- Email: [email protected]
Histiotypic organization and cell differentiation in rat retinal reaggregate cultures
Brain Researrh, 437 (1987) 298-308
298
Elsevier
BRE 13161
Histiotypic organization and cell differentiation in rat retinal reaggregate cultures Kimio Akagawa, David Hicks and Colin J. Barnstable Laboratory of Neurobi~Iog ;~'~The Rockefeller University, New York, N Y 10021 (U.S.A.)
(Accepted 9 June 1987) Key words: Photoreceptor; MOiler cell; Opsin; Monoclonal antibody; Immunocytochemistry; Electron microscopy; Cell interaction
Reaggregate cultures have been formed from cell suspensions of neonatal rat retinas. Histological sections of the ~eaggregates showed evidence of lamination with central rosettes formed around a lumen, a clear neuropil layer and an outer cellular layer. Each of the major retinal cell types, except ganglion ceils, could be positively identified using cell type-specific antibodies to label cryostat sections. Many of these were found to occupy positions within the reaggregates similar to those found in the intact retina. Electron microscopic observations showed abundant immature and mature synaptic endings within the neuropii layer, including a number of ribbon synapses. Examination of the rosettes showed an arrangement of Miiller glia and photoreceptors that closely resembled that of the intact retina. Within the lumen of rosettes, photoreceptors were found to contain stacks of disc-like membranes bounded by a plasma membrane, analogous to immature outer segments. The photoreceptors within rosettes also underwent molecular differentiation and expresscd an outer segment specific marker. The findings suggest that retinal cells have intrinsic properties that allow them to organize themselves into a correctly laminated structure and that particular cell interactions are necessary for continued differentiation of at least rod photoreceptors and MOiler cells. INTRODUCTION The extent to which genetic and epigenetic factors influence the generation and differentiation of the diverse sets of neuronal cell types of the vertebrate nervous system is poorly understood. It has been shown that some cells can express molecules characteristic of their mature phenotype before they have completed migration,-suggesting that determination of cell types begins in or near the ventricular zone of the developing tissue 9.21. The mechanistic relationship between the initial determinative events and the subsequent steps of development that can take place over several weeks remains unclear. The retina is a valuable model for the study of neural development for a number of reasons. It has only a very minor neural input through a small number of efferent fibers ~2. This means that until the formation of ganglion cell connections with central targets, early retinal development is essentially independent of
transsynaptic neural influences ~9. The stimulatory input of the retina comes from the detection of light by photoreceptors. Since the light transduction apparatus of the photoreceptors is one of the last steps of retinal differentiation, it is unlikely that the patterned sensory input has any role in the early formation of cell types or cell connections 16"25"37.The retina also has the advantages of being accessible to experimental manipulations, and of consisting of a relatively small number of neuronal cell types which are arranged in a laminated structure. A useful approach to study factors important in retinal development has been to isolate the tissue and study its developmental capacity in vitro, Monolayer cultures of retina have been used to study such properties as synz, m e formation by cholinergic amacrine cells 27, regeneration of processes by axotomized ganglion cells 3'1°'2° and the behavior of normal and activated glial cells 29. Reaggregate cultures have been used to provide a 3-dimensional in vitro environment
Correspondence: C,J. Barnstable, Laboratory of Neurobiology, The Rockefeller University, 1230 York Avenue', New York, NY
10021, U.S.A.
0006-8993/87/$03.50 O 1987 Elsevier Science Publishers B.V. (Biomedical Division)
299 for the cells z4. Aggregates of retina and of other r ,ain regions have been shown to undergo tissue sorting which led to the formation of layers that were thought to be analogous to those found in vivo 14"3°'32. Comparison of reaggregate and monolayer cultures has led to the suggestion that specific cell interactions are necessary for the expression of the glial enzyme carbonic anhydrase 2-'. although the mechanism of this effect is not yet known. One of the problems of using cultures of any heterogeneous tissue is that once removed from their normal environment, the identification of cell types becomes difficult. We have overcome this problem for rat retina by producing a series of monoclonal antibodies that selectively recognize molecules on each of the major neuronal and glial cell classes4-6. Using these reagents we have shown that each of the retinal cell types can be identified in monolayer cultures ~'2°. Although some of the cell types continued to differentiate in these cultures, the extent of differentiation was abnormal and several steps did not occur. Of particular relevance for the present study was th~ observation that photoreceptors remained alive for many weeks and continued to express the visual pigment protein opsin, but no rod outer segments were detected immunocytochemically. In this paper we have examined reaggregates of neonatal rat retinal cells. Using cell type-specific monoclonal antibodies, we have shown that the lamination in ~h~ aggregates does resemble that seen in vivo. In addition, both Miiller cells and photoreceptors acquire structural features not seen in either cell type maintained in monolayer culture. MATERIALS AND METHODS
Preparation of reaggregate cultures. Retinas were isolated from l-2-day-old Long-Evans rat pups without significant pigment epithelium contamination as described previously'. A single cell suspension was produced by mechanical trituration with a fire-polished pipet. The extent of dissociation was monitored microscopically and trituration was continued until no cell cluro-s of > I0 cells were found. Cells were suspended at a density of 5 × 104/ml in minimal essential medium (MEM) supplemented with 0.6% glucose, 5% rat serum, 3 mM taurine, 2 mM glutamine, 1 pg/mi DNase and 10 l~giml gentamycin. The
cell suspension was rocked at 30 cycles/min in an atmosphere of 7% CO,,, balanced air, with saturated humidity at 37 °C. The cultures were given fresh medium every 2 days and remained viable for at least two weeks. The speed of rocking determined reaggregate size. Under the conditions used, the reaggregates were spherical and of 300-800 /~m in diameter. Reaggregates of 500-700 lira were analyzed in the experiments reported here. Antibodies. All the antibodies used in this study have been described previously. Antibody HPC-1 reacts with retinal amacrine cells'~; 2G12, an antiThy-1, is specific for ganglion cells in retina~: RETPI is specific for rod photoreceptors 4 and reacts with a determinant on the N-terminal sequence of opsin 13"15. Antibody RHO-C7 was raised against bovine opsin but, unlike RET-P1, the determinant recognized is only found in outer segments and only appears when outer segment formation begins 6"~3 RET-G7 antibody recognizes a component of the cytoskeleton of gila but not neurons'. The antiserun: against the 140 kDa subunit of neurofilaments was a gift from Dr. C. Marotta, Harvard Medical School. bnmunocytochemistry. Reaggregates were rinsed 3 times with serum-free MEM and fixed with 4% paraformaldehyde in phosphate-buffered isotonic saline (PBS). During fixation the tissue was rocked to prevent any deformation of structure. After rinsing to remove fixative, groups of reaggregates were immersed in Tissue Tek II (Miles labs) and frozen, 15~m, cryostat sections were cut and mounted on gelatinized slides. Sections were preincubated with 5% normal goat serum in PBS to eliminate non-specific binding. Both primary and secondary incubations were carried out for 1-2 h at room temperature. Sections were mounted in PBS-glycerol (1:1) and viewed using a Zeiss microscope equipped with epifluorescence and appropriate combinations of filters. Electron microscopy. Reaggregate cultures were rinsed and fixed in 1% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.2 for 1 h at 4 °C. After washing in the same buffer for 1 h, tissue was postfixed in 1% OsO4 in 0.1 M cacodylate buffer. The reaggregates were again washed in the same buffer, dehydrated in ethanol and infiltrated with Epon/Araidite resin. Following polymerization, blocks were trimmed and sectioned. Initially, 1-!~m sections were stained with 0.1% Toluidine blue in I~- sodium borate and
300 inspected for the presence of rosettes. Once suitable areas were located, thin sectione (60-70 nm) were collected and examined on a Jeol 100S electron microscope. RESULTS
Formation ofreaggregates. The 1-2-day-old rat retina has an immature morphology and many of the cells do not yet have an extensive network of processes. With the mechanical dissociation used to prepare the cell suspension, those processes that were present were presumably broken off slate the cell suspension appeared as small (5-7 btm) spherical cells. Under the culture conditions ured, aggregation
could be detected microscopically within 30 min and by eye within 6 h. The rate of reaggregation was clearly dependent upon initial cell density and speed of agitation. At the cell density chosen a slow rocking speed of 30 cycles/min gave reaggregates whose mean diameter was approximately 500/~m. Higherspeed rotatory cultures (60-80 cycles/rain) gave almost entirely small cell clusters (data not shown). Ofto:'. !arger aggregates seemed to be formed by the fusion of two smaller aggregates since occasional bilobed structures were observed. Histology of reaggregates. Semithin sections of reaggregates clearly indicated cell sorting within the structures (Fig. 1A). The outermost layer of the reaggregate tended to consist of a morphologically ho-
A
.,>~
J
,I
•
,
•.
