Peripheral nerves travel to their targets along precise routes, and it is likely that different cues provide guidance at different stages of the journey. In a develop­ing chick limb, the cutaneous nerve fibres follow at first deep mixed nerve trunks, in company with motor axons; they branch from these trunks at predictable points and approach the skin; they then ramify profusely to form a plexus at a precisely defined depth beneath the ecto­derm, at exactly the same level as the blood vascular plexus. To analyse the role of signals from the target patch of skin in regulating cutaneous nerve develop­ment, we have ablated patches of dorsal wing ectoderm using short-wave ultraviolet irradiation at E4 (embry­onic day 4), approximately one day before nerves grow into the limb bud. The irradiated patches remain denuded of ectoderm for more than a week, by which time the cutaneous nerve plexus on the contralateral control side is well developed and can be revealed by whole-mount silver staining. Where the ectoderm has been ablated, no cutaneous nerve plexus forms, and the nerve branches that normally would have diverged from the neighbouring mixed nerve trunk to innervate the missing patch of skin are absent - ab initio, apparently. The routes of the mixed nerve trunks are not affected. Partial ablation of the territory of a cutaneous nerve branch often leads to loss of the whole nerve branch; the intact skin territory thus left vacant is invaded by ramifications from the remaining cutaneous branches, as expected if the normal extent of a cutaneous nerve’s territory is regulated by competition. Where there is an ectodermal lesion, cutaneous innervation stops precisely at its boundary, even though the vascular plexus extends for some distance beyond this margin, beneath the denuded surface. The data suggest that the embryonic skin is required firstly to trigger divergence of cutaneous nerve branches from the mixed nerve trunks, and secondly, once the nerve fibres have reached the skin, to supply a trophic cue (probably NGF) encouraging growth of a plexus; at the same time, the embryonic skin generates a signal inhibiting nerves from approaching closer than about 70 μm to the surface.

The innervation of a vertebrate limb consists of a set of mixed nerve trunks from which two main sets of branches diverge: those that go to muscles and those that go to the skin. We have previously examined the role of muscles in controlling routes of nerve outgrowth in the chick limb bud (Lewis et al. 1981): we found that in a limb with no muscle cells, the mixed nerve trunks develop more or less normally, but muscle nerve branches fail to develop. In this paper, we look to see whether the nerves depend on their other major target, the skin, in an analogous way.

There are more aspects to this question than meet the eye. This can be appreciated when one considers more carefully the pattern of cutaneous innervation and the routes that growth cones destined for the skin take through the limb on their way to the target. First, they must travel along deep routes in company with axons destined for muscles, as components of mixed nerve trunks. At certain predictable points along these trunks they may diverge to form a cutaneous nerve branch. If they take the branch route, they follow again a precise and predictable track towards the skin. On arriving at a certain specific depth beneath the epidermis, the growth cones change direction and approach the epi­dermis no closer, behaving as if excluded from the most superficial layer of mesenchyme, (paralleling phenom­ena seen in tissue culture - Verna (1985)). They then ramify profusely to form a cutaneous plexus at that interface between the exclusion zone and the mesen­chyme through which they have travelled. Each of these steps in the journey might be regulated in its own way by influences from the target patch of skin. The aim of this paper is to find out which stages of the journey in fact depend on influences from the target, and in what ways. Does the skin supply chemotactic influences to guide the growth cones (see, for example, Lumsden & Davies, 1983, 1986), or trophic influences to encourage their survival (see, for example, Hamburger & Yip, 1984; Davies et al. 1987; Rohrer et al. 1988), or both or neither? And why does the cutaneous nerve plexus form only at a particular depth beneath the embryonic epidermis (the ectoderm) and at precisely the same depth as the cutaneous plexus of blood vessels, which, likewise, seem to be excluded from the most superficial mesenchyme (Caplan & Koutroupas, 1973; Feinberg et al. 1983)?

We have tackled these questions by locally destroying the key component of the skin - the epidermis - and examining the resultant pattern of innervation. The skin is a composite organ, and our findings give some indications as to which of its components - in particular epidermis, dermis, and blood vessels - have important effects on the developing nerves.

Fertilized chicken eggs (White Leghorn x Rhode Island Red, from Park Farm, Oxfordshire) were incubated at 38±1°C and windowed in preparation for ectodermal irradiations at stage 23/24 (staging according to Hamburger & Hamilton, 1951), on embryonic day 4 (E4). Immediately prior to irradiation, the vitelline and amniotic membranes were torn with surgical forceps to expose the right wing bud. A circular patch of ectoderm measuring 0.5-1 mm in diameter and constituting about half of the dorsal limb surface was ir­radiated (see Fig. 1A), using a focussed beam of short-wave (254 nm) ultraviolet light from a low-pressure mercury dis­charge lamp (u.v. Mineralight UVSL25 Combination u.v. Lamp with the filter removed, from UV Products Inc., San Gabriel, Calif.). The beam was directed using front-surface aluminized mirrors to focus the image of an illuminated adjustable aperture onto the embryo (Fig. IB). For initial alignment of the system, we used a bright source of visible light. The short-wave u.v. lamp was then moved into place behind the aperture, blocking out the light from the bright light source and providing u.v. illumination instead for 5 min, which we have previously shown to be consistently effective in destroying the ectoderm (Martin & Lewis, 1986). Immedi­ately after irradiation 2-3 drops of Hanks BSS with antibiotics (penicillin: 50-100i.uml-1, streptomycin: 50-l00 μg) were dripped onto the wing and the egg was resealed before returning to the incubator.

