Cellular interactions and pattern formation in the development of the visual system ofDaphnia magna (crustacea, branchiopoda)

Cellular interactions and pattern formation in the development of the visual system ofDaphnia magna (crustacea, branchiopoda)

DEVELOPMENTAL BIOLOGY 73,206-238 (1979) Cellular Interactions and Pattern Formation in the Development of the Visual System of Daphnia magna (Crus...

24MB Sizes 0 Downloads 17 Views

DEVELOPMENTAL

BIOLOGY

73,206-238

(1979)

Cellular Interactions and Pattern Formation in the Development of the Visual System of Daphnia magna (Crustacea, Branchiopoda) I. Interactions

between

Embryonic

Retinular

Fibers

and Laminar

Neurons

E. R. MACAGNO Department

of Biological Received

Sciences,

March

Columbia

12, 1979; accepted

University, in revised

New form

York, April

New

York

10027

25, 1979

Retinular cells in the Daphnia compound eye were deleted at specified developmental stages and in adults to test the hypothesis that growing retinular fibers trigger the differentiation of laminar neurons, thereby recruiting them as synaptic targets and organizing them into a particular structural pattern. Deletion of retinular neurons was accomplished by producing small lesions with either ultraviolet or laser microbeams. The effects of the lesions on the eye and optic ganglion were assayed quantitatively using computerized techniques for the three-dimensional reconstruction of biological structures from serial light or electron micrographs. The following results were obtained: (1) In animals with embryonic retinular lesions, the number of laminar neurons found in adults was reduced and was roughly proportional to the number of surviving retinular cells which sent fibers to the lamina. When retinular lesions were made in adults, however, no evidence of laminar cell loss was found. (2) A reduced number of higher-order neurons (medullary neurons) was found only in animals with very large embryonic retinular lesions. (3) When retinular cells were deleted before their fibers could grow into the lamina and contact immature target cells, the laminar cells deprived of these contacts failed to differentiate morphologically before degenerating at late embryonic stages. (4) In a number of cases, optic cartridges were formed with abnormal numbers of retinular fibers and laminar neurons. These observations support our hypothesis and increase our understanding of the role of cellular interactions in the development of the arthropod visual system.

hagen, 1976) and in some crustaceans (Macagno et al., 1973; Nassel, 1976; Stowe, 1977). Much experimental work has been directed at determining just how these two arrays and their connections are formed. This and subsequent papers report and discuss the results of a series of experiments which I have carried out to explore the role of cell interactions in the genesis of the visual system in the branchiopod crustacean, Daphnia magna. The experiments discussed in this report were designed to test the hypothesis that growing retinular fibers, through specific interactions with immature laminar neurons, determine the fate of these and other neurons in the optic ganglion of Daphnia. During normal development, retinular fibers grow into the laminar anlage and se-

INTRODUCTION

The arthropod visual system, with its extraordinarily regular cellular and synaptic arrangements, has long been a favored experimental subject of neurobiologists interested in describing the developmental mechanisms that underlie pattern formation. Compound eyes and optic ganglia consist of a series of essentially two-dimensional arrays of closely packed, repeating units connected by precisely ordered fiber projections. Within each unit, characteristic patterns are found in the arrangements of cells, cellular organelles, tibers, and synaptic connectivity of the two most peripheral arrays, the retinula (the photoreceptors) and the lamina, are particularly well known in a number of insects (Meinertz206 0012-1606/79/1202ofS33$02.00/0 Copyright All rights

0 1979 by Academic Press, Inc. of reproduction in any form reserved.

E. R. MACACNO

Cellular

quentially contact immature cells which then become their adult synaptic targets (Lo Presti et al., 1973). This early contact, which includes the formation of transient gap junctions between fibers and immature cells (Lo Presti et al., 1974), occurs just before these cells begin to differentiate morphologically by extending processes into the area that will become the laminar neuropil. This report discusses the results of deleting retinular neurons at two embryonic stages (before and after their fibers grow into the laminar anlage) and in adults. Deletions were accomplished by producing small lesions at specific locations in the compound eye using either ultraviolet (uv) or laser microbeams. Observations of the effects of such lesions on the laminar and higher-order regions of the optic ganglion were made on experimental embryos and adults serially sectioned for light or electron microscopy. Quantitative measurements of cell numbers and cell distributions were carried out using computerized techniques for the analysis of serial-section micrographs (Levinthal et al., 1974; Macagno et al., 1976). Parts of this work have been reported previously (Macagno, 1977, 1978). MATERIALS

Daphnia

AND

METHODS

Cultures

Animals used in these experiments were selected from a culture maintained continuously at Columbia since 1970 (Macagno et al., 1973). The animals, originally derived from the parthenogenetic offspring of a single female, are kept in beakers at nearly constant density in constant temperature incubators under continuous fluorescent lighting. Those used in these experiments were kept at 24°C. Cultures are transferred aseptically three times a week into fresh minimal salt medium (Macagno et al., 1973) with vitamins and trace elements (Murphy, 1970), and are fed daily a constant amount of algae. The algae, Chlamydomonas rein-

Interactions

in Development

207

hardii, are grown in a medium containing sodium acetate as a carbon source, using stock cultures obtained from Carolina Biological Supply Company. Since under these growth conditions Daphnia reproduction is entirely by parthenogenesis, all animals used in these experiments were isogenic. Staging In constant lighting and temperature the length of the embryonic cycle in Daphnia is invariant. Oviposition can be timed accurately by observing the state of the ovaries with the unaided eye and by noting the shedding of the carapace, which occurs 10 to 15 min before the eggs are extruded into the brood pouch. The extrusion takes place in about 1 min, depending upon the number of eggs in a brood. The number of eggs varies with instar as well as with certain environmental parameters (food supply, degree of crowding, etc.). In fixed environmental conditions developmental events occur at consistent intervals after oviposition, so that staging can be done by the clock. There are no detectable differences between embryos that develop in vitro and those that develop in the maternal brood pouch. Embryonic development takes about 55 hr at 24”C, at which time animals extend their caudal fin and begin to swim vigorously; feeding begins a few hours later, although yolk granules are still visible and are not completely digested until about 75 hr. Pigment forms in an ocellus before it does in the compound eye and serves as a useful temporal marker. Morphological differentiation of eye and optic ganglion neurons occurs from about 28 hr onward. Throughout this paper times mentioned are measured from the extrusion of the eggs into the brood pouch. In practice, a large animal (about 4 weeks old) with darkened ovaries is isolated and observed until it lays eggs. Provided the brood contains more than 12 embryos, the animal is then placed back in the incubator until 20 hr after oviposition, when the em-

DEVELOPMENTAL BIOLOGY

208

VOLUME 73, 1979

bryos are removed from the brood pouch and allowed to continue their development in the incubator in vitro until they are irradiated.

Ultraviolet

Irradiation

Retinal lesions were made in individual embryos with an ultraviolet (uv) microbeam (see Fig. la). The design of this microbeam apparatus is quite similar to those of other investigators (Uretz and Perry, 1957; Starling, 1974). It consists, basically, of an American Optical microscope fitted with a 50-W mercury arc source for epiillumination. The light path has been modified so that the uv wavelengths in the mercury spectrum can reach the object on the microscope stage without being attenuated. A Beck reflecting objective (36x, N.A. 0.50) is used both to observe the specimen with transmitted light from a tungsten source and to form the focused image of an aperture located in front of the mercury lamp. The diameter of the beam cross section and its location at the focal plane can be adjusted by changing the size and position of this aperture. The aperture used throughout these experiments produced a beam cross section of 6 pm. Exposure to the beam is brought about by moving a rotating shutter by hand, with the exposure necessary for causing cell death determined empirically (see below). An embryo, held by gentle suction on the tip of a polished glass pipet, is positioned in the object plane with respect to a cross hair in an eyepiece. Positioning is accomplished by using the stage and focusing controls of the microscope and by rotating the pipet. It is necessary to anesthetize the embryos prior to irradiation since they occasionally make twitching movements. After irradiation the embryos are placed in individual depressions in plastic trays containing fresh medium and returned to the incubator. For each brood half of the animals are uv irradiated and the other half used as controls. Of the controls, half are set aside prior to irradiation

a

1

FIG. 1. (a) Ultraviolet microbeam apparatus used in these experiments.(1) Tungsten source for transmitted illumination. (2) Condenser. (3) Specimen chamber on the microscope stage. (4) Reflecting objective, X 36. (5) Half-silvered mirror. (6) Barrier filter. (7) Eyepiece with cross hair for locating the target. (8) Beam shutter. (9) Filter. (10) Pinhole aperture. (11) Quartz condenser. (12) Mercury arc source, 50-W HBO lamp. (b) Diagram of the retinular and optic ganglion anlage in a stage A embryo. The arrows indicate the directions of the uv microbeam that were used in these experiments. The broken lines outline the region of highest radiation flux and the area of the head in which lesions were made. The epidermis is the hatched area. This diagram represents a coronal cut through the middle of the retinula. MP, midsaggital plane, or plane of bilateral symmetry.

and not handled further until feeding. The other half of the controls are subjected to the same procedure as the experimental animals but exposed to the microbeam with the uv part of the spectrum strongly atten-

E.