~
~.,r ~ q
.
Fig. 1. A: semithin section of a single reaggregate stained with Toluidine blue• Layers of differently stained cells and neuropil can be clearly seen. This reaggregate probably arose by fusion of two initial reaggregates since two central rosettes can be seen (curved arrows). B" higher-power view of a portion of (A), showing the outer layer of cells with processes directed in towards the neuropil (curved arrow) and a rare mitotic figure (straight arrow). C: higher-power view of the rosette in (A) showing luminal material inside the darkly staining cell nuclei• Bar in A = 1001(m.
301 tions of 7-day reaggregates with a panel of cell typespecific monoc!onal antibodies. Amacrine and ganglion cells. An antibody against a membrane antigen of all amacrine cells, HPC-1, labelled cell bodies at the outer layer of the reaggregates and processes in the outer portion of the neuropil layer (Fig. 2). Although the intensi',y of labelling ~!i ii ~~ii i ~ ~ii!i ¸i~ii~¸iii~i!i!i~ii!~~I!~i!iii!i!ii~i~!!!i!~i!~ii!l~!!i i~i!i!!i!iii!i!~!ii!~iii!i!i~i!i~i¸~!!iii!i!~!¸ii~!ii!~~ii~i!~~ii!!~i i!~!ii~i~i!i!~ii!i!~iii~i !~!i i~i~!i!i!i~ili increased over a period of two weeks in culture, no HPC-i-positive cells could be detected in i!i!~ii!~ii!i~i!i!i!i~ii!ii~ili!i!i~ii~i!i!i~ii!i!~~i!!~i i!i!i~ii~!!~i !~!iii!~~i!!ii!i!ii!~i!i~!ii!~!~i~i!~ii!~!~iii~i~i!i!~li!i!~i~i!~i~!i!i~'Ii~i!~i~iI!~i!Ii!iI~!iIi!~i~i~il!iI~!i~!i~i!i!~i~!~i!l!i~i!'~iIiIl!ii~!i~I~i~i!i~i~!iI!~i!~iIi!~!~i!~i!i~~i!ii~i!~iil!~i~!!~i!ii~Sli!i!~i~!~i!i!i!~i~i!~i!~i~!i!i~!i~!i!~i!~i!~i~i!~i!i!i~li!~i!~i!i!~ii!~ii!~i~i~!i~!i~!~i i~!~i~i~!additional ~i !i!i¸i!~i!i!¸i!iil!~i!~ il!~ii!~iii~ !¸i!~i!il!i~ l!i¸!i~f¸i!i!i!~i!i!i~iii¸~iii~iii!!i!i~¸!ii~ii!~!ii:i¸:iIIiIi¸I!~i!~i!i!i~¸ii~i!ii!¸~i!¸i!~ li!~il:i!i!i!i!i~¸iii:!)i!~i!i~II !!iili¸lili~iii!i¸!il!ii!ii!li!i~i!i!i!i!i!~ii!ii!i!i~~!:i~!ilS !i!i!ii!~ii!i~!¸i~ii¸li!~i!i!~ii!i~¸i~i!i~i!i¸!i~!i!~i!i~i!ili~i~i!i!i!il!i~ the inner portions of the reaggregates by immunofluorescence. No ganglion cells were detected in the reaggregate cultures used in these experiments, as defined by lack of labelling with antibodies against the ganglion cell marker Thy-1. Photoreceptors. Opsin is one of the earliest appearing rod photoreceptor-specific molecules. Antibody RET-P!, an anli-opsin, labelled many cells throughout the central portions of the reaggregates (Fig. 3A,B). The most concentrated RET-P1 labelling was observed in structures that could be seen by phase-contrast illumination to correspond to the rosettes seen in the Toluidine-stained material~ When labelling with outer segment-specific monoclonal anFig. 2. Cryostat section of reaggregate labelled with amacrine tibodies was performed, a different pattern of labelcell-specific antibody HPC-1. A: fluorescence. B: phase-conling was observed, as shown for antibody RHO-C7 trast. Only the outer layer of cell bodies and the processes im(Fig. 3C,D). Small bright patches of label were demediately underneath are labelled. Bar in B = 190!~m. tected in structures corresponding to the c,:r, tral portions of the rosettes. No RHO-C7 labelling was mogeneous layer of cell bodies that had large nuclei, found away from the rosettes even though adjacent clear nucleoli and a thin rim of cytoplasm. Processes sections of the same reaggregate showed that many from some of these cells could be observed passing RET-Pl-positive photoreceptors were present into an inner neuropil layer (curved arrow Fig. I B). throughout the central regions. The thickness of this layer varied between reaggreHorizontal and Midler cells. Since no ganglion cells gates from 10% of the reaggregate diameter to being were found in these reaggregates, we were able to barely discernable. Inside the oeuropil layer was a use an antiserum against the 140-kDa subunit of neumore heterogeneous layer of cell bodies. Within this rofilaments to label horizontal cells. Rare single cells layer, occasional mitotic figures were found (straight could be observed that did not seem to have a consisarrow Fig. 1B). The other cell bodies within this layer tent position in the reaggregates (Fig. 4A). In particwere of variable size and staining intensity, suggestular, double-labelling with anti-neurofilament serum ing that they may consist of a variety of retinal cell and antibody RHO-C7 showed that the horizontal types. The innermost regions of the reaggregates cells were not located on the outer regions of the phoconsisted of clusters of small, densely staining, fusitoreceptor rosettes (Fig. 4B). Miiller cells were form cells (Fig. 1C). Frequently these cells were labelled with antibody RET-G7. Labelling was obgrouped in rosette-like structures around a central luser,'ed throughout the structures but two features men. To examine whether these differences obwere more strongly labelled (Fig. 5). The first was a served at the light microscopic level corresponded to dense label in the neuropil layer ill the outer portion different retinal cell types, and to examine whether of the reaggregate and the second was a series of rathe arrangement of these cell types corresponded to dial fibers that passed through the rosettes and ended that seen in the intact retina, we labelled cryostat sec-
302
Fig. 3. Cryostat section of reaggregate labelled with photoreceptor-specific antibodies. Fluorescence (A) and phase-contrast (B) views of a section labelled with RET-P1. Labelled cells are found throughout the inner portions of the reaggregate but clusters of higher intensity of labelling are also found. Fluorescence (C) and phase-contrast (D) of a section labelled with RHO-C7. Only small areas of tissue show fluorescence. These correspond to the luminal portions of rosettes (arrow in D). Bar = 100l~m. in a more diffuse label within the rosette lumen.
Electron microscopic observation of reaggregates. Examination of the region surrounding rosette formations in 7-day reaggregate cultures revealed elongated cell bodies containing irregular nuclei with clumped heterochromatin (Fig. 6A). Extending from these nuclei towards the central lumen of the rosettes a number of structural features could be observed.
The cell bodies were joined by an array of desmosome-like junctions resembling the in vivo outer limiting membrane (arrow in Fig. 6A and B). Between the nuclei and this outer limiting membrane, many axially oriented mitochondria were visible. Cellular protrusions containing an array of organelles including mitochondria, vesicles and centrioles projected into the central lumen (Fig. 6B). These structures of-
303
Fig. 4. Cryostat section of reaggregate labelled with both a rabbit anti-neurofilament serum to label horizontal cells and monoclonal antibody RHO-C7 to label photoreceptor rosette centers. A: a single horizontal cell (down arrows) in a peripheral region of a reaggregate. B: a photoreceptor rosette is also present in the section (up arrows), but remote from the horizontal cell. Bar in A = 50!~m.
ten terminated in a longitudinally oriented cilium, and bore a strong resemblance to developing photoreceptor inner segments in vivo. Between these inner segments, fine microvilli were observed projecting from the surfaces of cells packcti between the photoreceptors (Fig. 6B). These correspond to the Mtiller cell microvilli observed in the intact retina. Within the central lumen of rosettes, membranous expansions from some cilia were seen. These irregular structures were bounded by a single plasma membrane, and contained a small number of membranous discs (Fig. 6C). Occasionally, membrane profiles exhibiting a distinct stacking of expanded disc rims were detected (Fig. 6D). In addition, concentrations of amorphous membranous material were observed in the lumen and packed between photoreceptor cell bodies (Fig. 6E). Examination of peripheral areas of the reaggregates revealed numerous fibers and enlarged endings containing an array of vesicles, filaments and smooth c~sternae (Fig. 7A-C). Features of mature ,:,ynapt~c endings, such a~ membrane thickenings, were less commonly observed (Fig. 7C). In addit~.on, densely
Fig. 5. Cryostat section of reaggregate labelled with Miiller cell-specific antibody RET-G7. Labelling is apparent throughout the section but is concentrated in the b~md ol neuropil under the surface (arrow) and in the rosettes ~asterisk). Bar = 11~1 !tm.
stained structures resembling synaptic ribbons were occasionally seen (Fig. 7D).
