc l e ce ll has about five sarcomeres running obliquely for most of the c e ll ' s len g th. The cytoplasmic appearance of the tail's body muscles is much the same as the de s c ript ion given by Ware et al. (1 975) . The r econstructed region of the tail is served by the l ast two muscle cells of each q uadrant. The most po sterior pair of dorsal ~uscles i s atypical, because the dorsal quadrants merge posterior to the anus ( Figure ld). A single muscle cell from the right dorsal quadrant continues poster:i.ad as a single ';quadrant '' oppos ed to the two ventral quadrants . This situation l e nds the phasmidial region of t he taj_l an app e arance of three -f old symmetry . Nematode muscles are unusual in that they send musc l e 26 arms to the n erve cords to rec e ive innervation (Schne ider , 186 6 ; Rosenbluth, 1965). Inf. elegans the ventral muscle quadrants send arms to the ventral nerve cord for innervation, while the dorsal quadrants ~end arms to the dorsal nerve cord. All of the motoneurons which signal these muscles are located in the ventral nerve cord. The dorsal nerve cord contains no cell bodies, but receives all of its fib ers by commissure from the ventral cord. The organiza- tion of the ventral nerve cord has recently been described by White et al. (19 76). In the tail, the body muscle cells send out muscle arms which travel to the nerve cords for innervation . .. Neuromuscular junctions occur at specialized musc le plates. The muscle arms of paired muscle cells divide into fingerlike processes which interdi g itate to form the muscle plate (pairs of dorsal muscle arms go to the dorsal cord; pairs of ventral muscle arms go to the ventral cord). Motoneurons form point junctions which signal two adjoining musc le fingers. Paired muscle arms also form gap junctions between themselves in these regions. The nature of these interactions suggests that dorsal pairs a.ri.d ventral pairs of muscle cells generally receive synchronous input and maintain electrical synchrony by g ap junctions. Similar muscle arm interactions have been noted in As c ar is lumbricoides (Rosenbluth, 1965; Stretton, 1976). Heisblat, 27 Bye rl y and Russel l (1976) have r e corded intra c e llularly from the muscle bellies in Ascaris and found that the many muscle cells in any given re gion along a muscle quadrant maintain c l ose synchrony , and that the two ventral (o r dorsal) quadrants also maintain synchrony at any given r egion along th e animal ' s leng;th . In the posterior regions of Q. e l egans muscle p l ates and neuromuscular junctions are distributed sporadical l y along the l engths of the dorsal and ventral cords. The innervation of the defecation musc le s is describ ed i n another section of th i s paper . Th ese special ized muscles do not form t heir neuromuscular junctions in the nerve cords , no r with t he same rnotoneurons . NERVOUS SYSTEM OP TlIE TAIL : 'l'he n e rvous sys tern of Q. elegan~ has about 300 neurons , the majority being concentrated i n s everal l oosely structured gan gl ia in the head, between the first and second bulbs of the pharynx . The nerve rin g contains some ce l l bodies , and it receives six major sensory fiber bundle s from t he sense organs of the snout . A pair of lateral ganglia , a ventra l gang li on, and a retrovesicular gang lion lie jus t posterior to the nerve ring . The ventral nerve cord extends posteriad from t h e retrovesicular gangl ion. The ventral nerve cord has 65 c e ll bodjes distributed rather eve nly along it s l ength (Fi Gure 2a) . Ti1e ventral cord term i_nates :i.n the ta:Ll j_n th e 28 Fig~~~ 2 . ( a ) Feulgen - stained adult hermaphrodite, dark staining ( compact) cell bodies only , animal is lying on ventral side ; (b) schematic view of nervous system in the tail, cell body positions are indicated , do1·sal cord is displaced to r eveal ventral c ord , three commissures connect dorsal cord to ventral cord and lumbar ganglia ; (c) eight animals have been extensively sectioned , these series are mapped to show t h e region of the tail ' s nervous system which is included . 29 (a) ( b) DC vc ~ ( C) Bl 16 Bl 17 8120 8123 8124 8126 8127 8136 30 pre-anal ganglion (PAG) which contains twelve loos el y associated cell bodies (Figure 2b ). Two commissurcs run in ventrol atera J positions from the PAG posteriad to the paired lumbar gang lia, each of wh ich cont a ins tw e lve loos e ly associated c e ll bodies. The sensory cells of the phas raids have cell bodies in th e lumbar ganglia. A small dor so - rectal ganglion (DRG) li e s in a central position dorsal to the ana l hypoderma l ridge . The DRG has thr ee cell bodies , and each sends a fiber to the PAG through a circum-rectal commi s sure. A dorsal cord (which contains no cell bodi es ) extend s a lon g the l e ngth of the body opposite to the ventral cord and rec e ives all of its fibers by cornmissure from the ventral cord. The most anterior of these commissures comes from the r etroves icular g ang lion (White, 1975). The dorsal cord terminates j_n the tail as a pa:Lr of circumfe r e ntial comm i ss ures which merge with the lurnbar-PAG commissures . A pair of caudal nerves run from the pha s mids to the lumb a r gang lia, generally in a ventrolateral position . Several nerve fibers run past the phasmids to the extre me tail tip . A pair of la tera l n e rves runs anteriad from the lumbar gangl ion. These lat eral cords are ve ry small and their i soort ance i s unclear. The disposition of the tail's nervous system is shown in Fi g ur es le and 2b. The v entra l a nd dorsal cords are closely associated wi th the ve ntr a l a nd dorsal hypoderma l ridges (Whit e , 1975). 31 Each n e r ve cord run s as a ti ght bund l e on one side of the hyp o derma l ridge . In the t a il t he ve n tral cord and th e PAG always lie on the left side of the ventral hyp o de rma l ridge, while th e dorsal cord a lway s lie s on th e right side of the dor s al hypoderm a l ridge (Fi g ur e s lb,e). Commis s ure s also run in close assoc iati on with hypodermal ti ss u e , remaining betwe en th e body mus c l es a nd the thin hypodermal sheet in their path s from cord to cord. Th e lumbar ganglia a nd the lat e ral cords l ie in the large lateral hypodermal rid ge s. The dorso-r e ctal gan g lion lies against the anal hypodermal ridge , and it s commissures run b e tween t he de press or ani and the anal hypoclermal sheet . Nerve fibers and neuronal ce ll bodies lie in clo se proximit y to muscle cells alon g the l e n g th of the an.i mal , but th e nervous ti s sue seems to be isolat ed from the muscl e tis s ue b y a baseme nt membrane (Wh it e , 1975). This basement memb r a ne i s especially evi d e nt alon g th e vent r a l cord and t h e PAG. Motoneurons make direct contact wi t h muscle at special muscle p l ates along the nerve c ords . The moto - n e urom; make p oint junctions which a p p e a r to cont a c t two musc l e fin gers s imultaneously . Th e cytoplasmic spec i al i z a- tions at t hese neuromuscular jun ct ions a re ve r y similar to t he othe r sy n a p ses in the t ail . Th e pre-synaptic motoneuron has an electron dense deposit a l ong th e inner membrane at t he site of the junc t ion, but there is no post-synaptic 32 specialization in the muscle fin ge r. Clear round vesicles, approximately 25 nm. in diameter, are always evident. Th ere are genera lly severa l motoneurons which make contacts at each muscle plat e . The pre-anal ganglion is the only well -defined reg:ion of neuropil in the tail. The PAG contains about 100 point synaps es and a few gap junction s, occurring among forty unbranch ed fib ers . Outside of the PAG there are very few neur on-neuron contacts found in the tail. ganglia have no neuropil. The lumbar The cells of the lumbar ganglia send fib e rs into the PAG to participate in synapt ic interactions . Figure 3 shows the distribution of synaptic contacts alon g the len gth of the PAG in two animals. The anterj_or boundary of this region of neuropil also forms the anterior limit of many of the lumbar fibers which take part in the se contacts (compare Figure 3 with the fib er projections shown in Figure 6). The posterior bound of the PAG neuropil also corresponds to th e posterior limit of many interneurons which proj e ct into the PAG from the anterior ventral cord . The PAG n e uropil is much simpler in design than mos t regions of invertebrat e neuropil because there is no fiber branching and a minimum of specializations of any sort . There arc very few gap junctions formed between neuron s in th e tai l, a sharp distinction from earlier findin g s in the ventral cord (White, J.975; Whit~ ~t al., 33 Fi gure 3 . ( a ) Schematic view of tail nervous system showing cell body positions of neurons . (b) and (c ) Distribution of sy~apses in lumbar and pre -anal ganglia in tvrn ani;,1als , Bl26 2nd Bl 36 . Numbers of synapses are summe d over three micron int ervals . ..., J, .) 'f SYNAPSE DISTRIBUTION Lumbar Synapses PAG Synapses 10 B126 Cf) ~ 0 0 0. 0 NMJ C >, CJ) 5 10 B 136 0 ~ - - - - - - ~ ~ ~ - ~ ~ - - - ~ ~ - - . 7J~~..z;r/~-rrrr~0 NMJ D D = Chemica l Synaps es ~ = Neuromuscu lar Junct ion s (NMJ) in Ventral Cord = Electrical Synapses F i gure 3 5 35 1976) and in the nerve ring (Ware et al., 1975). The gap junctions which do exist in the PAO are between the fibers of bilaterally symmetric lumbar sensory cells (cells 17, 18, 27-30). There are a few large chemical synapses, but the great majority of the chemical synapses are small point synapses. Many neurons in the FAG signal the same fibers repeatedly over short distances by multiple point synapses. Multiple synapses are also common in the ventral cord (White et al., 1976). Most of the synapses are dyadic, with a single pre-synaptic fiber and two post-synaptic fibers (Figure 4). Fewer synapses involve one or three post-synaptic fibers. There is always a pre-synaptic darkening along the inner membrane, but there is no postsynaptic specialization. The synapses are en passant, and the pre-synaptic element is usually enlarged locally and filled with clear round vesicles. Most synapses have 25 nm. vesicles, but the lumbar sensory fibers (17, 18, 27-30) have 50 nm. vesicles with more irregular outlines. The fibers in the PAG neuropil also exhibit nonsynaptic interactions in the form of small (approximately 0.2 micron) membrane invaginations between neighboring fibers and between fibers and cell bodies (Figure 4). These invaginations are frequent in the neuropil, uncommo~ elsewhere in the tail. Similar structures have been reported previously in 36 _.. . . 1, F igure t . Typical synapse types in pre - anal ganglion. Open arrows indic a te chemical synapses ; most are dyads . arrow indic ates an e l ectrical synapse . j_ntercellular "invaginations ." Thin arrows indicate Magni.fica.tion Each panel i s rough ly 1.4µ x 1.0µ. Filled ~ 55 , OOOX . '37 , ' ... ~·-, f $# •• wi, vi; .. ~ ~.•·, ,,)~ ·~.:J i,. • .•,,.. • 1\1"., • .,f pi ,. ' ~ ~ M. : • .:~ •i ·,, '\ ·-?· . ·:'::,,-· ,'J' J. 1 :2..~~:- - ~~ ,~ ¥ f••,~~ :.°',; ... ' ;.,'.() -.:· \~ \ •. "• .. -~~. ' . ·f', ' t~ i· 38 other neural systems , u sually in association with coated Landis and Reese ( 1974 ) observed simi lar pro - vesicl es . tuber ances a sso ciated with coated vesicles which are formed betwe en n eural elements in th e glomerulus of the cerebellar cortex. Henkart (1975) ob se rve d coated ve s icle invagina- tions between glial e l ements (donor) and neural elements (rece iver ) in Aplysia ganglia. Henkart reports that the frequenc y of the se invaginations increases with increases in ca++ conc e ntration (or by subst itution of co++ or La+++ for ca++) . She suggests th at these invaginations may be the morpholo g ical basis for the transcellular transfer of macromolecules. Transfer of newly synthesized protein has been shown between Schwann cells and squid giant axon by Lasek et al. ( 197 L+) . Membrane j_nvaginati.ons are numerous in the n ema to de PAG b ut are infrequently associated with vesicles. No obvious pattern was evident in the distribu - tion of these structures among the neurons of the tail . Out side of the FAG, there are very f e w synaptic contacts in th e tail . A few electrical synapses and one class of chenical synapses are found in the lumbar ganglia . The dorsal cord contain s some neuromuscular junctions , but no other synaptic interactions . No synapses have been found near the phasmlds . SE NSORY ELE ffiliNTS : J\1ost of th e hermaphrodite I s sense organs are located in the head. The anterior tip of the animal is 39 innervated by two large amphids (each containing twelve ciliated dendritic endings) and by twenty-four other specialized dendritic endings which are assorted among sixteen papillae of sj_x types. 'l'he detailed morphology of these endings and their supporting cells has previously been described by Ware et al. (1975) and Ward et al. (1975). A pair of cervical papillae, the deirids, are positioned just posterior to the nerve ring, with senso ry endings embedded in the cuticle at the lateral lines. These organs are presumed to be pressure-sensitive, as they have no external opening (Ware et al., 1975). A pair of 11 post- deirids 11 may exist in the posterior half of the animal , also having sensory endings along the lateral lines. Sulston, Dew and Brenner (1975) have observed a pair of small dopamine-staining bodies in the posterior regions which they pre sume to be associated with the post-deirids. I have observed similarl y positioned groups of cell bodies in my feulgen-stained animals (Figure 2a). My serial reconstructions of the tail have not extended to this region. The nature of these receptors (?) remains obscure. The tail contains a pair of phasmids, each having two ciliated dendritic endings. No proprioceptors have been found in the posterior regions, nor have they been found elsewhere inf. elegans (White et al., 1976). no other 11 There are anal papillae 11 in the hermaphrodite tail, although 40 there are some non - papillary sensory endings near t h e phasmids. The phasmids are very similar to the amphids and papillary organs of the head . Each phas mid consists o f two ciliated dendritic endings within an extracellular pcc k et formed by a cap cell and a pocket cell (Fi g ure 5) ( Wa re et al ., 1975; also called the socket and s heath cell s by Ward et al ., 1975 ). A wing cell provides a wrapping p rocess along the surface of each phasmid. The extracellular surround of the p r1a smidial cilia is in contact with the exterior, much like the amphidial pocket . The phasmidial cilia have fi l ame n to u s endings which protrude very slightly from the phasmidial opening ( Figure 5a) . not have striated rootlets . These cells do Electron-dense materj_al seals the cap and pocket cells together, much the same as the sealing of the amphidial cap and pocket cells (Ware et al. , 1975 ) ( Figure 5b ). The cilia invade the pocket by means of a wrapping process of the pocket cell . The general cyto - plasmic appearance of the phasmidial pocket is very similar to that of the amphidial pocket, but there is no 11 fin g er c e 11 n associated with the phasmid (Hare et al., 1975) (Figure 5c ). The cell bodies of the wing, cap , and pocket cells lie just anterior to the phasmid (cells 1- 6 ) ( Figure 9 ). The cell bodies of the phasmjdial ciliated neurons are located in the anterior ends of the lumbar ganglia ( cell:s 27-30 ) (Figure 9 ) . 41 Fi g ur e 5 Sch e m~ tic drawing (c e nter) show s a long itudinal reconstruction of th e ri g ht pha s mid. ( a -f) Micro g r a ph s sho w transver se sections throug h diffe r e nt l e vels of the ' · ct an d GaL · 1 'cip; . pnasrrt:L .L ( a;' Pha s midial opening to exterior, filam e ntous material from ciliated neurons extends into this opening ( b) Phasmidial cro s s-section shows cilia enveloped by the ca~ cell, dark material insi~e cap cell cements it to the invading pocket cell , wing cel l is wrapped around cap cell (c) P.hasmidlal cross -- section shows cilia enveloped by pocket cell, w~ng cell is wrapped aro u nd pocket cell (d) shows ba s al body of ci l ium from cell 19 , whict is outsid e of the wing c~ll (e) shows fan - like ending of cilium fro~ cell 1 9, which is emb edd e d in s ide the ½ing cell ( f) s hows extreme tail tip of tl1e nematode , dendrites of cells 17,18 ar e envs lopeo. on l y in a li ttle cuticle ancl hypodermis . Posterior tip i s to the top of the drawing , lumbar ganglion is to the botto m of the drawing . to dr aw ing . Sca l e bar app l ies only to Nicro graphs are enl a rged approximately 13,000X . C - cuticle SC - subcuticul ar channel H - hypod e rmis Pc - pocke t cell Cp - cap c el l 42 There is an asymmetric ciliat e d dendritic endj_ng emb e dde d in the wi n g cell of t he rig ht phasmid . The cilium en ds as a b road fan - li ke ele ctron- dens e ex tension (Fi g ure s 5d, e ) . This e nd ing js not exposed to the e~te r ior , lying very c l o se to the c en tral axis of the t a il . This dendrite has no striat ed roo tl et , an d has i ts c el l bcdy in t he ri g ht lumb a r gan g lion ( cell 19 ) . The correspond::_~~g cell in the l ef t lumbar gan g lion h as no dendrite (cell 20) . This asymmetry ha s b ee n confirmed in six animals , al~ay s with the same handedness . Tvw lumb ar c e lls ( cells 17, 18 ) each have a poster ior process which travel s in a bundle with the dendritic processes of t he phasmidial a nd asymmetric cilia . Thes e two proc esses a re presumed to be se n so r y dendrites , a lthough th e y do not show any specialized e ndin gs . These two process es run in vent ro l a teral p ositions past the fhasrnids a nd continue posteriad for an other 65 microns to the extreme tip of the t ai l ( Fig ur es 2b an d 5g ). In these posterior r eg i on s the tail narr ows to a fine hair-like process which unde r go es fr e quent se vere bends in th e co u rs e of the n emato de 1 s a ctivit y , e13pe ci a lly durin g reversal . Th is hair-like process bends in response t o the s li gh t es t me chanic a l stim u lj_ , and I pr es ume t h at the den dri tes of cells 17 , 18 are se n s itive to the se me ch a nical st i muli . IV . OBSERVATIONS : DETAILED NERVOUS ANATOMY OF THE ADULT HERMAPHRODITE TA IL CELL IDEN TI F ICATION : NEURONAL MORPHOLOGIES AND POSITION S Preliminary s tud ies of f e ul ge n - st a ined worm s indicated th at the neuronal cell bodies of the tail are con s tant in nwnb e r an d stab l e in ge n eral confi g u ration . The lurGbar gang lia an d the PAG are vi sible in th ese pre p ara ti ons as separate collections of c e ll bodi es . This leve l of analysis obviously c a nnot serve to demonstrate the reproducibility of individua l n e uron i dent i ficatio n s . Reconst ruction of ser ial thin se ctions of the taj l has now confirmed th a t a complete list of individual ce ll t ypes c an be identified in each adult tail , a nd that each c el l type has a rel a tive ly stable cell body position and a