- Email: [email protected]
De novo formation of axon-like processes from axotomized retinal ganglion cells which exhibit long distance growth in a peripheral nerve graft in adult hamsters
Brain Research, 484 (1989) 371-377
371
Elsevier BRE 23450
De novo formation of axon-like processes from axotomized retinal ganglion cells which exhibit long distance growth in a peripheral nerve graft in adult hamsters* Eric Y.E Cho and Kwok-Fai So Department of Anatomy, Facultyof Medicine, Universityof Hong Kong, Hong Kong (Hong Kong) (Accepted 27 December 1988)
Key words: Peripheral nerve graft; Optic nerve crush; Axon-like process; Retinal ganglion ceil; Hamster
Damaged axons in the central nervous system of the adult mammal can be stimulated to regenerate extensively into a peripheral nerve graft. It was generally believed that the new axonal sprouts which extend into the graft arose from the injured proximal axonal stumps. However, when retinal ganglion cells of the adult hamster were axotomized by crushing the optic nerve and the proximal axonal stump was not in direct apposition to the graft, a new axon-like process could be seen to be emitted from either the cell soma or dendrite and extended in the graft for at least 1-2 cm. This axon-like process was distinct from the original injured axon which could still be seen to course towards the optic disc in the retina. Evidently, even a fully differentiated central nervous system neuron of the adult mammal retains a great degree of morphological plasticity so that if the original axon is discouraged to regrow after injury, other parts of the neuron can act as favourable sites for the sprouting of a new axon-like process.
The limited and abortive regrowth exhibited by the majority of central nervous system (CNS) axons of the adult mammal after injury is well documented 2'8'12. However, transplantation of a peripheral nerve (PN) segment to the damaged CNS was found to promote extensive axonal regrowth into the graft 1. For instance, in the rat 14 and hamster 15, it was demonstrated that damaged ganglion cell axons would regrow into a PN implanted into the retina. In these previous studies it was generally believed that the injured proximal axonal stump regenerates directly into the graft. We have now observed that when retinal ganglion cell axons in the adult hamster are damaged but the injured proximal stump is not in direct apposition to the PN graft, a new axon-like process can arise either from the cell body or dendrite to extend into the graft and regrow for a very long distance in it. In adult golden hamsters 4 - 6 weeks old, an autologous PN segment (common peroneal branch of the sciatic nerve) 2-2.5 cm long was implanted
into the superior temporal quadrant of the right eye 14. At 1-2 months post grafting, horseradish peroxidase (HRP, Sigma, 50% in saline) was applied to the cut end of the graft transected at 1-2 cm from the eye. Eighteen to 20 h later, the animal was perfused with 0.9% saline, the retina was removed and fixed in a phosphate-buffered fixative containing 1% paraformaldehyde and 1.25% glutaraldehyde for half an hour. After reacting the retina for H R P histochemistry using tetramethylbenzidine as the chromogen 1~, a sector of retrogradely labelled ganglion cells peripheral to the grafting site in the retina could be seen (Fig. 1A), the axons of which were damaged by the grafting procedure and have regrown into the graft 14'~5. In the present study, in addition to PN grafting, the optic nerve (ON) of the eye which received the PN transplant was crushed at 1 m m behind the globe at the same time when grafting was performed. W h e n H R P was applied to the graft as described above, in addition to the population of labelled ganglion cells peripheral to
* A preliminary report of some of the results has been published 4. Correspondence: E.Y.P. Cho, Department of Anatomy, Faculty of Medicine, University of Hong Kong, Hong Kong, Hong Kong. 0006-8993/89/$03.50 © 1989 Elsevier Science Publishers B.V. (Biomedical Division)
372
,6,
B
/ D
j Fig. 1. A: distribution of HRP-labelled ganglion cells (dots) in the superior temporal quadrant of a retinal whole mount from an animal with a PN implanted into the retina (postgrafting time = 28 days). B: same as A but from an animal with concurrent ON crush of the same eye with PN grafting (postgrafting time = 43 days). In A. labelled cells were only observed peripheral to the graft whereas in B an additional population of labelled cells central to the graft could also be seen. Most of the cells from the central population were located less than 500 ~tm from the grafting site although occasionally a few cells could be seen as far as 1.5 mm away. Their numbers range from 7 to 30 (n = 5). C: schematic diagram of the superior temporal quadrant of a retinal whole mount from an eye with ON crush in addition to PN grafting. Cells a-f represented the labelled cells seen after HRP application to the graft. Cells a, b and c have their axons being damaged by the ON crush and are collectively referred to as the central population whereas cells d, e and f, the peripheral population, have their axons being interrupted by the graft and with the injured proximal axonal stumps in direct apposition to it. D: distribution of labelled ganglion cells in the retina from an animal with ON crush and PN grafting (28 days post grafting). Immediately before HRP application to the graft, an incision (L) was made in the retina at a region between the graft-implantation site and optic disc. HRP was then applied to the cut end of the graft at 1 cm from the eye and the animal processed as in (A) and (B). G, graft-implantation site: X. optic disc. Scale bar. 1 mm (same for A. B and D).
