Early development and organization of the retinopetal system in the larval sea lamprey, Petromyzon marinus L

Anat Embryol (1995) 192:517-526 �9 Springer-Verlag 1995

Maria Celina Rodicio �9 Manuel A. Polnbal Ram6n Anad6n

Early development and organization of the retinopetal system in the larval sea lamprey, Petromyzon marinus L. An HRP study

Accepted: 3 May 1995

Abstract Development of the retinopetal system of the larval sea lamprey, Petromyzon marinus, was investigated following labelling of this system by injection of horseradish peroxidase into the orbit. This study extends our previous report on larval stages and provides a detailed description of the development of this system. We present quantitative and qualitative evidence suggesting that the retinopetal nuclei of Sch- ober's M2-M5 nucleus, the mesencephalic reticular area and the tectum arise sequentially in that order, that the three retinopetal nuclei originate from a common anlage in the ventricular zone of the mesencephalic tegmentum and that the retinopetal cell population increases throughout the larval period. No neuronal death was observed. We also describe and discuss the significance of a transitory phase of retinopetal cell differentiation characterized by the presence of ventricular dendrites. Finally, we compare the development of retinopetal and retinofugal systems.

Key words Visual system �9 Centrifugal fibres �9 Ontoge- ny �9 Mesencephalon �9 Agnatha

Introduction

From an evolutionary point of view, the organization of the nervous system of lampreys is of great interest because these animals belong to the earliest line of vertebrates, the Agnatha. In addition, lampreys have a micro- phagous larval form, the ammocoete, which undergoes a complex metamorphosis after five or more years of life (see Hardisty and Potter 1971). The eye begins its devel-

M. C. Rodicio - R. Anad6n (~) Departamento de Biologfa Fundamental, Universidad de Santiago de Compostela, E- 15706 Santiago de Compostela, Spain Tel.: (34) 81-563100 ext. 3296; FAX: (34) 81-596904

M. A. Pombal Departamento de Biologfa Celular y Molecular, Universidad de La Corufia, E-15071 La Corufia, Spain

opment during the embryonic period (Damas 1944), but only becomes functional at metamorphosis (Rubinson et al. 1977; De Miguel and Anad6n 1987; Rubinson 1990). This characteristic provides a unique opportunity to study some aspects of the development of the visual system.

The presence of retinopetal fibres was discovered at the end of the last century in birds (Cajal 1889), while Tretjakoff (1916) was the first to report their existence in lampreys. Since the 1970s, the study of the retinopetal system has made considerable progress due to the introduction of neuroanatomical tracer techniques. The presence of retinopetal fibres in both adult and larval lamprey has been confirmed with horseradish peroxidase (HRP) labelling and electron microscopy (Vesselkin et al. 1989; De Miguel et al. 1989). The central components of this system have been identified in adult lamprey using both retrograde transport (Vesselkin et al. 1980, 1984) and electrophysiological techniques (Vesselkin et al. 1984). These studies indicated that retinopetal neurons lie in both the mesencephalic M5 nucleus of Sch- ober (1964) and the adjacent mesencephalic reticular area (MRA). The organization of the retinopetal system in larval lampreys appears to be similar, although some retinopetal neurons have also been observed in the optic tecmm (De Miguel et al. 1990).

In this study of retinopetal neurons in the lamprey, we extend our previous observations to the earliest larval stages and provide qualitative and quantitative evidence of how the three retinopetal nuclei of larval lamprey arise. We also discuss the significance of the retinopetal neurons with ventricular dendrites.

Materials and methods

Larval sea lampreys (Petromyzon marinus L.) were caught in the Rivers Ulla and Mifio (NW Spain) and reared at room temperature (16-20~ in well-aerated tanks with a bed of fluvial sediment. As the age of the larvae at capture was unknown, we used body length as an indicator of developmental stage. Age and body length are correlated (Ooi and Youson 1976; Purvis 1979; Rubinson and

518

Cain 1989), although there is some size variability within each age class in natural populations. The present report is based on the study of 114 well-labelled larvae ranging from l l to 160 mm in length. As in a previous study of the oculomotor nuclei (Pombal et al. 1994), larvae were assigned to one of six size classes: class 1, from 11 to 20 mm (16 larvae); class 2, from 21 to 40 mm (12 larvae); class 3, from 41 to 60 mm (19 larvae); class 4, from 61 to 80 mm (21 larvae); class 5, from 81 to 100 mm (22 larvae), and class 6, from 101 to 160 mm (24 larvae). Larvae of more than 100 mm in length were grouped together because they may have started metamorphosis.