ve ry stable set of fiber proj e ctions . se t Great care was require d to mat ch the p roper of sensory fibers from t he l wnbar g anglia with their sy n ap tic a ll y active distal pro j ect ions in the PAG . Th e su cc ess of this effort potentiates the manipul at ion of identifi e d ne uron s in livi ng an imals , either by muta ti on or l aser in activati on . The unbunched distr i b uti on of the ce ll bodies in the t ai l makes this region particularly favorable for the se manipulations . Careful ident ifications by ce ll body pos ition also allows the recognition of daughte r cells from li neage cell divisions . Thi s sect ion is a catalo g ue of the morpho lo gica l charact e rist i cs which di stinguish each neuronal cell type in the tail . Most of the tail's cell s can be identified solely on the basis of their cell body position, cytoplasmic features , and proximal fiber projections . All supporting cells, hypodermal cells, gland c ells , muscle c el ls, g ut c ells , and most neurons can be quickly categorized on this basis . (C el l body positions of hypodermal cells, muscle cell s , and gut cells here are not shown in the followi n g fi gures for the sake of clarity.) There are some neurons, particularly lumbar neurons, which can be confused unless their distal fib e r projections are traced through the commissures to the PAG . The lumbar fiber projections are very character- istic for each cell type. The fiber positions withi n the commissures and within the PAG are so stable that each lumbar neuron can now be identified withi n the PAG without tracing b a ck to its cell body. Reconstruction of the lumbar-PAG commissures is particul ar ly difficult in sections cut perpendicular to the body axis , but I have now accomplished this task in several animals. The com.missures of the dorso-rectal ganglion are also very difficult to reconstruct in my sect ions, and I have been able to match these fibers to their ce ll bodies i n the DRG in only one anima l. However, the DRG fj _bers are easily distinguished and compared from animal to anima l within the PAO . These DRG commissures have also been traced once by Lois Edgar in another animal , and our results are in agreemen t . All of the following cell types have been found to be morphologically identical in two or more animals; especial l y in Bl26 and Bl 36 . Cell body posit ions are s urnmari zed in Figure 9, and Figures 6 , 7, and 8 contain schematic drawings of the processes of each identified cell . LUMBAR GANGLIA : The phasmidia l support cells have no anterior projections and are easily distinguished. There are three c el l types : wing cells (1,2), cap cells (3 , 4) and pocket cell s (5,6) . Their cell bodies lie in the extreme tail tj_p, posterior to the lumbar ganglia (Fi gure 9). Cells 7,8 have a gland-like appearance . They have lat e ral processes which extend along the lateral subcuticular channels . They may be analagous to similar cells noted in the anterior by White (1975). Their cell body positions are more variable than those of the 1 umbar neurons ( Figure 9) . Two pairs of cells produce the ventrolate ral cords. Cells 9 ,10 have characteristic cell body positions at the posterior li mi ts of the lumbar ganglia (Fi gure 9) . Their ven tro - lateral fibers are filled wi th tubules , and these fibers extend from the extreme tail tip anteriad throughout our sections . (The tubule-filled proj ection of cel l 9 can be see n in Figures 5b,c). Cells 11,12 have character i s tic cell body positions in the centra l lumbar ga ngl ia (F'j_g ure 9), and they have ventrolateral processes wh ic h trave l i n close associatio n with the processes of cel ls 9,10 . The processes of 11,12 are not tubu le - filled . Cells 13 , 14 lie dorsally in th e tail , very near the lumbar re g ion . These two ce lls h ave extens ive dorsolateral pro c esses , very much like th e ventrolateral pro c esses o f cells 9- 12. The proc esse s of c e lls 13 , 14 are not tubule - f ill ed , a nd they extend from t he tai l tip anteriad throug hout our sections . The followin g information about the lumb a r morphology of neurons 15 - 32 is probab l y suffi cient to identify them in the reconstruction of other adult tails . No ne of th is information is required to identify the processes of these c e ll s in reconstructions of the PAG . Some detai l s of fiber position for these ce l ls are s ummarized in Fi g ures 6, 9 , 10 an d 1 1 . Figure 6 . Lumbar Ganglion Cells Eac h panel dep::Lcts . t. a s c:,ema - 1.c view of the nervo us system of the ta il, s howing the cell body position ~, 2nd fiber projecLLons of a bilatera lly homolc gous pair of lumbar neurons . ( Exception: cells 19 and 20 are each unpair ed neurons . ) All of these cells s end processes through lumba r PAG commissures into the pre -anal ganglion ; only eight cells have anterior projection s wh i ch enter the ventral c o rd. Seven cells have posterior s ensory de nd rites. (2J.,22) have short axons in the lumbar ganglia. Two cells All cell body positJons and fiber positions are those of B136. 49 Fi ~ ure 6 50 The c ell bodies of c e lls 15,16 are located jus t posterior to the lumto.r e;ang lia proper . The extent of their fiber pro - jections into the PAG is show n in Figure 6a . The se cells have no sens ory specializations and their synapti c i nteracti on s ar e diffuse. The function of cells 15 , 16 is unclear . Cells 17,18 are presumed to be se nsory neurons, since they have nd e ndrite s n which extend to the extreme tip of the tail . The fib er projections of these ce ll s are sh own in Figure 6b. Th es e cells have extensi ve synaptic output in the FAG . Cell 19 i s th e asymme tric cilium senso ry n e uron , located in the right lumb ar gangli on . Th e c or r espon d i n g cell in the left lumbar ganglion, cell 20, ha s no sensory dendrite . Both c e lls have projections into the PAG, shown in Figure 6c . Only cell 19 h as many synaptic connections i n the PAG , but both cells project anter iad into the ventral cord . 19 j _·s Ce ll the only lumb ar sensory neuron which does se nd a process into the ventral c ord . The c ell body of cell 20 is n o t eas il y distinguished from those of cells 22 an d 24 - all 51 three cell bodies iie in simi lar dorsal positions within the left lumbar gang lion, b u t cell 20 i s generally posterior to 22 a nd 24 , an d it s process enters the lumbar- PAG commiss ure i n a more po sterior position . Cel l s 21,22 have cell bodies whic h lie dorsal l y within the lumbar ganglia (Figure 6d) . Their projections into the PAG do not partici pate in signific ant synaptic interact ions . How ever , cells 21 , 22 eac h have a short axo n which runs for a short distance anteriad a lon g the dorsal ma rgin of the lumbar gang lion . These axons become embedded i n-t he cell bodies of c e lls 23 ~24 , where they form a se ries of chemic a l synapses . Cel l s 23,24 l ie in dorsal positions with in the ltmiliar gan g li a . They have no dendrites within the lumb a r gang l ia , b ut send proce sses into the PAG which receive major synapti c inp ut there ( Figure 6e ) . The cell bodies of cells 23,24 receive dir ec t synaptic c ontact from th e proximal axons of cells 21 , 22 . Cells 23,24 are thought to be major intern e urons in the ventra l cord . The cell bodies of c e lls 25,26 a l so lie 52 in dorsal positions withj_n the lwnb ar gan g lia , very near those of cells 23 , 24 . Cel ls 25 , 26 often appear somewhat smal l er and rounder than neighboring lumbar cell bodies . These cells have no sensory de ndrites ; their projections into the PAG are shown in Figure Theh" proximal fib ers are easi ly distinguished as they enter the lumbar -- PAG commissures , since they form gap junctions there with the anterior processes of cells 9,10 , the lateral tubul e-fi l led fibers . The s i gnificance of these contacts is not c l ear . Ce l ls 25,26 are important interneurons withing the PAG . Cells 27,28 are phasmidial sensory neurons , whose dendrites form the more dorsa l cilia within the phasmids (Figure 6g ). Their cell bodies are somewhat larger than neighboring lwnbar cell bodies , and they lj_e dorsolateral l y in the lumbar ganglj_a . 'l'he se cell s have extensive synaptic output in the PAG . Their axons travel in extreme ventral positions within the lumbar-FAG cornrnissure s , under the nei ghbo ring fibers (Fi g u re 10 ). Cells 20 , 30 are phasmidial sensory n e uron s , whose dendrites form the ventral cilia with in 53 the phasmids . Their cell bodies have anterior positions within the l u_mbar gan g lia. The projections of these cells are shown in Figure 6h. They have ex t e nsive synaptj_c output in the PAG, quite distinct from that of cells 27,28 . Cells 31,32 have the mos t anterior cell bodies in the lumbar ganglia . They do not have sensory dendrit es , but they do se nd processes into the PAG a nd anteriad into the ventral cord, sh own in Fi g ure 6i . Their fibers do not t ake part in PAG synapses . The morphology of the se cells does not reveal any clear functional role . Figure 6 presents s chematic morphologies for all of the lumba r neurons which send projections into the PAG . (Cell body positions are those found in Bl36 .) All of the fiber projections have b~en confirmed in two or more animals. The cell body positions of the l umbar ganglia of both Bl26 and B136 c a n be compared in Figure 9 . PRE - ANAL GANGLION : The pre - anal gang lion cells are relatively simp le to identify in reconstructions of the PAG . Their cell bodies and their fibe r configurations within the PAG are all quite characteristic . Figure 7 displays the s chematic morp~ologies of these twelve cells jn two animals ( Bl26 and B1 36 ). age an d non - lineage cells . Th e cells are grouped into , . .1..ine - Some sp ecific features are not e asily depict e d on th e fi g ure ; these ar e listed below . Some of thes e c el ls have be en identifi e d as motoneurcns by their junctional contact s wi th musc l e ar~s from l ongitudinal b ody mus cles . Mos t of th e t we lve cells have rather sparse synaptic con ta ct s within the PAG; however , cell 43 i s a n active interne uron within the PAG . Specific fe at ures of some PAG cells include : Cel l 36 has a l arge crenated nucleus a nd a unique c e ll body po s ition at the posterior limit of the PAG. The dorsal process of cell 36 has motor output a nd run s in the DD position . Cell 37 h as a posteri or process wh ich continues in a ve nt ra l position a lon g th e ins ide surface of the an u s aft er a ll other proc esse s have commi ssured j_nto the lumbar g anglion . Cel l 38 has a dorsal process which has motor output and runs in the DA position . The ce ll ' s ve ntral proc ess run n ea r other c l ass A fibers . Cell 39 has a pro cess in the ventra l cord 55 1:.iigur e 7 . Pre - Anal Ganglion Cells Eac h panel d e picts tw o sc h emati c vi ews of th e nervous system of the tail, showing the cell body positions and fiber pro j ect i ons o f a pre-ana l g ang lion cell in two anima ls (Bl26 and B136) . The ce ll s shown in the l eft pane l s (a - f) are present at hatch in the pre - a n a l ga n g li o n . The c e lls shown i n t h e right panels ( g -1) are the daug ht e r s o f post h a tch lin eage cell divi sions . to the dorsal co rd. Four cells have commissur es Dashed lines indicate p r es umed p roject i o ns ( e . g . the dorsal cord of B1 26 has not been full y reconstruc t e d) . Two c el l s (37 , 40) have posterior proces s es which remain in lateral positions after other fibers have made circumfere ntial c ommissu r es ; t he se processes h ave not been fully reconstructed. 56 a g b h C d e f Ftg ur e 7 57 which has motor output an d runs in the VD position . Cell 40 has a rather larger cell body which typically sends out several eitra shor t processes into the PAG . Cell 41 has the largest cell body in the PAG, an unmi s takab le feature . Ce ll 42 has a distinctive posterior process which emerges from the cell body as a fl atte ned sheet . This process travels through the ri g ht FAG-lumba r co~nissure to the dorsal cord. Gap junctions with cell 9 are sometimes observed in the lumbar gang lion. The dorsal c o:cd process and the v en t ral cord proce s s both travel near other class A fibers , but cell 42 has no apparent motor output in the tail . Cell 43 is an inte rneuron . This cell body is positioned at the e xtreme right s ide of the PAG (s ee Figure 1 2 ). Cell 44 has its cell body in a central position , lying jus t dorsal to the PAG n e ur opi l . Cell 45 has its cell body displaced far to t he ri ght side of the PAG . Thi s cell body is somewhat flattened, especially at the anterior end . Ce ll 46 has a very round cell b ody at the 58 anterior end of the PAG n e uropil , and lies to the ri ght . Ce ll 47 has it s cell body anterior to the PAG neuropil . This cell body is very round and l ies to the ri gh t. The cell ' s proc es s es i n th e dorsal and ventral cords run near other class A fibers , but no motor output has been found . DORSORECTAL GANGLION : The dorsorectal ganglion has on l y th r ee cell bodies, each se ndin g a fiber into the PAG through one of the circumferential commis s ure s . The morpholo g ies of the DRG cel l s and their projections are shown in Figure 8. These data are t ak en fr om Bl26, whe r e th e circumf e rential commis s ures h ave been successfully r econ str ucted. commissure s cannot b e reliabl y traced in Bl36 . These Loi s Edgar first recon s truct e d these commissures in anot he r animal (unpublished observation s ) . her earlier r esults. My own findings in B126 confirm The fi be r from cell 33 is g r eat l y en l arged and filled with synaptic vesic l es as it crosses the ventral hypodermal ridges to enter th e PAG . This fib er 's size makes it easily di s tin g uished from the nei g h boring fib er of c e ll 35 . Ce ll 33 probably acts as a defecation motoneuron, form ing neuromuscu l ar jun cti ons with the musc l e a r ms of the defecation mu sc le s along the dorsal surface of the ventral hypod e rmal r idge . 59 Figure 8. Dor so r e ctal Ganglion Ce lls Each panel d ep icts a sc h ematic v i ew of th e n ervous system of the tail, show ing the c e ll body po sition a nd f ib e r projections of a dorsorectal ga nglion cell in 8136 . Th e same configuration of cell bodies a nd fibers has b ee n partially reconst r ucted in Bl26 and complete l y reconstruct e d in another animal by Lois Edgar . Each ce l l has a single proj e ction which e nt e rs the PAG by way of a circumr ecta l cornrnissure. ventral cord . All of these fibers travel anteriorly into the 60 r r I I ffl , .. l),.,., ,-,;, 61 ANTERIOR CELLS: Thirteen or more fibers entering the PAG from the ventral cord are presumably distal projections from a nterior neurons. Most of these fibers end blindly at the posterior limits of the PAG . One tc three others enter circumrectal comrnis s ures and continue posteriad along the anal ridge. Thirteen fibers have been identified in the PAG in both B126 a nd B136 . Each can be identified by its separate route through the PAG, its size, varicosities, cytoplasmic appearance, and synaptic connections (Figure 12). Four of these fjbers represent two pairs of major interneurons. These four fibers travel together in a central position within the PAG; a pair of larger fibers running ventral to a pair of very small fibers (Figure 19). The larger fibers have a very light cytoplasm and many swe llings filled with vesicles. The identity of these interneurons will be discussed further, but for the moment we will label these fibers as the a (large r) and 8 ( smaller ) interneurons . One of the remaining fibers is clearly a motoneurons with junctions along the ventral cord to the most posterior body muscles - this fiber runs in the "ventral B" position of White et al. (1976), and is labelled Bin subsequent figures. The fibers which enter the circumferential commissures have been labelled t, u, and v . Fib e r t passes under the DRG and continues posteriad along the anal hypodermal ridge. 62 This fiber has many varicosities along its len gth , especially along the anal ridge . this feature .) DRG . ( Lois Edgar first noted Fibers u and v seem to end blindly at the Vis present only in B136. Neither u nor v has been identified in the animal reconstructed by Lois Edgar . Fibers t and u receive significant synapti c input from sensory fibers in the PAG. None of these fibers app e ars to have any sy naptic output in the tai l . Fibers is very striking in appearan ce , containing elaborate membra nous inclus ions. This fiber h as no signifi- cant synaptic int e r a ctions a nd may not be a neuronal process. Fiber n rec e ives substantial synaptic contact from a PAG interneuron , and has no synaptic output in the tail . Three other fibers from the anterior , lab e lled p , q, and r, participate in very sparse synaptic interactions . A complete l is t of the cell s of the adult hermapl1ro ditc tai l is given in Appendi x I . The cell body pos itions in the tail are illustrated in Figure 9 , compari ng the recon s tructions of B126 and B1 36 . The relative cell body positions sh ow a certain degree of variabi l j_ty . This va ri abi lity has not endangered our identific at ions of individual cells because of the concurrent stability of fiber positions . Fib er positions are very stable and often 63 Figure 9 . Cell Body Positions in Two Animals ( B126 and Bl36) Each panel depicts a schematic view of the nervcus system of the t ail , showing the cel l body positions cf all neurons and s upporting c el ls in the tails of B126 (atove) and B136 (below). Ce ll bodies of supporting cells lie along the periphery of the lumbar gang lia . Cell bodies of one pair of bilaterally homologous neurons (cells 13,14) lie dorsally , outside of the lumbar ganglia . M - muscle cell body in ventral quadrant sensory neuron interneuron motoneuron n- other neuron \.0 Ci) '-S s:: Q-g '"lj :-1· ~ @] ::M .. .. :·r-,(: ~ I i i \ [217 rF ~ [fil !cl :·M·: ~ : .. ... : [HJ @] - ~~ ® @J [ill I \ /\fill[§_ - - 2 ~ , f u ~@] OJ [1] ~ 0~ iGIJ-2}.