304
Fig. 6. Electron microscope view of thin section through rosette. A: the cell bodies contained irregular nuclei with clumped heterochromatin (N). Mitochondria (M) collected luminally to the nuclei and the cell bodies were joined by desmosome-like junctions (arrow), B: higher magnification of the inner segment region showing the ,,ame junctions (arrow). Cellular protrusions into the rosette lumen often contained mitochondria, vesicles and centrioles (C). These structures often ended in a longitudinally oriented cilium. Between the photoreceptors microvilli like those of M~iller cells were observed (open arrowhead). C.D: within the lumen many photoreceptors ended in membranous expansions which contained small stacks of discs (arrow). E" in some areas irregular accumulations of membranes could also be seen. Bar in A and B = 1#m; in C.D and E = 0.5!+m.
305
f--
f,~lL,~"
II
.~.
,,, •
.
+,-
+ ++ ++
+.a
•
• '
++
~
. .: ~+.+
].~.
~
~
".
"
,
.+
+ +,A+,
+-~
o
i ~
+'-
~
,a,
"'
*
. . . . .
"
+
. +,
+
:.
+ ;m~
]
+
" :
:-.,;
~
+~'
+.. +,
,.
+++-
++, . ..,:,~"~+
-: + . . . . .
+
-
_-+.
+.
,7
...
,'¢
-
,
+,
. ,I+.
+ .
~],,%'~
~'~,,~
~
+
+ ..
...
+ + , + , ~ ~+~"
+..+
+ +++I+
-:~+~+
-. *'
....
,:,,"
.
:" . ++
.'.- .:.~ . . . . . +.
+
+, "- '
"-
•
+"
"'+~
-'._+~ . t - , + " + , +
"?
+.
' "+
'L+:+%++: m +
,
- +
Fig. 7. Four examples of svnaptic structures foand in the peripheral areas of the reaggregates. A.B: immature synaptic endings exhibiting numerous vesicles (~,'). mitochondria (m) and smooth cisternae (sc). C: more mature synapse showing clustering of vesicles (v)
near a membrane thickening (mt). D: this synapse possesses a densely staining synaptic ribbon (srJ characteristic of photoreceptor and bipolar cell synapses. All bars = 0.5 !+m.
DISCUSSION
Histiotypic cell sorting has been observed in a variety of reaggregate cultures 3u'32"33. Al t h o u g h the exact s e q u e n c e of reaggregate f o r m a t i o n is not cleat, electr on microscopic e x a m i n a t i o n of reaggregating chick retinal cells at different times after dissociation has led to the suggestion that the sequence of aggrega-
tioo is into r a n d o m groups of cells followed by polarization or sorting out 3°. This is s u p p o r t e d bv studies of the kinetics of aggregate f o r ma tio n zs. Suggestive evidence of sorting out of some cell types in cell clusters in m o n o l a v e r cultures of rat retina has also been obtained 2. The polarity of retired reaggregates can vary according to experimental conditions. Chick reaggre-
306 gates from E7 retinas were found to take on a 'lumicentric" organization analogous to the rat reaggreo gates described here 3°. Reaggregates of El4 chick retinal cells, on the other hand, formed 'axocentric' structures that seemed to be of the opposite orientation. A role for factors other than age has been suggested since reaggregates of chick retina were more highly organized in the presence of pigment epithelial cells 35"36. In addition, newborn rat retinal cells grown in serum-free medium developed a less marked but observable tendency to laminate in a ph6torereptor-outermost orientationS7. Although we have not fully investigated all these variables, we have added rat pigment epithelial cells to our reaggregates without any detectable effect upon the cellular organization (data not shown). In our experiments, the outer layer of cells consisted of amacrine cells. These formed a single layer rather than distributing themselves on both sides of an inner plexiform layer as occurs in vivo. It is possible that this different distribution is due to a lack of ganglion cells which might play an active role in determining inner retina structure. It is possible that the loss of ganglion cells is due to absence of either presynaptic or postsynaptic interactions. Although we have observed synapses in these reaggregates (see Fig. 7), we have not yet been able to assign the cell types involved. An alternative explanation for the absence of ganglion cells is that, as the most mature cells in the neonatal retina, they suffer the greatest physical damage during dissociation. A-type horizontal cells, as detected by labelling with neurofilament antisera, were not obviously organized in a laminar fashion. These cells are among the first generated in the retina ~8'3'. When the A-type horizontal cells are born, they leave the mitotic zone and position themselves in the middle of what appears to be an otherwise undifferentiated neuroblast layer. As they migrate they maintain a radial projection to the ventricular surface of the retina and it has been suggested that this could provide a mechanism by which these cells can sense retinal depth and thus control their position 6.7. This radial process has disappeared by birth, when the cells already have an extensive network of horizontal processes. Thus, dissociation and reaggregation of neonatal retinas may remove the normal mechanism of positioning of these cells which results in their random distribution
throughout the reaggregates. The molecular basis of cell-sorting in these cultures is not known. It has been shown that antibody fragments directed against a cell adhesion molecule can cause a decrease in the laminar order in reaggregates of chick retina, although the effects on cell differentiation and synapse formation were not measured 28. As well as lamination, molecules have been described in graded concentrations along the dorsoventral axis of the retina 1~'3~. It will be of interest to see whether retinal cells also sort out along, or re-establish, these axes in reaggregate cultures. We have previously shown in monolayer cultures of neonatal rat retina that rod photoreceptors remain as small spherical cells and do not form detectable outer segments I. In the reaggregates, on the other hand, those photoreceptors that form rosettes do differentiate further. The ultrastructural characteristics of these rosettes clearly showed features typical of differentiating photoreceptors in vivo, namely junctions of the outer limiti~Jg membrane, inner segments, cilia and outer segments ~6. The abundance of amorphous membrane material seen in the cultures may represent outer segment disc material which has failed to integrate properly in the absence of some factor. However, the presence of small stacks of discs with expanded rims indistinguishable from the in vivo arrangement 25, and the frequency of the packaging of membrane material within an enveloping plasma membrane, indicate that the cells can begin to elaborate a structurally recognizable outer segment. The ultrastructural differentiation correlated with biochemical differentiation since outer segment-specific antigens were detected in the luminal regions of the rosettes. As well as supporting photoreceptor differentiation beyond that seen in monolayer cultures, the reaggregates also showed more complete Miiller cell differentiation. In monolayer cultures, Miiller cells remain as flat fibroblastic cells ~. In the rosettes, Miil!er cells participate in the outer limiting membrane junctions and extend microvilli in between the photoreceptor inner segments as in in vivo. Immunocytochemical labelling of these cells suggested that some of the Miiller cell processes are arranged in a radial manner, although we have not yet been able to follow such processes across the whole extent of a reaggregate. The mechanisms controlling photoreceptor differ-
307 entiation in these cultures are not yet clear. It is possible that the reaggregates permit the production of growth factors that can influence outer segment formation. Eye-derived growth factor has recently been s,~,own to ~ind to "~-.~'......... +. . . . . --," segments, and may be involved in some aspect of their regulation :6. A further possibility is that reaggregates promote synapse formation and o u t e r segment formation is linked to this. A third possibility is that specific interactions between Miiller cells and photoreceptors are necessary. This last possibility is supported by the observations that R E T - P l - p o s i t i v e photoreceptors outside the rosettes do not form o u t e r segments (Fig. 3) and that in chick cultures gliotoxic agents prevent rosette formation 22. W h e t h e r the effect of the interac-
Our results show that the stages of photoreceptor differentiation defined in vivo can be separated by different culture conditions. This suggests that differentiation of a rod photoreceptor is not solely defined at the time of cell commitment but also depends upon later signals. With the availability of controlled culture conditions and suitable molecular probes, it will now be possible to define more rigorously the nature of these signals and the level at which they act.
tion is to provide a structural substrate for organelle m o v e m e n t and outer segment formation or involves more specific m e m b r a n e interactions or transfer of molecules through the o u t e r limiting m e m b r a n e
helpful comments on the manuscript and Peter Peirce for excellent photographic assistance. This work was supported by N I H Grants NS20483, EY05206, NS22789 and an Alfred P. Sioan Research Fellowship to C.J.B.