The larvae were anaesthetized in 0.1% tricaine methane sul- phonate (MS-222, Sigma). HRP (Serva) was applied with a 000 insect pin either into the eye following removal of overlying tis- sues (small larvae), or into the orbit following removal of the eye (large larvae). After labelling for 1 rain, the eye or the orbit was rinsed with Ringer's solution. Following a survival period of 1 to 6 days (according to larval size) at 4~ the larvae were deeply reanaesthetized and killed by decapitation. Brains were then fixed for 2-3 h in cold (4~ 1% glutaraldehyde and 1% paraformalde- hyde in 0.1 M phosphate buffer, pH 7.3. The brains were pro- cessed en bloc using a diaminobenzidine (DAB) method (Bourrat and Sotelo 1986) and postfixed in buffered 1% OsO 4, although osmification was omitted in some cases (see below). Most brains (91) were embedded in Spurr's resin and sectioned in the transverse or horizontal planes (30 pm thick; 85 and 6 larvae, respectively) after warming the block surface (West 1972). Some brains (23) were longitudinally divided into the two hemibrains, flat- tened on gelatinized slides and mounted whole. Neuron size was determined with a calibrated ocular micrometer. Labelled cells were counted at xl000 magnification, with no correction factor. For statistical comparison of numbers of labelled cells between larval size classes, we used Student's t-test. For topographical purposes and to look for pyknotic nuclei we used our laboratory collection of larval brains, cut at 10 gm and stained with cresyl violet or haematoxylin-eosin.

Results

The results of this study confirm a previous report of the presence of the retinopetal system in larval lamprey (De Miguel et al. 1990). Moreover, they show that retinopetal neurons of lamprey can be labelled only after the embryonic period and that the three retinopetal nuclei arise sequentially from the periventricular region to the lateral area. The nomenclature for retinopetal nuclei used in this study is adopted from Vesselkin et al. (1980), although slightly modified by De Miguel et al. (1990). Briefly, we use "M2-M5 nucleus" for the "M5 nucleus" of Vesselkin et al. (1980) and define a "retinopetal tectal nucleus" in accordance with its location in the optic tectum (De Mig- uel et al. 1990).

Technical comments

The application of HRP into the orbit after removing the eyeball, as performed in large larvae, labels the central visual system well. However, this procedure is not useful in small larvae, whose eyes are too small: in these larvae, HRP was applied into the eye itself. Both methods generally label the oculomotor nucleus in the mesencephalon, in addition to the visual system. Since the oculomotor nucleus is well-characterized in larvae (Pombal et al.

1994), it can be discerned from the retinopetal and retinofugal systems. For evaluation of retinopetal cell number, we used only specimens with well-labelled retinofugal projections. OsO 4 postfixation considerably intensi- fied the label, produced a brownish background and stained the lipid droplets of ependymal cells dark brown. Since M2-M5 neurons are closely grouped, the OsO 4 postfixation step was sometimes omitted in order to fa- cilitate cell counts.

Developmental stages of the retinopetal nuclei

The location and morphology of the retinopetal nuclei in larval lampreys are summarized in Fig. 1. Three developmental stages can be defined on the basis of the appearance of the retinopetal nuclei and the localization of their neurons: (a) larvae of classes 1 and 2, in which labelled cells were only observed in the M2-M5 nucleus (Fig. 1A,B); (b) larvae of classes 3 and 4, in which labelled cells were observed in both the M2-M5 nucleus and the MRA (Fig. 1C) and (c) larvae of classes 5 and 6, in which labelled neurons were also present in the ventral optic tectum (Fig. 1D).

M 2 - M 5 retinopetal nucleus

As reported previously (De Miguel et al. 1990), the M 2 - M 5 retinopetal nucleus lies in the medial mesencephalic tegmentum, about 25% of its cells being rostral to the first Mtiller cell (M2 nucleus) and the remaining 75% caudal (M5 nucleus).

We observed contralateral labelled retinopetal cells in this nucleus in only four of the 16 class-1 larvae whose optic nerves were well labelled. These cells were scarce (Table 1; Fig. 2A) and scattered in the tegmental grey matter at the level of the first Mtiller neuron and just dor- sal to it. No ipsilateral cells were observed in this size class.

In the remaining five size classes, most M2-M5 neurons were contralateral, although there were some ipsilateral cells (Table 1; Fig. 2B-D). In class-6 larvae, the M 2 - M 5 nucleus was separated from the ventricle by a thin fibrous layer and about three rows of periventricular cells (Fig. 2D). In classes 2-5, this periventricular zone also contained labelled cells (Fig. 3), which were less frequent in larger larvae (Table 2).

The morphology and dendrite distribution of M2-M5 neurons change considerably during development, suggesting that maturation of this nucleus is reached towards the end of the larval period. In class-1 larvae, M2-M5 neurons are round or oval and have no dendrites, whereas in the other classes they are bipolar, giving rise to lateral and medial dendrites. Most M 2 - M 5 neurons are oriented perpendicular to the ventricular surface (Figs. 3, 4) (though in larvae of classes 3-6 some cells lying close to the lateral area had their longest axis parallel to it; Figs. 3, 5). In larvae of classes 2-6, the lat-

Fig. 1A-D Schematic draw- ings of transverse sections of the larval lamprey mesencephalon showing the location of labelled retinopetal neurons after application of horseradish peroxidase into the orbit. A Lar- vae, 11-20 mm. B Larvae, 21-40 mm. C Larvae, 41-80 mm. D Larvae more than 81 mm in length. The dot- ted line indicates the limit between the grey and the white, and the broken line the tecto- tegmental limit (M5 M2-M5 nucleus, MRA mesencephalic reticular area, T optic tectum). Contralateral is on the right

1

.@ f i

."

