What constrains growing nerves to follow the paths they take during the development of peripheral nerve patterns? This paper examines two, related, topics concerning the pathways taken by sensory nerve fibres in the embryo chick wing: the constraints imposed on the nerves by limb tissues; and the timing of axon outgrowth. Sensory ganglia from 7-day-old chick embryos were grafted into younger host embryo wing buds which had been previously denervated. The resultant nerve patterns revealed that, first, nerve fibres could grow almost anywhere within the wing bud, with the exceptions of cartilage and a region just beneath the growing tip. Secondly, the younger the host wing bud at the time of grafting, the more likely the neurites were to form a thick fascicle which followed the limb’s normal nerve pathways. The wing apparently does not impose a rigid restraint on nerves to grow only along certain routes; however, if a nerve fibre reaches a normal nerve pathway, it prefers to follow it.
The wing nerve anatomy of chickens forms in a reproducible manner over a relatively short period of time during early development. The timing of the formation of this pattern is governed by the age of the tissues through which the nerves grow (Swanson & Lewis, 1982); and these tissues also determine the geometry of the nerve pattern. For example, cranial nerves growing into a wing bud grafted onto the jaw take on the nerve anatomy appropriate to the wing (Swanson & Lewis, 1982). The routes taken by nerves in limbs have been thus compared to ‘public highways’ defined by the limb tissues irrespective of the source of the nerves (Lewis, Al-Ghaith, Swanson & Khan, 1983).
This paper correspondingly tackles two main questions concerning the nerve highways. The innervation of limbs by nerves from foreign sites has shown that nerves are apparently restricted to follow set routes in the limb: but is this really a rigid constraint, or can nerves grow ‘across country’ away from the highways if the starting conditions are altered? The other question concerns constraints on the timing of axon outgrowth. Are the highways permanent once constructed, remaining capable of guiding nerve fibres? Or do the highways exist only for a short time to guide the first, pioneer fibres, with the fibres that grow into the limb later sticking to and being guided by those already present?
The experiments presented here were designed to try to examine these possibilities. Operations on embryo chicks were carried out in two steps. First, the normal nerve supply of embryo wings was destroyed (by ultra-violet irradiation of the early neural tube). Then, since no other histological marker of the highways is yet available, permitted routes for outgrowth were assayed by grafting foreign nerve cells into the uninnervated limb bud at a later stage. Specifically, sensory ganglia dissected from 7-day-old chick embryos were implanted into host wing buds a few days after the irradiation procedure. A wide range of host embryos was chosen, between stages 21 and 28, spanning the time at which nerves normally enter the limb and first begin to form the recognizable standard pattern. Host embryos were fixed 1 to 3 days later, and silver stained to reveal nerves that had grown out from the implanted ganglia.
The results of these double operations show that the ganglia survive and that neurites grow out from them into the uninnervated wings. Neurites were found to grow in regions of the limbs not normally encountered by nerves. In other words, the wing does not impose a rigid restraint on nerves to grow only along the usual highways. However, if a nerve fibre reaches a highway, it appears to prefer to follow it, fasciculating with other fibres to form a trunk-like bundle. This preference for remaining on highways is found especially in fibres growing distally, towards the tip of the wing; fibres growing proximally, towards the shoulder, seem to be shorter and more disorganized.
With reference to the question as to how long the highways are available as guides for growing axons, the interpretation of the results is not clear cut. Generally, the younger the age of the host wing, the more likely that distal-growing neurites will form a fascicle following the limb’s normal nerve trunk highways. And the closer to a highway a ganglion is grafted, the more likely it is that the normal path will be taken.
