Lampreys and hagfish, which together are known as the cyclostomes or 'agnathans', are the only surviving lineages of jawless fish. They diverged early in vertebrate evolution, before the origin of the hinged jaws that are characteristic of gnathostome (jawed) vertebrates and before the evolution of paired appendages. However, they do share numerous characteristics with jawed vertebrates. Studies of cyclostome development can thus help us to understand when, and how, key aspects of the vertebrate body evolved. Here, we summarise the development of cyclostomes, highlighting the key species studied and experimental methods available. We then discuss how studies of cyclostomes have provided important insight into the evolution of fins, jaws, skeleton and neural crest.

A small number of brain neurons project to caudal levels of the spinal cord in the larva of the teleost Brachydanio rerio. These cells were identified in animals 6 days after fertilization by backfilling with horseradish peroxidase following transection of the cord at the level of the cloaca. In preparations with the most labeled cells a total of 30-40 were present on each side of the midline. They were located within three regions of the brainstem: the midbrain nucleus of origin of the medial longitudinal fascicle (mlf), the hindbrain reticular formation, and the hindbrain vestibular nucleus. A total of 15 classes of cells could be distinguished by soma positions, dendritic fields, and axonal pathways. For some of these classes only one or two cells were usually present on each side of the brain. Most of the labeled cells contributed axons to the mlf ipsilateral to the soma; however, the Mauthner cells and three new types of hindbrain reticulospinal reticulospinal cells have decussating axons that enter the contralateral mlf. The observed distribution of labeled reticulospinal cells is similar to that previously described for large reticular cells in adult teleosts and to the system of identified Mauthner and Müller cells in the lamprey.

In the lateral edge of the "white matter" in the lamprey spinal cord, there is a group of nerve cells referred to as edge cells. The results of a combined physiological, light microscopical, and electron microscopical study suggest that these cells serve as intraspinal mechanoreceptors. Edge cells are depolarized on stretch of the lateral margin of the spinal cord, and they have nestlike ramifications in this region oriented in a rostrocaudal plane. These cells exhibit a close structural similarity with the crayfish stretch receptor.

The regeneration of large unmyelinated axons following transection of the spinal cord was studied in small larval sea lampreys (Petromyzon marinus). Individual Müller and Mauthner axons normally occur in a characteristic pattern in the spinal cord, but their positions were altered in the first several segments caudal to the lesion following regeneration. Müller axons grew out of the ventral tracts and sometimes looped back towards the brain or crossed the midline; maximum misdirection of axons occurred near the site of transection Mauthner axons frequently bifurcated. Despite the aberrant and incomplete regeneration of axons, the larvae exhibited normal coordinated swimming ,crawling, and coiling behavior.

The structural organization of the lamprey extratelencephalic forebrain is re-examined from the perspective of the prosomeric segmental paradigm. The question asked was whether the prosomeric forebrain model used for gnathostomes is of material advantage for interpreting subdivisions in the lamprey forebrain. To this aim, the main longitudinal and transverse landmarks recognized by the prosomeric model in other vertebrates were identified in Nissl-stained lamprey material. Lines of cytoarchitectural discontinuity and contours of migrated neuronal groups were mapped in a two-dimensional sagittal representation and were also classified according to their radial position. Immunocytochemical mapping of calretinin expression in adjacent sections served to define particular structural units better, in particular, the dorsal thalamus. These data were complemented by numerous other chemoarchitectonic observations obtained with ancillary markers, which identified additional specific formations, subdivisions, or boundaries. Emphasis was placed on studying whether such chemically defined neuronal groups showed boundaries aligned with the postulated inter- or intraprosomeric boundaries. The course of diverse axonal tracts was studied also with regard to their prosomeric topography. This analysis showed that the full prosomeric model applies straightforwardly to the lamprey forebrain. This finding implies that a common segmental and longitudinal organization of the neural tube may be primitive for all vertebrates. Interesting novel aspects appear in the interpretation of the lamprey pretectum, the dorsal and ventral thalami, and the hypothalamus. The topologic continuity of the prosomeric forebrain regions with evaginated or non-evaginated portions of the telencephalon was also examined.