/ / '

D

~

- ' , .

) )r

( 2

~IRRA %

�9 . . -

" ' . . . . . . . . . . .

519

Table 1 Number of labelled cells (mean+standard deviation) in the M2-M5, MRA and tectal retinopetal nuclei during the lamprey larval period (n=5, except in class 1, where n=4)

Nucleus M2-M5 nucleus MRA Tectum

Class Size Ipsi Contra Ipsi Contra Ipsi Contra

1 11-20 - 14.7+-6.4" - - 2 2 1 4 0 4.2+_2.0* 164.2+15.4" - - 3 41-60 6.8+-5.1" 268.4+_49.3* 0.8+_0.7* 14.2_+9.1" 4 61-80 14.2+6.2" 336.0+_42.3* 0.6+1.2* 115.0_+50.4" 5 81-100 40 .4_+25 .4 378.4_+75.5* 5.4+_2.8 217.6_+80.8 6 >101 42.4+_6.6 547.0+_62.2 4.6_+1.7 224.8+_35.3

m

0.8+_0.9* 25.4+7.8* 2.4+_2.3 265.0+_48.3

* < Mean cell number is significantly lower (P_0.05) than the class-6 mean for that nucleus (Student's t-test)

eral dendrites extend towards the meninges and then dorsally to terminate among retinofugal fibres of the axial optic tract (Fig. 3) and, in larvae o f more than 7 0 - 8 0 m m in length, among retinofugal fibres of the lateral optic tract. Some lateral dendrites also reach other areas o f the pretectum and optic tectum. In classes 2-5 , some contralateral M 2 - M 5 retinopetal neurons (up to 10% in class 2) have medial dendrites that cross the periventricular layer and protrude into the ventricle with a club-like appearance (Figs. 3, 5). The proport ion o f cells with ventricular dendrites decreases as the larvae grow (Table 2). In classes 3-4, some medial dendrites run either ventral- ly or dorsally, without reaching the ventricle. In classes 5 and 6, medial dendrites form part of the thin fibrous layer between the M 2 - M 5 nucleus and the periventricular cells (Fig. 4). The axons o f M 2 - M 5 cells are thin, gener-

ally arising f rom a lateral dendrite and running in the axial optic tract.

M R A retinopetal nucleus

Neurons of the M R A nucleus are located in the mesencephalic reticular area adjacent to the M5 nucleus and caudal to it (Figs. 6, 7). They were only observed in larvae o f classes 3 -6 (more than 40 m m in length; Fig. 2C-D) . As with M 2 - M 5 neurons, changes occur in the morphology and orientation of M R A cells during development. In class-3 larvae, all M R A retinopetal neurons were bipolar cells oriented perpendicular to the ventricle. Their lateral dendrites course and branch like those o f M 2 - M 5 neurons, and their medial dendrites run to the M 2 - M 5

520

Fig. 2A-D Transverse sections through the mesencephalon, showing labelled neurons after application of HRP into the orbit in a 15-ram larva (A), a 33-ram larva (B), a 61-ram larva (C), and a 85-ram larva (D) (I ipsilateral side, aster isk third Miiller cell, ar- Jvws oculomotor nucleus, open arrows meningeal melanophores, arrowheads ipsilateral retinopetal neurons). Other abbreviations as in Fig. 1. Note that the brain in Fig. 2C was not postfixed with os- mium tetroxide. Bars 50 ~tm (A), 100 ~tm (B-D)

nucleus. In classes 4 -6 , many M R A retinopetal neurons (between about 30% and 50%) have their perikarya located among fibres o f the axial optic tract (Fig. 6), their long axes being perpendicular to the long axes o f more medial M R A retinopetal neurons (Fig. 7); however, both the pattern o f dendrite arborization and the origin o f the axon appear to be similar. As in M 2 - M 5 neurons, the axon arises f rom a lateral dendrite.