R. MACAGNO

Cellular

uated by a glass filter (transmittance 0.07% at 260 nm, 4% at 280 nm, and 40% at 330 nm). Both control groups developed normally throughout these experiments. At about 60 hr, when Daphnia already swim vigorously, all animals are transferred to fresh medium and fed algae. Some animals are allowed to survive to the third instar, at which time they are tested for simple behavioral responsesto light and prepared for anatomical observations. Others are fixed shortly after irradiation to assay the character of the lesion.

Parameters of Irradiation 1. Developmental stage. Embryos at two developmental stages were used in these experiments. Stage A embryos were between 28 and 29 hr old, a time when the first retinular cells begin to show morphological differentiation (e.g., pigment granule formation and axonal extension) but have yet to contact laminar targets. Stage B embryos were between 42 and 44 hr old, by which time all retinular cells have made contact with their laminar targets. Older embryos and adults proved refractory to making lesions with the uv microbeam, principally because of the large amount of shielding pigment in the eye. Lesions in adults were made instead with a laser beam (see below). 2. Radiation dosage. The radiation dosage was varied by opening the beam shutter for times from 5 set to 2 min (no attempt was made to measure the absolute flux of radiation in the uv microbeam). Cell death resulted from exposures of 15 set or longer (except when filters were used; seeResults), with an increasing number of cells killed as the exposure time was increased. A possible reason for this dosage effect lies in the fact that, because the beam converges through other cells onto those that are specifically positioned at the minimum beam diameter, these other cells receive a proportionately lower flux of radiation. Since radiation effects are cumulative, those cells receive a lethal dosage only at longer exposures.

Interactions

in Development

209

3. Direction of the beam. In order to avoid producing any damage to the laminar anlage, which in the embryo is adjacent to the developing eye, experimental animals were oriented so that the beam passed almost entirely through the eye. (The geometry of the embryonic tissues and the beam orientation are illustrated in Fig. lb.) In some test animals the beam was focused at the eye but directed toward the developing lamina. Although no damage to the lamina was found in any of these test animals, all results discussed below were obtained from animals irradiated as indicated in Fig. lb. 4. Location of the lesion. The location of the beam was varied relative to the embryonic eye in order to produce lesions in groups of retinular cells at specific developmental stages. Lesions near the midplane clearly affected the column of nonneural cells, which is very prominent in this location during development (Lo Presti et al., 1973), as well as the retinular cells. 5. Postirradiation illumination. To test the possibility that uv-induced lesions could be affected by postirradiation illumination (i.e., by photoreactivation (Cook, 1970)), alternate irradiated embryos in one series were placed into a light-tight container in the incubator immediately after exposure to the uv microbeam. These embryos were transferred in weak red light and received no other light exposure (except from the microbeam) between the time they were positioned in the microbeam apparatus and the time they were fixed for histology. The other embryos in this group were placed in the incubator at a distance of 40 cm from a 20-W GE Cool White fluorescent light bulb. Except for this particular series, the postirradiation conditions were exactly the same for experimental and control embryos (i.e., continuous &orescent illumination at a constant flux). Laser Irradiation Eye lesions were produced in a number of Daphnia adults with an argon laser microbeam. Since the green lines of argon are

210

DEVELOPMENTAL BIOLOGY

strongly absorbed by the shielding pigment in the retinular cells, it was necessary to use very short exposures and an attenuated beam. No attempt was made to quantify the dosage or to orient the eye in a highly specific manner. Each adult was anesthetized and irradiated while lying on its side in a small drop of medium. In this orientation (see Fig. 2) the laser beam misses the optic and supraesophageal ganglia completely. After irradiation the animals were placed in fresh medium and allowed to survive for up to 3 weeks before fixation and analysis. Used in this manner, the laser microbeam is essentially like a very small

VOLUME 73, 1979

microcautery needle, and cell death is brought about by heating. Histology Experimental and control animals were fixed at specific times after irradiation using standard techniques (Macagno et al., 1973). Animals were cut in half and put into 1% osmium tetroxide in Dalton buffer, with somewhat lower salt concentration for embryos than for adults (see Lo Presti et al., 1973). Following a buffer rinse, specimens were block stained with 0.5% uranyl acetate in maleate buffer, dehydrated, and embedded in epoxy resin. Serial sections were cut

FIG. 2. Visual system of two adult experimental specimens irradiated at stage A, fixed, and mounted in epoxy prior to sectioning. CE, compound eye; EM, eye muscles; LA, lamina; ME, medulla; ne, ocellus or nauplius eye; RF, retinular fibers; SG, supraesophageal ganglion. The arrow in (b) indicates the site of the lesion in this specimen. After sectioning, the specimen in (a) was found to be missing 3 R-fibers and no L-cells, whereas the specimen in (b) was found to be missing 49 R-fibers and 31 L-cells. The lamina is clearly smaller in the specimen on the right side. Bar = 50 pm.

E. R.

MACACNO

Cellular

thick (1 pm) for light microscopy (LM) or thin (80 to 150 nm) for electron microscopy (EM) using a Sorvall MT2-B microtome. While cutting thick sections a few thin sections were taken through the optic fibers in order to assay fiber number and morphology at the EM level. Sections were usually cut perpendicular to the eye-ganglion (anterior-posterior) axis. Thick sections were stained with toluidine blue, thin sections with uranium and lead. Thick sections were photographed with 35-mm Kodak Panatomic-X film at various magnifications, using either Zeiss or Leitz photomicroscopes. Thin sections were photographed at low power (700 X) in a Zeiss EMS-S modified to accept a lOO-ft roll of Kodak AHU microfilm. Some thin sections, especially in neuropil areas, were photographed at higher magnifications (3000 X, 7000 x). At low power (700 X) most of the compound eye and optic ganglion of young adults fit within single 70-mm negatives, so that mosaics of photographs are not generally required for complete analysis of these two organs.

Computer

Reconstruction

In order to assay experimentally induced changes in cell number and arrangement, cellular morphology, and synaptic connectivity, serial micrographs of both thick and thin sections were aligned with respect to each other and rephotographed on 35-mm film strips using an image combiner (Levinthal et al., 1974). In these film strips frame number corresponds to position along the third spatial axis. The film strips, one for each specimen, were analyzed using threedimensional computer reconstruction techniques developed at Columbia University (Levinthal et al., 1974; Macagno et al., 1976). The architecture of the eye and of various components was obtained by tracing corresponding boundaries on each individual movie frame. The computer, using section thickness to transform frame number to a spatial coordinate (the Z-axis), can

Interactions

in Development

211

produce a display of the stored data at any viewing angle desired as well as stereopair displays. Cell counts and cell distributions were determined by placing a pointer on each cell nucleus, storing its position and diameter in three-dimensional space, and displaying the data as three-dimensional crosses or as spheres. Simple algorithms were used to check that no cell nucleus was recorded more than once. Multiple passes through the film strips by different operators were used to ensure that no cells were left uncounted. Individual neurons were recorded from films of thin serial sections by tracing the perimeter of each process on individual frames or by drawing a linear branching structure through the center of each process as it was followed from frame to frame of a film strip. RESULTS

The Number of Retinular vive uv Irradiation

Fibers That Sur-

Each retinular cell in the single, medially located, compound eye of Daphnia magna sends a fiber (R-fiber) to the optic ganglion. Since there are normally 8 retinular cells per ommatidium and 22 ommatidia (11 on each side, symmetrically located about the sagittal midplane), the optic nerve consists of 176 fibers. The R-fibers from each ommatidium, along with a glial process, are segregated into 22 individual bundles in the optic nerve (Wolff and Guldner, 1970; Macagno et al., 1973). In order to assay the number of deleted retinular neurons under different irradiation conditions, some experimental animals which had been irradiated at embryonic stage A (28-29 hr) were thin sectioned through the optic nerve. Figure 2 shows two specimens photographed in the block prior to sectioning. The R-fibers were counted from electron micrographs such as those shown in Fig. 3. The data for these experiments are presented in Table 1. It can be seen from the data on series 1, 3, and 4 that the extent of the lesion in-