()§~gg r~@i,,;;,, = @! A@_?] [§]___. lg [IT] )) y §] @J 14 @ @g @] ® ~ 41 @ @ ·@ 0\ .t:- 65 more useful than cell body position in making identifications. White et al. (1976) found in the anterior ventral cord off. ~legans that fiber po~itions are stable enough to identify individual types of motoneurons and interneurons over great distances from their cell bodies. The relative sizes of these fibers are also characteristic of each cell type. Fiber positions within the PAG are stable enough to identify all cells without retracing the comrnissures. It is not possibH:; to reproduce all of these data in· these .fev.J pages. In my reconstructions I relied upon cinematographic techniques to quickly explore the routes by which each fiber makes its way through the PAG. B136 was compared to B126 in this manner, and all fibers were identified unmistakably. Corresponding pairs of lumbar cell projections generally have symmetrical placements within the PAG, and some pairs have gap junction interactions with their homologues. Commissures to the dorsal cord are identical in the two animals examined. The dorsal cord of B136 has been reconstructed extensively, covering 80 microns anterior to the commissures from the lumb ar regions. Relative fiber positions in the dorsal cord are extremely stable over the entire reconstructed region. Four members of the dorsal cord come from identified PAG neurons by way of commissures in the tail. The rest of the dorsal cord fibers (approxi- 66 mately five others) must come from c omm i ss ure s a nt er ior to the reconstructed re gion ; addit ional PAG neurons which have ventral projections could be participating in some of these commjssures . Figures 10 an d 11 d is play the ordering of fibers in the lumbar-PAG co mmj_ssures at two levels, c omparing the organization in Bl26 and 8136. Figure 12 displays the disposition of fibers and cell bodies at a position mid\1ay along the PAG in the same two anima ls. Figures 19a,b s how the positions of the important motoneurons and j_nt ern eurons at the anterior end of the PAG in both animals. These positions sho uld be compared to the organization of the anterior ventra l cord , sho~n in Figure 19c (taken from ·White et al ., 1976) . SYNAP'I'IC ORGAN I ZA'l1ION OF TEE PRE-AlJAL GANGLION l\JEUROPIL: Complete surveys o f the PAG synapses were made in B126 and B136 by examining prints of high magnification EM photo graphs of every PAG section in each an imal. All neuronal cell bodies and processes had been previously identifi ed by cinematography , so that a ll participating elements in each synapse could be noted. Occasional sect ion s were missing, and occasional sections were relatively poor :Ln quality , but there were no serious gaps in either series. A li st of synap s es was made for each anima l, noting the pos ition of each synap s e by se ctjon number. List e d synapses show 67 Piber Pos i t:i.o ns in Lumbar - PAG Cor;--,m is su:'.'es (1) Figure 1 0. Left and ri g ht lumbar-FAG c ommis sure s 2.re sho wn for Bl26 (a ) and B136 ( b) at corresponding lev e ls in ea ch animal . These cross - sections depict onl y the extr eme ventral edge of the animal ( s ee Figure 1, panel (e), crosssection 5) . Magnific a tion is about Sooox . A tracing of the fibers of each commissure i s shciwn above each micrograph . Fiber positions of homologous cells are similar from side to side , and from animal to animal . Each cornmissure carries fib e rs from the lumbar ganglion ventral l y to pass t h e ventro-latera l cord (cells 9, 11 or 1 0 , 12) , the f i bers then pass under th e ventral body muscle quadrant and then anteriorly to join the PAG . Th e level shown here is the most ant er ior lumbar entrance t o the commissures, where the most anterior lumbar cells (27 - 32) are jus t entering the commissure (from the extreme left and extreme right ) . Because the fib e rs are cut ob l iquely in this region , they are commonly elongated and sornetimes found discontinuous l y at several points in the same section . Some PAG cells (36,38,42) u se th ese commissures to trave l to the dorsal cord . Other FAG cells (37,40) some - times continue pos teriorly near the positions shown , without p ass ing under the muscle cells . L R &<:, 28 f!Fj ~ ~@' L$? ~~..-,;;--,.<J&cx ~ ~ 24 a • ~;:;, 4• b 69 Figure 11. Fiber Position s in Lumb a r-PAG Commi ;s su:ce s \. . Left and ri g ht lurnb a r - P.A.G con!fn is s ures are s ho~.,m in B126 (a) a nd Bl36 (b) at corresponding levels in ea ch a nimal . Magnification is about 25,000X . These cross- sections ar e just ant e rior to th os e in Fi g ure 10 , near the extr e me vent ra l edg e of the a nimal . He r e the fibers are in an orderly b undle be tw een the pa ss ag e u n d e r th e mu s cl e quadrant and the e ntrance to the posterior li mits of the PAG. Lumbar fibers (15-32) are in relatively constant positions with respect to each other. Bilaterally ho gous pair s of fibers take similar positions in these commissures. Fib ers from PAG cells (36,37,38,40,42) are not rigidly ordered . 10 R L a ~ -~ h\ ,f, :• ·' 1c '~~ - ~:-~ • •:t ~ I • ,: b . .-~·! ( ' ' .. ' ... , ~ .t'" j';r:JJ ,H•·fi.j,• ' • ' '.)J.~ 71 Figure 1 2 . Fiber P._ositions in the Pre - Ana l G2.r.. g lion Fiber positions are compared at correspondin g l eve ls in B136 , ( a ) and ( b ), and B126 , 16 , 000X . Magnifj_cation is 2bout Each cross-section is from the extreme ventr a l edge of the anima l 7) . ( c ). ( see F i gure 1 , panel ( e) , cross-se ctio n Larger profi l es are those of cell bodies . Slight shifts in ce ll body positions appear very striking in these comparisons ( 40 vs . 41 ). H - hypodermis M - muscle cell ( ventral quadrant ) G - gut ce ll DMF - musc l e arms f rom defecation muscles 12 a b C 73 pre - synaptic specializations for three to ten successive sections (electron-dense ,deposits along the inside of the pre-synaptic membrane , u sually accompan i ed by a cluster of vesicles). A second list of poss i ble synapses was made, where two or three sections adjoining missing or damaged sections displayed indications of a pre-synaptic specialization. The pattern of these interattions is not noticeably different from the pattern of the primary synapse li st . Together these two li sts probably i n c lude eve ry synapse in the PAG for each animal. Figure 13 shows a chart of the synaptic patterns in Bl26 and Bl36 , where the numb e rs of synapses are compared for each possible combination of pre - and post - synaptic elements . Most synapses are dyadic ; these are li s t ed twice in separate bo xes , once for e ach post-synaptic element . Figure 13 include s a ll chemical synapses from both th e primary and secondary lists for each animal . Multiple point synapses are conrnon ; these are listed in Figure 1 3 as mult iple numbers - that i s ,a serial synapse made up of three point synapses is added into the listing as three separate synapses . Despite th ese corrections, Figure 13 shbuld not be interpreted as a repres entation of true electrophysio l o gical synapse strengths . The reconstructed se ries are qui te complete , so the observed variations in syn apse multiplicity are presumed to r epresent real variations . 74 Figure 13 . Chemical Synapses 47 X 47 Th i s chart displays every chemical synaptic c o ntact in the tail for two animals (see Figure 3) . Each pre-synaptic fiber is listed at the left, each post - synaptic fiber is listed above . Synapses for Bl26 are listed above the diagonal , for B136 below the diagonal , in each box - note the il l ustration in upper right corner . Each line indicates a single synaptic contact in one animal (pre+ post) . A dyadic synapse , with two post - synaptic fibers, is listed in each of two separate boxes, once for each post - synaptic fiber . Unidentified participants in synapses are lumped into the category "Others ." The numbers of contacts in the most h eavily filled boxes are as high as 12 vs . 13 , 10 vs . l J. , 11 vs. 12. All chemical synapses are located i n the pre - ana l ganglion except for the following contacts : 21 + 23 and 22 + 2 4, which are in the lumbar ganglia . 75 CHEMICAL SYNAPSES POST-SYN APT:C FIBER [8i2·571 [6,.;,&J 76 This chart is a simplificati on which neglects the sp a tial configuration of the synapses - these features will be dealt with separately . Comparison of the two a n imals reve als a very strong pattern of synaptic interactions. The pattern is clearly non-random , with 71-1% of the s ynapses falling into the same few classes (3 . 4%) in both animals . Seve ral synapse types are present in very high multiples in each animal (seve n or more times) . Roughly 26% of the synapses are scattered singly among the uncommon classes (96%) of interaction. These synapses may be important , or they may represent the background noise of mistaken synapse formation (It is not known whether a given class of interaction must be multiplyrepresented to have functional significance in the nematode ' s behavior . ) Some neurons are involved in many types of synaptic interactions j_n the PAG , while others are involved in few or no interactions . Individual patterns of interaction typify each neuron. Most bilaterally symmetric pairs of lumbar cells have rather similar patterns, and these patterns are recognized to be even more similar upon further analysis. input . Several PAG neurons receive similar patterns of A few neurons are active as both pre - and post - synaptic elements; these are the interneurons, a , 8 , (cells 25,26) , and x (cell 43). 0 ~ Cells 23,24 are labelled '7'7 they interneurons , d e spite their total lack of synaptic output in the PAG. (Nonethele ss , have patience and note the diverse and extensive senso ry input which they i nterneurons receive in the PAG . ) Beca us e the n e rve processes in t he FAG are generally quite s traight and unbranched , th e re is little topological possibility for se lf- sy napses. have been found. Indeed, no self- synapses However , there are frequent synapses between some bilaterally homologous n e ur ons , particularly among the lumbar sensory neurons . These sensory fj_bers also form a few g ap junctions among syrnmetrj_c pairs (Figure 14) . These sensory fibers travel through the PAG in close opposition to their symmetric partners, a nd all of them terminate without entering the ventral cord . Thus t he PAG represents the prime s ite of sensory processing for the sense organs of the tail . (Cell 19, the neuron with the asynm1etric cilium, does appear to send a process into the ventral cord , a possible exception to this rule . ) The synaptic output of the two t y pes of phasmidial ciliated neuron s are strikingly different an d diffe r e nt behavioral functions . may be serving very The synaptic output of sensory cells 17,18 somewhat resemble s the output of phasmidial cells 27 , 28 . Ve r y little feedbaek control impinges upon these sensory neurons from non - sensory c e lls . F igure 15 displays a somewhat s implified synapse chart 78 Figure 14 . Gap Junction s This cl1art displays every e lectric al ga p junction in the tail for two animals ( see Figure 3) . Dark shading indic ates a synapse in the pr e - ana l gang l ion . indicat es a s ynapse in the lumbar ganglion. Li gh t shading No more than one gap junction of a ~iven type is found per anima l. As in Figure 13, synapses for 8126 are shown above the diagonal , for B136 belo w the diagonal , in eac h box . All gap junctions are show n twice on this chart, since both partjcip at in g fibers are presumably both pre - and post synaptic element s here. 79 8126 GAP JUNCTIONS .Fi g u re lL! 80 Figure 15 . Simplified Synapse Chart 23 X 23 Th i s chart displays the s um of the chemica l synapses fo und in Bl26 and B136 among the active fibers . As in Figure 13, pre - synapti c fibers are listed at t he left , post - synaptic fibers at the top . Dyadic synapses are again listed twice, once for each post - synapt ic fiber. synapses are in the PAG . 6 - sensory cell 0 - inte rne uron 0 0 - motoneuron - other neuron AJl 81 CHEMICAL SYNAPSES @ 7 I 2 5 0 I 3 4 ., 2 2 Ill 2 3 I 2 0 I I 25 I 2 1 14 I 13 6 8 23 8 9 I 6 I 10 3 I 5 5 4 12 7 5 -r ::) 8 4 " 2 I () 0 " 4 I 3 8 I 2 8 I 2 6 2 8 3 4 8 I 2 4 3 4 I I I --f-- G ----· " - 4 3 9 4 4 3 10 I 2 I 2 2 5 6 I (> . I i---· - - - I 2 2 I 2 I 0 4 6 .. 3 4 I I 2 I " ·10 I 2 8 - - - ·- I 2 I 2 - .. .. I -- (l) (j) j I ! I ! . -- -·· r-- i ! · ·- I - - 0 I - I Q 0 . 0 ·- - - -0 ·- 82 for the most synaptically active neurons of the PAG. The data from Bl26 and Bl36 have been summed for this chart, using the numbers given in Figure 13. These few neurons form 85% of all the PAG synapses among themselves. Dyadic synapses are again represented twice, once for each postsynaptic element. The similarities between bilateral pairs of neurons are again evident. Ipsilateral connections are roughly twice as frequent as contralateral connections. Most of these left-right differences are probably a reflection of the lack of left/right mixing of lumbar fibers in the posterior half of the PAG , where the fibers retain some of their commissural organization. Even if the l eft/right differences do arise from passive causes, they could have some functional significance . This topic wi ll be pursued in another section. Figure 16a is a simplified synapt ic chart of the most active neuron classes , where the synapses of bilaterally homologous pairs of neurons have been combined. treatment ign ores left/right distinctions. This The averaged data from Figure 15 have been summed to obtain the numbers in Figure 16a. In this chart, the convergence of ill synapses into a very small number of classes is evident, with the a and y interneurons r e ceiving the bulk of all sensory input. Dyadic synapses are again represe nted twice , once for each post - synaptic element . The connectivity of 83 Fi g ure 16 . Chemical Synapses : Homologues Comb ined ( a ) This chart display s the s um of t he chemical synapses f ound in B126 and B136 among the mos t active fibers , where a l l contacts o f bilaterally h omol og ous fibers are combined . Homologous moto neurons a r e grouped as As before , pre - synaptic fibers are listed a t the left , post - synapt i c fibers at the top. As b efore , dyadic sy~a pses a r e l isted twice , once for each post-synaptic fiber . (b ) shows a wi ring diagram based upon th e data i n ( a ) . Heavy li nes i nd i cate t h e most co rn.man synaptic connectio n s (15 or mo r e occurrenc es in t h e two anima l s combined ). 84 CHEMICAL SYNA PSES: HOMOLOGUES C0 ~/181 NED (a) POST-SYNAPTIC ~M@0@@@@ 45 66 9 4 27 10 ~ 6 \1/b "' 12 II 6 6 32 15 ~ 15 22 7 4 3 II 17 @ 4 © © 13 ® (b) .Fi [-l; ur e 16 85 this simplified chart can be surru11arized by the wj_ring diagram shown in Figure 16b . ( 'l'he layout a nd syrnb ol s used in Figure 6b are those of White e t al . (1976) . Justifica- tion for borrowing th e ir nomenclature will be forthco □ ing.) A cursory g l ance at the wiring diagram reveals that there are several types of simple feedback loops involved in the initi al processing of se nsory signals by five interneuron types . Of these interneurons , the a , 6, y , and x 0 types have processes which proce ed ante riad from the PAG into the ventral cord . The feedback loo ps could be used as a n egative feedback circuit to filter sensory signals , or as a positive feedback circuit to prolong and enhance sensory s ignals . Alternately , the spatial configuration of synaptic contacts c ould be a controlling feature in using feedback loops to select certain timing features of sensory signals . The multiple routes by which sensory signals can be rout e d to the a and y interneurons suggest the potential significance of timing features . are discussed below . All of these subjects For the moment , we shall examine the spatial configuration of sy nap ses , which is important in many of these possibilities . Th e re are two major aspects to the spatial configuration problem : the spatial separation of individual contacts and the predominance of dyadic contacts . There are relatively few non - dyadic contacts ; about eight mo nads and 86 twelve triads per animal. The non-dyadic contacts do not follow any consistent pattern, but appear to be just aberrations from the dyadic norm. Other aberrant contacts are observed where hypodermal tissue substitutes for one of the two post-synaptic elements. Triads are generally difficult to assign, since there is generally one clear post-synaptic element and two other possible post-synaptic elements. In 75% of the triads, two of the post-synaptic fibers are from homologous cells. The complete synaptic chart of dyadic contacts requires three dimensions to properly di~play the possible combinations of contacts. In Figure 17 these data are presented as six different planes from an 8x8x8 chart of the dyadic contacts. (Two planes are devoid of data points and have been left out.) This data has been assembled from an aggregate of all the dyadic contacts in Bl26 and Bl36. (Now each synapse is listed once, ignoring the two possible permutations of the post-synaptic elements. I have not analysed the left/right ordering of these elements.) . This level of analysis produces a proliferation of possible s yn apse types (288), yet there is again a clear restriction of contacts into a small number of classes. This restric- tion appears to have important implications in both the functional circuitry and the developmental specification of synaptic contacts. 87 Figure 17 . Preferred Dyadic Synapse Combinations Six planes are shown from a three-dimensional chart of dyadic synapses. Each plane represents the summed synaptic output from a given cell type in Bl26 and Bl36; this presynaptic element is shown above its array of dyadic outputs~ Bilaterally homologous nerves and homologous motoneurons are grouped as single cell types. Each array of output is bordered by two li sts of possible post-synaptic e l ements . (Since these 8 x 8 arrays are redundant , half of eac h array is not shown; i.e.,A + B,C i s considered equ ival ent to A+ C,B). Each dyadic contact is shown onl y once on this chart; mona ds and triads are n ot includ ed . Two planes are not shown because two cell types have no synaptic output in th e PAG (cells y a n d @). 88 PREFERRED DYADIC SYNAPSE COMBINATIONS PRE- SYNAPTIC ELEMENT: a 8 X 89 Very few dyads involve two homologous post-synaptic elements. Post-synaptic elements are rarely homologues of the pre-synaptic fiber. Thus synapses can generally be B represented as A ➔ C where A> B, and Care each from different neuronal types. Beyond this gene r al restriction, each type of pre-synaptic fiber generally contacts only a few possible combinations of neuron types> and each type of neuron has a different spectrum of preferred combinations. Thre e sets of pre-synaptic elements contact an a interneuron at almost every dyadic (and triadic) synapse. Th e other three sets of pre-synaptic elements contact a y interneuron at almost every dyadic (and triadic) synapse . Most pre-synaptic elements show a very strong preference for one particular pair of post-synaptic targets. more than contacted: (In 75% of all triads, the same pair of targets are A ➔ C1 B ). C2 In a typical PAG dyad, A ➔ B C' the two post-synaptic e lements (B,C) are generally involved in another PAG synapse . B ➔ C. In this manner, the dyadic synapse mediates the simultaneous stimulation of two routes of information flow which converge at C: ,/1B~ A ~ C. These multiple routes from A to C may represent the morphological basis for processing information through timing comparisons and timing delays. The following multiple routes are common : 90 They interneurons are the targets of each multip le route of information flow. The a interneurons are most often the intermediate elements in those multiple routes. The timing f ea tures of sensory stimulation of the tail may be scanned by they interneurons, comparing direct sensory input with delayed input from the a interne urons. The functional possibilities of dyadic circuitry are discussed in a subsequent section. 