t h e graft, l a b e l l e d g a n g l i o n cells which lay in a z o n e
a c c o m p a n i e d by O N crush, w e k n o w that t h e i r axons
b e t w e e n t h e grafting site and optic disc as well as s o m e which s u r r o u n d e d the grafting site c o u l d also be o b s e r v e d (Fig. 1B). Since this a d d i t i o n a l p o p u -
w e r e not i n t e r r u p t e d by the g r a f t i n g p r o c e d u r e but r a t h e r w e r e d a m a g e d by the O N crush. W e d e s i g n a t e t h e s e cells as the central p o p u l a t i o n ( a l t h o u g h n o t all
lation o f cells was o n l y s e e n w h e n P N grafting was
o f t h e m lay c e n t r a l ( t o w a r d s t h e optic disc) to t h e
373 graft, see Fig. 1C) to distinguish them from that of the peripheral population which have their axons being intercepted by the graft and with the injured proximal stump in close apposition to it. In accordance with a previous study tS, it was originally interpreted that the axons of ganglion cells from the central population looped back into the retina after their severance in the ON and extended into the PN graft. Therefore, if immediately prior to application of HRP, a large incision was made in the retina between the grafting site and optic disc in a manner which would have severed the axons of ganglion cells from the central population before they entered the graft, the central population should fail to become labelled if their axons followed the course as described above. However, when animals (n = 4) with PN grafting accompanied by ON crush were treated in the above manner, the central population of labelled cells still persisted (Fig. 1D). This strongly suggests that other processes distinct from the originally injured axon in the ON have extended into the PN so that the central population could still be labelled even when the axon was interrupted immediately before HRP application to the graft. The detailed morphology of the regenerating ganglion cells was studied in order to see whether any other processes distinct from the original axon have sprouted into the PN graft. We adopted a reduced silver staining method on whole-mounted retina which was found to stain the regenerating ganglion cells (including the dendritic tree and axons) intensely. At 1-2 months post grafting, ganglion cells which have their axons extended into the graft were first labelled with a fluoresent dye (Fast Blue or Granular Blue, 3% solution in saline) applied to the graft at 1-2 cm from the eye. After a survival time of 4 days, the retina was dissected out and fixed in 2% paraformaldehyde (in phosphate buffer pH 7.4) for 1 h and then mounted in glycerol. Labelled ganglion cells were viewed under epifluorescence and their distributions recorded by photography. The retina was then fixed in 10% formol saline for 1 h and silver-stained according to the protocol of Leicester and Stone 9. Using this method, in the normal adult hamster retina, only the axons were revealed but the ganglion cell somata and the dendrites were not stained (except for some weak
staining associated with a few ganglion cells with the largest soma size). However, ganglion cells regenerating axons into a PN graft were found to be intensely stained, revealing the detailed patterns of their dendritic trees. The positions of the silver stained cells were compared with that of the fluorescent dye-labelled cells to identify those which have regenerated axons or processes into the graft. In retinas which were optimally stained with this silver method, we consistently observed some ganglion cells in the central population which had been labelled with the fluorescent dye (and hence have regenerated into the graft) to possess an axon-like process clearly distinct from the existing axon. Altogether we analysed 38 cells from 9 retinas with this type of peculiar morphology, 9 of which could not be shown to be labelled by the fluorescent dye. In 4 of these 9 cells (e.g. cell 3 in Fig. 2), the region of the retina where they were located was not photographed at the time when the positions of the fluorescent dye-labelled cells were being recorded. In others, the process may have extended into the graft but have not reached the application site of the dye. However, since the morphology of these 9 cells was similar to the other 29 (which were labelled by the fluorescent dye), we felt justified to include them in our analysis. The extra process originated most commonly from either the celt body (55% of the total of 38 cells) or a primary dendrite (26%), and in rare cases from a secondary (13%) or even higher order (5%) dendrite. An example of a retina from an animal which received PN grafting together with ON crush and examined at 56 days post grafting (Figs. 2 and 3) served to illustrate the appearance of these processes and their trajectories in the retina. Cells 1, 2, and 3 belonged to the central population since their axons were not interrupted by the graft and could still be seen running to the optic disc. In each cell, however, a separate process could be seen to arise from the cell body and extend to the grafting site in the retina. These processes were unbranched during their whole intraretinal course (although in 3 out of the 38 cells studied one or two very fine side branches could be seen to arise from the process), maintained a fairly uniform diameter without the tapering commonly associated with dendrites and thus resembled an axon in external appearance. Once they were emit-
374
A
B G
P
~
ODe
Fig. 2. A: camera lucida drawing of 3 silver-stained ganglion cells of the central population from an animal with ON crush and PN grafting (56 days post grafting). Cells 1 and 2 were labelled by the fluorescent dye Fast Blue applied to the graft (see Fig. 3)i In all 3 cells, an unbranched axon-like process (p) distinct from the axon (a) could be seen to arise from the cell body and extend towards the graft-implantation site (G) whereas the axon headed in the direction of the optic disc (OD). The process of cell t had a relatively uncomplicated trajectory to the graft but that of cells 2 and 3 took a tortuous course, displaying a series of loops along its way. In fact, 47% of the 38 cells we studied had the axon-like process exhibiting some sort of loops along its course to the graft and 53% of the cells had the process following a relatively undeviated intraretinal pathway towards the graft. Scale bar, 100/am. B: detailed tracing of cell 1 showing the origin of the axon-like process from the cell body. In some cases like in this cell, the process was slightly thicker than the axon but in others the two had comparable diameters, a, axon; p, process to graft. Scale bar, 50/tm.