521

Fig. 3 M2-M5 nucleus of a 33-ram larva, showing ventricular and lateral dendrites of retinopetal neurons. The arrow indicates a vertically oriented neuron. Note also the location of labelled fibres of the axial optic tract (asterisk). Bar 50 gm. Inset detail of two ventricular dendrites. Bar 10 gm

Fig. 4 Detail of the M2-M5 neurons of an 84-ram larva, showing the plexus of medial dendrites (arrowheads) lateral to the periventricular region (asterisk). Note a neuron giving rise to a medioven- tral dendrite (arrow). Differential interference contrast. Bar 20 gm

Table 2 Percentage of labelled retinopetal ceils with ventricular dendrites in Schober's M2-M5 nucleus in larval lamprey (n-5, except in class 1, where n=4)

Class 1 2 3 4 5 6

Size (inm) 11-20 21-40 41-60 61-80 81-100 >_101 Contralateral 0 9.62 3.35 1.13 1.90 0 Ipsilateral - 4.76 0 3.57 0.47 0

Tectal retinopetal nucleus

Labelled retinopetal neurons in the ventral optic tectum are present in classes 5 and 6 (Figs. 2D, 7, 8). These late-appearing retinopetal neurons are particularly f re- quent in class 6. They are found over the entire rostro- caudal extent of the ventral optic tectum and are located medial to the ventral portion of the lateral optic tract (Fig. 8). Their orientation, morphology, dendritic arborization patterns, and axon emergence patterns are similar to those of the most lateral MRA retinopetal neurons.

Estimation of cell population sizes

To investigate changes in cell population sizes during development, we used Student's t-test for pairwise comparison between larval size classes, as in a previous study of the oculomotor nuclei (Pombal et al. 1994). The number of labelled retinopetal neurons in each class is shown in Table 1, and the proportion of M2-M5 neurons with ventricular dendrites is shown in Table 2.

In all three nuclei, the number of retinopetal cells in the earliest class in which labelled neurons appear was very low in comparison with that in subsequent classes. Characteristically, the standard deviation within each class is high (Table 1). Despite this variation, Student's t- test indicates that, for all nuclei, the number of labelled cells was significantly higher in class 6 than in all other classes (Table 1); the only exceptions were the contralateral and ipsilateral MRA and the ipsilateral M2-M5 nuclei of class 5. This result strongly suggests that new axons reach the eye throughout the~larval period.

The number of contralateral M2-M5 retinopetal neurons increased steadily throughout the larval period, while the number of tectal neurons increased between classes 5 and 6. The number of ipsilateral retinopetal neurons in both the M2-M5 and MRA nuclei increased until class 5. At the end of the larval period, ipsilateral retinopetal cells represented about 5% of the total number of retinopetal ceils. In contrast to the M2-M5 and tectal nuclei, there does not appear to be any increase in the retinopetal population of the MRA nucleus after class 5. However, it seems probable that a proportion of the labelled neurons in the MRA of classes 4 and 5 were in

522

fact migrating to the tectum and/or that some of the labelled neurons in the M2-M5 nucleus of these size classes were migrating to the MRA. None of the three retinopetal nuclei showed signs of cell death during the larval period (i.e. there was no decrease in the number of cells, nor were pyknotic nuclei - indicators of apoptosis - observed in our sections stained with haematoxylin-eosin).

Discussion

Significance of the tectal population

The distribution of retinopetal neurons in class-6 larvae is similar to that reported by other authors in adult lampreys (Rep6rant et al. 1980; Vesselkin et al. 1980), suggesting that at this stage the development of the retinopetal system has been completed. These authors, however, did not identify a distinct tectal population of retinopetal neurons in L a m p e t r a f luv ia t i l i s . This may be due to the existence of between-species differences, to the im- portant differences in tectal development between larvae and adults (De Miguel and Anad6n 1987), or both. In teleosts, the existence of tectal retinopetal neurons was first suggested by Vanegas et al. (1973) on the basis of electrophysiological data. Tectal retinopetal neurons were later found with tracing techniques (Ebbesson and Meyer 1981, 1989; Meyer and Ebbesson 1981; Meyer et al. 1981; Ekstr6m 1984), but it has been suggested that the labelling of tectal cells with HRP in teleosts, as well as in frogs, could be due to transneuronal transport (Cra- pon de Caprona and Fritzsch 1983; Hughes and Hall 1986; Uchiyama 1989). Specifically, some authors consider that the use of dimethyl sulphoxide and lysolecithin as auxiliary reagents in HRP studies may enhance mem- brane permeability and, as a result, transneuronal labelling (Crapon de Caprona and Fritzsch 1983; Uchiyama 1989). Such reagents were not used in our experiments; however, there are more convincing reasons to suggest

Fig. 5 Transverse section through the contralateral M5 nucleus (M5) of a 56-mm larva, showing the presence of MRA neurons (arrowheads). Note a Mtiller cell (asterisk) and the oculomotor nucleus (white star). The large arrow indicates a ventricular dendrite and the small arrow a retinopetal cell with vertical orientation (V ventricle). Bar 25 gm

Fig. 6 Horizontal section through the tegmeutum of a 91-ram larva, showing the contralateral retinopetal nuclei. The arrow indicates the axons of the axial optic tract, open arrow lateral optic tract. Rostral is at the top of the figure. Bar 100 gm