'rhe spatial separation of the dyads may also be j_mportant. The complex possibilities offered by this variab le 91 are difficult to comprehend. Without the ability to do intracellular recording in this system, it i s not possible to eva luate the physiolo gica l constraints which rule the circuits. Non ethe l ess , the a int e rn eurons are favorably placed to display th e natur e of the problem . Fi gure 18 shows the placement of synaptic cont acts a long the two a fibers in Bl26 and Bl36. The synaptic ou t- put of each a fiber is shown above the fiber, a n d the synaptic input to each a fib er is shown below the fiber . The varicosities of the fibers are indicated; bending of these fib ers is insignificant. imme diate ly evident. Several spatia l features are The posterior connections show a strong ipsilateral b i as , whi l e the anterior connections seem evenl y mixed . Alpha fibers are pre-synaptic at varicosities spread intermittently along their l ength , wh ile th ey receive synaptic input more continuosly along their lengths. There are many clo se ly spaced reciprocal synapses with¢ interneurons which set up feedback loops with min imum time delays. Cont acts with y interneurons remain strongly ipsilatera l throughout the PAG, but these contacts do not demonstrate any obvious spatial pattern. Sensory input from cells 17 ,1 8,19 seems to be restricted to small r egions , but the larger number of contacts from phasmidial neurons 27,28 are spread along the lengt h of the a fiber s without an obvious pattern . Alpha fiber output to motoneurons seems 92 Figure 18. Synaptic Input and Output of a Inter neuro ns The spatial distribution of synaptic contacts involving a interneurons is shown schematically for Bl26 (a) and 8136 (b). The varicosities of each a fiber are shown schematically ; bending and intertwining of these fi bers are not significant and are not shown. Sensory inputs to the a interneurons are shown by short arro ws be low each fi be r . Other inputs to the a interneurons are sho wn by long arr ows below each fiber. Alpha interneuron synaptic outputs are show n by arrows above each fiber, showing al l post-synap tic members of each output . Most inputs to a interneurons involve concurrent inputs to other fibers - these other contacts are not shown . These fibers extend for the length of the PAG, and in one case a 2 extends somewhat further into a FAG-lumbar fiber bundle . 93 SYNAPTIC INPUT AND OUTPUT OF a INTERNEURONS 'P2 (a) 'P2 a, 'Pi A 'P2 'P i 'Pi =1Uf___l__i__L_ __-= l ~/ J~ / /~ l !1 ls I\~?1/1 rt 7 a2 c/>2 82 'Pi 15 'P2 'Pi XO 'P i r 'Pi 8i 'Pi ai A Y2 Y2 az y2 Yi A 82 n 16 16 'Pi c/>2 8i Y2 Y2 Y2 A JL-L. IVf~ 1\~~t ~ c/>2 ls ·1r 1 'Pi 8 2 (bl An l erior ~ Posterior Fi g ure 18 IOfL f - - ---'--- -----l 94 restricted to the a nt erior ha lf of the PAG , but th e numb er of these contacts is too var i ab l e for the spa ti a l c on figu rat i o n to be the factor of inte r es t . In fact , spatial c onfi g u rat ion has not emerged as a u sefu l tool in studying the ro le of the a int erneur ons within the PAG . FUR'l HER IDENTIFICA'TION OF TAIL NEURONS BY FIBER POSI'1 ION 1 1 AND LINEAG E : The neur ona l circuits which we hav e r e con- structed in th e n ema tode t ai l are only a part of the larger circuitry which governs the behavior of the whole animal. It is inconvenient to di sc uss the functional properties of t he PAG circuits without kno wi ng how these circuits are conne cted to the ventra l cord and the nerve ring. The comparable circuits in the v e ntral cord b e tween moton e urons and interneurons have rec e ntly been recon s tructed by White et ,al . (1976 ) . The lineage of many of the PAG neurons has b ee n identified r e cently by Sulston (19 76 ). Sulston, White and Edgar have recently fol l owed the lineage of many of the lumbar cells (unpublished observations ). The sensory inputs from the a nterior sense organ s into the ventra l cord circuitry are currently being inve s ti ga ted, a nd the ge neral r o u te of thes e inputs is known (White, p erso na l communica tion ). Reconstructions of the ventral cord have r e vealed a r emarkably consistent pattern of fiber distribution and synap t ic interactions along the length of the ventral and dorsa l cord. Thi s pattern has permitted White and his 95 colleagues to recognize all of the major inte rneurons and motoneuron clas se s in most regions along the length of the animal by reconstructing relatively short series . Only in the regions nearest to the PAG were they unable to recognize the same pattern . In the ventr~l cord there are 65 motoneuron cell bodies which fall into seven classes. Each cl ass has an in d ividu~l distribution of axonal and dendritic processes. Each class of motoneuron neurite travels in a unique position in the ventral and/or dorsal cord . In a given region along the animal, there is only one motoneuron of each cla ss which is synaptically active . Nei ghboring members of each class contract t heir homologues by ga p junctions at the distal ends of their neurites . There are ten interneurons of four classes which are functional along the ventral cord ( none in the dorsal cord ). Each interneuron cl ass tr a vels along the ventral cord in a uniqu e po s ition . The synaptic pattern by wh ich the in ter- neurons contact each other and the motoneuro n s i s es se ntially unchange d along the l ength of the cord, although various t ypes of connections do vary in re l ative frequency . The interneurons reported ly receive a great numb er of synap tic c ontacts in the nerve ring . White has designated the four cl asseG of interneurons as a, S, y, a nd 6 (Fi gure 19c). The a interneurons are a bilaterally sy~netri c pair of monopo l ar 96 cells, with cell bodies in the anterior reg ions of the lateral g ang lia in the h e ad . Their processes are very large and run down the center of the ventra l cord (White, 1975 ). The B interneurons are a bilaterally symmetric pair of monopolar cel l s, with cel l bodies in tl1e posterior regions of the lateral ganglia . Their processes are also large and run prominent l y above the a processes (White, 1975) . They interneurons do not have cel l bodies in the head, but appear to come from cell bodies somewhere in the tail of the animal . They processes run under the a processes in the ventral cord . There are four 8 inte r- neurons , two bilaterally symmetric pairs , with cell bodies in the anterior regions of the l ateral ganglia. The 8 processes are smaJ.ler in diameter and they run between the a and B processes in the ventra l cord . Two of the 8 processes terminate in the anterior ventral cord, while the other two c processes continue through the posterior ventral cord (White, 19751 White et al ., 1976) . INTERNEURONS : In my own reconstr·uctions of the PAG , I have noted the presence of four synaptica lly active int erneurons which ente r the PAG from the ventra l cord ; two large clear pr6cesses and two small processes which run just above them as they enter the PAG (Figures 1 9a,b ). The two large processes travel in a central positio n in the cord a nd I have identified them as the a interneurons of White et a l. 97 Figure T9. Fiber Positions in the Ventral Cord The fiber positions of prominent interneurons and motoneurons are shown in our most anterior reconstructions of Bl26 (a) and 8136 (b) and in a typical cross - section of the anterior ventral cord (c) (White et al., 1975) . fication in these tracings is about 25 , 000X . I12.gni - In the ve:,tral cord, major motoneuron classes (A, B,C,DD, VD) adopt characteristic positions with respect to each other and to the muscle plate (M); major interneuron:s adopt character:!_s•tic positions with respect to each other and have characteristic sizes ; a and a interneurons being larger than the 6 and y interneurons. In Bl36 (b), one of the proposed 6 interneuron s appears unnaturally large, due to a mitochondrion in the cytoplasm; this fiber is much smal ler along the rest of its length, in accordance with a typical o interneuron . 9. 8. ~ ~ I M D ~ c~ a ~I y B • w ~ · ,. . . . ,. - ,--'I CB (47) --------==::::: H b ~ 99 _ . (1976). The two smaller fibers are identified as the 6 interneurons of White ~t al . (1976) . No correlates to the B interneurons have been observed in the PAG ; I assume that the B interneurons terminate in the posterior ventral cord . Two bilaterally symmetric monop olar neurons from the lumbar ganglia (cells 23,24) send processes anteriad into the ventral cord just beneath the a fibers; I have identified these cells as they interneurons of White et al. ( 1976) . Figure 19 displays drawings of the relative fiber positions in the anterior PAG of Bl26 and Bl36 and drawings of the relative fiber positions in the ventral cord ( taken from White et al ., 1976 ) . The synaptic interactions of the identified a , 6 , and y interneurons in the PAG are not identical to their ventra l cord interactions, but they do' have some strong features i n common. In White et al . ' s (1976 ) reconstructions of the Ventral cord , class A motoneurons typically receive synaptic input only from a interneurons , 6 interneurons, and other class A motoneurons . In my reconstructions of the PAG , the a and 6 fibers contact cells 38 and 42 ( c l ass A motoneurons ), and cell 47 ( a suspected class A motoneuron ) . explain my (I wil l identifications of cells 38 ,4 2 , 47 shortly. ) Delta interneurons i n the ventral cord have only one other synaptic target , a interneurons . My PAG 6 fibers have only one other synaptic target, a fibers. The most striking 100 difference in the PAG is the multiplicity of In the ventral cord , th ere are many y a ➔ ➔ a y contacts . ➔ a contacts but no y contacts . PRE-ANAL GANGLION LINEAGE : Many of t he ventral cord moto - neurons d eve lop and become wi red into the ventral cord circuitry after the n e ma tode has hatched from the egg . The post - embryoni c development of these moton e urons from t hei r precursor cells has recently been reported by Sulston (1976 ). Each precursor cell undergoes a fixed set of ce ll divisions in every animal examined , and the progeny of these cell divisions can be recognized individually by Nomarski optics in the living animal. Reconstructions of identified cells in the anterior ventral cord have demonstrated that each developing cell is d esti ned to become a particular predictable type of motoneuron . Thirteen prec ur so r cells (PO to P12 ) go through these programmed cell divisions, usually to gi ve five motoneurons (VA,VB,VC,DAS ,VD) and a hypodermal cell; the motoneurons migrate into prescribed positions along the ventral cord and become wired in (Figures 21,22). Some li n e age cells undergo pro g rammed cell deaths and do not survive in the adu J.t anima l . Some of the most an terior l ineage cells which lie in the retrovesicular ganglio n do not assume the typical morphologies of other lineage c ells - some are not motorneurons at all . However, eve n these atyp ical cells do have reproducible fates . The 101 other motone urons of the ventral cord, which are present at hatch are called "J 11 cells and become the DA, DB, a,~d DD motoneurons in the adult v entral cord . The adult pre - anal ganglio n contains six lin eage cells from the most posterior pair of precursor cells, Pll and Pl2 (Sulston, 1976). The other six PAG cells are present at hatch , along with fifteen ventra l cord mo toneurons a~d twelve other retrovesicular cells . cell divisions of Pll and Pl2 . Sul sto n has fo ll owed the He reports that their cell divisions are atyp ic a l with a total of six programmed cell deaths occurring, plus the production of two extra hypoderrn.al cells. Figure 20 shows Sulston ' s drawing of the a 1ult PAG beside drawings of the PAG ' s of B126 and B136 . We can readily identify the lineage cells in the PAG as follows: Cell 37 comes from Pl2d-predicted to be a DAS rnotoneuron Cell 39 comes from Pl2e-pr e dicted to be a VD motoneuron Cell J-12 comes from Pl2a- pre dicted to be a VA motoneuron 45 comes from Plle- pr e dicted to be a VD motone uron Cell Cell J-16 comes from Plld-pred icted to be a DAS rnotoneuron Cell 47 comes from Plla-predicted to be a VA motoneuron Each of these motoneuron types can be immediately identi~ied by its morpholog y and pattern of synaptic i~put and nearornusc~tlar output . 'Ihe details of these morpho lo gies 102 £'.igure 20 . FAG Cell Body Positions - Lineage The cell body posit i ons of the pre - ana l ganglion cell bodies are sh own in l ongitudina l reconstructions of Bl26 (a) and Bl 36 ( b ) , and i n a " typical n animal ( c) follo v1ed oy Nomarsk i optics for lineage cell div i sions (from Su l ston , 1976). Six lineage cells can be identified , products of cell divisions by precursor cel ls Pll and P12 . 0 N (D (; rn ~~ ­ · •__; h -i (c) • ( b-'I (a) ®@® ® Pl2e Pl2a , G) Q O O ,)\ - 0 @ Pi 'le I Plld GJ C? 10 0 o~~ Pl2d ®@ ®® ~@@®G ® @®@® G:) @ ® PIia 9 ® ® w 0 f-' 104 Figure 21 . Motoneuron 'l'ypes - Morphol ,:-'gies Typical motoneuron morpho lo gies are shown for t he eight t y pes of ventra l cord motoneurons , displayed in schematic views of the h ead and the ventral cord (taken from White et al ., 1976 ). All motone uron s have their c ell bodies in the ventral cord . (C el l bodies are shown as dark dots .) Some ty p ically have anterior projections , others have posterior projections , some typ ic al l y have projections into the dorsal c o rd . J~. Ne uromuscular outputs are indicated by / 105 ~- --------= ---- VENTRAL -------------=:--------- 8 OORSAL A 8 OORSAL ~::] DORSAL AS DORSAL - VENTRAL Fi g ure 21 C D 106 Figure 22 .. Lineage Cell Divisions (a) Lineage cell divisions don a te a set of dau ghte r cells into each lumbar ganglion; here the set of' dau ghter' cells in the right lWTibar ganglion is shown, with their lineage. The left lumbar ganglion receives an exactly homologous set of daughter cells which become bilaterally homologous cells . Cell 19 is the daughter of a separate se t of cell divisions in the anterior re gions of the nematode, and mi g rates into the right lumbar ganglion. No cor respon d- ing cell migrates into the left lmrrba r ganglion (Edga r , White and Sulston , unpubli she d data ). H - hypodermal cell X - programmed cell death P - precursor cell (b) Lineage ce l l divisions donate two homologous sets of daughter cells into the pre-anal ganglion. Three prog rammed cell deaths occur in e&ch lineage; three daughters become potential motoneurons, types VA,DAS,VD ; one cell becomes a hypode rma l cell . 107 M a H 19 H 7 15 3 21 b C d e VA DAS VD Fi g ure 22 H 108 I ( ~~k~~ vLl- C:l.1 et al ., 1976). P~c m \J ~i~e ..1.. _ ..• . _ __ ,._, Each class cf motoneuron (A , B,D , AS) travels I in a s tandard positJon wit~in the nerve c ords (Fi g ure 1 9c) . The prefix V designates that the moton e uron forms neuro r.. :l nu_,.___) ,, cu 7 a -r• J l!)YlC 1- i ,,.v,s• (NM.T) in the v e ntral cord; the pre~ix D 0 J... - __ _ v _.,_ VJ.J. ... _ ..., de::;i 6 nate;3 NIU output in the dor~;a l cord . 'l'hw, a DB rr..otoneuron travc:: ls in the B p osition t o torm NMJ iL th e dorsal cord , 2nd has the t y pica l morpholo g y of a class B motoneuron . Class D motcneurons have neurites in both cord s , ma intainini a dendrite in the op posite cord to its o wn axon . The D dendrit e s as s ociate closely with the muscle plate and receive synaptic contact from class A and class B motoneurons - these contacts are simultaneous to the class A and B contacts to the muscle plates . To determine the true fates of the PAG lineage cel l s , I re - examined the positions in the cords wh i c h thei r neuri tes assu~ne, and comp a red their ce l l morphologies to those of the standard motoneuron classes ( see Figures 7 , 21). Cell 37 is predicted by lineage to become a DAS moto neuron . Cell 37 should have a short proximal dendrite in t he v e ntra l cord travel l ing anteriad to a commi s sare to the dorsal cord , where it should maintain an axonal process with NMJ o u tput . In fac t, cell 37 has o n ly a short proximal proce s s which appears to - end blindly withou t synapt ic cont a cts or commissural connections . Cel l 37 does n o t 109 appear to be a DAS motoneuron , nor any type of motoneuron in either animal. No other nearby PAG neuron maintains a DAS type proc ess eithe r, so I conclude that line age c e ll Pl2d does not fo llow the prescrib e d lin eage fate. Ce ll 46 i s al so predicted to b ecome a DAS motone uron. Cell 46 does send a process ant e riad into the v e ntral cord . No comrnissure to the d ors al cord is observed in either anim a l in our recon structions, but such a commiss ure could pos s ibly occur further anterior . A DAS type axon is obs e rved in th e dor sal cord of B136 which h as NMJ output and which extends fro m some commissure anterior to the reconstructed r egion. The process might really be a part of c e ll 46, or it mi gh t b e an extension of th e DAS motoneuron arisin g from lineage c e ll PlOd . No other nearby PAG c e ll app ears to be a DAS rnotone u ron . Th e fate of lineage cell Plld r emai n s indetermin ate , but is poss ibly in accordance with th e expe ct ed pro gram. Cell 39 is predicted by lin eage to become a VD motoneuron, with an anterior axon a nd an a nterior commissure to a dorsal d endrite . Ce ll 39 does send an axon anteriad a n d N~J output has bee n found in both animal s . No commissures have been found in my recon s tructions, but such a commiss ure may we ll oc c u r furth er anterior (th e f ig ures in White et al . (1976) stron g ly predict this ). A dor sal VD dendri te is evident in th e dor sal cord of B136 , coming from a comm i ssure 110 anterior to the reconstructed region . No other nea rb y FAG cell is a good candidate for a VD motoneuror1 . The f ~te of Pl2e remains inde termi~ate , but is very possibly in agreement with the expected program . Cel J. 45 is also expected to be a VD motoneuron . Cell 45 sends a process anteriad which is not in position as a class D fiber , but more like the positi o n of a class C fiber . No NMJ output is observed . One nearby PAG cel l (cell 41) migit make a better candidate for a possible VD motoneuron - although there is no strong evidence that cel l 41 is a motoneuron at all . cel l Cel l 41 cannot be mistaken for 45 in my reconstruct i ons , nor in the lineage pict u re of Sulston (1976 ) (Figure 16 ), since cell 41 has a much larger cel l tody than any other PAG cel l. lineage ~ell Pl le, cel l I conclude that does not fol l ow the normal l ineage program . Cell 42 is predicted to beco~e a VA motoneuron , with a ventral axon ar1terior to the ce l l body and a ventral dendrite posterior to the cell body . Cell 42 sends an anterior process into the ventral cord which is rough l y in the proper position to be a class A axon , but no NMJ output is observed in either anima l . Cell 42 also has a dendrite which extends posteriad from the cell body and which continues through the FAG , through the rig~t lumbar - PAG commissure and then trave l s dorsally to the dors a l cord . This dendritJc 111 process continues anteriad in the dorsal cord for at least eighty microns in a clas s A position without synaptic activity . This dendritic process does r e ceive prop er class A synaptic input fr om a and 6 fib ers in the PAG. Cell 42 may really be fol1owi.ng th 2 fate of becoming a Vt\. motoneuron. 