212

DEVELOPMENTAL BIOLOGY

creases with dosage, when all other parameters remain constant. For 20-, 30-, and 60set exposures to the uv microbeam, the mean loss of R-fibers was about 35,59, and 105 fibers, respectively. A significant variation from animal to animal was found, and is reflected in the large standard deviations, especially at high dosages. The strong effect of postirradiation illumination is clearly shown by a comparison between series 1 and 2. These two groups differ experimentally only in the postirra-

VOLUME 73, 1979

diation treatment. Immediately after exposure to the uv microbeam, series 1 specimens were placed at a standard distance from a fluorescent light source (see Materials and Methods) until they were fixed, whereas series 2 animals were placed in a light-tight box (in the same incubator) until fixation. (Normally the development and structure of retina and lamina in Daphnia are not significantly affected by lighting conditions during development (Macagno, 1975)J After a 20-set exposure, specimens

FIG. 3. Low-power electron micrographs of sections through the optic nerve at the base of the compound eye of a normal, (a), and two experimental adults, (b) and (c), irradiated at embryonic stage A. Some retinular fibers in the optic nerve can be seen to be clearly segregated into bundles of eight fibers. One such bundle, corresponding to the same ommatidium, has been circled in each micrograph. Retinular cell bodies are visible within the margin of the eye, which is indicated in (a) by arrowheads. The cells, as well as the fibers, contain shielding pigment granules. There are 176 fibers in (a), 173 in (b), and 155 in (c). The lesion in (b) was restricted to the deletion of three retinular cells from one ommatidium; there is one bundle, therefore, with five fibers, which is indicated by the arrow. The lesion in (c) deleted two ommatidia completely and three partially. Bars (lower right) = 10 pm.

FIG. 3 b and c 213

DEVELOPMENTAL BIOLOGY

214

TABLE

VOLUME 73, 1979 1

NUMBEROFR-FIBERSINEXPERIMENTALANIMALSASAFUNCTION Series

No. of animals

Controls 1 2 3 4 5 6

10 6 7 6 4 9 6

Length of exposure 20 20 20 30 60 15 30

set set set see set min min

OFIRRADIATIONEXPOSURE

Filter

Postirradiation illumination

Mean No. of R-fiib;sdper

SD”

Glass6 None None None None Glass? Glass*

Yes Yes No Yes Yes Yes Yes

176 140.8 103.6 117.3 71.2 171.6 145.6

0.0 11.1 12.5 9.7 23.1 6.8 9.4

a Standard deviation of the mean number of optic fibers. b Transmittance 0.07% at 260 nm, 4% at 280 nm, and 40% at 330 nm.

in series 1 lost about 35 fibers, series 2 about 73. It is apparent from this result that photoactivatable repair mechanisms which can strongly affect the recovery of individual cells from uv radiation damage are present in Daphnia. This suggests that the damage could be largely in the DNA of the cell (Cook, 1970). In order to test further whether photons at 260 nm, the wavelength of maximum absorption by DNA, were mainly responsible for the cell deletions, a glass filter was used in series 5 and 6 (as well as in the controls) to attenuate the beam. This filter has a transmittance of 0.07% at 260 nm (DNA absorption peak) and 4% at 280 nm (protein absorption peak). In the controls (20-set exposure) no cell loss was detected. On the average about 4 retinular neurons were deleted per specimen for a 15min exposure (series 5), and 30 for a 30-min exposure (series 6). If we compare series 1 (20-set exposure, no glass filter) and series 6 (MOO-set exposure, glass filter), we see that the number of retinular cells deleted is about the same (35 and 30, respectively) despite a loo-fold difference in the time of -. FIG. 4. Electron (b) after irradiation photoreceptor cells within the cytoplasm staining, degenerating debris remain within boundary of the eye

exposure. Since the glass filter transmits less than one part in a thousand at 260 nm, we would expect that fewer cells would be deleted if all damage were caused by 260nm radiation. It is apparent that longer wavelengths also play a part in producing the deletion of retinular cells. The exact nature of the lesion remains, of course, unknown. Time Course of w-Induced Degeneration

Retinular

Cell

A group of 30 animals irradiated at embryonic stage A were fixed at short times (30 min to 5 hr) after irradiation to ascertain how rapidly the affected retinular cells degenerate. By about 2 hr after irradiation, cells at the lesion site are clearly degenerating (see Fig. 4a). The cytoplasm of these cells is condensed and stains strongly with heavy metals. Debris from degenerating cells is found within phagocytic vacuoles in the cytoplasm of nearby glial cells. Four to 5 hr after irradiation few signs of degeneration remain; there is some debris, largely within glial cells at the midplane (see Fig. 4b). Observations on a few serially sec-

micrographs showing degenerating retinular cells in irradiated embryos, 2 hr (a) and 4 hr at stage A. The nauplius eye (NE) can be seen in both micrographs at lower left, with already containing shielding pigment granules. (a) Degenerating cells (arrows) can be seen of phagocytic cells, adjacent to their nuclei (n). Particularly noticeable are the darkmitochondria (m). (b) Four hours after irradiation only small pieces of dark-staining a column of glial cells along the midplane of the eye (arrows). Arrowheads denote the anlage. Bars = 5 pm.

E. R. MACACNO

Cellular Interactions

FIG.

4 a and b

in Development

215

216

DEVELOPMENTAL BIOLOGY

tioned specimens indicate that the growth of fibers of irradiated retinular cells either fails to start or is rapidly stopped, so that these fibers never contact cells in the lamina. As the irradiated cells degenerate, their positions are rapidly occupied by adjacent survivors which move into the vacated area as the eye coalesces into a smaller structure. Retinular cells irradiated at stage B, after all fibers have reached the lamina, also appear to degenerate rapidly. Within a few hours all signs of degeneration have disappeared from the eye, which again has coalesced into a smaller structure.

Lack of Direct Radiation Optic Ganglion

Damage

to the

A critical requirement for these experiments is that no lesions of the optic ganglion be produced directly by the experimental procedures. This is not a problem at all in adults, where the eye is at a considerable distance from the optic ganglion. In the embryo, however, the developing optic ganglion is adjacent to the eye- Some orientations of the embryo with respect to the microbeam allow radiation to enter the optic ganglion after passing through the epidermis and eye anlage. Because the shorter uv wavelengths (up to about 280 nm) are strongly absorbed by macromolecules in the intervening tissue, a very small fraction of this component of the radiation will reach the inner layers. Absorption by tissue is about 90% per 10 pm at 260 nm (Jagger, 1967). Since the epidermis’ is at least 5 pm thick and the embryonic eye at least 25, at most a part in a thousand of the 260-nm radiation would reach the optic ganglion. As noted above, however, wavelengths longer than 260 nm, which are less strongly absorbed by tissue, appear to contribute to the cell deletion in the eye. Tests were therefore performed to determine if these wavelengths could, in fact, cause damage in the optic ganglion. Three stage A embryos were irradiated for 60 set in an orientation such that the beam, while being focused at

VOLUME 73, 1979

the eye, would enter the optic ganglion after passing through eye and epidermis. In none of these embryos was any direct damage to the cells of the optic ganglion apparent. For further safety, however, embryo orientations which minimized exposure of the optic ganglion to the microbeam (see Fig. lb) were used in all the experiments discussed below.

Structure imental

of the Lamina

in Mature

l&-per-

Animals

In normal Daphnia adults the 22 bundles of R-fibers project onto 22 structural units in the lamina. These units, called optia cartridges, include 5 laminar neurons each. for a total of 110 laminar neurons (Macagna et al., 1973). Each R-fiber bundle enters the lamina through a ring formed by the five laminar neurons, as shown schematically in Fig. 5a. Once beyond the layer of laminar cell somata, the R-fibers expand and branch out, forming synaptic junctions with the monopolar extensions of the laminar neurons. Retinular processes terminate in the laminar neuropil; laminar processes extend into the medullary neuropil (Macagno et al., 1973; Macagno and Levinthal, 1975). In order to ascertain the effects of retinular cell deletion on the lamina, 13 adults uv irradiated as stage A embryos, 4 as stage B embryos, and 2 laser irradiated as adults were serially sectioned through the eye and optic ganglion for light or electron microscopy. A few were sectioned only through the optic nerve and optic ganglion. Movies of aligned serial sections were made for all series and analyzed with the computer to obtain cell counts and cell distributions in the eye and lamina, as well as a count of the fibers in the optic nerve. Stage A lesions. The deletion of retinular cells before their axons make contact with laminar neurons has profound effects on the lamina. First, the size of the lamina is reduced, as in the micrographs of sections near the surface of the lamina shown in Figs. 5c and d. Second, the number of laminar neurons is also reduced, in proportion

Cellular

E. R. MACAGNO

Interactions

TABLE NUMBERS

OF R-FIBERS

AND L-CELLS

2

IN EXPERIMENTAL R-fibers

217

in Development

ANIMALS

R-fiber bundles

IRRADIATED

STAGE A

R-cells

Controls*

176

176

22

22

110

0.625

5.00

n.d. n.d.’ n.d. 127

163 153

C d

153

20 21

17 15

94 99 105

127

16

n.d.