Fig. 7 Whole mount of the brain of a 138-mm larva, showing the relative positions of the M2-M5 (M2-M5 out of focus), MRA (MRA) and tectal (7) retinopetal neurons. The arrow indicates the first MUller cell (out of focus). The shaded area on the right (asterisk) corresponds to the area occupied by fibres of the lateral optic tract. Rostral is at the top of the figure. Bar 100 gm

Fig. 8 Transverse section through the mesencephalon of a 142- mm larva, showing the location of the tectal retinopetal neurons below the retinofugal projection (outlined arrow). Other abbreviations as in Fig. 1. Bar 100 gm

523

that the labelling of tectal neurons observed by us was not the result of transneuronal transport. First, in whole mounts, axons of these neurons could be traced to the axial optic tract (De Miguel et al. 1990; present results). Second, these neurons form a well-defined population at the ventrolateral edge of the tectum and are very similar in appearance to the lateral neurons of the MRA. Third, if labelling were due to transneuronal transport, other re- tinorecipient tectal cells would be expected to have been labelled; no such labelling was observed. Moreover, transneuronal labelling with HRP was not observed in several other studies of cranial nerves in larval lampreys (Homma 1978; Anaddn et al. 1989; De Miguel et al. 1990; Gonzfilez and Anad6n 1992, 1994; Pombal et al. 1994), despite characteristically intense labelling of primary sensory fibres. Accordingly, we consider the labelled tectal cells to be part of the retinopetal system.

Origin of the retinopetal nuclei

A number of observations strongly suggest that the three retinopetal nuclei of lampreys originate from the same anlage in the ventricular zone. First, the location of the earliest labelled retinopetal neurons in the ventricular layer, their round shape and their lack of dendrites strongly suggest that these cells lie on their region of origin. Second, some labelled cells in the M2-M5 nucleus of larvae of classes 2 to 5 have ventricular processes that contact the ventricle in the area in which the earliest retinopetal cells arise. Similar cells in other developing nuclei of larval lampreys have been interpreted as differen- tiating neuroblasts (Rodicio et al. 1992; Pombal et al. 1994): their presence may therefore indicate the location of a nuclear anlage (see below). Third, in size classes in which the MRA and tectal retinopetal nuclei are present, the distribution and dendritic orientation of retinopetal cells are strongly suggestive of a well-defined migration route extending from the M2-M5 nucleus towards the MRA and tectal retinopetal nuclei. Fourth, no clear-cut distinction can be made between lateral MRA retinopetal neurons and ventral tectal retinopetal neurons either on the basis of morphology or location. These observations strongly suggest that neurons of the three retinopetal nuclei arise from a common anlage in the region of the M2-M5 nucleus.

In birds, the isthmo-optic nucleus is thought to arise in the alar plate of the isthmic region (Clarke 1982). Uchiyama (1989) includes the retinopetal nuclei of lamprey in the isthmo-optic nucleus type of retinopetal system. However, the close topographical association between the common anlage of the lamprey retinopetal nuclei and the first Mtiller cell in the mesencephalic tegmentum (present results) strongly suggests that the retinopetal nuclei of lamprey originate in the basal plate, rostral to the isthmus. This may suggest that the retinopetal nuclei of lampreys and birds may not be homolo- gous. Retinopetal tegmental neurons have also been found in primitive bony fishes (Meyer et al. 1983) and

524

reptiles (Ferguson et al. 1978; Schnyder and Ktinzle 1983; Sch0tte and Weiler 1988); however, it is unclear whether they arise from the basal or alar plates. Further studies of the region of origin of isthmic retinopetal nuclei are required to resolve this problem.

Retinopetal nuclei of larvae and the development of retinal innervation

The HRP techniques used in the present study do not allow determination of the birth dates of the first retinopetal neurons, although neuron birth must of course occur before the first retinopetal axons reach the eye. However, such techniques allow elucidation of the time course of retinal innervation by centrifugal fibres. Our results suggest that centrifugal innervation of the retina begins in larvae of about 20 mm: the first labelled retinopetal neurons are observed in these larvae (only 4 of the 16 class- 1 larvae investigated had labelled retinopetal cells, and in all 4 larvae such cells were very scarce). Interestingly, all class-1 specimens (except those in which the optic nerve was not successfully labelled) showed well-labelled retinofugal fibres in the optic chiasm and optic tract, and in none of these larvae was there labelling of retinopetal nuclei without labelling of retinofugal projections. It is worth noting that eye development in lamprey larvae is bimodal (see De Miguel and Anad6n 1987). The primary (central) retina, with differentiated photoreceptors, is present in all larvae studied and is of similar appearance in larval size classes 1, 2 and 3. A secondary (peripheral) retina gradually grows around the primary retina from about 60 mm length onwards, but differentiation of photoreceptors in this area does not occur until metamorphosis. Thus, the presence in classes 1 to 3 of retinofugal projections, the use of the same labelling procedure in the three classes and the observed changes in the number and appearance of neurons strongly suggest that the lack of labelled retinopetal cells in class-1 larvae should not be attributed to failure of the method, but rather to imma- turity of the retinopetal system.