~.) u
V + I . ~ J A ~ . ~ , , . . l..~ L ' L . , ~
v,,.4
+ ~.
junctions remains to be elucidated.
REFERENCES 1 Akagawa, K. and Barnstable, C.J., Identification and characterisation of cell types in monolayer cultures of rat retina using moneclonal antibodies, Brain Research, 383 (1986) 110-120. 2 Akagawa, K. and Barnstable, C.J., Identification and characterisation of cell types accumulating GABA in rat retinal cultures using ,:ell type-specific monoclonal antibodies, Brain Research. 408 (1987) 154-162. 3 Armson, P. and Bennet, M.R., Retinal ganglion cell cultures of high purity: effects of target tissues on cell survival, Neurosci. Lett., 38 (19+33) 181-186. 4 Barnstable, C.J., Monoclonal antibodies which recognise different cell types in the rat retina, Nature (London), 286 (1980) 231-235. 5 Barnstable, C.J., Immunological studies of the retina. In J. Brockes, (Ed.), Neuroimmunology, Plenum, New York, 1982, pp. 183-214. 6 Barnstable, C.J., Monoclonal antibodies as molecular probes of the nervous system. In T. Springer (Ed.), Hybridoma Technology in the Biosciences and Medicine, Plenum, New York, 1985, pp. 269-289. 7 Barnstable, C.J. and Constantine-Paton, M., Initial events of lamination in the mammalian retina, Soc. Neurosci. Abstr., 10 (1984) 787. 8 Barnstable, C.J. and Drfiger. U., Thy-l: a ganglion cellspecific marker of rodent retina, Neuroscience, 11 (1984) 847-855. 9 Barnstable, C.J., Holstein, R. and Akagawa, K., A marker of early amacrine cell development in rat retina, Dev. Brain Res., 20 (1985) 286-290. 10 Cohen, J., Burne, J.F., Winter, J. and Bartlett, P., Retinal ganglion cells lose response to laminin with maturation, Nature (London). 322 (1986) 465-467.
ACKNOWLEDGEMENTS We would like to thank Dr. Charles Marotta for the anti-neurofilament serum, Dr. Janet Sparrow for
11 Constantine-Paton, M., Blum, A.S., Mendez-Otero, R. and Barnstable, C.J., A cell surface molecule distributed in a dorsoventral gradient in the perin~tal rat retina, Nature (London), 324 (1986) 459-462. 12 Dr~iger, U.C., Edwards, D.L. and Barnstable, C.J., Antibodies against filamentous components in discrete cell types of the mouse retina, J. Neurosci., 4 (1984) 2025-2042. 13 Fekete, D.M. and Barnstable, C.J., The subcellular localisation of rat photoreceptor-specific antigens, J. Neurocvtol., 12 (1983) 785- 803. 14 Garber, B.B. and Mo.¢cona, A.A., Reconstruction of brain tissue from cell suspensions. I. Aggregation patterns of cell dissociated from different regions of the developing brain, Dev. Biol., 27 (1972) 217-234. 15 Hargrave, P.A., Adamus, G., Arendt, A., McDowell, J.H., Wang, J., Szary, A., Curtis, D. and Jackson, R.W., Rhodopsin's amino terminus is a principal antigenic site, Exp. Eye Res., 42 (1986) 363-373. 16 Hicks, D. and Barnstable, C.J., Lectin and antibody labelling of developing rat photoreceptor cells: an elecm~+n microscope immunocytochemical study, J. Neurocvtol., 15 (1986) 219-230. 17 Hicks, D. and Molday, R.S., The development of rat retinal neurons in cell culture, J. Neurochem., 41 (1983) suppl. $84. 18 Hinds, J.W. and Hinds, P.L., Early development of amacrine :ells in the mouse retina: an electron microscope, se~ rial section analysis, J. Comp. Neuroi.. L79 (1978) 277-300. 19 Jacobson, M., Developmental Neurobiology, Plenum, New York, 1978. 20 Leifer, D., Lipton, S.A., Barnstable, C.J.. and Masland, R.H., Monoclonai antibody to Thy-I enhances regeneration of processes by rat retinal ganglion cells in culture,
308
Science, 224 (1984) 303-306. 21 Levitt, P., Cooper, M.L. and Rakic, P., Early divergence and changing proportions of neuronal and glial precursor cells in the primate cerebral ventricular zone, De/: Biol., 96 (1983) 472-484. 22 Linser, P. and Moscona, A. ~ I~duction of glutamine synthetase in embryonic net~ral :eLina: Iocalisation in Miiiler fbers and dependence on cell interactions, Proc. Natl. Acad. Sci. U.S.A., 76 (19'79) 6476-6480. 23 Magnani. J.L., Thomas~ W.A. and Steinberg, M.S., Two distinct adhesion mechanisms in embryonic neural retina cells. I. A kinetic analysis, Dev. Biol., 81 (1981)96-105. 24 Moscona, A.A. and Housman, R.E., Biological and biochemical studies on embryonic cell-cell recognition. In J.W. Lash and M.M. Burger (Eds.), Cell and Tissuc Interactions, Raven, New York, 1977, pp. 173-185. 25 Nir, I.; Cohen, D. and Papermaster, D.S., Immunocytochemical localization of opsin in the cell membrane of developing rat photoreceptors, J. Cell Biol. 98 (1984) 1788-1795. 26 Plouet, J., Mascarelli, F., Lagente, O., Dorey, C., Lorans, G., Faure, J.P. and Courtois, Y., Eye derived growth factor: a component of rod outer segment implicated in phototransduction? In B. Ehinger (Ed.), Signal Systems in the Normal and Degenerating Retina, Elsevier, Amsterdam, in press. 27 Puro, D.G., Battelle, B.A. and Hansmann, K.E., Development of cholinergic neurons of the rat retina, Dev. Biol., 91 (1982) 138-148. 28 Rutishauser, U., Thiery, J.P., Brackenbury, R. and Edelman, G.M., Adhesion among neural cells of the chick embryo. III. Relationship of the surface molecule CAM to cell adhesion and the development of histiotypic pattern~, J.
Cell Biol., 79 (1972) 371-381. 29 S~rthy, P.V., Establishment of Miiller cell cultures from adult rat retina, Brain Research, 337 (1985) 138-i41. 30 Sheffield, J.B. and Moscona, A.A., Electron microscopic analysis of aggregation of embryonic cells: the structure and differentiation of aggregates of neural retina cells, Dev. Biol., 23 (1970) 36-61. 31 Sidman, R.L., Histogenesis of mouse retina studied with thymidine-H3. In G.K. Smelser, (Ed.), Structure of the Eye, Academic~ London° !96!, p p 487-.506. 32 Steinberg, M.S., The adhesive specification of tissue selforganisation. In T.G. Connelly et al. (Eds.), Morphogenesis and Pattern Formation, Raven, New York, 1981, pp. 179 -203. 33 Townes, P.L. and Holfreter, J., Directed movements and selective adhesion of embryonic amphibian cells, J. Exp. Zool., 128 (1955) 53-118. 34 Trisler, G.D., Schneider, M.D. and Nirenberg, M., A topographic gradient of molecules in retina can be used to identify neuron position, Proc. Nat. Acad. Sci. (U.S.A.), 78 (1981) 2145-2149. 35 Vollmer~ G. and Layer, P.G., An in vitro model of proliferation and differentiation of the chick retina: coaggregates of retinal an0 pigment epithelial cells, J. Neurosci., 6 (1986) 1885- 1896. 36 Volimer, G., Layer, P.G. and Gierer, A., Reaggregation of embryonic chick retina cells: pigment epithelial cells induce a high order of stratification, Neurosci. Lett., 48 (1984) 191-196. 37 Weidman, T.A. and Kuwabara, T., Postnatal development of the rat retina. An electron microscopic study, Arch. Ophthalmol., 79 (1968) 470-484.