ted from the ganglion cell, these processes would either head straight towards the grafting site with only minor deviations from the main direction as in cell 1, or they could take complicated and tortuous courses in the retina before reaching the graft finally (cells 2 and 3). In most instances the axon-like processes were confined to the inner plexiform layer (IPL) throughout its intraretinal course to the graft (74% of the 38 cells), while a small number of them (22%) originally residing in the IPL m o v e d up to the nerve fibre layer as they approached the grafting site. In this respect it is different from the axon which is always located in the nerve fibre layer. W h e n e v e r regenerating ganglion cells in the central population were found to possess such extra axon-like processes, only one could be observed to arise from each cell and no matter how tortuous their initial trajectories were, their final destination would invariably be the grafting site. However, not all ganglion cells from the central population were shown to exhibit this morphological peculiarity when examined by the silver staining method. This is because the majority of regenerating ganglion cells
from the central population tended to be located in a region near to the graft where a great multitude o f dendritic trees and the associated ganglion cell bodies would be stained, making the definite identification of any extra axon-like processes an extremely difficult task. Therefore only cells located about 300/~m or more from the grafting site could be reliably analysed but at this distance and beyond the number of labelled cells of the central population was very small. We also observed some cells from the central population with no axon seen running to the optic disc but only an axon-like process extending into the graft. Since this process coursed in the IPL rather than the nerve fibre layer (the normal location of ganglion cell axons), we suspected it was in fact newly generated to extend into the graft whereas the orginal axon had degenerated after ON crush. Thus, although we do not see the presence of extra axon-like processes occurring in the majority of cells in the central population in the silver-stained preparations due to the above mentioned factors, it seems logical to conclude that many of the regenerating cells in this population possess them if we
Fig. 3. A: photomicrograph showing the 3 silver-stained cells in Fig. 2. The axons of the 3 cells (1, 2 and 3) were traced by arrows. The axon-like process of cells 1 and 2 were indicated by asterisks (*) while the course of the axon-like process of cell 3 was followed with arrowheads. G, graft-implantation site; OD, optic disc. Scale bar, 100 pm. B: photomicrograph at a higher magnification showing cells 1 and 2. In both cells the axon (a) and the process (p) to the graft arose at a site on the soma distinct from each other. Scale bar, 50 pm (same for B and C). C: cells 1 and 2 were retrogradely labelled by the fluorescent dye Fast Blue applied to the graft at 1.5 cm from the eye and photographed before silver staining was performed.