MATERIALS AND METHODS
Fertilized chicken eggs (White Leghorn × Sykes Tinted, from Needle Farm, Elstree, Herts.) were incubated at 38° ± 1°C and windowed at about stages 12 to 14 (host embryos) or about stage 32 (donors). The vitelline and amniotic membranes were carefully torn to expose the embryos, which were then staged according to the timetable of Hamburger & Hamilton (1951). The neural tube of the intended host embryos was irradiated over a length extending from somite 9 to at least somite 22 with a focussed beam of ultra-violet light from a high-pressure mercury discharge lamp (see Swanson & Lewis, 1985, for details). The beam was positioned so that only the neural tissue was irradiated, avoiding the adjacent somite blocks. The exposure time of 15 to 20 seconds was determined by pilot experiments to be that required to destroy the neural tube totally, while doing as little damage as possible to the underlying and surrounding tissues and major blood vessels. After the embryos had been irradiated a few drops of Hepes-buffered Hanks balanced salt solution with antibiotics (penicillin; 50–100i.u./ml; streptomycin; 50-100Mg/ml) were added to the eggs before the windows were resealed, and returned to the incubator until the next stage of the experiment.
Donor embryos at stage 32 were quickly removed from their eggs, rinsed in 0-9 % saline and placed in a saline bath at room temperature for dissection. The head and viscera were removed and the sensory ganglia exposed from the ventral side by carefully removing the overlying connective tissue and muscles with fine forceps. To facilitate removal of the ganglia, the legs were pulled gently away from the body. Ganglia from brachial, thoracic, lumbar and sacral levels were used in the experiments. No obvious differences in the results were seen with respect to these different sources; the ganglia from the lumbar levels were used most often, since these are the most numerous and the easiest to dissect out. The isolated ganglia, with only short lengths of connecting rami, were kept in either 0-9 % saline or medium 199 (Gibco) at room temperature or 38°C until required for grafting, which was normally done no more than one and a half hours after removal.
The irradiated (aneural) host embryos (between stages 21 and 28) were prepared to receive a graft by making a short slit in their right wing bud with sharpened tungsten needles, taking care not to penetrate through the full depth of the limb bud. A ganglion was then simply pushed into the slit; the pressure of the tissue next to the wound was generally sufficient to hold the graft in place, although some grafts needed to be secured by a fine platinum wire pin.
The ganglia were each grafted into one of three standard positions in the host limbs, but the exact position in which they ultimately came to lie was variable. Three general ‘parking areas’ were noted. Two of these, the wrist and elbow, are regions whose rudiments are at first relatively large but subsequently grow much less than the neighbouring portions of the limb bud. The third ‘parking area’ was in the upper arm, in the region of the triceps muscle. Within each of these areas the exact position of the graft was variable. For example, in the elbow region some of the ganglia were to be found between the proximal ends of the radius and the ulna, some near the posterior margin of the wing, and some more distal, towards the middle of the forearm. The positions of ganglia in the wrist and upper arm were similarly variable.
Host embryos were then returned to the incubator for one or more days until they had reached stages 28 to 32, when they were fixed and processed for silver staining according to the procedure described previously (Lewis, 1978; Lewis, Chevallier, Kieny & Wolpert, 1981). Specimens were examined under a dissecting microscope (Zeiss) and drawings made of the resultant nerve patterns with the aid of a camera-lucida attachment (Zeiss). Some specimens were subsequently processed for routine wax-embedded histology.
The abbreviated names of nerve branches referred to in the present paper are as follows:
DC Al, dorsal cutaneous, alar web; DC Elb, dorsal cutaneous, elbow; DC Int, dorsal cutaneous, interosseous; DC prox uln, dorsal cutaneous, proximal branch along ulnar border (this is synonymous with the proximal branch of n. radialis lateralis) ; UMD, muscle nerve branch to ulnimetacarpalis dorsalis.