'I'he aberrant d c,rs al cornmj ssure may be due to the boundary problems associated with the cell body's extreme posterior location - whic h does not allow th e normal room for a posterior dendrite. Similar boundary pro8lems appear to occur for the most anter:ior VB moton eur ons in the retrovesicular ganglion, whose anterior dendrit es trav el aberrantly to the dorsal cord and/or the lateral gangJia ( see the f:igures in White et al . ( 197 6)). l\'o other PAG ce 11 nearby to cell 42 appears to b e as g ood a candidate for a VA motoneuron. The fate of Pl2a remains indeterminate, but is probably in accordance with the ~xpectcd program . Cell U7 is also predicted to be a VA motoneuron. L1 7, like cell 1+2) has a dor sa l dendrite . Cell comm:Ls sur2 fro m. a posterior The cornmh,sure tr2,vels around the left side of th~ body and the dorsal process travels as a class A fiber in the dors?t.l cord without c=rny synapt:lc in teractio ns. The po s terior dendrite does rec e ive cla ss A l• D p -.J +-v from fLbers in the PAG . Cell 47 s e nds a pro ces s anteriad in the ventral co r d in the cla s1, A posit i on, but no NIU output is cb1"erved l~ either animal. Ce ll 47 h2s a un ique cell bady position 112 which precludes mistaking its lineage. I conclude that the fate of lin eage cells is indeterminate, but possibly in agreement with the expected fate. Of the six lineage cel l s in the PAG, tNO are clearly no t following their predicted fates, and may not even be motoneurons. Two other cells have aberrant dorsal fibers, but thes e cells could still be following their predicted fates within the con:ines of certain.boundary problems. The remaining two ce ll s may be correctly following their predicted fates, but I cannot follow their processes far enough anterior to be certain. Six other PAG cells are present at hatch. Some of these cells might be expected to become DA, DB, or DD motoneurons in the adult , if they are analogous to the "J" cells of the ventral cord . Cell 36 does resemble a neuron, with dorsal D type NMJ output. DD moto- However, cell 36 does not have a class D dendrite in the ventral cord. 38 appears to be a DA motoneuron. Cell Cell 38 has class A NMJ output in the dorsa l cord and maintains a process in the ventral cord in the class A position. Cell 38 receives s y naptic input which diff ers somewhat from the VA fibers (42,47). None of the other four cells (40,41,43,44) have dorsal commissures of NMJ output in my reconstructions. No ne of the s e four cells app ea rs to be a moton e uron . 43 i s clearly an intern e uron , and has b e en l a bell e d X· Cell This 113 interneuron has not been mentioned as an important element in the anterior ventral cord (White et al., 1976), althou g h this cell does send a process anteriad into the ventral cord. The PAG has no VB lineage cells, since Pllb and P12b undergo programmed cell deaths (Figure 22) . have not been found either. DB motoneurons The PAG represents a boundary problem where class B motoneurons would have no room to send a functional posterior axon anyway ·(Sulston, _1976). There does exist a VB type fiber at the most posterior NMJ regions in the ventral cord which presumably is a process from PlOb in the posterior ventral cord. This fiber enters the PAG in a class B position and ends blindly at the posterior end of the PAG. the PAG. This fiber receives no synaptic input in The most posterior S ➔ B synapses are expected to occur anterior to the PlOb cell body (White et al., 1976; Sulston, 1976). This interaction may somehow help to signal the S fibers to end without reaching the PAG. A DB type axon with mu output is fou~d in the dorsal cord of Bl36 . This presumably ts a process from a anterior to the reconstructed region. 11 J" cell No DB type · moto- neu.rons are found in the PAG, but this is not unexpected becau se of the bounda ry problems involved. To suITLmarize, the general organization of rnotoneuron typ e s at the most posterior neuromuscular junction regions 114 in the ventral and dorsal cords does not appear to be any different from the organization reported previously for the anterior· regions of the nematode cords (W~ite et al ., 1976). All of the same motoneuron classes are observed to be in th eir proper positions at the muscle plate, and many of these motoneurons receive typica l synaptic inputs from id entified interneurons . LUMBAR LINEAGE : Many of the cells of the lumbar regions of the tail are not present at hatch , but develop from precursor cells during the first larval stages . A series of ordered cell divisions of two bilaterally homologous precursor cells produces two bilaterally homologous sets of cells whi ch migrate into position in the lumbar regions to b e come hypodermal cells, supporting cells and neuro~s in t he adult tail. The lineage of these cells has recently bee n observed by John Sul ston and John White by Nomarski optics in developing animals (unpublished data ). Lois Edgar has sectioned two animals in which the cell lineages had been followed . She has partially reconstructed the lumbar regions and identified th e c~ll bodies of the lineage cel ls ( unpublished data ). I have been fortunate to compare her reconstru ctions to my own work on the lumbar regions. Together , Lois Edgar and I have been able to identify al l of the lineage cells in the lumbar regions and determine the fate of each daugh ter cell. The lineage cells of each 11 5 lumba:~ region are derived fro1r; identical sets of cell divisions from single precursor cells, an d bilatera lly homologous daughte r cells cs.n be unambj_guo usly assigned as bilateral l y homologous pairs of a kno wn c e 11 type . 'i l1e l ineage cel l division patter0 is shown in Figure 22 , and the fate of each daughter cell i s sho~n An additio n al unpair e d lumbar ce::.l is the daughter ce 11 from a separa te set of lj_n ea.ge c ell divj_siom, . Tll:J.s daughter cell migrates :Lnto the tai1 from a small set of li n eage cells ~h ich originate near the center of the nematode . Two other daughters of thi ~ lineage Lligrate into th~ posterior latera l ganglion ( Lois Edgar , personal communication from John Wnite) . The daughter cell which mi gra t es into the tail has been id entified by cell bcdy position as ce1J. 19, in the right lumto.r ganglion . Cel l 19 , o f c ourse , is the most st riking asy imnetric cell in the tail , ha~ing a se n sory ciliated dendrite embedded in the wing ce ll of the right phasrn id . The indepe nd ent linc:ag e of c2J. l 19 demonstrates that •:: el l s 19 ~20 s hould not h: co:n s ideJ ed to 1 be bilateral l y homolog o u s in any sense. Instead, ce ll 19 and c ell 2 0 are now recognized as two asymme tric unpaired lumbar neurons . Cel l 2 0 is pre::;ent at hatch , wh:Lle cell 19 develops a~ter migrating into the tail post - hatch . has no clear functional ro le in tho adult tail . Cel1 2 0 116 In examining the set of cells which are present in the luE·1bar ganglia at hatching, one immediately notes that the two pairs of major interneurons of the adult tail (y and ¢) are present in the first larval stage. The phasmidial sensory neurons, 27-30, are also present~ although some of the phasmidial support cells are absent. Thus the pri:oary elements of the senaory circuits in the PAG are present in both the juvenile and the adult. 117 y. DISCUSSION COMPARATIVE ANATOMY: The tail of Q_. elegans is strikingly similar in general organization to the tails of several larger nematodes (Chitwood and Chitwood, 1950). Particu- larly interesting are the close similarities with the tail of Ascaris, a large parasitic nematode which has already been studied extensively and has favorable prospects for intracellular recording of its neuronal activity. At the level of gross anatomy, Goldschmidt (1908) and Chitwood and Chitwood (1950) found that the female tail of Ascaris has the same set of ganglia (two lumbar, one pre-anal, and one dorsorectall as reported above for the hermaphro~ite tail of c. elegans. More detailed comparison is complicated by the fact that accurate cell counts have not been reported for Ascaris, and by the fact that the light-microscope methods used in Ascaris do not permit the following of fine fibers. However, the approximate Ascaris cell counts reported (Bullock and Horridge, 1966) are close to those found iD Q_. elegans (e.g. pre-anal ganglion: 12; lQmbar ganglia: Ascaris 11, C.elegans Ascaris 6 each, Q_. ele g ans 12 each) and the real degree 9f correspondance may indeed he very high. More detailed study of Ascaris might establish whether homology is sufficiently great to permit electro phy sio lo g ical testing in Ascaris of functional hypotheses 118 generated by the synaptic anatomy of C.elegans. CHEMOSENSORY TRANSDUCTION IN THE TAIL: Q_. elegans is known: to have a variety of chemosensory responses (Ward, 1973; Dusenbery, 1974; Dusenbery et al., 1975; Dusenbery, 1976). The great concentration of sensory receptors in the snout (Ward et al., 1975; Ware et al., 1975), particularly the concentration of amphids and internal labial papillae in which ciliated sensory dendrites are directly exposed to the exterior, strongly suggests that anterior receptors play a role in chemosensory responses. However, the phasmids of the tail also contain ciliated sensory dendrites directly exposed to the exterior, raising the possibility - of tail receptor involvement as well. On the snout the close place- ment of the bilaterally or hexaradially symmetrical homologues argues against their differential use to detect gradients ; instead it seems most likely that these homologues act in concert. On the tail, receptors are also very closely spaced, but between head and tail there is a sufficiently large spatial separation (approximately 1.5 mm. in adult Q_. elegans) to suggest the possibility that headtail comparison might be used in chemosensory responses. Just how the comparison might be carried out is uncertain, but it is noteworthy that t wo of the major interneuron classes which receive sensory input in the tail (a and o) also receive input from some snout receptors with 119 exposed sensory de ndrite s (inner labial papill ae ; White , personal communicatj_on ) ( Fi gu r e 23c ). Head-tail comparison s carried out throu gh th ese j_ntern e urons could r eprese nt a comp onent of chemosensory r espo nses quit e di st inct fro m the mechanism by which an anim a l orients accu rate l y up a gradient. Fo r instance, White (1975) h as s ugges ted that accurate ori e ntation may b e b rought about b y stre n gthening body contractions coincidenta lly with head movements up a gradient . White has arg ued that such stre n gthe nin g might be c a rri ed out usin g B int erne urons , which r ece ive strong input from the ciliat e d neurons of the amphids , putative che morec ep tor s of the snout (Figure 23c ). But independen t of thi s r e fi ned ori en tation me chani sm there could a l so be a cruder one b ased on h ead-tai l comparisons of chemical stimuli , servin g t o ge t the an imal going in rough l y the right direction. (The r elative ly small n umb e r of possible chemosensory cells in th e tail would be in keeping with a crude orientation mechanis m. ) Whether th e tail is inde e d sensitive t o chemical stimuli , whethe r h ead - tail comparison s 2re indeed made , and if so , whet h er the a and o interne uron s mediate the compari son s a r e al l matters which r emain to be r eso lved . Another po ss ibility , ~ prio ri somewhat les s l ikely , is that chemosensory ce ll s in the t ai l mi ght b e u sed to guide backward movement , in simp li f i ed ana lo gy to th e rol e played 120 by snout receptors in forward movement . However , there are no known chemical orientation respons es during backward movement. Pre liminary ex p eriments with a reversing mutant have shown that an animal moving backwards does not orient to a salt gradient (Hodgk i n and Lew:Ls , unpublished observa-tions). Further studies of backward motion in thes e mut ants might b e worthwhile . Crude guidance of movement by head-tail comparisons may involve some broad- range chemotactic stimulus . However, the large number of s eparable chemotactic responses (nin e or more ) and the many ge netic a lly di s tinct chemos ensory mutants ( six or more) (Duse nbery, 1976) argue that the two pairs of phasmidial ciliat e d neurons are probably too few to be useful in most of th e n emato de ' s chernotactic b eh avior s . The tv!elve pairs of closely-spaced arnphidial neurons , which are we ll-expos e d to the exterior , probably detect chemical gradients by temporal comparisons, indepen d ent of any input fr om th e tail . 11 The nematode exhibits a pronounced searchin g 11 motion of th e head which is though t to increa se the amphidial capabilities in detecting chemical gradient s ( Ward , 1973) . Ea ch bilaterally homologous pair of amphidial neurons might be specialized to utilize a different receptor , g ivin g the nematode a broad ran ge of se n s ory capab:Lli ti.es . ,I 121 MECHANOSENSORY TRANSDUCTION IN THE TAIL : According to their placement in the animal , the variou s apparent mech a norecepto rs of C. elegans can be divided into two classes. Th e fir st are the groups at the extremes of the animal ( head and tail), which have been thought to mediate r esponses to externally applied touch, and the second are the deirids and post-deirids , more centrally placed and confin ed to the lateral lines . The placeme nt of the deirid s and post- deirid s suggests that they may serve to indic ate how tightly th e animal is held down by surface t ensi on; the lateral lines are in direct cont act with the surface on which the animal moves and are therefore most subject to mechanical deformation when the animal is held down. ( Indeed , C. elee_;ans changes its behavior very dramatically wh e n suspended in liquid, and prelj_minary observations suggest also that C . elegans can di s tinguish between areas of an agar surface with different degrees of hydration.)In this study, the po st - deirids were not studied because they lie just anterior to the sectioned region (probably associated with a small group of cell bodies shown in Figure 2a). However , two cel l s of the lateral cord, ce l ls 9 ,10, might be carrying sensory signals from the post-deirids to the¢ interneurons, with which they make extensive gap junc tions; a more anteriorly extended set of sections would be required to explore this possibility. (Hesse (1892) previously noted 122 a posterior p r ojection from the post - deirids into the l ateral cord in As~aris megalocephala) . The more posterior receptor group of the tail includes the phasmidial endings (cells 2 7- 30 ), the buried ending ( cell 19) , and a pair of fine posterior processes ( cel l s 1 7, 1 8) . The mechanical responses thought to be mediated by one or more of these receptor types pose an interesting prob l em ; forward moving responses can be generated by touch to any part of the posterior part of the body , even though the apparent mechanoreceptors are confined to the tail tip . The necessary transfer of mechanical energy to the receptors might be accomplished by the cuticle itself , but an alter n ative possibi l ity presents itself . The phasmids l ie at the posterior ends of large fluid - filled l ateral channels , whose function is uncertain . (A chain of gland- like cells lying along these channels have been suggested to secrete their contained fluid (White, 1975)) . In the region of the ph2.smids, the tail is narrm·red considerably and the channels are relatively large , suggesting that the channels may serve to focus pressure waves from distant mechanical stimuli onto the phas mids . I n terestingly , the buried ending of cell 19 may also be in a position to be stimulated by such The dendrites of cells 17 , 18 do not appear to be located in a sensitive position for stimuli transduced by 123 the lat era l channe l s . These dendrites extend posteriorly for another 65 microns past the phasmids. At their posterior limits these fibers are sheathed in a very f ine cuticular process, l ess than four microns thick (Fi gu re 5f) . This process is dragged behind the animal in a thre ad- like fashion . Each time the animal reverses, this proce ss is bent severely , usually forming a sharp 180° bend , and othe r mechan ical stimul i batter it abo ut very easi ly. Although these dendrites show no cytoplasmic specializa t ion s , it seems likely that they are responsive to mechanical stimulation of the tail tip . Mechan ical stimulation of the tail causes striking behavioral responses . An an im a l which is stationary o r moving backwards wil l stop and move rap idly forward in response to tail tap and an animal moving fo rward wi ll increase its forward speed . Habituation to tail tap may occur after twenty or more repeated st imuli (1/second) . Mechanical stimulation of the posterior half of the body will cause similar responses . Mechanical stimulation of the snout of the anterior half 6f the body wil l cause the animal to move backwards . Backwards motion is not as coordinated nor as fast as forward motion in response to tail tap . Habituation to head tap also occurs after repeated stimulation . Mutants insensitive to li gh t touch of the head have been reported by Hodgkin and Lewis 124 (unpublished observations). I have isol ated several similar mutants which are insensitive tc light touch of the tail . Study of these mutants is continuing . All of the sensory n euro ns of th e ta il may be mechan o- se nsory , utilizin g a vari ety of methods for transducing mechanical stimuli . The c e ll bodies of these neurons are favorably positioned for laser inactivation st udi es . DEVELOPMENT OF THE TAIL: CHANGES AFTER HATCH : The hermaphrodite tail does undergo some developmental changes af ter hatching , but these changes are relatively minor in comparison to the extensive changes post - hatch in the ma le tail of C. elegans . The hermaphrodite ' s PAG has six lineage cells from the Pll and Pl2 precursor cells . Four of these lineage cells may become motoneurons in the adul t tail in a ccordance with their line age fates. The other t wo lineage cells do not appear to be functional neurons. The PAG has six other cells which are present at hatching . Two of these cells are motoneurons