79

0.607 0.614 0.686 0.622

4.95 4.70 5.00 4.94

F

126 122

126 122

15 16

n.d.

73 81

0.598 0.643

4.87 5.06

E

n.d. 116 n.d.

111 116

14 18

102

14

69

0.750 0.658 0.676

4.83 5.21 4.93

j k

89 87

89 87

12 14

129 9 n.d.

73 87

i

57 68

0.640 0.782

4.75 4.86

1 m nd

96 47

70 39

9 5

45 22

0.643 0.564

5.00 4.40

0

0

0

-

-

Normal bundles”

L-cells

AT EMBRYONIC

Specimen

L-cells/Rfibers (Rl)

L-cells/ b%%

E

n Bundles with eight optic fibers. ’ Five controls were analyzed. ’ Not determined. ’ Only the left half; exact counts specimen.

of right

8

n.d. n.d. 0

side not possible

to the number of retinular cells deleted. The numerical data for the 13 animals in these experiments are presented in Table 2. Computer reconstructions from which these data were obtained are shown in Fig. 6. A plot of the number of laminar neurons versus the number of retinular fibers appears in Fig. 7. It is evident from these data that, over a wide range of retinular cell deletions, the number of laminar neurons found in the adult is roughly proportional to the number

because

0

series of sections

is incomplete

for this

of R-fibers that reach the lamina (Fig. 7). There are, however, significant and interesting deviations of the ratios Rl (L-cells/ R-fibers) and R2 (L-cells/R-fiber bundles) (see Table 2) away from their normal values of 0.625 and 5, respectively. For example, a value of Rl greater than 0.625 (found in eight cases) implies that at least some cartridges have fewer than eight R-fibers per five L-cells. A value of Rl smaller than 0.625 (found in five cases) implies that there are more R-fibers per L-cell than normal in

FIG. 5. (a) Diagram of the surface of the Daphnia optic lamina, indicating the arrangement of the 11 optic cartridges on each side of the midplane in an adult. (b-d) Low-power electron micrographs of corresponding regions of the lamina in three serially sectioned animals. All micrographs have the same magnification and orientation, with dorsal up and ventral down. The position of the midplane is indicated by arrowheads. The bar in (b) corresponds to 20 pm. (b) Control animal. Twenty-two bundles of retinular fibers can be found, each with 8 fibers. Because the laminar surface is curved, only some laminar neurons are visible, mostly near the midplane. Because the plane of section dips further into the ganglion on the left, some bundles on the right have yet to join the lamina. Bundles on the right have been lettered a through k. (c) Experimental adult, irradiated at embryonic stage A. (This is specimen a in Table 2.) The lesion, made at the midplane, deleted one ommatidium completely on each side. Those on the right are labeled a through j. As determined from serial sections, bundle a contains only six fibers, g only seven, and the rest the normal complement of eight. (d) Experimental adult irradiated at embryonic stage A. (This is specimen h in Table 2.) The lesion was entirely on the right side, so that the left side shows a normal set of eleven cartridges. Only three bundles, labeled a, b, and c, can be found on the right side, which is obviously smaller than the control left side. Bundle c has seven optic fibers and the others eight, as determined from serial sections.

218

DEVELOPMENTAL BIOLOGY

a

FIG. 5 aandb

VOLUME 73, 1979

. /

_- .FIG. 5 c and d 219

220

DEVELOPMENTAL

BIOLOGY

VOLUME

73, 1979

d FIG. 6. Computer reconstructions of eye and optic ganglion of control and experimental animals. (a) Contour tracings of a control animal, viewed from the side, as in the whole mounts of Fig. 2. (b) Location of the cell nuclei of the same animal as (a). ce, compound eye; la, lamina; me, medulla; sg, anterior portion of supraesophageal ganglion. (c) Same as (b), but rotated 90’ for a view along the dorsoventral axis. (d-f) Experimental animals, showing the contours of the eye as well as the location of each retinular, laminar, and medullar cell nucleus. The size of the cross corresponds to the size of the nucleus. Note that the medtdlar cells are clearly separated into right (rme) and left (lme) groups. Data for specimens d, e, and fin Tables 2 and 5 were obtained from these reconstructions. Magnification the same in (a-c); bar in (c) = 50 pm. Magnification in (d-f) the same; bar in (f) = 50 pm.

some cartridges. The number of abnormal bundles is also indicative of the extent of deviation of Rl from the control value. The values of Rl and R2 can be determined, in principle, for each cartridge in each specimen as well as for each specimen as a whole. This, however, requires careful assessment of synaptic connectivity in EM serial sections. It is the only way to determine with certainty which cartridge an L-

cell belongs to when cartridge structures are disrupted. Because this is a very timeconsuming process, the cartridges in only five experimental specimens have been analyzed in this manner. The results, which are presented in Table 3, indicate that, in animals with relatively small lesions, bundles of eight R-fibers do not recruit and maintain more than five L-cells, although extra L-cells have been made available by

E. R. MACAGNO

Cellular

Interactions

in Development

R-fibers

FIG. 7. Plot of the number of laminar cells (L-cells) versus the number of retinular fibers (R-fibers) in control and experimental adults irradiated as stage A embryos. The line, drawn arbitrarily with a slope of % and intersection at the origin, is meant to show the deviation of individual cases from a purely linear relation between L-cells and R-fibers.

the deletion of other ommatidia. In two animals with extensive lesions (specimens h and i, Table 3), however, some cartridges were found with an extra L-cell. It is also apparent from these data that bundles with fewer than eight fibers can maintain a normal complement of five L-cells, and in a few cases six. Two cases in other specimens, both with bundles of only two fibers, have only two L-cells associated with them. One of these cases is illustrated in Fig. 8d, along with other examples of abnormal cartridges. Two other cases are worthy of special note. One fiber bundle on the left side in specimen b (Table 3) has no L-cells associated with it. A second fiber bundle, on the right side of specimen a (Table 3), has 14 fibers and forms a cartridge with 5 Lcells (see Fig. 8b). This supernumerary bundle appears to arise from the fusion of two bundles from adjacent ommatidia with partial deletions of retinular cells. Further description and a discussion of synaptic con-

nectivity in these abnormal cartridges will appear in a later report. In a few cases where a large, medially located lesion was made, as in specimens 1 and m in Table 2, a number of retinular cells that survive uv irradiation differentiate partially (they form pigment granules, for example) but do not send axons to the lamina. This is perhaps because the medial glial substrate on which axons travel to the lamina (Lo Presti et al., 1973) is also partially destroyed by this type of lesion though damage to the retinular cells cannot be entirely discounted. In these cases, the number of L-cells that survive is apparently determined by the number of R-fibers that get to the lamina, and not by the number of surviving retinular cells. This further supports the hypothesis that direct interaction with R-fibers is required for L-cell survival. Stage B lesions. The deletion of embryonic retinular cells after their axons have

222

DEVELOPMENTAL

BIOLOGY

TABLE 3 NUMBERS OF R-FIBERS AND L-CELLS IN OPTIC CARTRIDGES IN EXPERIMENTAL ANIMALS Specimen”

Side

R-fibers

L-cells

8 7 6 14 8 163

5 4 5 5 5 99

8 1 1 1 9 20

8 7 5 8 5 153

5 4 0 5 4 93

8 1 1 9 1 20

8 7 6 5 8 153

5 5 5 5 5 105

4 1 1 4 11 21

8 8 8 a 7 111

6 5 6 5 6 73

1 10 1 1 1 14

8 7 8 8 6 102

5 6 6 5 6 69

6 3 1 2 1 13

a

Left

Right Total b

Left

Right Total C

Left

Right Total h

Left Right

Total i

Left Right

Total n Same

as in Table

Number of cartridges

2.

contacted the immature laminar neurons also leads to the degeneration of the latter. The data, which closely resemble those for stage A lesions, are shown in Table 4. The loss of L-cells is roughly proportional to the number of R-cells deleted. Adult lesions. In contrast to the strong effects of R-cell deletion in the early embryonic stages, similar operations using laser irradiation in adults do not lead to the loss of L-cells, at least within a period of a few weeks. Figure 9 shows sections through