376 also take into account the result of the experiment of making a lesion between the optic disc and grafting site before H R P application to the graft. What is the signal for inducing the formation of these axon-like processes? Some clues may be gained by considering the situation in animals which received a PN graft but with no ON crush. We have examined the morphology of over 100 regenerating ganglion cells in these animals at 2-5 weeks post grafting using the same silver staining method and not one was seen to possess any extra axon-like processes. Since the axons of these ganglion cells were intercepted by the graft and directly opposed to it after injury, the damaged proximal axonat stump could presumably respond to the favourable PN environment with little delay to regrow into it 3. In contrast to this, the central population of ganglion cells (in animals with ON crush and PN grafting performed together) would have their injured proximal axonal stump residing in the ON where the environment for regrowth was certainly not optimal. However, it can be envisioned that some sort of trophic influence emanating from the PN has already signalled the cell body to turn on its cellular machinery required for the initiation of axonal regrowth, such as the synthesis of new membrane constituents or growth associated proteins 5. If the proximal axonal stump in the ON cannot act as a suitable target for responding to or incorporating the newly synthesized constituents they may be diverted to other regions of the cell such as the soma or primary dendrite to generate new axon-like processes which can then be attracted towards the PN graft. The viability of the proximal axonal stump after injury has been suggested to be important for the distribution of sprouting over the neuronal surface 6'13. The generation of supernumerary axons in spinal motoneurons of the cat after their axons in the peripheral nerve have been damaged and pre-
vented to regenerate may also be explained similarly 7. It seems that even mature neurons in the mammalian CNS like the retinal ganglion cell still possesses a remarkable degree of morphological plasticity such that the basic neuronal geometry can be profoundly altered as it regenerates after injury. Other examples include the sprouting of axon-like processes from the dendrites of axotomized motoneurons in the cat 1° or formation of a supernumerary axon from the cell body of the motoneuron after a distal axotomy in the cat 7. Our results also indicate that these novel axon-like processes are attracted to grow towards and into the PN graft, unlike the supernumerary axons of cat motoneurons which exhibit variable trajectories in the spinal cord 7. Moreover, similar to injured axons regrowing in a PN graft t, the axon-like processes observed in our present study can also elongate for a considerable distance (1-2 cm, the distance from the eye at which HRP or the fluorescent dye is applied to the graft) in the graft. The fact that not more than one axon-like process is generated suggests that the intrinsic mechanisms regulating its formation may be similar to that operating in the differentiation of ganglion cells to the mature state whereby only one axon is formed. Even then, however, the ganglion cell now faces the problem of having to sustain apparently 2 axons which is not in keeping with the rules of its developmental program. Therefore it remains to be seen whether the original damaged axon will ultimately be eliminated and if the cellular events which occur in a normal axon such as the transport of synaptic vesicles and propagation of action potentials are still operating in the damaged axon which has resisted degeneration.
1 Aguayo, A.J. Axonal regeneration from injured neurons m the adult mammalian central nervous system. In C.W. Cotman (Ed.), Synaptic Plasticity, Guilford Press. New York, 1985, pp. 457-484. 2 Berry, M., Regeneration of axons in the central nervous system. In V. Navaratnam and R.J. Harrison (Eds.), Progress in Anatomy, Vol. 3. Cambridge Univ. Press. 1979. pp. 213-233. 3 Cho, E.Y.P. and So, K.-E, Rate of regrowth of damaged retinal ganglion cell axons regenerating in a peripheral
nerve graft in adult hamsters. Brain Research. 419 (1987) 369-374. 4 Cho, E.Y.P. and So, K.-F.. Growth of newly formed neuronal processes into a peripheral nerve graft stimulated by axotomy of retinal ganglion cells in the adult hamster, Hong Kong Soc. Neurosci. Abstr.. 10 (1988) 56. 5 Grafstein. B.. The retina as a regenerating organ. In R. Adler and D.B. Farber (Eds.), The Retina: a Model for Biology Studies. Part 2. Academic Press. New York. 1986. pp. 275-335.
This study was supported by the Croucher Foundation of Hong Kong.
377 6 Hail, G.E and Cohen, M.J., Extensive dendritic sprouting induced by close axotomy of central neurons in the lamprey, Science, 222 (1983) 518-521. 7 Havton, L. and KeUerth, J.-O., Regeneration by supernumerary axons with synaptic terminals in spinal motoneurons of cats, Nature (Lond.), 325 (1987) 711-714. 8 Kiernan, J.A., Hypotheses concerned with axonal regeneration in the mammalian nervous system, Biol. Rev., 54 (1979) 155-197. 9 Leicester, J. and Stone, J., Ganglion, amacrine and horizontal cells of the cat's retina, Vision Res., 7 (1967) 695-705. 10 Lind,, H.. Risling, M. and Cullheim, S., ~Dendraxons' in regenerating motoneurons in the cat: do dendrites generate new axons after central axotomy?, Brain Research, 358 (1985) 329-333. 11 Mesulam. M.-M., Tetramethyl benzidine for horseradish peroxidase neurohistochemistry: a non-carcinogenic blue
12
13
14
15
reaction product with superior sensitivity for visualizing neural afferents and efferents, J. Histochem. Cytochem., 26 (1978) 106-117. Ram6n y Cajal, S., Degeneration and Regeneration of the Nervous System (R.M. May, transl.), Harrier Press, New York, 1928, reprint 1957. Roederer, E. and Cohen, M.J., Regeneration of an identified central neuron in the cricket. I. Control of sprouting from soma, dendrites and axon, J. Neurosci., 3 (1983) 1835-1847. So, K.-E, and Aguayo, A.J., Lengthy regrowth of cut axons from ganglion cells after peripheral nerve transplantation into the retina of adult rats, Brain Research, 328 (1985) 349-354. So, K.-E, Xiao, Y.-M. and Diao, Y.-C., Effects on the growth of damaged ganglion cell axons after peripheral nerve transplantation in adult hamsters, Brain Research, 377 (1986) 168-172.