A total of 229 embryos was irradiated and received ganglion grafts. A total of 112 embryos (49 %) survived both operations, of which 61 (54 % of the survivors, or 27 % of the initial total) were stained well enough to reveal that neurites had grown from the grafted ganglia and that there was no residual host innervation, or very little. The presence of residual innervation in some host wings was probably due to slight variations in the relative positions of the embryos and the irradiation apparatus; the lamp and its optics had to be set up and aligned at the start of each series of irradiations, and the size and shape of the image of the ultra-violet beam had to be tailored to fit each embryo. In the set of specimens selected for analysis, the residual host innervation, when present, extended only into the most proximal region of the limbs (usually only the triceps muscle nerve branch and its accompanying cutaneous nerve branch DC Elb were to be seen – Fig. 1A), and in no limb of the set selected did fibres from the host encounter those from the graft, nor is it likely that the distal parts of the limb had received even transitory innervation by host fibres.
The range of combinations of stages of the embryos at the grafting and fixing steps of experiments is given in Table 1, along with the numbers of embryos of each combination used in the analyses of results. Embryos are designated by a pair of numbers, ‘H/F’, to identify to which group they belong; the first number, ‘H’, represents the stage at which the embryo was used as a host to the implanted ganglion, and the second number, ‘F’, represents the stage at which that embryo was fixed. It can be seen from the table that some ‘H/F’ combinations are represented by only one successfully denervated and well-stained host limb.
Because of the thinness of the neurites and their three-dimensional pattern of growth, clear photographs were difficult to obtain and so most specimens were drawn under the dissecting microscope with the aid of a camera-lucida attachment. This projects the volume of tissue through which the neurites are growing onto a two-dimensional sheet, but at the same time allows one to record all the visibly stained fibres. However, the level of magnification attainable with the dissecting microscope (up to × times 50) did not always allow the resolution of single fibres; when checked at higher magnifications using a compound microscope ‘single’ fibres often revealed themselves as small bundles of two or more neurites. Single lines in the drawings do not, therefore, necessarily represent single nerve fibres. Analysis of the patterns is restricted mainly to those nerve fibres growing through the deep tissues of the limbs since it is here that the pattern is most stereotyped and reproducible in normal wings; cutaneous nerve branches, although they normally diverge from the main deep nerve trunks at specific places, do not have very specific terminal branching patterns. Moreover, since in the experimental limbs the majority of grafts sent out neurites which were restricted to growing on the dorsal side of the wings, the analysis is of the dorsal patterns. From the few cases where fibres grew on the ventral side, however, it appears that the conclusions for the dorsal part of the pattern are applicable to the ventral side also.
Effects of the ultra-violet irradiation
Figure 1 shows the effects of the ultra-violet irradiation procedure on the gross structure of embryos. In the particular silver-stained wing of such an embryo shown in Fig. 1A, a slight residual innervation from the host is visible, demonstrating the effectiveness of the staining procedure. No wing nerves were seen in the majority of specimens.
Figure IB shows the effect of a completely successful irradiation as seen in cross section through the trunk. The surface ectoderm has healed over and the spinal cord is missing. The underlying notochord seems not to have been damaged and adjacent somite blocks are still intact. Compare this with Fig. 1C which depicts a section of the spinal cord from a similar level in a normal embryo.
Histological appearance of grafted ganglia
Figure 2 shows the appearance of a ganglion in a wing which was fixed for sectioning 60 h after grafting. The ganglion shows many stained nerve cell bodies, as well as many nuclei of surrounding satellite cells. The ganglion itself shows little sign of mixing with the surrounding host tissues. There were also no obvious signs of degeneration, although the relative proportion of nerve cell bodies seems less than in intact ganglia.
Directions and routes of outgrowth of neurites from implanted ganglia
Because of the variability of the location of ganglion parking areas, every limb may be thought of as a unique case; but each limb in a given ‘H/F’ group presents a pattern of neurites, of which the overall form is comparable to and not much different from that of other limbs whose grafted ganglia had come to lie in the same general area. Figures 3A and 3B show two wings from the group 24/32, in which the neurites had approximately 72 h to grow out from the time the ganglion was grafted to the time the wing was fixed, and the ganglia had come to lie just distal to the elbow near the posterior edge of the wing. The neurite patterns, though different in their details, show some general similarities. First, most of the fibres have grown distally towards the wrist and hand; fewer fibres have grown proximally into the upper arm and, in the limb depicted in Fig. 3A especially, the proximally directed fibres appear to make a more diffuse pattern than those which have grown distally. Secondly, the neurites which have grown out distally have initially grown from the grafted ganglia as a widely fanned array, but these loosely associated fibres have gradually come together to form a thick fascicle which lies along the path taken by the dorsal nerve n. radialis profundus in the normal wing (Fig. 3D).