VOLUME

73. 1979

the optic ganglion of a laser-irradiated ammal which was fixed a week after the irradiation. Most of the R-fibers have degenerated, leaving space in the ganglion which is then filled by glial processes, though a number of degenerating retinular terminals are still visible. In two specimens where serial section reconstructions were used to count the number of L-cells, 109 and 110 cells were found 3 weeks after the operation. The number of viable R-fibers remaining in these two specimens were 104 and 120, respectively. There are no signs of degeneration of the laminar neurons in either of these two specimens. Effects on the Higher-Order Optic Ganglion

Regions

of the

The deletion of photoreceptors in development also affects the structure of the higher-order region of the optic ganglion, the medulla. However, the effect is different from and not as profound as that in the lamina. There appears to be a definite threshold before loss of medullary cells (Mcells) begins, as the data below indicate, and total degeneration of the medulla does not occur, even when the lamina has completely disappeared. A small number of adults irradiated as stage A embryos were serially sectioned for light microscopy through the entire optic ganglion. Computer reconstructions and cell counts of both lamina and medulla were made in these cases. The medulla normally has approximately 330 neurons (about 165 each on the left and right sides) (Macagno and Levinthal, 1975). There is a small uncertainty (generally of two to four neurons) in counting cells in the medulla at the light microscopic level because in the region where the optic ganglion merges with the supraesophageal ganglion a few cells could be assigned to either ganglion. (At the EM level the position of the dendritic field can be easily determined, making such a choice unambiguous.) The effects of R-cell deletion, however, fall well beyond this range of

E. R. MACACNO

Cellular

uncertainty. The data for four experimental specimens and one control appear in Table 5. Computer reconstructions of experimental and control animals are shown in Figure 6. In specimen d the lesion was made entirely on the right side. R-cell and L-cell loss on the right side is over 50%, but no effect is apparent in the number of M-cells. In specimen 1, the lesion includes both sides, but is larger on the right. R-cell and L-cell loss is about 40% on the left, but again the number of M-cells is normal. On the right side, however, almost 80% of the R-cells and L-cells are gone, and now a 35% loss of M-cells is also apparent. This observation is supported by the measurements in specimen m, in which a medial lesion has caused extensive cell loss on both the left and right sides. In this case, M-cell loss occurs also on both sides. Although complete elimination of R-cells in one half of an eye leads to complete loss of that side’s lamina, this is not the case for the medulla. In specimen n there are some 30 medullar neurons remaining on the right side even though the right halves of eye and lamina are not there at all. Sections through this specimen are shown in Fig. 10. A more detailed determination of M-cell survival as a function of L-cell number requires further reconstructions and will be reported in the future. The Role of R-Fibers in the Differentiation, Proliferation, and Degeneration of Laminar Neurons In general, animals irradiated at stage A (28-29 hr) show no signs of degeneration of L-cells until about 48 hr (see Fig. lla). This is some 10 hr after the 38-hr stage, when the last retinular fiber bundles would normally arrive at the lamina (Lo Presti et al., 1973). However, in one case out of seven examined, degeneration has taken place as early as 45 hr, as shown in Fig. llb. Degenerating laminar cells are found adjacent to the midplane in all these animals, indicat-

Interactions

in Development

223

ing that those that degenerate are probably those which are last available for contact and recruitment by the ingrowing retinular fibers (see Macagno, 1978). A section through a 49-hr embryo irradiated at stage A is shown in Fig. llc. Evidence at the EM level of cell degeneration in these animals consists mostly of the presence of membrane-bound, deeply staining cellular debris, though occasionally whole cells which stain heavily and are very shrunken can also be found. Darkly staining cells can also be found with the light microscope, though sometimes they are difficult to identify (see, however, Fig. 11~). Observations on 20 experimental animals fixed at various stages between 42 and 55 hr indicate that L-cells do not all degenerate at one particular time, but instead over a period of 5 to 10 hr. Few signs of degeneration are found in control animals fixed at these stages, indicating that normal overproduction of laminar cells is probably small. L-cells which are not contacted by Rfibers fail to differentiate processes. There is no evidence of formation of neuropil in regions near the midplane if R-fibers are absent (see Fig. 12). Moreover, no signs of degenerating processes can be found in the laminar neuropil when the extra L-cells degenerate, although conceivably such processes could have been retracted before degeneration began. In two serially sectioned 45-hr embryos irradiated at stage A, no processes were found in three extra L-cells that were reconstructed with the computer. It is reasonable to assume that, in general, L-cells begin their morphological differentiation only after being contacted by Rfibers. In addition, a small (10 to 15%) increase in cell size which accompanies early differentiation fails to occur in the untouched cells. The results obtained so far do not permit me to state conclusively whether the usual number of laminar cells is born and available for recruitment in animals with experimentally reduced eyes. Since unrecruited

FIG 22

E. R. MACAGNO

Cellular

Interactions

TABLE NUMBERS

OF R-FIBERS

AND L-CELLS

IN EXPERIMENTAL

Specimen

R-cells

R-fibers

Controls* 0 P q r

176 nd.’ n.d. n.d. nd.

176 151 137 135 133

R-fiber bundles

225

in Development

4 ANIMALS

IRRADIATED

Normal bundles”

L-cells

AT EMBRYONIC

STAGE B

L-cells/Rfibers (Rl)

L-cells/ bundle

WW

22 20 17 18 18

22 17 15 14 14

110 99 84 90 88

0.625 0.656 0.613 0.667 0.662

5.0 4.95 4.94 5.00 4.89

D Bundles with eight R-fibers. ’ Two controls analyzed. ’ Not determined.

L-cells degenerate over a period of some hours, it is impossible to determine, in an animal fixed at a particular stage, how many cells have degenerated or will degenerate. There is some evidence, albeit somewhat indirect, that fewer L-cells are available for recruitment in laminar anlagen of lesioned animals than in controls. Animals fixed before extra L-cells begin to degenerate but after the arrival of all surviving Rfibers (38- to 40-hr stage) have smaller laminar cross sections on the irradiated than on the control side. One such animal is illustrated in Fig. 13. Although unrecruited L-cells are available, the lamina is smaller than normal and the number of cells is less than expected. Two explanations for this reduced number of cells can be proposed. It is possible that fewer cell divisions have taken place on the affected side, suggesting that interaction with R-fibers determines the rate or number of mitotic divisions of neuroblasts or ganglion mother cells. Alternatively, L-cells born when other L-cells are already waiting at the midline for Rfibers which do not arrive (or arrive late)

may migrate into other regions of the brain, thereby reducing the number of extra cells to be found. The present data do not allow an unambiguous choice between these possibilities. Degeneration of L-cells in animals irradiated at stage B (42-44 hr) has not been examined in detail in these experiments. I have, however, looked in some specimens at the location of degenerating L-cells following stage B lesions. One would expect to find, in this case, that L-cells in the cartridges corresponding to the deleted ommatidia would be those that degenerate. Degenerating L-cells should also be found away from the midplane, provided the lesion is away from the midplane. That this is in fact the case is illustrated by the micrographs from a serially sectioned embryo shown in Fig. 14. DISCUSSION

Regulation Ganglion

of Celt Number

in the Optic

The most obvious effect of experimentally reducing the size of the eye is the

FIG. 8. Electron micrographs illustrating the same laminar cartridge in four adults. (a) Control animal. The retinular fibers have been numbered 1 through 8. Four of the five laminar cells (L) appear in this section; the fifth is located further into the ganglion. (b) Experimental animal in which two bundles have fused to form a single one with 14 fibers, numbered 1 through 14 on this micrograph. The five laminar cells (L) associated with this cartridge are seen in this section. (c) In this experimental animal this cartridge has five fibers (numbered 1 through 5) and four laminar cells (L). (d) In this experimental animal the cartridge has only two fibers (I and 2) and two laminar cells (Ll and L2). Other laminar neurons from adjacent cartridges are also labeled (L). Serial section reconstructions show that only Ll and L2 receive synaptic connections from fibers 1 and 2. Scale bars represent 5 pm. The bar in (d) applies also to (b) and (c).

FIG. 9 226

E. R. MACACNO TABLE

Cellular

5

L-CELLS,AND M-CELLS IN EXPERIMENTALANIMALSIRRADIATEDAT EMBRYONICSTAGEA

NUMBERS

Specimen”

OF R-FIBERS,

Side

R-fibers

L-cells

M-cells’

Left Right

88 88

55 55

160 166

d

Left Right

88 39

55 24

169 165

1

Left Right

51 19

31 14

163 107

m

Left Right

24 15

12 10

120 133

n

Left Right

0 n.d.