This paper reviews the development of our research on the motor consequences of Mauthner cell function and related brainstem neurons. These cells activate fast-start responses such as seen in fishes escaping from predatory attacks. Our goal was to devise a neuroethological theory of fish escape that accurately reconciled the underlying neural function with a correct concept of the motor act. The identified neuron concept of invertebrates greatly influenced the initial studies. Horseradish peroxidase technology allowed us and other workers to identify principal neurons in the brainstem escape system. Digital imaging technology permitted adequate kinematic characterization of the behavior. Resulting experiments showed that Mauthner system demonstrates two general principles of motor organization:(1) the Mauthner cell is a command-like higher order neuron that serially outputs to a lower level central pattern generator; and (2) the Mauthner cell participates in a larger parallel, brainstem escape network. In this network, we showed that the spatio-temporal pattern of activity codes the timing and magnitude of agonist and antagonist trunk muscle contractions during the behavior. Because the approach angle of the stimulus determines these parameters, we were able to discover the overall sensorimotor relationship between stimulus angle and motor output. This relationship is given as a set of descriptive equations written in terms of stimulus angle, magnitude and timing variables of trunk muscle contractions, and resulting escape trajectory. The equations unify the apparent variability of C-start movement patterns into a single, quantitative theory. Recent studies by other workers show how this concept can make accurate predictions about the underlying neural processes, even at the level of the single, identified cell.

Among the advantages offered by the lamprey brainstem and spinal cord for studies of the structure and function of the nervous system is the unique identifiability of several pairs of reticulospinal neurons in the brainstem. These neurons have been exploited in investigations of the patterns of sensory input to these cells and the patterns of their outputs to spinal neurons, but no doubt these cells could be used much more effectively in exploring their roles in descending control of the spinal cord. The variability of cell positions of neurons in the spinal cord has precluded the recognition of unique spinal neurons. However, classes of nerve cells can be readily defined and characterized within the lamprey spinal cord and this has led to progress in understanding the cellular and synaptic mechanisms of locomotor activity. In addition, both the identifiable reticulospinal cells and the various spinal nerve cell classes and their known synaptic interactions have been used to demonstrate the degree and specificity of regeneration within the lamprey nervous system. The lack of uniquely identifiable cells within the lamprey spinal cord has hampered progress in these areas, especially in gaining a full understanding of the locomotor network and how neuromodulation of the network is accomplished.

The organization of the accessory optic and pretectal circuits is surveyed in a variety of poikilotherms and the conclusion drawn that a common organizational pattern exists. This pattern consists of the presence of three optic pretectal nuclei located rostrocaudally in lateral-superficial, central and dorsomedial (or periventricular) positions. Furthermore, a well-defined basal optic tract and ventrolaterally placed terminal field at the level of the oculomotor nucleus appears to be a consistent feature among bony fish, amphibians and reptiles. The possible role of these circuits in in visuomotor behaviors is discussed.

The spinal nerves in amphioxus are compared with the spinal and cranial nerves in lampreys. The dorsal spinal roots in amphioxus are similar to the mixed sensory and motor dorsal roots of many cranial nerves in lampreys but not to the purely sensory dorsal spinal roots in lampreys and gnathostomes. Likewise, cranial nerves V, VII, IX and X in lampreys, and all spinal nerves in amphioxus, lack a separate ventral motor root which is a constant feature of all spinal motor roots in lampreys and other vertebrates. Based on these similarities and differences, it is proposed that cranial and spinal nerves in craniates are independently derived serial homologs of elements of an amphioxus-like ancestral pattern. Further evolution involved the addition of neural crest-derived ganglia to most cranial and all spinal nerves, and the addition of placodally derived ganglia to many cranial nerves. The possible homology of ocular motor nerves is discussed but cannot be resolved owing to the absence of these nerves in hagfishes, which are the only relevant outgroup.

Modern views of agnathan phylogeny consider Petromyzoniformes and Myxiniformes to belong to distinct classes that diverged from a common ancestor at a remote period, perhaps in the lower Cambrian, greater than 600 million years ago. Both are more primitive than elasmobranchs, holocephalans and bony fishes. Myelin is well developed in elasmobranchs and other fishes but was reported to be lacking in the spinal cord of lampreys. In order to search further for possible early myelin in some part of the nervous system of one of the agnathan stems, or for further evidence that it first appeared in chondrichthians, we extended the sampling to many parts of the brain and cord of hagfish. Transmission electron microscopy was used as a nearly ideal criterion. We find no trace or forerunner of the spiral, multilaminate glial wrapping. Many axons are embedded within one or more glial cells, like unmyelinated fibers in other vertebrates, or lie contiguously in bundles without an obviously complete glial investment. True myelin must be presumed to have been invented within the vertebrates, in ancestors of the living cartilaginous fishes after the agnathans branched from the vertebrate stem.