The observed changes in the number of labelled retinopetal cells during development are summarized in Table 1. Standard deviation within each size class was high, suggesting that not all the neurons present were labelled (as expected, given the labelling method used) and/or that variation among individuals is high. Statisti- cal analysis, however, showed that the number of labelled cells in class-6 larvae was significantly higher than that in other classes (with the exception of the contralateral and ipsilateral MRA and ipsilateral M2-M5 nuclei of class-5 larvae). This supports the hypothesis that the number of centrifugal fibres reaching the retina increases steadily during the larval period. The development of the lamprey retina during the larval period, however, takes place in two clearly distinct stages. During the first stage (between hatching and about 60 mm), retinal growth is negligible, while during the second stage, it is explosive (De Miguel and Anad6n 1987; Rubinson

and Cain 1989). The time course of retinopetal innervation does not, therefore, parallel that of overall retinal growth. Ultrastructural studies of large larvae have revealed the existence of synaptic contacts between the centrifugal fibres and ganglion cells (De Miguel et al. 1989): whether retinopetal innervation and the development of ganglion cells are related needs further investi- gation.

Significance of retinopetal cells with ventricular processes

It is currently believed that before the onset of differentiation, vertebrate neuroblasts first lose ventricular attachments and then sprout axons (Cajal 1909). From studies of the lamprey, there is growing evidence that neuron differentiation does not always conform to this model, in that axons may grow from cells that still retain their ventricular attachments (Nakao and Ishizawa 1987; De Mig- uel et al. 1990; Rodicio et al. 1992; Pombal et al. 1994; present results). There have also been a few reports of a similar phase in the differentiation of chick embryo motoneurons (Tello 1923; Heaton et al. 1978). The occa- sional presence of ventricular contacts in developing vertebrates contrasts strikingly with the frequent reports of a similar type of neuron in adults, the cerebrospinal-fluid- contacting neurons (see Leonhard 1980, for a compre- hensive review). Such neurons are generally considered to be of a primitive type. The presence of ventricular- dendrite-bearing neurons in some neural systems of larval lamprey may be an ancestral character that was wide- ly extended in primitive chordates. In support of this hypothesis, most neurons of adult amphioxus, including the giant Rohde cells, contact the cerebrospinal fluid (Meves 1973; Ruiz and Anad6n 1989).

Although cells with ventricular processes have been observed in the larval M2-M5 nucleus (De Miguel et al. 1990), a striking result of the present study was the relative abundance of such cells in class-2 larvae (Table 2). This result suggests that all M2-M5 neurons may pass through a transient phase of this type. The apparent absence of such cells in class-1 larvae, however, is intrigu- ing: one possibility is that they are so scarce that the probability of finding them in such a small neuronal population is very low (see Table 1).

Does cell death occur during development of the retinopetal nuclei in lamprey?

The isthmo-optic nucleus of the developing chick goes through a phase of neuronal death before hatching (Clarke et al. 1976; Clarke and Cowan 1976; Clarke 1992): a large proportion of neurons (about 56% of cells) undergoes apoptotic changes and dies. Cell death, however, does not appear to occur in the retinopetal nuclei of either teleosts (Crapon de Caprona and Fritzsch 1983; Rusoff and Hapner 1990) or larval lampreys (present re-

sults). We do not know the number of retinopetal cells in adult Petromyzon marinus, but the number of M2-M5 and MRA neurons in premetamorphic P. marinus (about 560 and 190, respectively) is not greatly different from that in adult Lampetrafluviatilis (about 590 and 230, respectively; Vesselkin et al. 1980). These figures, however, do not include tectal neurons (about 265 in P. marinus), which have not been detected in the two other lamprey species investigated to date (L. fluviatilis, Vesselkin et al. 1980; lchthyomyzon unicuspis, Fritzsch and Collin 1990). Evaluation of the size of the retinopetal population in adults of P. marinus is clearly required to deter- mine whether neuronal death occurs in the retinopetal nuclei during or after metamorphosis.

Conclusions and prospective

The study of the retinopetal nuclei of lampreys throws light on several interesting features of their development including: (a) the diversification of three well-characterized nuclei from a single anlage; (b) the presence of an inside-out gradient of development; (c) the late development of centrifugal innervation of the retina with respect to the development of the first retinofugal pathways; (d) the presence of a transient phase of neuron differentiation; (e) the probable absence of cell death. However, our results do not allow determination of the stage at which proliferation of these cells occurs: birth-dating studies of these nuclei are required to investigate whether cell proliferation extends to phases in which neuron differentiation is observed, or whether it occurs only during a limit- ed period in lamprey development. The results of such studies should allow a more precise comparison of development of the retinopetal nuclei in lamprey with that in other vertebrates.