and may undergo some r ew iring post hatch, much like the ' J ' cells of the ventral cord (Sul st on , 1976) . Three of the other four original PAG cells see m to lack neuronal function in the adult tail . Perhaps these cells have more import a nt roles in the male tai l - the ma l e tail has a pre- ana l gangli on which contains a similar number of cell bodies while mediating a muc h more complex se t of behaviors . Alternately , they may have some function 125 in the juvenile tail which is no longer required in the adult. The lumbar ganglia in the hermaphrodite gain many new lineage cells post-hatch. Many of these cells appear to be involved in sensory transduction. porting cells for the phasmids . Cells 1-4 are sup- Cells 7,8 are postulated to be important in sustaining sensory transduction in the subcuticular channels. in the adult. Cells 17,18,19 are sensory . neurons Many cells analogous to these line age cells are known to be present in the cephalic regions , in asso ciation with other sensory ganglia (Ware et al., 1975 ). One wonders which of the cephalic cells are also derived from post - hatch cell divisions. Each of these lineage cells could be adding to the sensory capabilities of the adu lt nematode. No behavioral changes are known to occur post- hatch in the hermaphrodite which would correspond to the addition of these cells, but a more careful search might be fruitful . Three other pairs of lumbar lin eage cells become neurons whic h have few apparent functions in t he adult . Cells 11,12 are associated with the ventro-lateral cords. These cells have no apparent synapti c interactions in the tail, and have no sensory specializations . Cells 15,16 have no sensory dendrites, and have a smal l number of synapti c interactions in the PAG. Cells 21,22 have no sensory 126 dendrites and no synaptic inputs, but they hav~ a single major synaptic output to the cell bodies of the a interneurons - the only chemical synapses of the lumbar ganglia. Presumably this synaptic input to a could have some pacemaker function in the adult. Overall, these three pairs of lineage neurons seem rather useless in the adult. Perhaps they are just a vestige of some of the ifilportant lineage changes in the male tail. The male nematode is most easily identified and distinguished from the hermaphrodite by the specializations of its tail. The male tail undergoes elaborate changes during the third and fourth larval stages to form the copulatory bursa. Before these changes take place, the male nematode is not visibly distinguishable from the hermaphrodite, and the juvenile male's behavior is apparently no different from that of the juvenile hermaphrodite. The adult male performs complex searching behaviors and mating behaviors while moving backwards, guided by sensory struct•;_res of the bursa. The layout of t h e nervous system in the male tail is rather similar to the layout of the hermaphrodite tail, but the number of sensory structures is greatly increased. There is also a proliferation of specialized muscle cells. The neuropil of t h e male PAG is enlarged, but the PAG contains only a few extra cell bodies (Hall, unpublished observations). The number of inter- 127 neurons in the male tail canno t be assessed fr~m my preliminary reconstructions. The nature of these lineage ch~nges are currently being examined morphologically by Albertson~ Wnite and Sulston (personal communication). DE"VELOPMENT OF THE TAIL: RULES OF FIBER GROWTH AND SYNAPSE FOFr-1ATION: Synaptic organization in the tail of C. elegans reveals a striking constancy. Synapse formation is restricted to a lxlx40 µ3 region of neuropil. In the pre- anal ganglion there are about forty unbranched neurites which form dyadic synapses among themselves, and a very small number of synapse types are heavily favored. Moto- neuron processes within the PAG are strictly.dendritic, having no synaptic outputs. The interneurons have single processes which often function both pre- a."'ld postsynaptically, generally without any spatial distinction within the PAG (probably common in invertebrate ganglia, as found in the lobster by King (1976ab)). Since all of these processes are unbranched, they can achieve their constant synaptic organization only by maintaining close proximity to their synaptic partners as t h ey travel through th e FAG . The set of genetic specification s which govern fi b er growth in the PAG may intersect wit h the set of specifications which govern synapse formation. Previous studies of the nematode's ven tral cord have alr e ady sho wn that the important interneuro ns travel in 12 8 close proximity to the active motoneuron axon s a nd den drites, whi le other inactive fibers are g rouped at the periphery of the ve ntral cord. The active fibers all maintain rigid spatial relationships among themselves and wit h the muscle endp l ates throughout t he length of t he ventral cord (White et al ., 1976 ). Th ese spatial rela- tionships are not unrelated to the synaptic organization of the ventra l cord . However, no genetic factor has yet been shown to govern the organization of the ventral cord or the formatio n of synapses. Ordered patterns of fiber growth also appear to underlie the organization of the PAG . Each lumbar ganglio n contains nine cells which send s in gle fibers in an ordered bundle into the PAG . The mo re active lumbar fibers associate with one another before r eaching the PAG . Withi n the PAG the lumb ar fibers intermingle in a stereotyped fashion with the ventral cord interneurons . Active sensory fibers and interneurons are closely g rouped in the c ente r of the PAG , whi l e the inac tive fibers travel a long the periphery . Bilaterally symmetric pairs of sensory neurons generally run in clos e opposition through the PAG , with the interneurons clustered around them . As in the ventral cord, pre - synaptic fibers display local swellings which bring them into close proximity with the proper pairs of post-synaptic fibers. 129 The anterior bound of the PAG corresponds to the limits of fiber growth for many 10mb a r c e lls. The posterior bound of the PAG corresponds to the limits of fiber growth for many of the ventral cord interneurons (~hose cell bodies are in the lat e ral ganglia in the head) . Within these bounds synapses are distributed rather evenly ; outsi de these bounds synapses are sparse . Almost every PAG synapse involves both lumbar ganglion cells and lateral g anglion interneurons. These synapses can only occur in this s hort region where the distal processes overlap. Dyadic chemical synapses are the major form of synapse in the PAG, but they are not reported to be as common in the ventral cord . The ventral cord has a much higher frequency of electrical synapses (White et al. , 197 6) . The classes of chemical synapses which are typical in the ventral cord are not very common in the PAG and some classes are entire l y ab se nt in the PAG. The interactions of the a and y interne urons with each other are particularly striking in this regard. In the PAG, dyadic a-+ y s ynapses are common, while in the ventra l cord there are many y-+ a synapses instead. Some factor specific to the PAG may exe rt a general influence which regulates fiber growth a nd the probability of synapse formation (p erhaps including the suppression of s ome specific synapse types). The higher synapse density within the PAG ma y be a function of the unique collection of fibers which are brought toge ther here . 130 Our studies of synapses in the tail of Q. elegans reveal that chemical synapses are far more common than gap junctions (electrical synapses). Among the chemical synapses, a majority are dyadic - each pre-synaptic fiber contacting two post-synaptic fibers at the same point. A continuous synaptic cleft is bordered by all three fibers. No morphological specializations identify either postsynaptic fiber. An intracellular electron-dense tuft is associated with the pre-synaptic membrane, and a cluster of synaptic vesicles is generally apparent in the pre-synaptic· terminal. rrhe pre-synaptic tuft genera lly "points II between the two post-synaptic elements. These dyadic synapses appear to play a central role in the functional organization of the tail's circuitry, and probably reflect important developmental processes in synapse specification. Several types of dyadic synapses have previously been noted in other organisms, both in invertebrate and vertebrate nervous systems. Ribbon synapses in the vertebrate retina are dyadic, and comprise 10% of all the contacts in the inner plexiform layer of the retina (Dowling and Boycott, 1966; Dowling and Werblin , 1969). Ribbon synapses are also found in invertebrates, as in the ocellus of the fleshfly, ~- peregrina (Toh and Kuwahara, 1975), but these synapses are not a ll dyadic. The ribbon synapse is characterized by an electron-dense nribbonn which lies 131 adjacent to the synaptic cleft in the pre - sy n aptic termina l. All of the cell membranes at a ribbon synapse are rath er electron-dense, and the spacing between cells is scsewhat wider at the cleft. ribbon. Synaptic vesic le s cluster along the In the fleshfly, the pre-synaptic ribbon often appears T-shaped in cross-section (Toh a nd Kuwabara, 1975). Similar ribbon dyads have a lso been observed in the second optic ganglion of the blowfly (Smith, 1966), in the fly lamina (Trujillo-Cenoz, 1965b; Boschek, 1971), and in the optic lamina of the lobster (H amori and Horridge, 1966). The button synapse is another common form of dyadic synapse, best characterized in the ocellus of the dragonfly (Dowling and Chappell, 1972}. The button sy naps e is similar in appearance to the dyadic contacts in Q. ele gans . The button synapse has an elect ron-dense tuft attached to the pre-synaptic membrane, adjacent to the synaptic cleft. The cell membranes at a button synapse are relatively electrondense, and the synaptic cleft is often somewhat wider and more uniform than the unspecialized intracellular gap . Synaptic veslcles cluster in the pre-synaptic terminal near the tuft. Similar tufted dyadic synapses have recently been described in the lob ster stomatogas tric ganglion, where they comprise the predominant class of chemical synaptic cont acts (King , 1976a). Tufted dyadic contacts have also been reported in th e first synapt ic relay of the spider eye 132 (Trujill o-Cenoz, 1965a) and in the neural pl e xus of the lateral eye of Limulus (Whitehead and Purple , 1970). Most of these examples of dyadic synapses involve sensory gangli a , generally at a primary level of sensory processing. Many of these dyadic synapses are be lie ved to be involved in reciprocal synapses, similar to the reciprocal connection which I have found in the nematode tail. Re9iprocal dyads have previously been found in the vertebrate retina (Dow lin g an d Boycott , 1966; Dowling, 1970), in the spider visual system (Trujillo-Cenoz, 1965e ), in the Limulus plexus (Whitehead and Purple, 1970), in the dragonfly ocellus (Dowling and Chappell, 1972), and in the fleshfly ocellus (Toh and Kuwabara , 1975). Chappell and Dowling (1972) suggest that the reciprocal synap se in visual systems could mediate lateral inhibition among sensory cells to enhance image discrimin a tion or mediate "on" or 11 off" re spo nses - behavioral functions which are well known from electrophysiology of the same visual systems. Reciprocal dyadic synapses in the nematode tail may prove to be the morphologic al basis for med iating "on 11 or sensory stimulation of the tail. 11 off 11 responses to This subject i s discussed further in another section. The functional status of individual synapses cannot be asse ssed on morphological gro und s . In synapse counts I have included some synapses which c a n be found on only a f ew 133 sections near breaks in the series. Most of the counted synapses have groupings of synaptic vesicles nearby. Previous studies by Mark (1970) and Cass, Sutton, and Mark (1973) have suggested that some morphologically intact synapses can be functiona l ly inactive in the frog. It is possible that some of the PAG synapses could also be nonfunctional. Most PAG synapses are grouped into a small number of highly repeated classes. These multiple connec- tions may well be required for a given class of synapses to be functionally significant. There is also a dearth of functional data concerning dyadic synapses, and I cannot be certain whether both post-synaptic members of a dyadic synapse actually receive functional input. All of the eight cell types which are involved in most of the PAG dyads are known to have significant reproducible input and output, which lessens the chance that any of the eight types might lack an active functional role in these synapses. Dyadic synapses in the nematode tail generally involve three non-homologous fibers, producing muliple routes of information flow. Most of these dyads involve a few preferred combinations of fibers. Strict rules for dyad formation may also govern the synapses of other organisms, but these rules can only become apparent when total reconstruction is successful (all fibers must be identified). Few regions of neuropil have yet been studied so 135 of neurons appear to participate in the same sets of synaptic interactions, providing the same ce llul ar signal when acting post-synaptically and requiring the same pairs of cellular signals to act pre-synaptically. There is no evidence that individual s ynapses are uniquely specified . Instead, the formation of. dyadic synapses in the PAG appears to be a probabilistic phenomenon . Synapses occur intermitt en tly in regions where the proper set of fibers are grouped together . Certain dyads are much more common than others de spite appare ntly similar morphological chances for interaction . Each dyadic type appears to have a se parate probability of formation . Dy ads of a given type are not generally spaced apart in an organized pattern, and dy ads of different types are distrib uted independently of one another. formation in the PAG are unknown . The dynamics of sy napse J.1hese dyads may be formed 1 only at certain times in development, or they may be continuously formed and destroyed . The dynamics of synapse formation may be a very important feature in understanding the synapse pattern found in the adult . The orderly growth of the nerve fibers into a stereotyped array may precede synapse formation, limiting the number· of po ss ible sy n apse combinations perhaps involving a transient recognition process similar to that observed in the Daphnia visual system (LoPre st i, 136 Macagno and Levinthal, 1973). Proper groups of pre - a nd post-synaptic fibers might grow together as they enc~unter one another, increasing their chances for interaction and excluding many other possible combinations. Alternat e ly, fiber positions could be altered after synapse formation is relatively complete, with the completed synapses causing certain groups of fibers to hang together during a nonspecific adjustment of relative fiber positions. The forma - tion of a dyadic synapse may require the pre -s ynaptic fiber to identify two proper synaptic fibers within a broad array of choices, or the post - synaptic fibers may use a prior recognition system to pair off before searching together for a proper pre-synaptic fiber. How many cellular recognition factors must be involved in speci fy ing the dyadic synapses of the eight major PAG ce ll types? If eight different factors are secreted , one per cell type, then each pre - synaptic fiber would have to independently assign probabilities of synapse formation to thirty~six possible pairs of dyadic partners - a cumbersome task. However Figure 17 clearly indicates that most pre-synaptic fibers choose a very restricted set of preferred dyadic contacts. There appear to be two different levels of specific control : first, there is a very strong requirement for one particular type of post - synaptic fiber (B) to be involved in every dyad - perhaps donating a 137 primary factor (b); second, there is a restricted set of post-synaptic fibers which 3.re acceptable as the other member of the dyad (C) - perhaps donating a supplemental factor (c) . We have previously noted that 90 - 95% of all dyad s do include the proper donor of primary factor (b). '11 he proper supplementary factor (c) appears to be donated in only about 61% of the dyads, as can be seen in the following listing : TABLE 1 (Data from Pigure 17) (B) o/. ,o of DYADS POST(C) % of DYADS a y 83 ¢ 67 ¢ a 100 0 65 27 , 28 a 98 y 70 0 a 100 A 67 X y 83 27,28 67 17 , 18 y 96 a,x or A? 32? Average 95% PRE - w_ POST- 61% Cells 17,18 clearly do not follow these rules so well , as their second post - sy naptic fibers (C) are not so strict l y specified ( see Figure 17) . Seven different types of fibers are li s ted above as donating recognition factors ; the eighth cell type , cel l s 17,18 , receives very few synaptic - inputs 13 8 and may not supply a functional recognition factor . (Since the se are lineage cel l s , their lack of synap tic inputs might also refl e ct a c essation of synapse formation by the other, non-lineage cell types when cells 17 , 18 wire into the PAG . ) Fewer than seven se p a rate reco g nition factor s may be required, s ince some cell types appear to r e ceive si~ilar cla sses of input : note the similarities in input toy and 6 , and to¢ and A, in Figure 17. The 6 fibers may be donating the same reco gnj_tion factor as the y .fibers , but in lesser concentrations , or supplying a very similar factor . Likewise , the¢ fibers and A fib ers may be donating the same factor in slight l y different form . These similarities are not entirely convinci n g on the basis of so little data , and no compelling , quantitative model of dyadic sy napse formation has yet emerged . Efforts are continuing along these l ines. A small nwnb e r of cellular recognition factors may be uti l ized to go v e rn the form a tion of most of the synapses in the PAG . 'I'h ese same factor s could also be i mportant in the order l y fiber growth which must form the PAG . A mutation of one of these factors c ould cause multiple morphological alternations and associated behavioral ch a nges (such as para lysis or severe uncoordination) , poss ibly with lethal consequences. It ma y prove to be very difficult to i so late viab l e mutants which will be u sef ul for exploring the role s of th ese f a ctors . 