The tendency to follow normal highways seen when neurites grow out distally from ganglia in the elbow region as described above, is in contrast to the behaviour of neurites which grow out from ganglia which come to lie in the wrist area. Fig. 3C shows one such case, from the group 26/31, for which about 54 h elapsed between grafting and fixing. In this limb it can be seen that neurites have grown out from the ganglion in many different directions — there is no indication of the direction of the original severed roots. Neurites have grown through cutaneous tissue as well as deep within the limb. A feature of this specimen, in comparison to those described above, is that most of the neurites appear to have grown further proximally, into the forearm, than distally, into the hand. Another striking feature is that, although fibres can be seen to grow side by side, separated by only a few pm, they have not formed tight fascicles of many fibres. Furthermore, in all limbs examined, it was apparent that, no matter how extensive neurite growth was in proximal levels, there were no fibres to be seen in the most distal region, to a depth of about 200–300 μm just beneath the distal tip of the wing.
Extent and rate of outgrowth of neurites
Figures 4A,B show the neurite patterns of two limbs, both of which had received a graft at stage 26, but which were fixed at different stages, 28 and 32. That is, the limbs were left for about 20 and 60 h respectively. After 20 h only a few short neurites can be seen. After 60 h more neurites have grown out from the graft, and to greater distances. Fig. 4C illustrates this graphically for different lengths of time and different embryos. The distance travelled by the longest visible neurite was measured in each of 26 wings and plotted against the time elapsed between grafting and fixing. The 26 wings selected were those in which the staining was sufficient to be reasonably certain that the full extent of the neurites had been stained. At least one limb was selected from each of the different ‘H/F’ combinations and the ganglia were found in all the different positions described, so that there are many sources of variation besides the time elapsed. Nonetheless, the data points show a significant correlation (r = 0·88, d.f. =24, P< 0·001) between time and distance. The rate of neurite extension, estimated by the slope of the line, is about 0·72 mm per day. There appears also to be a delay of about 10 h before any neurite extension begins, presumably due to some trauma associated with the operation.
Timing of formation of the ‘highways’
If, as reasoned earlier, highways are constructed by the limb for nerves to grow along, can we discover anything about how long the highways remain available as a guiding influence for nerves? Are the highways durable once constructed, and available subsequently for nerve fibres to grow along even in cases where the pioneer growth cones were prevented from making their entry at the normal time? To try to answer this question, the neurite patterns of limbs which received grafted ganglia at progressively older stages were examined.
Figure 5 shows the appearance of the paths taken by neurites from ganglia grafted into limbs at a series of different stages, from stage 21 to stage 27, and left for at least 48 h before fixing. There seems to be a general trend such that the later the ganglion is implanted, the less closely do its neurites approximate to the normal pattern. This point can best be made by contrasting individual specimens.
Figure 5A shows the appearance of a 21/29 limb, that is, a stage-21 host fixed at stage 29. The graft has come to rest in the upper arm and has put out fibres through both the deep and the cutaneous tissues of the limb. The deep nerves distal to the graft have followed closely the path taken by the limb’s normal main dorsal nerves. The fibres have formed a tight fascicle along the normal route of n. brachialis superior in the upper arm (compare Fig. 3D), from which two cutaneous branches closely resembling DC Al and DC Int have diverged and grown down into the forearm. The main bundle of fibres meanwhile continues distally, branching at the elbow to follow along the normal route of n. radialisprofundus in the forearm, and sending off a side branch towards the ulnar border of the wing, corresponding to n. radialis lateralis. On reaching the wrist the fibres following the route of n. radialis profundus fan out just as in a normal wing. In addition, numerous small fibres are seen to grow in other directions from the ganglion, but none travel very far, especially those growing proximally, towards the shoulder. In all, this specimen presents a neurite pattern which closely resembles that of a normal wing, except for the absence of any clear muscle nerve branches.