Acknowledgement This work was supported by a grant (XUGA 20014B94) from the Xunta de Galicia (Spain).

References

Anad6n R, De Miguel E, Gonzfilez MJ, Rodicio MC (1989) HRP study of the central components of the trigeminal nerve in the larval sea lamprey: organization and homology of the primary medullary and spinal nucleus of the trigeminus. J Comp Neu- rol 283:602--610

Bourrat F, Sotelo C (1986) Neuronal migration and dendritic maturation of the medial cerebellar nucleus in rat embryos: an HRP in vitro study using cerebellar slabs. Brain Res 378:69- 85

Cajal SR (1889) Sur la morphologie et les connexions des 616ments de la retine des oiseaux. Anat Anz 4:111-121

Cajal SR (1909) Histologie du systbme nerveux de l'homme et des vert6brds. Maloine, Paris (reprinted by CSIC, Madrid 1955)

Clarke PGH (1982) The generation and migration of the chick's isthmic complex. J Comp Neurol 207:208-222

Clarke PGH (1992) Neuron death in the developing avian isthmo- optic nucleus, and its relation to the establishment of functional circuitry. J Neurobiol 23:1140-1158

Clarke PGH, Cowan WM (1976) The development of the isthmo- optic tract in the chick, with special reference to the occur-

525

rence and correction of developmental errors in the location and connections of isthmo-optic neurons. J Comp Neurol 167:143-164

Clarke PGH, Rogers LA, Cowan WM (1976) The time of origin and the pattern of survival of neurons in the isthmo-optic nucleus of the chick. J C omp Neurol 167:125-142

Crapon de Caprona M-D, Fritzsch B (1983) The development of the retinopetal nucleus olfactoretinalis of two cichlid fish as revealed by horseradish peroxidase. Dev Brain Res 11:281- 301

Dumas H (1944) Recherehes sur le d6veloppement de Lampetra fluviatilis L. Contribution h l'~tude de la cdphalogenbse des vert6brds. Arch Biol (Paris) 55:1-284

De Miguel E, Anad6n R (1987) The development of retina and the optic tectum of Petromyzon marinus L. A light microscopic study. J Hirnforsch 28:445-456

De Miguel E, Rodicio MC, Anad6n R (1989) Ganglion cells and retinopetal fibers of the larval lamprey retina: an HRP study. Neurosci Lett 106:1-6

De Miguel E, Rodicio MC, Anad6n R (1990) Organization of the visual system in larval lampreys: an HRP study. J Comp Neu- rol 302:529-542

Ebbesson SOE, Meyer DL (1981) Efferents to the retina have mul- tiple sources in teleost fish. Science 214:924-926

Ebbesson SOE, Meyer DL (1989) Retinopetal cells exist in the optic tectum of the steelhead trout.Neurosci Lett 106:95-98

Ekstr6m P (1984) Central neural connections of the pineal organ and retina in the teleost Gasterosteus aculeatus L. J Comp Neurol 226:321-335

Ferguson JL, Mulvany PJ, Brauth SE (1978) Distribution of neurons projecting to the retina of Caiman crocodilus. Brain Be- hav Evol 15:294-306

Fritzsch B, Collin SP (1990) Dendritic distibution of two populations of ganglion cells and the retinopetal fibers in the retina of the silver lamprey (Ichthyomyzon unicuspis). Visual Neurosci 4:533-545

Gonzfilez MJ, Anad6n R (1992) Primary projections of the lateral line nerves in larval sea lamprey, Petromyzon marinus L. An HRP study. J Hirnforsch 33:185-194

Gonzfilez MJ, Anad6n R (1994) Central projections of the octaval nerve in larval lamprey: an HRP study. J Brain Res 35:181- 189

Hardisty MW, Potter IC (1971) The behaviour, ecology and growth of larval lampreys. In Hardisty MW, Potter IC (eds) The biology of lampreys, vol l. Academic Press, London, pp 85-125

Heaton MB, Moody SA, Kosier ME (1978) Peripheral innervation by migrating neuroblast in the chick embryo. Neurosci Lett 10:55-59

Homma S (1978) Organization of the trigeminal motor nucleus before and after metamorphosis in lampreys. Brain Res 140:33- 44

Hughes TE, Hall WC (1986) The transneuronal transport of horseradish peroxidase in the visual system of the frog (Rana pipi- ens). Neuroscience 17:507-518

Leonhard H (1980) Ependym und circumventriculfire Organe. In: Oksche A, Vollrath L (eds) Handbuch der mikroskopischen Anatomie des Menschen, vol 4. Nervensystem, Part 10, Neu- roglia I. Springer, Berlin Heidelberg New York, pp 177- 666

Meves A (1973) Elektronenmikroskopische Untersuchungen tiber die Zytoarchitektur des Gehirns yon Branchiostoma lanceola- turn. Z Zellforsch 130:511-532