139 The study of recessive lethal mutations is cu rrently quite difficult in the nematode. Future work with l ethal mutants may become more feasible with the advent of ba lancer mutants which could maintain these genes as heterozygotes (Herman et al., 1976). The simplicity of the defecation circuit in the PAG suggests that this system may be amenable to genetic dissection. The defecation circuit is monosynaptic: cell 33 + defecation muscles. factor(s) Mutations of the recogn ition responsible for these synapses might be expressed as relatively simple morphological variants of cell 33 and the defecatioh muscle arms, with associated deficits in defecation behavior. Better characterization of defecation mutants could be useful in this regard. Balancer mutants again could be helpful in maintaining "constipated 11 mutants . FUNC'rI0NAL ANAYLYSIS OF '11HE: PRE-ANAL GANGLION CIRCUI'rRY: Most of the PAG circuitry serves primarily to dir ec t sensory information to ve ntral cord. interneuro ns, with resulting behavioral respons es by the whole animal. To ana l yse the function of the PAG circuitry, one must first consider the function of the ventral cord int ern e urons and motoneurons , which control the animal ' s body motion . The organization of the ventral cord circuitry has rec e ntly been reported by White (1975) and White at al. (1976). Their descriptions of the motoneuron and interneuron types h a ve been g ive n in 140 previous sections. The basic wiring di agram of the ventral cord circuitry is shown in Figure 23b (White et a l., 1976). The various synapse types are distributed sparsely along the ventral cord. Some synapse types are more commcn in anterior regions of the cord, others in posterior regions . Some of these synapse types are conspicuously missing in the PAG. Analysis of any part of the circuitry of _Q. elegans is hampered by the inability to obtain electrophysiological recordings of n e uronal activity or muscle activity. The large parasitic nematode, Ascaris, is more amenable to electrophysiology and numerous studies have been made upon Ascaris muscle. The struct ure of Ascaris muscle is sur- prisingly similar to that of _Q_. elegans a nd the layouts of the nervous systems of the two s pecies are very sim ilar. It is expe cted that the physiology of As~aris muscles is much the same as in C. elegans . The physiology of Ascaris muscle has been studied by del Castillo and coworkers (DeBell, Castillo and Sanchez , 1963; de l Castillo and Morales, 1967; de l Castillo, de Me llo and Morales, 1967) , and more recently by Stretton ( stretton , 1976; personal communications) and by We isblat, Byerl y and Russe ll (1976). Acaris musc l e displays myogenic activity, even when all nerve act5.vity has been blocked. Modulation of thi s spo ntan eous muscular activity can be achieved by 141 application of neurotransmitters to the muscle plate : acetylcholine is excitatory and GABA is inhibitory. Gap junctions between ad joining mus cle arms permit adjoining muscles to maintain close synchrony . Waves of muscular contraction can possibly be propagated along the length of the animal eve n in the absence of nervous activity . Crofton (1966) was the first to sugges t that the nematode ' s normal body motion is produced by myogenically propagated muscular contraction . The nervous system may act only to modulate the spontaneous muscle activity. Proprj_oception is probably stretch- activate d, and is probably mediated by the muscles themselves or by the motoneurons of the ventral and dorsa l cords . Current experiments confirm that muscular acti vity is highly sens itive to stretching and bending (Wei sblat , Byer ly and Russell, unpublished observations ). White et al. (1976) suggest that the B and A moto - neurons are best situated to be excitatory and inhibitory motoneurons respectively, one using acetylcholine and the other u sing GABA as neurotransmitters . Class D motoneurons are driven exc lus ively by the A and B motoneurons (Fi gure 23b ) and have NMJ output to the set of body muscles opposing their dendritic region , s uggesting that they set up recipro cal inhibition of opposing sets of muscles - allowing the dorsal side of the animal to relax as the ventral side contacts an d vice versa (White, 1975). White suggests that 142 this reciprocal inhibition is required to generate the initial set when body mot ion is begun, but that it is unimportant during the self-propagated motion thereafter. Forward and backward motion are both achieved by sinusoidal waves of muscular contractions. The nematode lies on on e side so that the dorsal body muscles work in opposition to the ventral body muscles. The l eading end of the body gene rally appears to be undergoing stronge r contractions than the trailing end. White et al. (1976) sugges t th at this gradient of muscle activity is induced by a gradient of motoneuron activity , with muscle oscillations at the more active end of the body entraining oscillations of the remaining body muscles to become the leading end. In this model, sensory stimuli could produce behavioral reversal by acting upon the interneurons of the ventral cord to reverse the gradient of motoneuron act ivity. Gradients of motoneuron activity must be established by gradie nts of intern euron activat ion, wh i ch in turn might be achieved either by gradients of specific synapse types between interneurons and motoneurons or by other mechanisms . White et ~l. (1976) mention only a few synapse gradients along the ventral cord; notably B ➔ a and y ➔ a synapses are mo re common in the posterior ventral cord. No other synapse gradien ts are reported to exis t between interneurons and motone uro ns . One might expect that 8 intern euron output 143 would be more concentrated in the a nterior ventral co rd because there a re four 6 interneurons there , vers u s two in the posterior ventral cord, b ut no gradient invol ving the 6 i n tern e urons has been reported. Among other possible mechanisms for activation grad ie nts, one mu s t conside r the possibility of pa ss ive de cr ementa l spread of activity (without action potentials) in ventral cord interneurons . This mec hani sm would provide the ve nt ra l cord wi th grade d sensory activation (only) . Nematode nerves h av e not yet been impaled reliably with microelectrodes , s o there is no applicable data concerning the ge neration of action potentials . Furthermore , the nematod e nervous sy stem i s so small that g r a ded pot en tials could con ceivab l y be utili ze d to perform much of the system ' s information processing . The space constant (A) of C. el egans nerve fibers c annot be determined directly~ but est im a tes can be made using data from other systems and the ge n eral equation : • RRm A = /(2R· ) where R = radius of fiber ( cm ) l 2 ~ = membrane resi s tance ( St cm ) Ri = cytop las m re s istivity ( St cm) 144 Table 2 shows the expected range of the space constant .for the 0.2-0.4µ (di ameter ) nematode nerves in the P.A.G , assuming that the cable properties are similar to those of .four se lected invertebrate preparations. TABLE 2 If Nematode Cable Properties Cable Prop~rties Space Constant (Expected Range in C. elegans) Squid Nerve 100-140µ Katz (1966) Lobster Nerve 125-180µ Katz (1966) Crab Nerve 200-290µ Katz (1966) Cockroach Giant Axon 50-70µ Pichon (1969) = Data Source These estimates vary rather widely and they may not be valid since the reference preparations are all quite large compared to the very fine fibers of the nematode. However, they do give some indication that synaptic interactions within the PAG may be quj_te capable of integrating sensory data with graded potentials (since most PAG synapses are within 0.5-30µ of each other, probably l ess than the space constants of the fibers involved). King (1976b) has previously suggested that similar neurites in the lobster stomatogastric gang lion, which also lack clearly separate dendritic and axonal diff erentiation , also perform normal synaptic integration of passively conducted potentials. 145 Preliminary evidence (White, 1975) suggests that chemo- sensory signals transduced by amphidial neurons are conducted by a multisynaptic route to the S interneurons (Figure 23c). White suggests that the B motoneurons are excitatory, stimulating stronger contractions of the body muscles to speed the animal's forward motion during chemotaxis. The circuit: amphidial neurons t St t B motoneurons could produce stronger contractions each time the head moves up the gradient of attractant, thus keeping the animal oriented to the gradient. White also found evidence that mechanosensory neurons in the head direct signals by a multisynaptic route to the a and 8 interneurons (Figure 23c). White presumes that the class A motoneurons are inhibitory. The circuit: + + + cephalic mechanosensors ➔ a,8 ➔➔ A motoneurons could mediate reversal of the animal in response to head tap by inhibiting the motion of th e anterior body muscles through the A motoneurons . My own data show that mechanosensory s timulus of the tail is routed particularly to the a, y, and 8 interneurons, with no S interneurons involvement (Figure 23a). They i~terneurons do happen to have synaptic targets similar to those of the 6 interneuro ns in the ventral cord (Figure 23b) (White et ~l., 1976). Whe n all o f these studies on the nervous circuitry of C. elegans are combined, one can b egin to speculate on the 1 46 Fi g ure 23 . Wiring Diagrams (a) Wiring diagram for the pre - anal ganglion . common connecticns s h own as heavy arrows . Mos t (b) Wiring dj_ag ram for the ventra l cord (h'hit e et a l., 1976) . (c) Wiring diagram (partial) for sensory in puts to ventral c o rd interneurons in the head (Wh ite , personal comm unication). Ph - phasmidi al neurons (27 , 28) TT - tail tip mechano se n so rs (17,18) As - asymmetric mechanosensor (19) A,B , D - motoneurons Am - amph i gial neurons (dotted line indicates output from finger and wing cells) Ce - mechanosensory ( a nd chemosensory?) neurons from papillae in the head RI - nerve ring inte r n e urons OI - nother 11 interne urons 147 a C Figure 23 148 possible functions of the ven tr a l cord interneurons in mediati n g behavioral responses to sensory stim ul ation . In mak ing these speculations, the following assumpt i ons are used : 1. Mu s cl e cells prop agate contactile waves along the len gt h of the animal by their musc le arCT interactions with nei ghb orin g musc le cells . 2. Muscle cell s produce spike activity independently , in response to stretch activation . 3. a ctivity : Three classes of mo ton e uron s mo dulate muscle Class A motoneurons are inhibitory . Class B motoneurons are excitatory . Cl ass D motoneurons se t up reciprocal inhibition of p a rall e l dor sa l and ventral musc l e ce l ls . Cl ass D motoneurons must be e xcit ed by Cl ass A and inhibited by Class B motoneurons , or v ice versa . 4. I n the ventral c ord : + A a,o ➔ + S , y ->- B + S , y + o, ( Synapse gradient s make these more com.rno n in pos ter ior regions .) B -+ A 5. Gap junctions b e t wee n n eighb oring motoneurons of a given clas s are s ufficient to maintain synchrony among them, but not sufficient to prop agate waves of excitation or inhibition without decreme nt . 149 When these assumptions are taken into account , there are two general alternatives which could explain the operation of the ventral cord interneurons. Alternative 1. Sensory stimuli which activate the S or y inierneurons automatically produce forward motion. Sensory stimuli which activate the a or 8 interneurons automatically produce backward motion . The following excitatory and inhibitory connections would be likely: amphidia l chemosensors + ➔ s + ➔ -", foward motion B ··~ + c~., 8 + A cephalic mechanosensors ➔ ➔ tail mechanosensors + -➔ y + ➔ B l-; a,8 ----7 backward motion ==--1 forward motion inhibits backward motion By what means could Sandy interneurons promote forward motion? They might have more S,y + ➔ B synapses in the anterior end to selectively excite the anterior body muscles. However, no synapse gradient of this type has been observed (Whit~ et al. 3 1976). synapse gradients favor more S,y rior ventra l cord. Instead , the observed ➔ a synapses in the poste - These synapse gradients might mediate selective inhibition of the posterior body muscles if the a interneurons are excited to subthreshold l evels which are still sufficient to cause a t A stimulation in the posterior regions. If the stimulation of a interneurons did reach 150 threshold, their action potential would cause general inhibition of the whole animal and failure to .mov -"' orv1 ard . e 1 By what means backward motion? could a and o interneurons promote + A synapses in They might have more a,o ➔ the anterior ventral cord to selectively inhibit the anterior body muscles . However, no synapse gradient of this type has been noted (White et al . , 1976). Alternative 1 will on1y be a useful hypothesis if new synapse gradients can be found in the ventral cord, otherwise it seems unsatisfactory. Alternat~ve 2 . Mechanosensory stimuli to the head and tai l are compared by the a and 8 interneurons. The end of the animal receiving stronger mechanical stimuli becomes relatively quiescent through local a,o i A inhibitory effects. These comparisons could be accomplished best if the a _and o interneurons conduct signals only by passive spread without action potentials . The a and o interneurons are postulated to be quite in e xcitable, with thresholds to o high to allow action potentials despite massive multiple inputs. Sandy interneurons could set the general level of excitation of the nematode's behavior . These interneurons respond to sensory stimulation by carrying signals with little or no decrement (by action potentials) to the entire ventral cord. 151 The following connections would be likely : + amphid chemosensors -+ cephalic mechanosensors tail mechanosensors 8 + -+ B + + + -+ CY, , O -+ -+ lt - A + y -+ B + A -- CY, , O -+ . ,., s tronger contractions ,local inhibition of anterior muscles ,, stronger contractions )local inhibition of posterior muscles In this model, the observed gradients for + a S,y-+ synapses could serve to reinforce the dominance of the anterior body musc l es . (This dominance is known to cause the animal to strongly favor forward motion .) dominance of 8 , y The pre- ta contacts in the posterior ventral cord should lead to selective inhibition of posterior body muscles by + A contacts . a-+ Alternative 2 is dependent upon the differing thresholds and conduction properties of the interneurons , although it does not violate any general principles of neurophysi ology . The detailed behavior of the hermaphrodite is not well known , but this alternative is sufficien t to explain the known responses in a general manner . In this model , the feedback circuits of the PAO can mediate filtering or enhancement of sensory stimulation of the tail. Also> the y interneurons could be useful in guiding backwards motion towards an attractive chemosensory stimulus transduced by the phasmids , in the same manner that the S interneurons are thought to guide forward chemotaxis . This same circuit 152 could be elaborated in the ma l e tail to provide a range of attractive responses during backward motion . Accepting Alternative 2 as th e more probable hyp ot hesis , one makes the following additional assumptions ; 6. 7. In the PAG : a and + y,a,o (17,18,19,27,28) ->- a ,o ➔ o interneurons have no action + A a :t ¢ ,Y ¢ ➔ + y,a ,o potentials ~ but passively conduct graded s e nsory signals along their l engt h. Sandy interneurons generate action potential s in response to sensory stimuli . The additional assumptions are redundant in some For instance, the properti es of the 8 inter- respects . n e urons may be somewhat different from the a interneuro ns wi t hb u t a l tering the genera l mode of operation of the whole system . The interaction between a and¢ interneurons is postulated to be posit i ve feedba ck, but could inste ad be performing more compl ex senso ry filterin g . Inhibitory feed - back and the proper timing de l ays in the aC::¢ circuit might me diate an 11 on 11 or noff 11 re s ponse . The operation of the¢ int ern eurons is quite striking. Their synaptic o utput goes to very s imil ar targets when compared to the output of the sensory cells (17,1 8 , 19 , 27 , 28 ). The interaction of the¢ interneurons with the lat era l 153 cords (cells 9~l0) suggests that the¢ interneurons may convey sen s ory signals from the post-dejrids to the same set of interneurons which receive other sensory input in the tail. If these¢ outputs are excitatory, they could be responsible for the animal's responses to mechanical stimulation of the post-deirids. These responses are similar to the animal's responses to tail tap. There are multiple feedback loops, multiple point synapses and multiple routes leading to the excitement of the ventral cord interneurons a, y, and o. When one assumes that most or all of the major FAG synapses are excitatory, as in Alternative 2, the response to any tail stimulus will be maximized. The multiple synapses and multiple routes to y excitation insure that these fibers reach their threshold for producing act~on potentials (subthreshold y excitation will be counterproductive to the generation of forward motion). The formidable multiplicity of inputs to a inter- neurons may reflect their relative inexcitability. Highly repeated synapses may also reflect the need for a "safety factor" to insure proper function in a system with many mistaken connections. Multiple routes of a and y interneuron excitation could also mediate more complex s ensory filtering, using timing features of the stimuli . Inhibitory circuits might act to restrict the sensitivity of the interneurons to stimu li with 154 the proper intensity. The spatial configuration of the feedback loops is so variable that timing features ~ould seem to be muddled in these circuits. There is no data concerning the range of stimulus sensitivity possessed by the tail. Without better physiological data , these specu- lations are not worth carrying further. Our assumptions about the PAG circuitry presume a predominance of excitatory connections. invertebrate systems , notably In some other in the lobster stomato- gastric ganglion (Mullaney and Selverston, 1974a,b; Selverston and Mullaney , 1974), inhibitory connections are known to be very commonplace. The true circumstances in nematode neuropil are unknown, but I could interpret the PAG circuitry in similar terms without tearing apart al l of the functional interactions . (For instance , the B moto- neurons could be spontaneously active inhibitory motoneurons. Bandy interneurons could respond to sensory activiation by s e nding inhibitory signals to the B moto neurons, with the net effect of increasing muscle excitation.) Modelling all of the PAG connections as inhibitory synapses is rather cumbersome , so I have neglected this analysis until I have strong8r reasons to do so. This hypothetical model of the nematode ' s nervous system does not seem to make full u se of the possibilities 1 55 provided by the dyadic synapse. Multipie routes of infor- mation flow and simultaneous synaptic signals offer a wealth of potential timing features. Computer modell~ng of the complete nematode wiring would probably be required to fully assess these capabilities, but these efforts are premature until more animals have been reconstructed tci better judge the variability of the spatial separation of dyads. Even in these two animals , B126 and B136, the degree of variability seems rather large to allow for sophisticated timing mechanisms, especially since there are so few synapses involved. The random addition or loss of just a few synapses of a given type might be sufficient to greatly up s et the normal timing delays. Com.plex timing features are we l l - known in the thirty - cell itomatogastric ganglion of the lobster (Morris and Maynard, 1970 ; Mul l aney and Selverston, 1974b ), but these few neurons uti l ize one million synapses in the operation of the gangl ion (estimated by King, 1976a ). 