0 nd.

-30 n.d.

Control

n Same as in Table 2. ‘There is an uncertainty of a few cells in the counting of medullary cells (M-cells) (see text).

resulting reduction in the size (hypoplasia) of the optic ganglion. This phenomenon takes place not only in Daphnia, but also in the visual centers of Drosophila (Power, 1943; Hinke, 1961), dragonflies (Mouze, 1974, 1978), and locusts (Anderson, 1978), as well as other insects. Partial or total removal of the embryonic eye in vertebrates can also have profound effects upon the size of brain areas that receive visual input (Jacobson, 1978). In Daphnia the present data indicate unambiguously that hypoplasia of the optic ganglion is a consequence not only of the loss of the volume occupied by the incoming retinular terminals but also of the loss of ganglionic neurons. To some extent the vol-

Interactions

in Development

227

ume loss is also due to atrophy of the surviving ganglionic neurons (Macagno, unpublished observations). The number of neurons in the Daphnia lamina is roughly proportional to the number of photoreceptors. The coefficient of proportionality has an approximate value of 5/s,reflecting the fact that the deletion of an entire ommatidium (hence eight photoreceptors) always results in the loss of five laminar neurons. At a finer level of analysis, one finds that the partial deletion of an ommatidium does not lead to a proportional loss of L-cells in the corresponding cartridge. Rather, there is a threshold at about five fibers, since five or more fibers can recruit and maintain five (and sometimes six) L-cells. Bundles with fewer fibers generally have fewer than five L-cells in their cartridges. An even clearer example of a threshold effect is seen in the medulla. Loss of medullary neurons does not begin until a major fraction of the retina (about 80%) has been deleted. Even a total deletion of the retina, which causes a complete loss of the lamina, does not totally eliminate the medulla. One possible reason for the different responses of lamina and medulla could be the range of inputs received by the different types of neurons. L-cells receive synaptic inputs almost exclusively from one ommatidium. They certainly interact with only one ommatidium early in development. Medullary neurons, however, receive inputs not only from many different ommatidia but also from other medullary neurons as well as from centrifugal fibers. The threshold for

FIG. 9. Light micrographs from a serially sectioned, laser-irradiated adult, a week after the lesion was made. The lesion almost completely destroyed the right half of the eye, leaving the left side almost intact. Arrowheads denote the midplane. (a) Section through the optic nerve. Most fibers on the right have degenerated, but some are still in the process, as indicated by the darkly staining debris. (b) Section at the beginning of the lamina. Few signs of degeneration are found, though most fiber bundles on the right side have disappeared (cf. left side). (c) Section through laminar neuropil. Many degenerating processes can be seen on the right, mostly retinular terminals. Though it is impossible to say that no degenerating profiles belong to laminar cell processes, no laminar cell somata are found displaying signs of degeneration. Moreover, cell counts show that all laminar neurons are still there. (d) Section through the medulla. Few signs of degenerating processes are seen at this level. All cell somata appear normal. Bar in (d) = 50 pm.

FIG. 10 228

E. R. MACAGNO

Cellular

survival may be determined, therefore, by the requirement for a minimum number of contacts with other neurons during development. This is an interesting possibility which merits a more quantitative analysis than can be carried out on the basis of the present observations on Daphnia. In Drosophila, genetic mutants carrying reduced retinal fields also have smaller than normal visual centers (Power, 1943; Hinke, 1961). Power showed that the volume of the optic glomeruli was linearly related to the number of ommatidia in his mutants, a result later confirmed by Hinke. Although the optic lamina was found to be completely absent in eyeless mutants, higher-order regions were less affected and were reduced only by 57 to 85%. These results agree with those on Daphnia. The only discrepancy would appear to be the lack of a threshold effect in the Drosophila medulla, but it must be kept in mind that Power was measuring total volume changes, and not cell numbers. The linear decrease in medullar volume which he measured could be due largely to the loss of incoming retinular and laminar fibers, and not primarily to the atrophy or degeneration of medullary neurons. The decrease in the number of neurons in the visual centers could occur in three different ways. First, the normal number of cells might be present, but those that do not achieve a sufficiently strong interaction with other neurons would degenerate. Evidence of increased degeneration has been found not only in experimental Daphnia, but also in

Interactions

in Development

229

the dragonfly Aeschna cyanea (Mouze, 1978) and the locust Schistocerca gregaria (Anderson, 1978). However, in none of these cases can it be said from the data that all cells which would normally arise have actually done so. Second, the cell birth rate or the absolute number of cell births could be modulated by the demand provided by incoming Rfibers. The evidence for this feedback mechanism in Daphnia is not particularly strong, consisting only of gross measurements of a decrease in the volume of the optic ganglion anlage with respect to controls after R-cell deletion, but before L-cell degeneration begins. Accurate cell counts as a function of time have not been carried out to resolve this point. Anderson (1978) reports that in S. gregaria the production of new laminar cells proceeds autonomously. She observed no significant change in the mitotic rate of the laminar anlage between controls and animals deprived of retinal innervation. Mouze (1978), however, proposed that in Aeschna the R-fibers may control the total number of precursors, hence modulating the number of cells that give rise to ganglion cells and not their mitotic activity. Third, cells not recruited could migrate to other regions of the brain, a possibility which is difficult to test experimentally. However, the shape of these cells and their lack of any morphological characteristics of migrating cells, such as pseudopodia, make this alternative unlikely. In general, I would expect the first possibility to be the correct explanation of the

FIG. 10. Light micrographs from a serially sectioned adult uv irradiated at embryonic stage A. The lesion completely deleted the left side as well as part of the right side of the eye. (a) Section through the remaining part of the right side of the eye (RE), which has shifted toward the middle of the head. A lens (L), as well as a number of retinular cells with pigment granules, can be seen at this level. (b) Section through the right lamina (RL) and right medulla (RM). No evidence of a left lamina or left medulla can be found. (c) Section 20 am more posterior than that in (b). The right medulla is still visible, but now the left medulla (LM) has appeared, no connection of the LM to the eye or right side of the optic ganglion can be found throughout the serial sections. (d) Section about 20 pm more posterior than that shown in (c). Both the left and right medulla have joined the supraesophageal ganglion (SG). IN, intestine; LIC, left intestinal caecum; RIC, right intestinal caecum. Bar in (d) corresponds to 50 pm, and applies to all four micrographs.

230

DEVELOPMENTAL

BIOLOGY

FIG.

VOLUME

11

73, 1979

E. R. MACACNO

Cellular

observations, due to the prevalence of cell death as a general developmental mechanism for the regulation of cell number. However, only additional accurate, quantitative observations will tell us whether the other explanations are also applicable.

The Period of Trophic Interaction R-Fibers and L-Cells

between

The results of the experiments described above show that there is a period during Daphnia development when the laminar neurons require contact with retinular fibers in order to survive. The deletion of a fraction of the retinular cells, and hence of their fibers, between 28 and 44 hr brings about the degeneration of a proportional fraction of the L-cells. Similar experiments in the adult indicate that the L-cells lose their lability and can survive at least for a few weeks after deletion of their R-fiber inputs. How does this apparently critical period relate to differentiative events in the Daphnia visual system? As reported previously (Lo Presti et al., 1973), the initial contacts between growing R-fiber bundles and L-cells occur sequentially between 29 and 37.5 hr, with the most lateral bundles arriving first at the laminar anlage, those most medial arriving last. Following this initial interaction, which includes the transient appearance of gap junctions between fibers and L-cells (Lo Presti et al., 1974), the L-cells begin to extend processes which grow along with the R-fibers into the region which will become the laminar neuropil, forming the optic car-

Interactions

in Development

231

tridges. Prior to this initial interaction, Lcells are devoid of processes and are morphologically undifferentiated. By 37.5 hr the more lateral fibers and L-cell processes have branched and have what look like nascent synapses between them. This takes place in more medial cartridges somewhat later, so that by 44 hr the most medial groups are at this stage of differentiation. Also at about 44 hr, the oldest cartridges have begun to contact medullar processes, and synaptic contacts with other neurons begin to appear. Synapse formation with medullar cells has not yet been studied in detail. It would seem, therefore, that the period of lability of the L-cells extends at least until synapse formation between R-fibers and L-cells has begun and probably beyond. This is based exclusively upon morphological evidence of synapses between these elements. No data pertaining to the time at which synapses become functional are presently available for this system.