Meyer DL, Ebbesson S OE (1981) Retinofugal and retinopetal connections in the upsidedown catfish (Synodontis nigriventris). Cell Tissue Res 218:389-401

Meyer DL, Fiebig E, Ebbesson SOE (1981) A note in the recipro- cal connections between the retina and the brain in the puffer fish (Tetraodonfluviatilis). Neurosci Len 23:111-115

Meyer DL, Gerwerzhagen K, Fiebig E, Ahlswede F, Ebbesson SOE (1983) An isthmo-optic system in a bony fish. Cell Tis- sue Res 231:129-133

526

Nakao T, Ishizawa A (1987) Development of the spinal nerves in the lamprey. I. Rohon-Beard cells and interneurons. J Comp Neurol 256:342-355

Ooi EC, Youson JH (1976) Growth of the opisthonephric kidney during larval life in the anadromous sea lamprey, Petromyzon marinus L. Can J Zool 54:1449-1458

Pombal MA, Rodicio MC, Anad6n R (1994) Development and organization of the ocular motor nuclei in the larval sea lamprey Petromyzon marinus L.: an HRP study. J Comp Neurol 341:393-406

Purvis HA (1979) Variations in growth, age at transformation, and sex ratio of sea lampreys re-established in chemically treated tributaries of the upper Great Lakes. Great Lakes Fish Com- mun Tech Rep 35:1-36

Rep6rant J, Vesselkin NP, Ermakova TV, Kenigfest NB, Kosareva AA (1980) Radioautographic evidence for both orthograde and retrograde axonal transport of labeled compounds after intra- ocular injection of [3H] proline in the lamprey (Lampetra fluviatilis). Brain Res 200:179-183

Rodicio MC, De Miguel E, Pombal MA, Anad6n R (1992) The origin of trochlear motoneurons in the larval sea lamprey, Petro- myzon marinus L. An HRP study. Neurosci Lett 138:19- 22

Rubinson K (1990) The developing visual system and metamorphosis in the lamprey. J Neurobiol 21:1123-1135

Rubinson K, Cain H (1989) Neural differentiation in the retina of the larval sea lamprey (Petromyzon marinus). Visual Neurosci 3:241-248

Rubinson K, Ripps H, Witkovsky P, Kennedy MC (1977) Retinal development in the lamprey, Petromyzon marinus. Soc Neuro- sci Abstr 3:575

Ruiz MS, Anad6n R (1989) Some observations on the fine struc- ture of the Rohde cells of the spinal cord of the Amphioxus, Branchiostoma lanceolatum (Cephalochordata). J Hirnforsch 30:672-677

Rusoff AC, Hapner SJ (1990) Development of retinopetal projections in the cichlid fish, Herotilapia multiespinosa. J Comp Neurol 294:431-442

Schnyder H, Ktinzle H (1983) The retinopetal system in the turtle Pseudemys scripta elegans. Cell Tissue Res 234:219-224

Schober W (1964) Vergleichend-anatomische Untersuchungen am Gehirn der Larven und adulten Tiere von Lampetra fluviatilis (Linn6, 1758) und Lampetra planeri (Bloch, 1784). J Hi- rnforsch 7:107-209

Schtitte M, Weiler R (1988) Mesencephalic innervation of the turtle retina by a single serotonin-containing neuron. Neurosci Lett 91:289-294

Tello JF (1923) Les diffdrenciations neuronales dans l'embryon du poulet, pendant les premiers jours de l'incubation. Tray Lab Rech Biol Univ Madrid 21:1-93

Tretjakoff DK (1916) The sense organs of the lamprey (Lampetra fluviatilis). University of Novorossijsk, Odessa

Uchiyama H (1989) Centrifugal pathways to the retina: influence of the optic tectum. Visual Neurosci 3:183-206

Vanegas H, Amat J, Essayag-Millan E (1973) Electrophysiological evidence of tectal efferents to the fish eye. Brain Res 54:309-313

Vesselkin NP, Ermakova TV, Rep6rant J, Kosareva AA, Kenigfest N (1980) The retinofugal and retinopetal systems in Lampetra fluviatilis. An experimental study using radioantographic and HRP methods. Brain Res 195:453-460

Vesselkin NR Repdrant J, Kenigfest NB, Miceli D, Ermakova TV, Rio JP (1984) An anatomical and electrophysiological study of the centrifugal visual system in the lamprey (Lampetrafluviat- ilis). Brain Res 292:41-56

Vesselkin NP, Rep~rant J, Kenigfest NB, Rio JR Miceli D, Shupliakov OV (1989) Centrifugal innervation of the lamprey retina. Light- and electron microscopic and electrophysiological investigations. Brain Res 493:51-65

West RG (1972) Superficial warming of epoxy blocks for cutting of 25-150 gm sections to be resectioned in the 10-90 nm range. Stain Technol 47:201-204

Early development and organization of the retinopetal system in the larval sea lamprey, Petromyzon marinus L

Documents