Fig. 1.

(A) Photograph of a stage 23 chick embryo showing the region of dorsal wing ectoderm that is irradiated (bright spot) to create a permanently denuded patch. Scale bar: 2 mm. (B) Diagram of the apparatus used to irradiate the limb bud (after Martin & Lewis, 1986).

Fig. 1.

(A) Photograph of a stage 23 chick embryo showing the region of dorsal wing ectoderm that is irradiated (bright spot) to create a permanently denuded patch. Scale bar: 2 mm. (B) Diagram of the apparatus used to irradiate the limb bud (after Martin & Lewis, 1986).

After 24 h, the wings of some irradiated embryos were fixed in 1/2 strength Karnovsky’s fixative (Karnovsky, 1965), dehy­drated and critical-point-dried in preparation for scanning electron microscopy, to confirm that a patch of ectoderm was indeed missing.

The majority of irradiated embryos were left until stage 35/36 (H10), when they were either fixed in 1/2 strength Karnovsky’s fixative, embedded in Araldite, sectioned and stained with toluidine blue, or fixed in Bodian’s fixative and processed by an adapted Bodian’s silver method to reveal nerves (Lewis et al. 1981). Drawings were made using a camera lucida. The position and extent of limb surface denudation at this stage were ascertained by direct inspection of the whole mount with oblique lighting. The effects of the missing patch of ectoderm on the pattern of cutaneous innervation were examined by comparing right and left wing whole mounts, cleared in methyl salicylate and viewed under a dissecting microscope set up for dark-field illumination. Selected specimens were also examined more carefully under the compound microscope with interference contrast optics and/or oblique epi-illumination. To make the edge of the ectoderm surrounding the lesion more clearly visible, we often found it helpful to heighten refractive-index contrasts by adding ethanol to the clearing medium, in a proportion of roughly 1:1.

A number of control stage 35/36 wings were also examined in Araldite sections and as skin whole-mount preparations to determine the precise spatial relationship between dermis and cutaneous neural and vascular plexuses. To make the vascula­ture easily visible, we injected Indian ink (Pelikan No. 17) into the bloodstream via one of the vitelline vessels immediately before fixation of the embryo.

Nomenclature of cutaneous nerves is as in Swanson (1985): DC Al: dorsal cutaneous nerve of the alar web, DC Int: dorsal cutaneous nerve of the interosseous region, DC Uln: dorsal cutaneous nerve of the ulnar region, consisting of a proximal (prox) and a distal (dist) branch, DC Elb: dorsal cutaneous nerve of the elbow region.

The findings of this paper are based on comparisons between normal chick wings and those lacking a patch of ectoderm on part of their dorsal surface. Previous studies (Roncali, 1970; Bennett et al. 1980; Swanson & Lewis, 1982) have charted the normal development of the main nerve trunks in the chick wing and the time of appearance of their cutaneous branches. Nerves begin migrating into the wing bud at about stage 25 (E4-5), when silver-stained whole mounts reveal a pattern of fibres originating from 4 or 5 spinal roots and terminat­ing as a broad bundle in the presumptive brachial plexus region (Fig. 2A). For the hindlimb, it has been shown that during passage through the plexus region axons become sorted out according to their peripheral desti­nations (Lance-Jones & Landmesser, 1981). Sensory growth cones are among the pioneers to invade the limb proper (Landmesser & Honig, 1986). By stage 29 (E6-5), the four main cutaneous nerve branches that will innervate the dorsal forearm region are present (Fig. 2B). DC Al will innervate the alar web, while DC Int will innervate the middle (interosseous) region of the forearm; DC Elb supplies a region of skin overlying the elbow; DC Uln divides as it reaches the posterior (ulnar) edge of the limb and sends out a proximal and a distal branch. Each of these cutaneous nerve branches ramifies beneath the skin in such a way that the whole dorsum of the limb becomes innervated (Fig. 2C). Although there is some slight overlap at the edges of the territories, where small nerve twigs can often be seen to have fibre contributions from two neighbouring nerve branches, there is in general a well-defined frontier between the territory of one cutaneous nerve branch and that of another (Fig. 3). To analyse the role of ectoderm in establishing this pattern, we must begin by looking more closely at the normal behaviour of the nerves and of other tissues in the neighbourhood of the ectoderm.

Fig. 2.