Triggering of L-Cell Differentiation and Their Recruitment by Growing R-Fibers It was suggested in earlier reports (Lo Presti et al., 1974; Levinthal et al., 1975), on the basis of observations of normal development, that R-fibers might trigger the differentiation of L-cells. This could be accomplished with a signal transmitted via transient gap junctions found between the lead-fiber growth cone and L-cells during the first contact between them. The present experiments demonstrate that this contact is necessary and that a triggering signal for

FIG. 11. (a) Electron micrograph of a section through the anterior surface of the lamina in a 42-hr embryo which was irradiated at stage A. LE (RE), left (right) half of eye; LL (RL), left (right) lamina; arrowheads indicate the midplane. Circles, in LL indicate two retinular fiber bundles which are not found in the RL. The laminar cells labeled L have not been recruited by any retinular fibers, but have yet to degenerate. (b) Electron micrograph of a section through a 45-hr embryo irradiated at stage A. Nomenclature same as in (a). Arrowheads indicate the midplane of the lamina. Here the arrows point out debris of degenerating laminar cells. (c) Light micrograph of a section at the same level as in (a), but in a 49-hr embryo. Nomenclature same as in (a), with the addition of SG, supraesophageal ganglion. Arrowheads indicate the midplane of the lamina. Arrows here point out pycnotic cells on the LL. Bars correspond to 10 am in (a), 5 in (b), and 25 in (c).

232

DEVELOPMENTAL

BIOLOGY

L-cell differentiation must exist. Following deletion of R-fibers, those L-cells which are not touched by growing R-fibers fail to elaborate processes for a period of some hours after they would usually do so, although otherwise they appear normal until

VOLUME

73, 1979

they degenerate. No further data on the nature of the triggering signal have yet been obtained. It is clear that this triggering mechanism can serve both to recruit L-cells for a partitular cartridge and to coordinate the mat-

FIG. 12. (a) Low-power electron micrograph of a 49-hr embryo which had been irradiated at stage A. In this coronal section, the retinula (ret), the lamina (lam), and the bilateral medulla (lm, left medulla; rm, right medulla) can be readily identified. A small lesion was made at stage A on the left side of the developing eye, at the site marked by an arrowhead. A degenerating cell can be seen clearly in the left medulla (marked by an open arrow). le, lens; rh (with arrow), rhabdom; lip, lipid droplet. The bar at lower right is 10 pm long. (b) Higher magnification micrograph of the same section as in (a), showing the lamina (LLAM, left half of lamina; RLAM, right half of lamina). The midplane is indicated by the two arrows. The extensive neuropil on the right side of the lamina is labeled and delineated by the box. The neuropil on the left side is almost completely absent, with only a small area showing at lower left. Arrowheads indicate debris from degenerating laminar cells. The bar at lower right is 10 am long.

_-___--

--.FIG. 12b 233

234

DEVELOPMENTAL BIOLOGY

VOLUME 73, 1979

FIG. 13. Micrograph through the lamina of a 40-h embryo with a lesion entirely The cross section of the left lamina (LL) is noticeably smaller than that of the right to be a result of fewer available laminar cells rather than of cell death, which would later. Arrows indicate the midplane. Bar = 10 pm.

uration of R-fiber terminals and L-cell processes.Hence, the appropriate projection of R-fiber bundles onto laminar cartridges can be achieved without a requirement for the R-fibers to search for L-cells with a specific label (Macagno, 1978). However, a very interesting question remains to be answered: Why does a bundle of R-fibers normally

FIG. 14. Electron micrographs Debris of degenerating laminar which are sections close to the second cartridge, away from the (c) and (d). LL (RL), left (right) has been circled in (b-d). Same

on the left side of the eye. lamina (RL). This appears not occur until some hours

recruit only five L-cells? In some animals with large retinular deletions, a number of bundles were identified which form cartridges with six L-cells. Presumably, many immature L-cells were left unrecruited in these animals because the number of Rfibers was so extensively reduced. Does availability of target cells, therefore, mod-

from a serially sectioned 49-hr embryo irradiated on the left side at stage B. cells from a group near the midplane is indicated by the arrows in (a) and (b), anterior surface of the lamina. Debris of degenerating laminar cells from a midplane, is pointed out by arrows in the more posterior sections illustrated in lamina. The midplane is indicated by arrowheads. The same group of R-fibers magnification on all four micrographs. Bar in (d) = 5 Nrn.

E. R. MACAGNO

Cellular

FIG.

Interactions

14

in Development

235

236

DEVELOPMENTAL BIOLOGY

ulate the number recruited? A more careful assessment of the availability of immature L-cells in normal and experimental animals is necessary before this question can be answered affirmatively. An alternative explanation, since a number of the cartridges with six L-cells have only seven R-fibers, is that the irradiation and partial deletion of cells from ommatidia affect their ability to count to five. However, the fact that some bundles of eight fibers in regions not directly affected by the irradiation also form cartridges with six L-cells argues against this alternative. Very little direct evidence from other invertebrates has been published concerning the differentiation of laminar cells deprived of retinular inputs. Anderson (1978)) for example, reports that laminar cells in the locust S. gregaria do not seem to differentiate but instead degenerate in the absence of innervation from a growing retina. However, it is very difficult, if not impossible, to detect at the light microscope level whether the fairly small laminar cells elaborate fibers before degenerating at some later time. In some insects the laminar neuropil normally develops during or soon after the arrival of retinal innervation (Nordlander and Edwards, 1969; TrujilloCenoz and Melamed, 1973; Meinertzhagen, 1973), though careful tests of the causal relationship between these events are lacking. Other evidence comes from gross measurements of laminar neuropil volume. Mouze (1974), for example, has measured a reduction of around 10% in neuropil volume when retinal innervation is withheld from the larval optic lobes of the dragonfly, Aeschna. Since such reductions can be accounted for by the loss of the volume that would have been occupied by ingrowing retinular fibers, one should only consider such observations as suggestive of the hypothesis that deprived laminar cells fail to begin their morphological differentiation. There are also examples where the behavior of target neurons deprived of their

VOLUME 73, 1979

inputs during development is different from that of the neurons in the arthropod optic lamina. One such example is the deafferented antennal lobe of the moth Manduca sexta. Second-order neurons in these lobes seem to differentiate (at least in part) and to survive without antenna1 fibers ever reaching them (Sanes et al., 1977; Hildebrand et al., 1979). It is possible, however, that primary sensory fibers from other appendages also terminate in this region and affect its maturation (Strausfeld, 1976)) though current efforts have failed to find such fibers in Manduca (J. Hildebrand, private communication). A second example is the embryonic avian optic tectum. When an eye is removed in a 2- to 3-day-old chick embryo, the tectal cells which would receive input from optic fibers proliferate and differentiate autonomously, producing a normal-looking tectum (Kelly and Cowan, 1972). By the 14th embryonic day, however, a progressive atrophy of the tectum is established, with massive degeneration of the deafferented target cells. Furthermore, variations are also encountered when examining the behavior of neurons deprived of targets rather than inputs. Some neurons differentiate and survive without their targets. This is the case, for example, with most invertebrate photoreceptors (Mouze, 1974; Anderson, 1978; see Anderson for other references) and antennal receptors (Sanes et al., 1976). Other neurons, when deprived of their targets, will differentiate but later degenerate. Particularly well-studied casesof this category are neurons of the chick ciliary ganglion, which develop fully functional synaptic inputs before their death (Landmesser and Pilar, 1974), the avian isthmo-optic nucleus (Clarke et al., 1976), the chick spinal cord (Hamburger, 1975), and the optic ganglion of the butterfly (Nordlander and Edwards, 1968). It is apparent from these examples that the roles of cellular interactions in the survival and differentiation of neurons are

E. R. MACAGNO

Cellular

complex and varied. Understanding these roles will require considerably more refined experimental observations and the extensive comparison of developmental events in different species. The Role of the Eye in the Formation of the Visual System The experimental results reported here demonstrate unambiguously that growing photoreceptor fibers play a critical role in the formation of the optic ganglion in Daphnia. In an ordered spatiotemporal sequence, retinular fibers recruit a specific number of immature laminar cells per growing ommatidial bundle and trigger their differentiation. The retinular fibers thereby impose their own modular organization upon the lamina and, through their action on the lamina, affect the higher-order regions of the ganglion. In addition to pattern regulation, the retinular fibers may directly contribute to regulation of the size of the ganglion by affecting the total number of immature laminar cells available for recruitment by R-fibers. It is probable, however, that size regulation is largely achieved through cell death: Extra laminar cells which are found in animals with retinal deletions degenerate when they are not recruited by retinular fibers. In a previous report (Macagno, 1977), evidence of abnormal branching patterns and synaptic connections in animals with retinular cell deletions was presented, indicating that the specific pattern of differentiation of individual laminar cells (L-cells) can be affected by adjacent retinular fiber (R-fiber) groups when variations in retinal pattern have been experimentally induced. This, taken with the fact that a trophic relation between R-fibers and L-cells exists over a period of at least 10 to 12 hr during embryogenesis, suggests that R-fibers may not just trigger a predetermined differentiative pathway in recruited L-cells, but may, in fact, modulate it. In other words, the interaction may not only be trophic but also instructive.