298
Elsevier
BRE 13161
Histiotypic organization and cell differentiation in rat retinal reaggregate cultures Kimio Akagawa, David Hicks and Colin J. Barnstable Laboratory of Neurobi~Iog ;~'~The Rockefeller University, New York, N Y 10021 (U.S.A.)
(Accepted 9 June 1987) Key words: Photoreceptor; MOiler cell; Opsin; Monoclonal antibody; Immunocytochemistry; Electron microscopy; Cell interaction
Reaggregate cultures have been formed from cell suspensions of neonatal rat retinas. Histological sections of the ~eaggregates showed evidence of lamination with central rosettes formed around a lumen, a clear neuropil layer and an outer cellular layer. Each of the major retinal cell types, except ganglion ceils, could be positively identified using cell type-specific antibodies to label cryostat sections. Many of these were found to occupy positions within the reaggregates similar to those found in the intact retina. Electron microscopic observations showed abundant immature and mature synaptic endings within the neuropii layer, including a number of ribbon synapses. Examination of the rosettes showed an arrangement of Miiller glia and photoreceptors that closely resembled that of the intact retina. Within the lumen of rosettes, photoreceptors were found to contain stacks of disc-like membranes bounded by a plasma membrane, analogous to immature outer segments. The photoreceptors within rosettes also underwent molecular differentiation and expresscd an outer segment specific marker. The findings suggest that retinal cells have intrinsic properties that allow them to organize themselves into a correctly laminated structure and that particular cell interactions are necessary for continued differentiation of at least rod photoreceptors and MOiler cells. INTRODUCTION The extent to which genetic and epigenetic factors influence the generation and differentiation of the diverse sets of neuronal cell types of the vertebrate nervous system is poorly understood. It has been shown that some cells can express molecules characteristic of their mature phenotype before they have completed migration,-suggesting that determination of cell types begins in or near the ventricular zone of the developing tissue 9.21. The mechanistic relationship between the initial determinative events and the subsequent steps of development that can take place over several weeks remains unclear. The retina is a valuable model for the study of neural development for a number of reasons. It has only a very minor neural input through a small number of efferent fibers ~2. This means that until the formation of ganglion cell connections with central targets, early retinal development is essentially independent of
transsynaptic neural influences ~9. The stimulatory input of the retina comes from the detection of light by photoreceptors. Since the light transduction apparatus of the photoreceptors is one of the last steps of retinal differentiation, it is unlikely that the patterned sensory input has any role in the early formation of cell types or cell connections 16"25"37.The retina also has the advantages of being accessible to experimental manipulations, and of consisting of a relatively small number of neuronal cell types which are arranged in a laminated structure. A useful approach to study factors important in retinal development has been to isolate the tissue and study its developmental capacity in vitro, Monolayer cultures of retina have been used to study such properties as synz, m e formation by cholinergic amacrine cells 27, regeneration of processes by axotomized ganglion cells 3'1°'2° and the behavior of normal and activated glial cells 29. Reaggregate cultures have been used to provide a 3-dimensional in vitro environment
Correspondence: C,J. Barnstable, Laboratory of Neurobiology, The Rockefeller University, 1230 York Avenue', New York, NY
10021, U.S.A.
0006-8993/87/$03.50 O 1987 Elsevier Science Publishers B.V. (Biomedical Division)
299 for the cells z4. Aggregates of retina and of other r ,ain regions have been shown to undergo tissue sorting which led to the formation of layers that were thought to be analogous to those found in vivo 14"3°'32. Comparison of reaggregate and monolayer cultures has led to the suggestion that specific cell interactions are necessary for the expression of the glial enzyme carbonic anhydrase 2-'. although the mechanism of this effect is not yet known. One of the problems of using cultures of any heterogeneous tissue is that once removed from their normal environment, the identification of cell types becomes difficult. We have overcome this problem for rat retina by producing a series of monoclonal antibodies that selectively recognize molecules on each of the major neuronal and glial cell classes4-6. Using these reagents we have shown that each of the retinal cell types can be identified in monolayer cultures ~'2°. Although some of the cell types continued to differentiate in these cultures, the extent of differentiation was abnormal and several steps did not occur. Of particular relevance for the present study was th~ observation that photoreceptors remained alive for many weeks and continued to express the visual pigment protein opsin, but no rod outer segments were detected immunocytochemically. In this paper we have examined reaggregates of neonatal rat retinal cells. Using cell type-specific monoclonal antibodies, we have shown that the lamination in ~h~ aggregates does resemble that seen in vivo. In addition, both Miiller cells and photoreceptors acquire structural features not seen in either cell type maintained in monolayer culture. MATERIALS AND METHODS
Preparation of reaggregate cultures. Retinas were isolated from l-2-day-old Long-Evans rat pups without significant pigment epithelium contamination as described previously'. A single cell suspension was produced by mechanical trituration with a fire-polished pipet. The extent of dissociation was monitored microscopically and trituration was continued until no cell cluro-s of > I0 cells were found. Cells were suspended at a density of 5 × 104/ml in minimal essential medium (MEM) supplemented with 0.6% glucose, 5% rat serum, 3 mM taurine, 2 mM glutamine, 1 pg/mi DNase and 10 l~giml gentamycin. The
cell suspension was rocked at 30 cycles/min in an atmosphere of 7% CO,,, balanced air, with saturated humidity at 37 °C. The cultures were given fresh medium every 2 days and remained viable for at least two weeks. The speed of rocking determined reaggregate size. Under the conditions used, the reaggregates were spherical and of 300-800 /~m in diameter. Reaggregates of 500-700 lira were analyzed in the experiments reported here. Antibodies. All the antibodies used in this study have been described previously. Antibody HPC-1 reacts with retinal amacrine cells'~; 2G12, an antiThy-1, is specific for ganglion cells in retina~: RETPI is specific for rod photoreceptors 4 and reacts with a determinant on the N-terminal sequence of opsin 13"15. Antibody RHO-C7 was raised against bovine opsin but, unlike RET-P1, the determinant recognized is only found in outer segments and only appears when outer segment formation begins 6"~3 RET-G7 antibody recognizes a component of the cytoskeleton of gila but not neurons'. The antiserun: against the 140 kDa subunit of neurofilaments was a gift from Dr. C. Marotta, Harvard Medical School. bnmunocytochemistry. Reaggregates were rinsed 3 times with serum-free MEM and fixed with 4% paraformaldehyde in phosphate-buffered isotonic saline (PBS). During fixation the tissue was rocked to prevent any deformation of structure. After rinsing to remove fixative, groups of reaggregates were immersed in Tissue Tek II (Miles labs) and frozen, 15~m, cryostat sections were cut and mounted on gelatinized slides. Sections were preincubated with 5% normal goat serum in PBS to eliminate non-specific binding. Both primary and secondary incubations were carried out for 1-2 h at room temperature. Sections were mounted in PBS-glycerol (1:1) and viewed using a Zeiss microscope equipped with epifluorescence and appropriate combinations of filters. Electron microscopy. Reaggregate cultures were rinsed and fixed in 1% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.2 for 1 h at 4 °C. After washing in the same buffer for 1 h, tissue was postfixed in 1% OsO4 in 0.1 M cacodylate buffer. The reaggregates were again washed in the same buffer, dehydrated in ethanol and infiltrated with Epon/Araidite resin. Following polymerization, blocks were trimmed and sectioned. Initially, 1-!~m sections were stained with 0.1% Toluidine blue in I~- sodium borate and
300 inspected for the presence of rosettes. Once suitable areas were located, thin sectione (60-70 nm) were collected and examined on a Jeol 100S electron microscope. RESULTS
Formation ofreaggregates. The 1-2-day-old rat retina has an immature morphology and many of the cells do not yet have an extensive network of processes. With the mechanical dissociation used to prepare the cell suspension, those processes that were present were presumably broken off slate the cell suspension appeared as small (5-7 btm) spherical cells. Under the culture conditions ured, aggregation
could be detected microscopically within 30 min and by eye within 6 h. The rate of reaggregation was clearly dependent upon initial cell density and speed of agitation. At the cell density chosen a slow rocking speed of 30 cycles/min gave reaggregates whose mean diameter was approximately 500/~m. Higherspeed rotatory cultures (60-80 cycles/rain) gave almost entirely small cell clusters (data not shown). Ofto:'. !arger aggregates seemed to be formed by the fusion of two smaller aggregates since occasional bilobed structures were observed. Histology of reaggregates. Semithin sections of reaggregates clearly indicated cell sorting within the structures (Fig. 1A). The outermost layer of the reaggregate tended to consist of a morphologically ho-
A
.,>~
J
,I
•
,
•.