156 SUMMARY 1. All of the cells in the C. elegans tail can be individually identified and are basically identical in several animals from an isogenic strain . 2. Sensory neurons, motor neurons, and interneurons can be identified and classified by their characteristic morpho logies and sy naptic connections . 3. Complete synaptic interactions among identified cells in a sensory - motor ganglion can be catalogued and compared from animal to anima l. 4. Synaptic patterns are stable with minor variations perhaps reflecting the probabilistic nature of synapse formation . Twenty - five per cent of al l synapses are apparently single random 11 mi s takes . 11 11 Important 11 synapse types are hi gh l y repeated . 5. The dyadic synapse is the predominant class of chemical synapse in the tail . This synapse is the morphological basis for the simultaneous stimulation of multiple routes of information flow in sensory processing , probably as a means of sensory filtering or sensory enhancement . 6. Dyadic synapses di sp lay very strict rules of synapse 157 formation . A. very few types of dyads are favored over all others . Two types of interneurons (a and y) are always the primary synaptic targets . 7- Projections from anterior interneurons can be identified by their characteristic fiber positions and synaptic int eractions. They interneurons , with cell bodies in the lumbar ganglion, can also be identified by their fiber positions in the anterior end of the reconstructed region . 8. Rather complete wiring diagrams are now available for the tai l and the ventral cord ; the wiring diagram for the cephalic region i s not known in deta il. These v,Ti ring diagrams allow us to exp l ore the behavior of the nematode in terms of defined motoneurons , interneurons , and sensory inputs . 9- Lin eage cel l s in the lumbar gangli a have been identified. Many of these cells are involved in sensory transducion in the adult nematode . 10. Lin eage cells in the pre - ana l gang lion have been identified . Two of these cells do not follow the gene ral pattern for their l i neage ; four other lineage cells become motoneurons, possib l y of the proper classes for their lineage. 158 11 . Phasmidial cilia (c el ls 27- 30) , the asymmetric cilj_um (cell 19 ), and the unciliat ed dendrites of se nsory cells 17 and 18 may all be sensit ive to mechanic a l stimuli . Sp e cific chemosensory respons es have not yet been identified . Several different mean s o f sensory trans - ducticin may be utilized in the r eceptio n of mechanica l stimuli by these neurons . 15 9 REFERENCES Albertson, D.G . and Thomson, J.N . (1976) The pharynx of Caenorhabditis elegans, Proc . Roy . Soc . London, Series B , in press. Benzer, S. (1971) From gene to behavior., JAMA , 218, 1 018 . Bird, A. F. (1971) The Structure of Nematodes., Academic Press, New York . Boschek, C.B . (1971) On the fine structure of the periphera l retina and lamina ganglionaris of the fly., Musca domes ti ca , Z . Sel l forsch . Mikrosk . Anat. , 11 8, 369 . Brenner., S . ( 1 973) The genetics of behavior , Br. Med . Bulletin , ~ , 269. Br enn er, S . C,J-9 7 4 ) The genetics of Caenorhabditis e l egans , Genetics , 7 7 , 7L Bul l ock , T. H. and Horridge , G. A. (1966) Structure and Function in the Nervous System of Invertebrates , W.H. Freeman and Co ., San Francisco ., 2 vol . Cass , D. T . ., Sutton , T. J . and Mark , R . F. ( 1973) Competit i on between nerves for functional connexions with Axolotl muscles, Nature , 243 , 201 . Chappell ., R . L . and Dowling , J . E . (1972) Neural organization of the median ocellus of the dragonf l y, I. Intracellular e l ectrical activity , J. Gen . Physiol . , 60,121 . Chitwood , B.G . and Chitwood, M. D. (1950) An Introduction to Nematology , Mo numental Printing , Baltimore . Chitwood, B.G. and Wehr, E.E . (1934) Th e value of cephalic structures as characters in nematode classification , with special reference to the superfamilf S~i~uroidea, Z. Parasitol . , l , 273 . Crofton, H . D. (1966) Ne~atodes , Hutchinson University Library , London. DeBell , J.T . , Castillo , J . del, and Sanchez , V. L . (1963) Electrophys iolo gy of the somatic muscle cells of Ascaris luIT,bricoides., J . Cell. Comp . Physiol . , ~ , 159. del Castillo, J ., de Mello, W. C., and Morales , T . (1967) The initiation of action potentials in th e somat ic musculatu:'.:"e of Ascaris lU:mbricoides, J . Exp . Biol . , ~ , 263. 160 del Castillo, J. and Morales, T. (1967) The electrical and mechanical a ctivity of the esophageal cell of Ascaris J.umbicoides , J . Gen . Physio l., 2.Q_, 603 . Dowling, J . E . (1 97 0) Organization of vertebrate retinas , I nvest . Dphthalmo l., ~, 655 . Dowling, J . E . a nd Boycott, B . B . (196 6) Organizat ion of the primate retina : e l e ct ron microscopy, P~6c . Roy . So c., London, Series B, 166, 80 . Dowling, J.E. a nd Ch appel l, R.L. (1972) Neura l organization of the median oce llus of the dragonfl y , I I. Synapt ic str uctur e , J. Gen. Physiol ., ~ , 148. Dowling , J . E . and Werb lin, F . S . (1969) Organiz ation of the re tina of the mudpuppy, Necturu s mac ulo sa , I. Synaptic structure, J. Neurophysi.ol ., E , 315. Dusenbery , D.B . (1973) Counter current separation: A new method for s tudyin g behavior in small aq u at ic organisms , PNAS, 70, 1349. Dusenbery, D.B. (1974) An alys i s of chemotaxis in the nematode Caeno rhabditis e l egans by c ounterc urrent separation , J. Exp . Zool ., 188, 41. Dusenbery, D. B . (1976) Genetic di sse ction of chemotaxis of the nem atode Cae norhabditi s e l egans, I. Chemo t ac tic behavior of mutants alt e r e d in their response to NaCl, in press. Dus enbery , D. B. , Sherida n, R.E., and Ru sse ll, R.L. (1975) Chemot axis - defe ctive mutant s of the nematode, Caenorhabditis e le gans , Genetics , So , 297 . Goldschmidt , R . (1903) Histo l ogische untersuchungen an nematoden., zoo1. · J a hrb·. · Anat ., 18, 1. Goldschmidt, R. (1908) Das nervensystem van Ascaris lumb ricoides und me gaToc ephala, I., Ztschr. Wiss. · Zool., 2..Q_, 7 3. Go l dschmidt , R . (1909) Idem. II., Ztschr . Hiss . Zoo l.., ~., 306. Goldschmidt , R . 256. (1910) Idem . III., Festschrift Hertwig., ~, 161 Hamori , J . and Horridg e , G. A. lamina , J . Cell S~i ., ! , 257 . Harris_, W. Caltech . (1966 ) The lobster optic (1976) Color vision irt Dros ~p hi l a , Ph . D. thesis , Harri s , W. , Stark , W., a nd Wa lker , J. (1975 ) Ge netic dissection of the photoreceptor system in the compound eye of Dro s ophila rnelano g aster, J . Physiol . , in press. Hartline , D. K. and Maynard, D. M. (1975) Motor patterns in the stomatog astric gan g lion of the lob s t e r Panulirus arg u s , J . Exp . Biol. , 62, 405 . · Hedgecock, E . M. and Ru s sell, R. L . (1 975) Norma l and mutant thermotaxis in the nematode , Cae rtorhabditis elegans , PNAS , JJ._ , 4 06 1. Henkart , M. (1 975 ) A morpholo g ical basis for transcellul ar transfe r of macromolecul es , Neu~osci~rtce Abgt~acts 1, no. 1172 , 762 . Herman, R.K ., Albertson , D.G ., and Brenne r, S . (1 976 ) Chromosome rearrangements in Cae:: norhabditis el e gans , Genetics , fil, 91 . Hesse , R. (1892) Ueber das nervensystem von Ascaris megal o~ cephal a , Z . Wiss . Zool. , 54 , 548 . Hodgkin , J. A. (19 74) Genetic and anatomica l aspects of t h e Caenorhabditis eTegans male , Ph.D . thesis , Cambridge . Hot ta , Y. and Benze r, S . ( 1972 ) Mapping of behavior i n Drosophila mosaics , PNAS , 67 , 1156 . Hughes , G. M. and Tauc , L . (1962 ) Aspects o f the organization of central nervous pathways in ApTysia depilans , J . Exp . · Biol ., -12_, 45. Jan , V. N. and Jan , L.Y . (1976) A s ynaptic transmiss i on mutant of Drosophil a melano g aster , in press . Jansen , J.K.S . and Nicholls , J . G. ( 1972) Regeneration a n d changes in synaptic connections between individual nerve cells in the central nervous system of the leech , PNAS , §_9 , 6 36 . 162 Kankel, D.R. and Hall, J.C. (1 975) Fate mapping of nervous system and other tissues in genetic mosaics of Drosbphila melahogaster, Develop. Biol., in press. Kandel, E.R., Frazier, W.T., Waziri, R. , and Coggeshall, R.E. (1967) Direct and comJnon connections among identified neurons in ApTysia, J. Neurophysiol., lQ_, 1352. Katz, B. New York. ( 19 66) Nerve, Muscle and Synapse, McGraw-:Hill, Kennedy, D., Selverston, A.I., and Remler, M.P. (1969) Analysis of restricted neuronal networks, Science, 164, 1488. King, D.G. (1976a) Organization of crustacean neuropil, I. Patterns of synaptic connections in lob ster stomatogastric ganglion, J. N~uro~ytol., ~, 207. r King, D.G. (1976b) Organization of crustacean neuropil, II. Distribution of synaptic contacts on identified motor neurons in lob ster stomatogastric gang lion, J. Neurocytol., 2, 239. Kristan, W.B., Jr., Stent, G.S ., and Ort, C.A. (1974a) Neuronal control of swimming in the medicinal leech, I. Dynamics of the svrirnrning rhythm, J. Comp. Physio l., ~, 97. Kristan, W.B., Jr. , Stent, G.S., and Ort, C.A. (1974b) Neuronal control of swimming in the medicinal le ech , - III. Impulse patterns of the motor neurons, J. Comp. Physiol., 94,155 . Landis, D.M.D. and Reese, T .S. (1974) Differences in membrane structure between excitatory and inhibitory synapses in the cerebellar cortex, J. Comp. Neurol. , 155, 9 3. Lasek, R.J., Gainer, H.~ and Przybylski, R.J. (1974) Transfer of newly synthesized proteins from Schwann cells to the squid giant axon, PNAS, 71, 1188. Le v inthal, C. and Ware, R.W. (1972) Three-dimensional reconstruction from serial sections, Nature, 236, 207. Lo-::>s s , A. (1905) The anatomy and life history of Agch ylostoma duodenale , Rec . Egypt . Gov ' t. School Med ., l, i. 163 LoPresti , V.~ Macagno , E . R., and Levinthal, C. (1973) Structure and development of neuronal connections in iso geni c organisms : Cellular interactions in the development of the optic lamina of Daphnia , PNAS , 70, 4 33 . Macagno , E . R., LoPrest i, V., and Levinthal , C. (1973) Structure and development of neuronal connections in iso genic organisms : Variations and similarities in the optic system of Daphnia magna, PNAS, IQ , 57 . J\•I ark , R. (1970) Chemospecific synaptic repression as a possible memory store , Nature , 225 , 178. Martini , E . (1903) Ueber furchung und gastrulation be i Cucullanus elegans , Ztschr . Wiss . Zoo 1., ~ , 50 1. Martini , E . (1906) Ueber s ubcuticula and seitenfelder einiger nematoden. I., Ztschr . Wiss . Zool ., 81 , 699 . Marti ni, E . (1 909 ) Idem . IV ., Ztschr . Wiss . Zoo l., .2.J.., 535. Martini , E . (1916) Die anatomie der Oxyuris cu~vula , Ztschr . ·wiss . Zool., 116, 13 7. Maynard , D. M. 193 , 59 . (1972) Simpler networks , Ann . N. Y. Acad . Sci ., Maynard , D. M. and Selverston , A.I. (1975) Organization of the stomatogastric gan g lion of the spiny lobster, IV . The pyloric system, J . Comp. Phys iol, 100, 161. Morris, J. and Maynard , D. M. (1970) Recordings from the stomatogastric n ervo us system in intact lobsters , Comp . Biochem . Physi ol., TI, 969 . Mul laney , B. and Se l verston , A.I . . (1974a) Organization of the stomatogastric ganglion of the spiny lobst er , I . Neurons driving the lat era l teeth , J . Comp. Physiol ., 21._ , 1. Mullaney , B. and Selverston , A.I. (1 974b ) Organization of the stomatogastric ganglion of the sp iny l obster , III . Coordination of the two subsets of the gastric system , J . Comp . Physiol ., 91, 53 . Nicholls , J . G. and Baylor , D.A . (1968) Specific modalities and receptive fi e lds of sensory neurons in the central nervous system of the l eech , J . Neurophys iol., 31, 740 . 164 Nicholls , J . G. and Purves , D. (1970) Mo no synap tic che mica l and electrical connexions between sensory and motor cells in the central nervous system of the leech, J . Physiol . , 209 , 647 . Pichon, Y. (1969) Ph.D . thesis , U. of Rennes ; in Parnas , I . and Dagan , D. ( 1971 ) Functiona l organization of giant axons , Adv . Ih~~d~ Physidl ., ~, 95 . Ort , C. A., Kristan , W. B ., Jr ., and Stent , G. S . (1974) Neuronal control of swimming in the medicinal leech, II . Iden tification and connections of motor neurons, J. Ccmp . Physiol ., ~ , 121 . Reynolds , E . S . (1963) The use of lead citrate at high pH as an electron-opaque stain in e l ectron microscopy , J . Cell Biol ., 1 7 , 208. Rosenbluth , J . (1965) Ultrastructure of somatic muscle cells in Ascaris lumbricoj_de s , J . Cell Biol . , ~, 5 79 . Schneider , A. (1866 ) Monographie der nematoden , Berlin . Selverston , A. I . and Kennedy , D. (1969) Structure and function of ide ntified nerve cells in the crayfish , Endeavo ur , 28 , 107 . Selverston , A. I . and Mul l aney, B . (1974 ) Organization of the stomatogastric ganglion of the spiny lob ster , II . Neurons driving the medial tooth, J. Comp . Physi61., 91, 33 . Smith , D. S. (1966 ) The organization of the insect n europ ile ; in Wiersma , C. A. G. (Ed . ) , Invertebrate Ne~vous Systems , U. Chicago Press , Chicago, 79 . Stretton , A. O. W. (1976) Anatomy and development of the somatic musculature of the nematode Asc·aris , in press . Strumwasse r , F. (1963) A circadian rhythm of activity an d its en dog en ous origin in a neuron, Fed. Proc . (FASEB) , 22, 220 . Strumwasser , F. (1967) Types of information stored in single neurons ; in Wiersma , C. A.G. (Ed . ) , Ihvertebrate Nervous Systems , U. Chicago Press , Chicago , 291 . 165 Strumwasser, F. (196 8) Membrane and intracellular mechanisms governing endogenous activity in neuron s ; in Carlson, F.D. (Ed.), Physi·ologicaT and Biochemical Aspects of Nervous Inte gration , Prentice:-Hall, New Jersey, 329. Stuart, A.E. (1970) Physiological and morpholog ical properties of motoneurons in the central nervous system of the leech, J. Physiol., 209, 627. Sulston, J.E. (1976) Post-embryonic development in th e ventral cord of Caenorhabditis elegans, Proc. Roy. Soc. London, Ser~~~ B, in press. Sulston, J.E. and Brenner, S. (1974) The DNA of Caenorhabditis elegans, Genetics, 77, 95. Sulston; J.E., Dew, M., and Brenner, S. (1975) Dopaminerg ic neurons in the,. nematode, Caenorhabditis elegans, J. Comp. Neural., 16 3, 215. _ Toh, V. and Kuwabara, M. (1975) Synaptic organization of the fleshfly ocellus, J. Neurocytol., i, 271. Trujillo-Cenoz, 0. (1965 a ) Some aspects of the structural organization of the arthropoid eye, CSH Symp. Quant. Biol., 30, 371. Trujillo-Cenoz, 0. (1965 b) Some aspec ts of the structual organization of the intermediate retina of Dipterans, J. UltrastrU:ct. Res., 13, 1. Voltzenlogel, E. (1902) Untersuchungen ueber den ana~ tomischen und histologischen bau des hinterendes von Ascaris megalocephala 1.md Ascaris lumbricoides, Zool. Jahrb., Abt. Anat ., 16, 481. -Ward , S. ( 1973) Ch emotaxis by the n ematode Caenorhabditis' elegans: Identifications of attractants and analysis of th e response by use of mutant s , PNAS, IQ, 817. Ward , S., Thomson, J.N., White, J.G., and Brenner, S. (1975) Electron microscopical recon s truction of the anterior sensory a natomy of the nematode, Caenorhabditis elegans, J . Comp . Neurol., 160, 313. 166 Ware, R . W., Clark, D., Crossland , K. , and Russell , R.L. (1 975 ) The nerve ring of t he nematode Caenorhabditis e l egans : Sensory input and motor output , J . Coh1p . Neural., 162 , 71. Ware , R. W. arid LoPresti , V. (1975 ) Three -di mensional reconstruction from serial sections, Tnt. Rev. Cytol ., 40 , 325 . Weisblat , D.A ., Byerly , L., and Russe ll, R. L . (1 976) Ionic mechanisms of electrical activity in somatic musc l e of the nematode Ascaris lumbricoides, J . Comp. Physi ol., in press . Weisblat , D.A. and Russel l, R.L. (1976) Propagat i on of electrical activity in the nerve cord and musc l e sy n cytium of the nematode Ascaris lwnbricoides , J. Comp. Physio l., in press. White , J . G. (1975 ) The structure of the ventral cord of Caenorhabditis elegans ., Ph.D . thesis, Cambridge . White , J.G ., Southgate., E. , Thomson , J.N., and Brenner ., S. (1976 ) The structure of the ventral nerve cord of Caenorhabditis e l egans, Proc . Roy . Soc. London , Series B, in press . Whitehead , R. and Purple, R. L . (1970 ) Synaptic organization in the neuropil of the lateral eye of Limul us , Vision Res ., 10, 129. Wiersma , C.A.G. (1 952) Neurons of arthropods , CSH Symp . Qu a nt. Bio l., 17, 155. Wiersma , C.A.G. and Bush , B . M.H. (1 963 ) Functional neuronal connections betwee n t he thoracic and abdomin a l c ords df the crayfish , Procambarus clarkii , J . Comp. Neura l., 121~ 207. Wiersma , C.A.G. and Hug h es , G. M. (1 96 1) On the functional anatomy of ne u ronal units in the abdominal cord of the crayfish , Procambarus clarkii (Girard), J . Comp. Neural . , 116 , 209 . Wiersma , C.A.G. and Ikeda , K. (1964) Int e rneurons commanding sw immeret movements in the crayfish , Procambarus clarkii (Girard), Comp. Biocherri . Physio l., g , 509 . Wiersma., C.A .G., Ripley, S.H ., and Christen se n, H.B. (1 955 ) The central representation of sensory stimulation in the crayfi sh ~ J. Cell. Comp. Ynysiol., ~ , 30 7 . 167 • APPENDIX I: Cells in the Hermaphrodite Tail (C ompare with Figures 6.,7J8,9) Accessory Cells Cell No . L eft Side 1 2 Right Side Lineage Wing Cell L212 Wing Cell L212 Cap Cell 3 4 L211 L211 Cap Cell Pocket Ce ll 5 Pocket Ce ll 6 Lateral Gland 7 8 Lateral Gland Ll221 Ll221 Lumbar Neurons Tubule Filled, Lateral Cord 9 10 Tubule Filled, Latera l Cord 11 12 Lateral Cord Latera l Cord Dorso-l a tera l Cord Neuron 15 16 L22212 Dor s o-l a teral Cord 13 14 L222 l2 Neuron Ll2 22 Ll222 168 Cell No . Left Side 17 18 L22211 Asymmetric Cilium (Sensory) Migrant Cell Neuron (Possible Pacemaker for 23) 1221 1221 y 1 Interneuron y 2 Interneuron ¢1 Interneuron 25 26 L22211 Neuron (Possible Pacemaker for 24) 23 24 Tail Tip Dendrite (Sensory) Unpaired Neuron 21 22 L1neage Tail Tip Dendrite (Sensory) 19 20 • Right Side ¢ 2 Interneuron Phasmidial Cilium A 27 28 Phasmidial Cilium A Phasmidial Cilium B 29 30 Phasmidial Cilium B Neuron 31 32 Neuron Dorsorectal WeU:rons 33 Defecation Motoneuron (?) 34 Neuron 35 Neuron I 169 PAG Neurons Lineage Gel l No . 36 DD Motoneuron (?) J cell (?) 37 Neuron Pl2d 38 DA Motoneuron J cell ( ? ) 39 VD Motoneuro n (.?) Pl2e 40 Heuron 41 Neuron 42 VA Motoneuron ( ? ) Inactive 43 X Interneuron 44 Neuron 45 Neuron Plle 46 DAS Motoneuron (?) Pll d 47 VA Motoneuron (?) I nactive Pl la Pl2a