371
Elsevier BRE 23450
De novo formation of axon-like processes from axotomized retinal ganglion cells which exhibit long distance growth in a peripheral nerve graft in adult hamsters* Eric Y.E Cho and Kwok-Fai So Department of Anatomy, Facultyof Medicine, Universityof Hong Kong, Hong Kong (Hong Kong) (Accepted 27 December 1988)
Key words: Peripheral nerve graft; Optic nerve crush; Axon-like process; Retinal ganglion ceil; Hamster
Damaged axons in the central nervous system of the adult mammal can be stimulated to regenerate extensively into a peripheral nerve graft. It was generally believed that the new axonal sprouts which extend into the graft arose from the injured proximal axonal stumps. However, when retinal ganglion cells of the adult hamster were axotomized by crushing the optic nerve and the proximal axonal stump was not in direct apposition to the graft, a new axon-like process could be seen to be emitted from either the cell soma or dendrite and extended in the graft for at least 1-2 cm. This axon-like process was distinct from the original injured axon which could still be seen to course towards the optic disc in the retina. Evidently, even a fully differentiated central nervous system neuron of the adult mammal retains a great degree of morphological plasticity so that if the original axon is discouraged to regrow after injury, other parts of the neuron can act as favourable sites for the sprouting of a new axon-like process.
The limited and abortive regrowth exhibited by the majority of central nervous system (CNS) axons of the adult mammal after injury is well documented 2'8'12. However, transplantation of a peripheral nerve (PN) segment to the damaged CNS was found to promote extensive axonal regrowth into the graft 1. For instance, in the rat 14 and hamster 15, it was demonstrated that damaged ganglion cell axons would regrow into a PN implanted into the retina. In these previous studies it was generally believed that the injured proximal axonal stump regenerates directly into the graft. We have now observed that when retinal ganglion cell axons in the adult hamster are damaged but the injured proximal stump is not in direct apposition to the PN graft, a new axon-like process can arise either from the cell body or dendrite to extend into the graft and regrow for a very long distance in it. In adult golden hamsters 4 - 6 weeks old, an autologous PN segment (common peroneal branch of the sciatic nerve) 2-2.5 cm long was implanted
into the superior temporal quadrant of the right eye 14. At 1-2 months post grafting, horseradish peroxidase (HRP, Sigma, 50% in saline) was applied to the cut end of the graft transected at 1-2 cm from the eye. Eighteen to 20 h later, the animal was perfused with 0.9% saline, the retina was removed and fixed in a phosphate-buffered fixative containing 1% paraformaldehyde and 1.25% glutaraldehyde for half an hour. After reacting the retina for H R P histochemistry using tetramethylbenzidine as the chromogen 1~, a sector of retrogradely labelled ganglion cells peripheral to the grafting site in the retina could be seen (Fig. 1A), the axons of which were damaged by the grafting procedure and have regrown into the graft 14'~5. In the present study, in addition to PN grafting, the optic nerve (ON) of the eye which received the PN transplant was crushed at 1 m m behind the globe at the same time when grafting was performed. W h e n H R P was applied to the graft as described above, in addition to the population of labelled ganglion cells peripheral to
* A preliminary report of some of the results has been published 4. Correspondence: E.Y.P. Cho, Department of Anatomy, Faculty of Medicine, University of Hong Kong, Hong Kong, Hong Kong. 0006-8993/89/$03.50 © 1989 Elsevier Science Publishers B.V. (Biomedical Division)
372
,6,
B
/ D
j Fig. 1. A: distribution of HRP-labelled ganglion cells (dots) in the superior temporal quadrant of a retinal whole mount from an animal with a PN implanted into the retina (postgrafting time = 28 days). B: same as A but from an animal with concurrent ON crush of the same eye with PN grafting (postgrafting time = 43 days). In A. labelled cells were only observed peripheral to the graft whereas in B an additional population of labelled cells central to the graft could also be seen. Most of the cells from the central population were located less than 500 ~tm from the grafting site although occasionally a few cells could be seen as far as 1.5 mm away. Their numbers range from 7 to 30 (n = 5). C: schematic diagram of the superior temporal quadrant of a retinal whole mount from an eye with ON crush in addition to PN grafting. Cells a-f represented the labelled cells seen after HRP application to the graft. Cells a, b and c have their axons being damaged by the ON crush and are collectively referred to as the central population whereas cells d, e and f, the peripheral population, have their axons being interrupted by the graft and with the injured proximal axonal stumps in direct apposition to it. D: distribution of labelled ganglion cells in the retina from an animal with ON crush and PN grafting (28 days post grafting). Immediately before HRP application to the graft, an incision (L) was made in the retina at a region between the graft-implantation site and optic disc. HRP was then applied to the cut end of the graft at 1 cm from the eye and the animal processed as in (A) and (B). G, graft-implantation site: X. optic disc. Scale bar. 1 mm (same for A. B and D).