~
~.,r ~ q
.
Fig. 1. A: semithin section of a single reaggregate stained with Toluidine blue• Layers of differently stained cells and neuropil can be clearly seen. This reaggregate probably arose by fusion of two initial reaggregates since two central rosettes can be seen (curved arrows). B" higher-power view of a portion of (A), showing the outer layer of cells with processes directed in towards the neuropil (curved arrow) and a rare mitotic figure (straight arrow). C: higher-power view of the rosette in (A) showing luminal material inside the darkly staining cell nuclei• Bar in A = 1001(m.
301 tions of 7-day reaggregates with a panel of cell typespecific monoc!onal antibodies. Amacrine and ganglion cells. An antibody against a membrane antigen of all amacrine cells, HPC-1, labelled cell bodies at the outer layer of the reaggregates and processes in the outer portion of the neuropil layer (Fig. 2). Although the intensi',y of labelling ~!i ii ~~ii i ~ ~ii!i ¸i~ii~¸iii~i!i!i~ii!~~I!~i!iii!i!ii~i~!!!i!~i!~ii!l~!!i i~i!i!!i!iii!i!~!ii!~iii!i!i~i!i~i¸~!!iii!i!~!¸ii~!ii!~~ii~i!~~ii!!~i i!~!ii~i~i!i!~ii!i!~iii~i !~!i i~i~!i!i!i~ili increased over a period of two weeks in culture, no HPC-i-positive cells could be detected in i!i!~ii!~ii!i~i!i!i!i~ii!ii~ili!i!i~ii~i!i!i~ii!i!~~i!!~i i!i!i~ii~!!~i !~!iii!~~i!!ii!i!ii!~i!i~!ii!~!~i~i!~ii!~!~iii~i~i!i!~li!i!~i~i!~i~!i!i~'Ii~i!~i~iI!~i!Ii!iI~!iIi!~i~i~il!iI~!i~!i~i!i!~i~!~i!l!i~i!'~iIiIl!ii~!i~I~i~i!i~i~!iI!~i!~iIi!~!~i!~i!i~~i!ii~i!~iil!~i~!!~i!ii~Sli!i!~i~!~i!i!i!~i~i!~i!~i~!i!i~!i~!i!~i!~i!~i~i!~i!i!i~li!~i!~i!i!~ii!~ii!~i~i~!i~!i~!~i i~!~i~i~!additional ~i !i!i¸i!~i!i!¸i!iil!~i!~ il!~ii!~iii~ !¸i!~i!il!i~ l!i¸!i~f¸i!i!i!~i!i!i~iii¸~iii~iii!!i!i~¸!ii~ii!~!ii:i¸:iIIiIi¸I!~i!~i!i!i~¸ii~i!ii!¸~i!¸i!~ li!~il:i!i!i!i!i~¸iii:!)i!~i!i~II !!iili¸lili~iii!i¸!il!ii!ii!li!i~i!i!i!i!i!~ii!ii!i!i~~!:i~!ilS !i!i!ii!~ii!i~!¸i~ii¸li!~i!i!~ii!i~¸i~i!i~i!i¸!i~!i!~i!i~i!ili~i~i!i!i!il!i~ the inner portions of the reaggregates by immunofluorescence. No ganglion cells were detected in the reaggregate cultures used in these experiments, as defined by lack of labelling with antibodies against the ganglion cell marker Thy-1. Photoreceptors. Opsin is one of the earliest appearing rod photoreceptor-specific molecules. Antibody RET-P!, an anli-opsin, labelled many cells throughout the central portions of the reaggregates (Fig. 3A,B). The most concentrated RET-P1 labelling was observed in structures that could be seen by phase-contrast illumination to correspond to the rosettes seen in the Toluidine-stained material~ When labelling with outer segment-specific monoclonal anFig. 2. Cryostat section of reaggregate labelled with amacrine tibodies was performed, a different pattern of labelcell-specific antibody HPC-1. A: fluorescence. B: phase-conling was observed, as shown for antibody RHO-C7 trast. Only the outer layer of cell bodies and the processes im(Fig. 3C,D). Small bright patches of label were demediately underneath are labelled. Bar in B = 190!~m. tected in structures corresponding to the c,:r, tral portions of the rosettes. No RHO-C7 labelling was mogeneous layer of cell bodies that had large nuclei, found away from the rosettes even though adjacent clear nucleoli and a thin rim of cytoplasm. Processes sections of the same reaggregate showed that many from some of these cells could be observed passing RET-Pl-positive photoreceptors were present into an inner neuropil layer (curved arrow Fig. I B). throughout the central regions. The thickness of this layer varied between reaggreHorizontal and Midler cells. Since no ganglion cells gates from 10% of the reaggregate diameter to being were found in these reaggregates, we were able to barely discernable. Inside the oeuropil layer was a use an antiserum against the 140-kDa subunit of neumore heterogeneous layer of cell bodies. Within this rofilaments to label horizontal cells. Rare single cells layer, occasional mitotic figures were found (straight could be observed that did not seem to have a consisarrow Fig. 1B). The other cell bodies within this layer tent position in the reaggregates (Fig. 4A). In particwere of variable size and staining intensity, suggestular, double-labelling with anti-neurofilament serum ing that they may consist of a variety of retinal cell and antibody RHO-C7 showed that the horizontal types. The innermost regions of the reaggregates cells were not located on the outer regions of the phoconsisted of clusters of small, densely staining, fusitoreceptor rosettes (Fig. 4B). Miiller cells were form cells (Fig. 1C). Frequently these cells were labelled with antibody RET-G7. Labelling was obgrouped in rosette-like structures around a central luser,'ed throughout the structures but two features men. To examine whether these differences obwere more strongly labelled (Fig. 5). The first was a served at the light microscopic level corresponded to dense label in the neuropil layer ill the outer portion different retinal cell types, and to examine whether of the reaggregate and the second was a series of rathe arrangement of these cell types corresponded to dial fibers that passed through the rosettes and ended that seen in the intact retina, we labelled cryostat sec-
302
Fig. 3. Cryostat section of reaggregate labelled with photoreceptor-specific antibodies. Fluorescence (A) and phase-contrast (B) views of a section labelled with RET-P1. Labelled cells are found throughout the inner portions of the reaggregate but clusters of higher intensity of labelling are also found. Fluorescence (C) and phase-contrast (D) of a section labelled with RHO-C7. Only small areas of tissue show fluorescence. These correspond to the luminal portions of rosettes (arrow in D). Bar = 100l~m. in a more diffuse label within the rosette lumen.
Electron microscopic observation of reaggregates. Examination of the region surrounding rosette formations in 7-day reaggregate cultures revealed elongated cell bodies containing irregular nuclei with clumped heterochromatin (Fig. 6A). Extending from these nuclei towards the central lumen of the rosettes a number of structural features could be observed.
The cell bodies were joined by an array of desmosome-like junctions resembling the in vivo outer limiting membrane (arrow in Fig. 6A and B). Between the nuclei and this outer limiting membrane, many axially oriented mitochondria were visible. Cellular protrusions containing an array of organelles including mitochondria, vesicles and centrioles projected into the central lumen (Fig. 6B). These structures of-
303
Fig. 4. Cryostat section of reaggregate labelled with both a rabbit anti-neurofilament serum to label horizontal cells and monoclonal antibody RHO-C7 to label photoreceptor rosette centers. A: a single horizontal cell (down arrows) in a peripheral region of a reaggregate. B: a photoreceptor rosette is also present in the section (up arrows), but remote from the horizontal cell. Bar in A = 50!~m.
ten terminated in a longitudinally oriented cilium, and bore a strong resemblance to developing photoreceptor inner segments in vivo. Between these inner segments, fine microvilli were observed projecting from the surfaces of cells packcti between the photoreceptors (Fig. 6B). These correspond to the Mtiller cell microvilli observed in the intact retina. Within the central lumen of rosettes, membranous expansions from some cilia were seen. These irregular structures were bounded by a single plasma membrane, and contained a small number of membranous discs (Fig. 6C). Occasionally, membrane profiles exhibiting a distinct stacking of expanded disc rims were detected (Fig. 6D). In addition, concentrations of amorphous membranous material were observed in the lumen and packed between photoreceptor cell bodies (Fig. 6E). Examination of peripheral areas of the reaggregates revealed numerous fibers and enlarged endings containing an array of vesicles, filaments and smooth c~sternae (Fig. 7A-C). Features of mature ,:,ynapt~c endings, such a~ membrane thickenings, were less commonly observed (Fig. 7C). In addit~.on, densely
Fig. 5. Cryostat section of reaggregate labelled with Miiller cell-specific antibody RET-G7. Labelling is apparent throughout the section but is concentrated in the b~md ol neuropil under the surface (arrow) and in the rosettes ~asterisk). Bar = 11~1 !tm.
stained structures resembling synaptic ribbons were occasionally seen (Fig. 7D).