At the opposite extreme, Fig. 5E shows a limb of the 26/31 group. In this case neurites have grown out from the grafted ganglion in the proximal forearm. In the deep fibres illustrated there is no evidence of any tight fasciculation. Rather, fibres can be seen to have grown out roughly in parallel, and in proximity to one another, but for the most part not in contact. Although the neurites have not grown in a tightly fasciculated manner, some of those which have travelled distally and reached the wrist have begun to fan out towards the posterior side in much the same way as do normal nerve fibres. Figures 5B, C and D illustrate intermediate cases, that had received their ganglion grafts respectively at stages 22, 23 and 24.
Figures 5C and D, where the grafts have ended up in the elbow region, illustrate two further points: first, the neurites which have grown proximally towards the shoulder have not gone so far as those growing in the opposite direction; secondly, these proximally directed neurites have not formed any tight fascicles at or near those positions where a nerve trunk would normally be found. Neurites originating from ganglia at the wrist level, such as those shown in Fig. 3C, show contrasting behaviour in the first respect, but similar behaviour in the second: the neurites from the wrist level ganglia have grown mostly in a proximal direction, but in no case have these proximally directed neurites formed a tight fascicle; rather, fibres grow alongside one another, separated by 10 to 20 μm, meeting to form small bundles and then separating again as they grow up the forearm between the radius and ulna. Unfortunately, none of the grafts into host younger than stage 25 was found in the wrist region, so the behaviour of fibres at the earlier stages cannot be assessed.
Other specimens (not illustrated) of the same ‘H/F’ combination confirm these general conclusions. In general, it seems that the older the host is at the time of grafting the less tightly fasciculated and the less ‘normal’ in appearance will be the resultant pattern of neurites. If grafted later than about stage 25–26 the neurites tend not to follow the limb’s normal nerve highways. Neurites growing proximally tend not to form thick fascicles and not to follow the normal nerve highways.
Muscle nerve branches
Evidently, grafted sensory neurites can grow out along the limb’s normal nerve highways (as well as along abnormal paths), but do they also branch off from the nerve trunks to innervate individual muscles? Specifically, do they form muscle nerve branches at the expected places? The answer appears to be no. Muscle nerve branches do not form from the trunks made by the sensory nerve fibres. However, although no neurites were found to grow into muscles in this way, they were found to grow through muscle tissue (Fig. 6). These neurites need not have grown into already formed muscles, but are more likely to have grown through the premuscle mass before muscle differentiation.
Sensory ganglia from 7-day-old chick embryos grafted into denervated wing buds of 4- to 6-day-old chick embryos survive and send out neurites from highly abnormal starting points into the developing host wing tissues. These sensory neurites can grow into and through both deep and cutaneous tissues of the wings and they can do so in the complete absence of any motor innervation. The neurites appear to be almost unrestricted in the paths along which they can grow; they do not grow only along the paths taken by nerves in a normal wing. If neurites reach tissues which do lie on the paths of the normal nerves, they prefer to continue along these paths and often fasciculate with other neurites to form a bundle which resembles closely a normal nerve trunk. Although fibres from the grafted ganglia can form these bundles, and can innervate the cutaneous regions of limbs, branches to individual muscles from the main trunks are not seen.
Evidently, the normal nerve guidance system does not represent an absolute restriction on the paths that sensory neurites may take: these are capable of growing out along other routes also. Nevertheless, the neurites do appear to be debarred from some paths. A major obstacle is the cartilage of the skeleton (Thurston, 1982). In particular, the radius appears to obstruct the paths of neurites which have come from grafted ganglia located in the posterior half of the wing. These fibres never grow through the skeletal element nor do they grow round it to innervate, for example, the deep tissues of the alar web. Fibres growing cutaneously, on the other hand, do not experience any such obstructions; as in normal development, they can extend almost anywhere in the skin.