t h e graft, l a b e l l e d g a n g l i o n cells which lay in a z o n e
a c c o m p a n i e d by O N crush, w e k n o w that t h e i r axons
b e t w e e n t h e grafting site and optic disc as well as s o m e which s u r r o u n d e d the grafting site c o u l d also be o b s e r v e d (Fig. 1B). Since this a d d i t i o n a l p o p u -
w e r e not i n t e r r u p t e d by the g r a f t i n g p r o c e d u r e but r a t h e r w e r e d a m a g e d by the O N crush. W e d e s i g n a t e t h e s e cells as the central p o p u l a t i o n ( a l t h o u g h n o t all
lation o f cells was o n l y s e e n w h e n P N grafting was
o f t h e m lay c e n t r a l ( t o w a r d s t h e optic disc) to t h e
373 graft, see Fig. 1C) to distinguish them from that of the peripheral population which have their axons being intercepted by the graft and with the injured proximal stump in close apposition to it. In accordance with a previous study tS, it was originally interpreted that the axons of ganglion cells from the central population looped back into the retina after their severance in the ON and extended into the PN graft. Therefore, if immediately prior to application of HRP, a large incision was made in the retina between the grafting site and optic disc in a manner which would have severed the axons of ganglion cells from the central population before they entered the graft, the central population should fail to become labelled if their axons followed the course as described above. However, when animals (n = 4) with PN grafting accompanied by ON crush were treated in the above manner, the central population of labelled cells still persisted (Fig. 1D). This strongly suggests that other processes distinct from the originally injured axon in the ON have extended into the PN so that the central population could still be labelled even when the axon was interrupted immediately before HRP application to the graft. The detailed morphology of the regenerating ganglion cells was studied in order to see whether any other processes distinct from the original axon have sprouted into the PN graft. We adopted a reduced silver staining method on whole-mounted retina which was found to stain the regenerating ganglion cells (including the dendritic tree and axons) intensely. At 1-2 months post grafting, ganglion cells which have their axons extended into the graft were first labelled with a fluoresent dye (Fast Blue or Granular Blue, 3% solution in saline) applied to the graft at 1-2 cm from the eye. After a survival time of 4 days, the retina was dissected out and fixed in 2% paraformaldehyde (in phosphate buffer pH 7.4) for 1 h and then mounted in glycerol. Labelled ganglion cells were viewed under epifluorescence and their distributions recorded by photography. The retina was then fixed in 10% formol saline for 1 h and silver-stained according to the protocol of Leicester and Stone 9. Using this method, in the normal adult hamster retina, only the axons were revealed but the ganglion cell somata and the dendrites were not stained (except for some weak
staining associated with a few ganglion cells with the largest soma size). However, ganglion cells regenerating axons into a PN graft were found to be intensely stained, revealing the detailed patterns of their dendritic trees. The positions of the silver stained cells were compared with that of the fluorescent dye-labelled cells to identify those which have regenerated axons or processes into the graft. In retinas which were optimally stained with this silver method, we consistently observed some ganglion cells in the central population which had been labelled with the fluorescent dye (and hence have regenerated into the graft) to possess an axon-like process clearly distinct from the existing axon. Altogether we analysed 38 cells from 9 retinas with this type of peculiar morphology, 9 of which could not be shown to be labelled by the fluorescent dye. In 4 of these 9 cells (e.g. cell 3 in Fig. 2), the region of the retina where they were located was not photographed at the time when the positions of the fluorescent dye-labelled cells were being recorded. In others, the process may have extended into the graft but have not reached the application site of the dye. However, since the morphology of these 9 cells was similar to the other 29 (which were labelled by the fluorescent dye), we felt justified to include them in our analysis. The extra process originated most commonly from either the celt body (55% of the total of 38 cells) or a primary dendrite (26%), and in rare cases from a secondary (13%) or even higher order (5%) dendrite. An example of a retina from an animal which received PN grafting together with ON crush and examined at 56 days post grafting (Figs. 2 and 3) served to illustrate the appearance of these processes and their trajectories in the retina. Cells 1, 2, and 3 belonged to the central population since their axons were not interrupted by the graft and could still be seen running to the optic disc. In each cell, however, a separate process could be seen to arise from the cell body and extend to the grafting site in the retina. These processes were unbranched during their whole intraretinal course (although in 3 out of the 38 cells studied one or two very fine side branches could be seen to arise from the process), maintained a fairly uniform diameter without the tapering commonly associated with dendrites and thus resembled an axon in external appearance. Once they were emit-
374
A
B G
P
~
ODe
Fig. 2. A: camera lucida drawing of 3 silver-stained ganglion cells of the central population from an animal with ON crush and PN grafting (56 days post grafting). Cells 1 and 2 were labelled by the fluorescent dye Fast Blue applied to the graft (see Fig. 3)i In all 3 cells, an unbranched axon-like process (p) distinct from the axon (a) could be seen to arise from the cell body and extend towards the graft-implantation site (G) whereas the axon headed in the direction of the optic disc (OD). The process of cell t had a relatively uncomplicated trajectory to the graft but that of cells 2 and 3 took a tortuous course, displaying a series of loops along its way. In fact, 47% of the 38 cells we studied had the axon-like process exhibiting some sort of loops along its course to the graft and 53% of the cells had the process following a relatively undeviated intraretinal pathway towards the graft. Scale bar, 100/am. B: detailed tracing of cell 1 showing the origin of the axon-like process from the cell body. In some cases like in this cell, the process was slightly thicker than the axon but in others the two had comparable diameters, a, axon; p, process to graft. Scale bar, 50/tm.