304
Fig. 6. Electron microscope view of thin section through rosette. A: the cell bodies contained irregular nuclei with clumped heterochromatin (N). Mitochondria (M) collected luminally to the nuclei and the cell bodies were joined by desmosome-like junctions (arrow), B: higher magnification of the inner segment region showing the ,,ame junctions (arrow). Cellular protrusions into the rosette lumen often contained mitochondria, vesicles and centrioles (C). These structures often ended in a longitudinally oriented cilium. Between the photoreceptors microvilli like those of M~iller cells were observed (open arrowhead). C.D: within the lumen many photoreceptors ended in membranous expansions which contained small stacks of discs (arrow). E" in some areas irregular accumulations of membranes could also be seen. Bar in A and B = 1#m; in C.D and E = 0.5!+m.
305
f--
f,~lL,~"
II
.~.
,,, •
.
+,-
+ ++ ++
+.a
•
• '
++
~
. .: ~+.+
].~.
~
~
".
"
,
.+
+ +,A+,
+-~
o
i ~
+'-
~
,a,
"'
*
. . . . .
"
+
. +,
+
:.
+ ;m~
]
+
" :
:-.,;
~
+~'
+.. +,
,.
+++-
++, . ..,:,~"~+
-: + . . . . .
+
-
_-+.
+.
,7
...
,'¢
-
,
+,
. ,I+.
+ .
~],,%'~
~'~,,~
~
+
+ ..
...
+ + , + , ~ ~+~"
+..+
+ +++I+
-:~+~+
-. *'
....
,:,,"
.
:" . ++
.'.- .:.~ . . . . . +.
+
+, "- '
"-
•
+"
"'+~
-'._+~ . t - , + " + , +
"?
+.
' "+
'L+:+%++: m +
,
- +
Fig. 7. Four examples of svnaptic structures foand in the peripheral areas of the reaggregates. A.B: immature synaptic endings exhibiting numerous vesicles (~,'). mitochondria (m) and smooth cisternae (sc). C: more mature synapse showing clustering of vesicles (v)
near a membrane thickening (mt). D: this synapse possesses a densely staining synaptic ribbon (srJ characteristic of photoreceptor and bipolar cell synapses. All bars = 0.5 !+m.
DISCUSSION
Histiotypic cell sorting has been observed in a variety of reaggregate cultures 3u'32"33. Al t h o u g h the exact s e q u e n c e of reaggregate f o r m a t i o n is not cleat, electr on microscopic e x a m i n a t i o n of reaggregating chick retinal cells at different times after dissociation has led to the suggestion that the sequence of aggrega-
tioo is into r a n d o m groups of cells followed by polarization or sorting out 3°. This is s u p p o r t e d bv studies of the kinetics of aggregate f o r ma tio n zs. Suggestive evidence of sorting out of some cell types in cell clusters in m o n o l a v e r cultures of rat retina has also been obtained 2. The polarity of retired reaggregates can vary according to experimental conditions. Chick reaggre-
306 gates from E7 retinas were found to take on a 'lumicentric" organization analogous to the rat reaggreo gates described here 3°. Reaggregates of El4 chick retinal cells, on the other hand, formed 'axocentric' structures that seemed to be of the opposite orientation. A role for factors other than age has been suggested since reaggregates of chick retina were more highly organized in the presence of pigment epithelial cells 35"36. In addition, newborn rat retinal cells grown in serum-free medium developed a less marked but observable tendency to laminate in a ph6torereptor-outermost orientationS7. Although we have not fully investigated all these variables, we have added rat pigment epithelial cells to our reaggregates without any detectable effect upon the cellular organization (data not shown). In our experiments, the outer layer of cells consisted of amacrine cells. These formed a single layer rather than distributing themselves on both sides of an inner plexiform layer as occurs in vivo. It is possible that this different distribution is due to a lack of ganglion cells which might play an active role in determining inner retina structure. It is possible that the loss of ganglion cells is due to absence of either presynaptic or postsynaptic interactions. Although we have observed synapses in these reaggregates (see Fig. 7), we have not yet been able to assign the cell types involved. An alternative explanation for the absence of ganglion cells is that, as the most mature cells in the neonatal retina, they suffer the greatest physical damage during dissociation. A-type horizontal cells, as detected by labelling with neurofilament antisera, were not obviously organized in a laminar fashion. These cells are among the first generated in the retina ~8'3'. When the A-type horizontal cells are born, they leave the mitotic zone and position themselves in the middle of what appears to be an otherwise undifferentiated neuroblast layer. As they migrate they maintain a radial projection to the ventricular surface of the retina and it has been suggested that this could provide a mechanism by which these cells can sense retinal depth and thus control their position 6.7. This radial process has disappeared by birth, when the cells already have an extensive network of horizontal processes. Thus, dissociation and reaggregation of neonatal retinas may remove the normal mechanism of positioning of these cells which results in their random distribution
throughout the reaggregates. The molecular basis of cell-sorting in these cultures is not known. It has been shown that antibody fragments directed against a cell adhesion molecule can cause a decrease in the laminar order in reaggregates of chick retina, although the effects on cell differentiation and synapse formation were not measured 28. As well as lamination, molecules have been described in graded concentrations along the dorsoventral axis of the retina 1~'3~. It will be of interest to see whether retinal cells also sort out along, or re-establish, these axes in reaggregate cultures. We have previously shown in monolayer cultures of neonatal rat retina that rod photoreceptors remain as small spherical cells and do not form detectable outer segments I. In the reaggregates, on the other hand, those photoreceptors that form rosettes do differentiate further. The ultrastructural characteristics of these rosettes clearly showed features typical of differentiating photoreceptors in vivo, namely junctions of the outer limiti~Jg membrane, inner segments, cilia and outer segments ~6. The abundance of amorphous membrane material seen in the cultures may represent outer segment disc material which has failed to integrate properly in the absence of some factor. However, the presence of small stacks of discs with expanded rims indistinguishable from the in vivo arrangement 25, and the frequency of the packaging of membrane material within an enveloping plasma membrane, indicate that the cells can begin to elaborate a structurally recognizable outer segment. The ultrastructural differentiation correlated with biochemical differentiation since outer segment-specific antigens were detected in the luminal regions of the rosettes. As well as supporting photoreceptor differentiation beyond that seen in monolayer cultures, the reaggregates also showed more complete Miiller cell differentiation. In monolayer cultures, Miiller cells remain as flat fibroblastic cells ~. In the rosettes, Miil!er cells participate in the outer limiting membrane junctions and extend microvilli in between the photoreceptor inner segments as in in vivo. Immunocytochemical labelling of these cells suggested that some of the Miiller cell processes are arranged in a radial manner, although we have not yet been able to follow such processes across the whole extent of a reaggregate. The mechanisms controlling photoreceptor differ-
307 entiation in these cultures are not yet clear. It is possible that the reaggregates permit the production of growth factors that can influence outer segment formation. Eye-derived growth factor has recently been s,~,own to ~ind to "~-.~'......... +. . . . . --," segments, and may be involved in some aspect of their regulation :6. A further possibility is that reaggregates promote synapse formation and o u t e r segment formation is linked to this. A third possibility is that specific interactions between Miiller cells and photoreceptors are necessary. This last possibility is supported by the observations that R E T - P l - p o s i t i v e photoreceptors outside the rosettes do not form o u t e r segments (Fig. 3) and that in chick cultures gliotoxic agents prevent rosette formation 22. W h e t h e r the effect of the interac-
Our results show that the stages of photoreceptor differentiation defined in vivo can be separated by different culture conditions. This suggests that differentiation of a rod photoreceptor is not solely defined at the time of cell commitment but also depends upon later signals. With the availability of controlled culture conditions and suitable molecular probes, it will now be possible to define more rigorously the nature of these signals and the level at which they act.
tion is to provide a structural substrate for organelle m o v e m e n t and outer segment formation or involves more specific m e m b r a n e interactions or transfer of molecules through the o u t e r limiting m e m b r a n e
helpful comments on the manuscript and Peter Peirce for excellent photographic assistance. This work was supported by N I H Grants NS20483, EY05206, NS22789 and an Alfred P. Sioan Research Fellowship to C.J.B.