Another region of the wing bud from which nerves are apparently prohibited from growing is a zone about 200–300 μm behind the tip. Nerves do not enter this area in either normal or experimental wings. Up to about stage 29, this region contains mesenchymal cells maintained in a special undifferentiated state by the influence of the apical ectodermal ridge (Summerbell, Lewis & Wolpert, 1973). The ridge itself also maintains a shallower region of avascular tissue about 100 μm thick directly beneath it (Feinberg, Repo & Saunders, 1983). The inability of nerve fibres to invade the deep sub-tip zone is similar to the inability of myogenic cells to migrate into it in the stage-25 wing bud (Newman, Pautou & Kieny, 1981). This evidence points once again to the limb tissues being a major factor controlling the development of the patterns of the other, invading, tissues.
Although neurites can grow elsewhere in the limbs, if they reach a normal highway they seem to prefer to follow it distally, fasciculating with other neurites along the same route. Furthermore, as well as offering an environment favourable to neurite growth, the tissues lying between the radius and ulna appear to favour fasciculation of fibres growing distally; this was the region in which the largest bundles of neurites were found, along roughly the normal highway towards the wrist. The degree of fasciculation along this path appears also to depend on the direction of neurite growth: deep neurites growing proximally from ganglia in the wrist do not show so strong a tendency to fasciculate as those growing distally from the elbow, although they grow far enough and close enough to one another to do so. Proximally growing neurites seem unable to recognize or follow a highway, whereas once distally growing neurites encounter a highway it seems particularly difficult for them to diverge from it. The mechanisms of fasciculation however, seem not as efficient as in the normal development of nerve trunks. In general, neurites tended to form only small bundles when growing through tissues not normally encountered, as opposed to the bigger bundles that grew only along the standard nerve routes in a normal wing. After various transplantations of amphibian limbs, Piatt (1942, 1952) and Deck (1955) also reported the preponderance of smaller nerve bundles over larger nerves.
Given that the limb appears to provide cues which normally guide nerve fibres along standard routes, but which are not apparently obligatory, for how long are these cues available to the neurites? In order to guide the pioneer fibres the routes must be present at early stages, when the pioneers grow out. The cues need not in principle really be present any longer than the time it takes to guide the pioneers, since the pioneers can guide subsequent axons by fasciculation. The results seem to indicate that the guidance cues are, indeed, present only transiently, for about a day. The interpretation is complicated by the fact that the implanted ganglia were found in different positions and that a wide range of host stages were used. However, if the sequence of limbs illustrated in Fig. 5 is examined, it will be seen that, in limbs in which the ganglia were found in, or near, the elbow (Figures 5 C,D,E and F), the older the host the less the neurite patterns resemble the normal. Neurites growing from ganglia grafted into the elbow region after about stage 26 appear not to follow highways. This suggests that in progressively older host wings neurites growing out from the same position experience conditions which are less and less likely to allow them to grow along the routes taken by the limbs’ normal nerves. From this it is inferred that, although the neurites are capable of growing long distances in the older wings, the highways become progressively less well defined.
Given that sensory neurites in the absence of motor axons can grow along paths resembling normal highways in some cases, and can diverge from those highways in a more or less normal fashion to form cutaneous nerve branches, why is it that none of the sensory neurites travelling along deep bundles were seen to form muscle nerve branches? Although the neurites seem not to be barred from growing through muscle tissue (Fig. 6), it appears that the sensory innervation of muscles needs to be guided into the muscles from nerve trunks by motor axons. The same dependence is demonstrated in wings innervated by normally located sensory ganglia but deprived of motor innervation by destruction of the ventral part of the neural tube (Swanson, 1983; Swanson & Lewis, in preparation).
I thank Dr Julian Lewis for his encouragement, guidance and advice. King’s College, London provided a Tutorial Studentship.