ted from the ganglion cell, these processes would either head straight towards the grafting site with only minor deviations from the main direction as in cell 1, or they could take complicated and tortuous courses in the retina before reaching the graft finally (cells 2 and 3). In most instances the axon-like processes were confined to the inner plexiform layer (IPL) throughout its intraretinal course to the graft (74% of the 38 cells), while a small number of them (22%) originally residing in the IPL m o v e d up to the nerve fibre layer as they approached the grafting site. In this respect it is different from the axon which is always located in the nerve fibre layer. W h e n e v e r regenerating ganglion cells in the central population were found to possess such extra axon-like processes, only one could be observed to arise from each cell and no matter how tortuous their initial trajectories were, their final destination would invariably be the grafting site. However, not all ganglion cells from the central population were shown to exhibit this morphological peculiarity when examined by the silver staining method. This is because the majority of regenerating ganglion cells
from the central population tended to be located in a region near to the graft where a great multitude o f dendritic trees and the associated ganglion cell bodies would be stained, making the definite identification of any extra axon-like processes an extremely difficult task. Therefore only cells located about 300/~m or more from the grafting site could be reliably analysed but at this distance and beyond the number of labelled cells of the central population was very small. We also observed some cells from the central population with no axon seen running to the optic disc but only an axon-like process extending into the graft. Since this process coursed in the IPL rather than the nerve fibre layer (the normal location of ganglion cell axons), we suspected it was in fact newly generated to extend into the graft whereas the orginal axon had degenerated after ON crush. Thus, although we do not see the presence of extra axon-like processes occurring in the majority of cells in the central population in the silver-stained preparations due to the above mentioned factors, it seems logical to conclude that many of the regenerating cells in this population possess them if we
Fig. 3. A: photomicrograph showing the 3 silver-stained cells in Fig. 2. The axons of the 3 cells (1, 2 and 3) were traced by arrows. The axon-like process of cells 1 and 2 were indicated by asterisks (*) while the course of the axon-like process of cell 3 was followed with arrowheads. G, graft-implantation site; OD, optic disc. Scale bar, 100 pm. B: photomicrograph at a higher magnification showing cells 1 and 2. In both cells the axon (a) and the process (p) to the graft arose at a site on the soma distinct from each other. Scale bar, 50 pm (same for B and C). C: cells 1 and 2 were retrogradely labelled by the fluorescent dye Fast Blue applied to the graft at 1.5 cm from the eye and photographed before silver staining was performed.