~.) u
V + I . ~ J A ~ . ~ , , . . l..~ L ' L . , ~
v,,.4
+ ~.
junctions remains to be elucidated.
REFERENCES 1 Akagawa, K. and Barnstable, C.J., Identification and characterisation of cell types in monolayer cultures of rat retina using moneclonal antibodies, Brain Research, 383 (1986) 110-120. 2 Akagawa, K. and Barnstable, C.J., Identification and characterisation of cell types accumulating GABA in rat retinal cultures using ,:ell type-specific monoclonal antibodies, Brain Research. 408 (1987) 154-162. 3 Armson, P. and Bennet, M.R., Retinal ganglion cell cultures of high purity: effects of target tissues on cell survival, Neurosci. Lett., 38 (19+33) 181-186. 4 Barnstable, C.J., Monoclonal antibodies which recognise different cell types in the rat retina, Nature (London), 286 (1980) 231-235. 5 Barnstable, C.J., Immunological studies of the retina. In J. Brockes, (Ed.), Neuroimmunology, Plenum, New York, 1982, pp. 183-214. 6 Barnstable, C.J., Monoclonal antibodies as molecular probes of the nervous system. In T. Springer (Ed.), Hybridoma Technology in the Biosciences and Medicine, Plenum, New York, 1985, pp. 269-289. 7 Barnstable, C.J. and Constantine-Paton, M., Initial events of lamination in the mammalian retina, Soc. Neurosci. Abstr., 10 (1984) 787. 8 Barnstable, C.J. and Drfiger. U., Thy-l: a ganglion cellspecific marker of rodent retina, Neuroscience, 11 (1984) 847-855. 9 Barnstable, C.J., Holstein, R. and Akagawa, K., A marker of early amacrine cell development in rat retina, Dev. Brain Res., 20 (1985) 286-290. 10 Cohen, J., Burne, J.F., Winter, J. and Bartlett, P., Retinal ganglion cells lose response to laminin with maturation, Nature (London). 322 (1986) 465-467.
ACKNOWLEDGEMENTS We would like to thank Dr. Charles Marotta for the anti-neurofilament serum, Dr. Janet Sparrow for
11 Constantine-Paton, M., Blum, A.S., Mendez-Otero, R. and Barnstable, C.J., A cell surface molecule distributed in a dorsoventral gradient in the perin~tal rat retina, Nature (London), 324 (1986) 459-462. 12 Dr~iger, U.C., Edwards, D.L. and Barnstable, C.J., Antibodies against filamentous components in discrete cell types of the mouse retina, J. Neurosci., 4 (1984) 2025-2042. 13 Fekete, D.M. and Barnstable, C.J., The subcellular localisation of rat photoreceptor-specific antigens, J. Neurocvtol., 12 (1983) 785- 803. 14 Garber, B.B. and Mo.¢cona, A.A., Reconstruction of brain tissue from cell suspensions. I. Aggregation patterns of cell dissociated from different regions of the developing brain, Dev. Biol., 27 (1972) 217-234. 15 Hargrave, P.A., Adamus, G., Arendt, A., McDowell, J.H., Wang, J., Szary, A., Curtis, D. and Jackson, R.W., Rhodopsin's amino terminus is a principal antigenic site, Exp. Eye Res., 42 (1986) 363-373. 16 Hicks, D. and Barnstable, C.J., Lectin and antibody labelling of developing rat photoreceptor cells: an elecm~+n microscope immunocytochemical study, J. Neurocvtol., 15 (1986) 219-230. 17 Hicks, D. and Molday, R.S., The development of rat retinal neurons in cell culture, J. Neurochem., 41 (1983) suppl. $84. 18 Hinds, J.W. and Hinds, P.L., Early development of amacrine :ells in the mouse retina: an electron microscope, se~ rial section analysis, J. Comp. Neuroi.. L79 (1978) 277-300. 19 Jacobson, M., Developmental Neurobiology, Plenum, New York, 1978. 20 Leifer, D., Lipton, S.A., Barnstable, C.J.. and Masland, R.H., Monoclonai antibody to Thy-I enhances regeneration of processes by rat retinal ganglion cells in culture,
308
Science, 224 (1984) 303-306. 21 Levitt, P., Cooper, M.L. and Rakic, P., Early divergence and changing proportions of neuronal and glial precursor cells in the primate cerebral ventricular zone, De/: Biol., 96 (1983) 472-484. 22 Linser, P. and Moscona, A. ~ I~duction of glutamine synthetase in embryonic net~ral :eLina: Iocalisation in Miiiler fbers and dependence on cell interactions, Proc. Natl. Acad. Sci. U.S.A., 76 (19'79) 6476-6480. 23 Magnani. J.L., Thomas~ W.A. and Steinberg, M.S., Two distinct adhesion mechanisms in embryonic neural retina cells. I. A kinetic analysis, Dev. Biol., 81 (1981)96-105. 24 Moscona, A.A. and Housman, R.E., Biological and biochemical studies on embryonic cell-cell recognition. In J.W. Lash and M.M. Burger (Eds.), Cell and Tissuc Interactions, Raven, New York, 1977, pp. 173-185. 25 Nir, I.; Cohen, D. and Papermaster, D.S., Immunocytochemical localization of opsin in the cell membrane of developing rat photoreceptors, J. Cell Biol. 98 (1984) 1788-1795. 26 Plouet, J., Mascarelli, F., Lagente, O., Dorey, C., Lorans, G., Faure, J.P. and Courtois, Y., Eye derived growth factor: a component of rod outer segment implicated in phototransduction? In B. Ehinger (Ed.), Signal Systems in the Normal and Degenerating Retina, Elsevier, Amsterdam, in press. 27 Puro, D.G., Battelle, B.A. and Hansmann, K.E., Development of cholinergic neurons of the rat retina, Dev. Biol., 91 (1982) 138-148. 28 Rutishauser, U., Thiery, J.P., Brackenbury, R. and Edelman, G.M., Adhesion among neural cells of the chick embryo. III. Relationship of the surface molecule CAM to cell adhesion and the development of histiotypic pattern~, J.
Cell Biol., 79 (1972) 371-381. 29 S~rthy, P.V., Establishment of Miiller cell cultures from adult rat retina, Brain Research, 337 (1985) 138-i41. 30 Sheffield, J.B. and Moscona, A.A., Electron microscopic analysis of aggregation of embryonic cells: the structure and differentiation of aggregates of neural retina cells, Dev. Biol., 23 (1970) 36-61. 31 Sidman, R.L., Histogenesis of mouse retina studied with thymidine-H3. In G.K. Smelser, (Ed.), Structure of the Eye, Academic~ London° !96!, p p 487-.506. 32 Steinberg, M.S., The adhesive specification of tissue selforganisation. In T.G. Connelly et al. (Eds.), Morphogenesis and Pattern Formation, Raven, New York, 1981, pp. 179 -203. 33 Townes, P.L. and Holfreter, J., Directed movements and selective adhesion of embryonic amphibian cells, J. Exp. Zool., 128 (1955) 53-118. 34 Trisler, G.D., Schneider, M.D. and Nirenberg, M., A topographic gradient of molecules in retina can be used to identify neuron position, Proc. Nat. Acad. Sci. (U.S.A.), 78 (1981) 2145-2149. 35 Vollmer~ G. and Layer, P.G., An in vitro model of proliferation and differentiation of the chick retina: coaggregates of retinal an0 pigment epithelial cells, J. Neurosci., 6 (1986) 1885- 1896. 36 Volimer, G., Layer, P.G. and Gierer, A., Reaggregation of embryonic chick retina cells: pigment epithelial cells induce a high order of stratification, Neurosci. Lett., 48 (1984) 191-196. 37 Weidman, T.A. and Kuwabara, T., Postnatal development of the rat retina. An electron microscopic study, Arch. Ophthalmol., 79 (1968) 470-484.