Interactions

in Development

237

Observations of a less detailed nature on a number of insects are in basic agreement with those on Daphnia. The parallels among results with different species suggest that the mechanisms used by Daphnia to assemble its visual system have a general validity, at least among invertebrates. I thank Bonnie Peng, Elizabeth Starch, Robert Campbell, and Marvin Siegfried for technical assistance and Linda Sproviero for typing the manuscript. I also thank G. Fischbach, E. Holtzman, and R. Schehr for critical readings of the manuscript. The laser irradiation of Daphnia adults was carried out at Cornell University and the NIH with the kind help of J. Schlessinger. This work was supported by Grants NS11738 and NS14946 from the Public Health Service. The computer work was carried out at the Columbia Computer Graphics Facility which is supported by Grant RR0042 from the Public Health Service. REFERENCES ANDERSON, H. (1978). Postembryonic development of the visual system of the locust, Schistocerca gregaria. I. Patterns of growth and developmental interactions in the retina and optic lobe. J. Embryol. Exp. Morphol. 45,55-83. CLARKE, P. G. H., ROGERS, L. A., and COWAN, W. M. (1976). The time of origin and the pattern of survival of neurons in the isthmo-optic nucleus of the chick. J. Comp. Neurol. 167,125-142. COOK, J. S. (1970). Photoreactivation in animal cells. In “Photophysiology” (A. C. Giese, ed.), Vol. 5, pp. 191-233. Academic Press, New York. HAMBURGER, V. (1975). Cell death in the lateral motor column of the chick embryo. J. Comp. Neurol. 160, 535-546. HILDEBRAND, J. G., HALL, L. M., and OSMOND, B. C. (1979). Distribution of binding sites for 125-I-labeled alpha-bungarotoxin in normal and deafferented antennal lobes of Manduca se&a. Proc. Nat. Acad. Sci. USA 76,499-503. HINKE, W. (1961). Das relative postembryonale Wachstum der Hirnteile von Culex pipiens, Drosophila melanogaster und Drosophila-Mutanten. Z. Morphol. Oekol. Tiere 50,81-118. JACOBSON, M. (1978). “Developmental Neurobiology,” 2nd ed. Plenum, New York. JAGGER, J. (1967). “Introduction to Research in Ultraviolet Photobiology.” Prentice-Hall, Englewood Cliffs, N.J. KELLY, J. P., and COWAN, W. M. (1972). Studies on the development of the chick optic tectum. II. Effects of early eye removal. Brain Res. 42, 263-288. LANDMESSER, L., and PILAR, G. (1974). Synapse formation during embryogenesis of ganglion cells lack-

238

DEVELOPMENTAL

BIOI .OGY

ing a periphery. J. Physiol. (London) 241.715-736. LEVINTHAL, F., MACAGNO, E. R., and LEVINTHAL, C. (1975). Anatomy and development of identified cells in isogenic organisms. Cold Spring Harbor Symp. Qua&. Biol. 40, 321-331. LEVINTHAL, C., MACAGNO, E. R., and TOUNTAS, C. (1974). Computer aided reconstruction from serial sections. Fed. Proc. 33,2336-2340. Lo PRESTI, V., MACAGNO, E. R., and LEVINTHAL, C. (1973). Structure and development of neuronal connections in isogenic organisms: Cellular interactions in the development of the optic lamina of Daphnia. Proc. Nat. Acad. Sci. USA 70,433-437. Lo PRESTI, V., MACAGNO, E. R., and LEVINTHAL, C. (1974). Structure and development of neuronal connections in isogenic organisms: Transient gap junctions between growing optic axons and lamina neuroblasts. Proc. Nat. Acad. Sci. USA 71, 1098-1102. MACAGNO, E. R. (1975). Structure and synaptic connectivity of the photoreceptors in Daphnia raised in complete darkness. Neurosci. Abs. 1,98. MACAGNO, E. R. (1977). Abnormal synaptic connectivity following UV-induced cell death during Daphnia development. In “Cell and Tissue Interactions” (J. W. Lash and M. M. Burger, eds.), pp. 293-309. Raven Press, New York. MACAGNO, E. R. (1978). A mechanism for the formation of synaptic connections in the arthropod visual system. Nature (London) 275,318-320. MACACNO, E. R., and LEVINTHAL, C. (1975). Computer reconstruction of the cellular architecture of the Daphnia magna optic ganglion. Annu. Proc. EMSA 33, 284-285. MACAGNO, E. R., LEVINTHAL, C., TOUNTAS, C., BORNHOLDT, R., and ABBA, R. (1976). Recording and analysis of 3-D information from serial sections. In “Computer Technology in Neuroscience” (P. B. Brown, ed.), pp. 97-112. Wiley, New York. MACAGNO, E. R., Lo PRESTI, V., and LEVINTHAL, C. (1973). Structure and development of neuronal connections in isogenic organisms: Variations and similarities in the optic system of Daphnia magna. Proc. Nat. Acad. Sci. USA 70,57-61. MEINERTZHAGEN, I. A. (1973). Development of the compound eye and optic lobe of insects. In “Developr..mtal Neurobiology of Arthropods” (D. Young, ed.), pp. 51-104. Cambridge Univ. Press, London. MEINERTZHAGEN, I. (1976). The organization of perpendicular fiber pathways in the insect optic lobe. Phil. Trans. Roy. Sot. London Ser. B 274,555-596. MOUZE, M. (1974). Interactions de l’oeil et du lobe optique au tours de la croissance post-embryonnaire

VOLUME

73, I979

des insectes odonates. J. Embryol. Exp. Morphol. 31,377-407. MOUZE, M. (1978). Role des fibres post-retiniennes dans la croissance du lobe optique de la larve d’Aeschna Cyanea Mull (Insecte Odon..le). WiZhelm Roux Arch. Entwicklungsmech. Organismen 184,325-350. MURPHY, J. S. (1970). A general method for the monoxenic cultivation of the Daphnidae. Biol. Bull. 139, 321-332. NASSEL, D. R. (1976). The retina and retinal projection on the lamina ganglionaris of the crayfish Pasifustacus leniusculus (Dana). J. Comp. Neurol. 167, 341-360. NORDLANDER, R. H., AND EDWARDS, J. S. (1968). Morphological cell death in the postembryonic development of the insect optic lobes. Nature (London) 278, 780-781. NORDLANDER, R. H., and EDWARDS, J. S. (1969). Postembryonic brain development in the monarch butterfly Danaus plexippus plexippus L. II. The optic lobes. Wilhelm Roux Arch. Entwicklungsmech. Organismen 163, 197-220. POWER, M. E. (1943). The effect of reduction in numbers of ommatidia upon the brain of Drosophila melanogaster. J. Exp. Zool. 94,33-71. SANES, J. R., HILDEBRAND, J. G., and PRESCOTT, D. J. (1976). Differentiation of insect sensory neurons in the absence of their normal synaptic targets. Develop. Biol. 52, 121-127. SANES, J. R., PRESCOTT, D. J., AND HILDEBRAND, J. G. (1977). Cholinergic neurochemical development of normal and deafferented antenna1 lobes during metamorphosis of the moth, Manduca se&a. Brain Res. 119,389-402. STARLING, D. (1974). A simple ultraviolet microbeam apparatus. J. Microsc. (London) 100,331-336. STOWE, S. (1977). The retina-lamina projection in the crab Leptograssus variegatus. Cell Tissue Res. 185, 515-525. STRAUSFELD, N. J. (1976). “Atlas of an Insect Brain.” Springer-Verlag, Berlin/New York. TRUJILLO-CENOZ, O., and MELAMED, J. (1973). The development of the retina-lamina complex in muscoid flies. J. Ultrastruct. Res. 42, 554-581. URETZ, R. B., and PERRY, R. P. (1957). Improved UV beam apparatus. Rev. Sri. Znstrum. 28,861. WOLFF, J. R., and GULDNER, F. H. (1970). Uber die Ultrastruktur des “Nervus opticus” und des Ganglion opticum I von Daphnia pulex. Z. Zellforsch. Mikrosk. Anat. 103, 526-543.