376 also take into account the result of the experiment of making a lesion between the optic disc and grafting site before H R P application to the graft. What is the signal for inducing the formation of these axon-like processes? Some clues may be gained by considering the situation in animals which received a PN graft but with no ON crush. We have examined the morphology of over 100 regenerating ganglion cells in these animals at 2-5 weeks post grafting using the same silver staining method and not one was seen to possess any extra axon-like processes. Since the axons of these ganglion cells were intercepted by the graft and directly opposed to it after injury, the damaged proximal axonat stump could presumably respond to the favourable PN environment with little delay to regrow into it 3. In contrast to this, the central population of ganglion cells (in animals with ON crush and PN grafting performed together) would have their injured proximal axonal stump residing in the ON where the environment for regrowth was certainly not optimal. However, it can be envisioned that some sort of trophic influence emanating from the PN has already signalled the cell body to turn on its cellular machinery required for the initiation of axonal regrowth, such as the synthesis of new membrane constituents or growth associated proteins 5. If the proximal axonal stump in the ON cannot act as a suitable target for responding to or incorporating the newly synthesized constituents they may be diverted to other regions of the cell such as the soma or primary dendrite to generate new axon-like processes which can then be attracted towards the PN graft. The viability of the proximal axonal stump after injury has been suggested to be important for the distribution of sprouting over the neuronal surface 6'13. The generation of supernumerary axons in spinal motoneurons of the cat after their axons in the peripheral nerve have been damaged and pre-
vented to regenerate may also be explained similarly 7. It seems that even mature neurons in the mammalian CNS like the retinal ganglion cell still possesses a remarkable degree of morphological plasticity such that the basic neuronal geometry can be profoundly altered as it regenerates after injury. Other examples include the sprouting of axon-like processes from the dendrites of axotomized motoneurons in the cat 1° or formation of a supernumerary axon from the cell body of the motoneuron after a distal axotomy in the cat 7. Our results also indicate that these novel axon-like processes are attracted to grow towards and into the PN graft, unlike the supernumerary axons of cat motoneurons which exhibit variable trajectories in the spinal cord 7. Moreover, similar to injured axons regrowing in a PN graft t, the axon-like processes observed in our present study can also elongate for a considerable distance (1-2 cm, the distance from the eye at which HRP or the fluorescent dye is applied to the graft) in the graft. The fact that not more than one axon-like process is generated suggests that the intrinsic mechanisms regulating its formation may be similar to that operating in the differentiation of ganglion cells to the mature state whereby only one axon is formed. Even then, however, the ganglion cell now faces the problem of having to sustain apparently 2 axons which is not in keeping with the rules of its developmental program. Therefore it remains to be seen whether the original damaged axon will ultimately be eliminated and if the cellular events which occur in a normal axon such as the transport of synaptic vesicles and propagation of action potentials are still operating in the damaged axon which has resisted degeneration.
1 Aguayo, A.J. Axonal regeneration from injured neurons m the adult mammalian central nervous system. In C.W. Cotman (Ed.), Synaptic Plasticity, Guilford Press. New York, 1985, pp. 457-484. 2 Berry, M., Regeneration of axons in the central nervous system. In V. Navaratnam and R.J. Harrison (Eds.), Progress in Anatomy, Vol. 3. Cambridge Univ. Press. 1979. pp. 213-233. 3 Cho, E.Y.P. and So, K.-E, Rate of regrowth of damaged retinal ganglion cell axons regenerating in a peripheral
nerve graft in adult hamsters. Brain Research. 419 (1987) 369-374. 4 Cho, E.Y.P. and So, K.-F.. Growth of newly formed neuronal processes into a peripheral nerve graft stimulated by axotomy of retinal ganglion cells in the adult hamster, Hong Kong Soc. Neurosci. Abstr.. 10 (1988) 56. 5 Grafstein. B.. The retina as a regenerating organ. In R. Adler and D.B. Farber (Eds.), The Retina: a Model for Biology Studies. Part 2. Academic Press. New York. 1986. pp. 275-335.
This study was supported by the Croucher Foundation of Hong Kong.
377 6 Hail, G.E and Cohen, M.J., Extensive dendritic sprouting induced by close axotomy of central neurons in the lamprey, Science, 222 (1983) 518-521. 7 Havton, L. and KeUerth, J.-O., Regeneration by supernumerary axons with synaptic terminals in spinal motoneurons of cats, Nature (Lond.), 325 (1987) 711-714. 8 Kiernan, J.A., Hypotheses concerned with axonal regeneration in the mammalian nervous system, Biol. Rev., 54 (1979) 155-197. 9 Leicester, J. and Stone, J., Ganglion, amacrine and horizontal cells of the cat's retina, Vision Res., 7 (1967) 695-705. 10 Lind,, H.. Risling, M. and Cullheim, S., ~Dendraxons' in regenerating motoneurons in the cat: do dendrites generate new axons after central axotomy?, Brain Research, 358 (1985) 329-333. 11 Mesulam. M.-M., Tetramethyl benzidine for horseradish peroxidase neurohistochemistry: a non-carcinogenic blue
12
13
14
15
reaction product with superior sensitivity for visualizing neural afferents and efferents, J. Histochem. Cytochem., 26 (1978) 106-117. Ram6n y Cajal, S., Degeneration and Regeneration of the Nervous System (R.M. May, transl.), Harrier Press, New York, 1928, reprint 1957. Roederer, E. and Cohen, M.J., Regeneration of an identified central neuron in the cricket. I. Control of sprouting from soma, dendrites and axon, J. Neurosci., 3 (1983) 1835-1847. So, K.-E, and Aguayo, A.J., Lengthy regrowth of cut axons from ganglion cells after peripheral nerve transplantation into the retina of adult rats, Brain Research, 328 (1985) 349-354. So, K.-E, Xiao, Y.-M. and Diao, Y.-C., Effects on the growth of damaged ganglion cell axons after peripheral nerve transplantation in adult hamsters, Brain Research, 377 (1986) 168-172.