Camera-lucida drawings of the developing dorsal cutaneous nerve branch patterns of a series of normal chick wings: (A) Stage 25+ showing nerve fibres originating from 5 spinal roots and terminating in a broad bundle, the presumptive brachial plexus. (B) Stage 30. Only the cutaneous nerve branches serving the dorsal forearm (solid lines) and the proximal parts of the dorsal mixed nerve trunk (broken lines) are shown. (C) Stage 35. For identification of the cutaneous nerve branches DC Al, DC Int, DC Uln (prox and dist) and DC Elb, see Fig. 3. Scale bars: 1 mm.

Fig. 2.

Camera-lucida drawings of the developing dorsal cutaneous nerve branch patterns of a series of normal chick wings: (A) Stage 25+ showing nerve fibres originating from 5 spinal roots and terminating in a broad bundle, the presumptive brachial plexus. (B) Stage 30. Only the cutaneous nerve branches serving the dorsal forearm (solid lines) and the proximal parts of the dorsal mixed nerve trunk (broken lines) are shown. (C) Stage 35. For identification of the cutaneous nerve branches DC Al, DC Int, DC Uln (prox and dist) and DC Elb, see Fig. 3. Scale bars: 1 mm.

Fig. 3.

Diagram of a stage 35 wing highlighting the four cutaneous nerve branches serving the dorsal forearm (DC Al, DC Int, DC Uln prox and dist, DC Elb) and their approximate territories (shaded areas).

Fig. 3.

Diagram of a stage 35 wing highlighting the four cutaneous nerve branches serving the dorsal forearm (DC Al, DC Int, DC Uln prox and dist, DC Elb) and their approximate territories (shaded areas).

Blood vessels and nerves normally form plexuses at precisely the same level beneath limb ectoderm

For present purposes, the embryonic skin can be regarded as having four significant components: epider­mis (i.e. ectoderm), dermis (or prospective dermis), blood vessels and nerves. In principle, the patterning of any of these components could depend on any of the others. An examination of the normal pattern suggests some possible dependencies and rules out certain others.

The dermis - the layer of relatively dense connective tissue just beneath the epidermis - begins to become distinct from the deeper mesenchyme of the limb bud at about stage 28 (E6) (Sengel, 1976, and our own observations). The cell population density in the pro­spective dermis rises gradually and, by stage 36 (E10), the dermis is quite sharply defined as a layer of cells that are much more closely packed and less elongated than those of the loose connective tissue just beneath. The depth of the dermis at stage 36 (E10) varies from about 70 μm on the dorsal side of the wing to about 30 μm on the ventral side (Fig. 4).

Fig. 4.

(A) Transverse Araldite section of a stage 36 wing at midforearm level. On the dorsal side, the neurovascular plexus (arrows) lies at the boundary between the dense dermal mesenchyme and the deeper, loose hypodermal mesenchyme, while on the ventral side, where the dermal layer is shallower, the neurovascular plexus lies well within the loose mesenchyme. (B) and (C) Araldite sections of dorsal (B) and ventral (C) surfaces of a stage 36 wing viewed at higher magnification. In both cases nerves (thin arrows) and blood vessels (fat arrows) form their plexuses at precisely the same level beneath the epidermis, but at the dorsal surface these plexuses lie at the junction of dermis and hypodermis and at the ventral surface they lie at least 15 μm deep to the dermis. Scale bars: (A) 500 μm; (B) and (C) 100 μm.

Fig. 4.

(A) Transverse Araldite section of a stage 36 wing at midforearm level. On the dorsal side, the neurovascular plexus (arrows) lies at the boundary between the dense dermal mesenchyme and the deeper, loose hypodermal mesenchyme, while on the ventral side, where the dermal layer is shallower, the neurovascular plexus lies well within the loose mesenchyme. (B) and (C) Araldite sections of dorsal (B) and ventral (C) surfaces of a stage 36 wing viewed at higher magnification. In both cases nerves (thin arrows) and blood vessels (fat arrows) form their plexuses at precisely the same level beneath the epidermis, but at the dorsal surface these plexuses lie at the junction of dermis and hypodermis and at the ventral surface they lie at least 15 μm deep to the dermis. Scale bars: (A) 500 μm; (B) and (C) 100 μm.

Already at stage 18 (E3), when limb outgrowth is just beginning, well before there is any sign of a distinct dermal layer, blood vessels, though plentiful in the core of the limb bud, are excluded from a zone just beneath the epidermis (Caplan & Koutroupas, 1973; Feinberg et al. 1983). By stage 25, the vessels have begun to proliferate immediately beneath the avascular zone to form a plexus of innumerable anastomosing capillaries at that level. This plexus becomes more clearly defined as development proceeds, and is clearly revealed by marking the vessels by an injection of Indian ink (Fig. 4A). On the dorsal side of the limb, the plexus lies at the interface between the dermis and the loose deep connective tissue (Fig. 4B). This might be taken to suggest either that the dermis controls the location of the blood vessel plexus or that both are controlled by the same cue from the ectoderm. On the ventral side of the limb, however, the plexus lies in the loose mesen­chyme about 15 /an or more deep to the dermis (Fig. 4C), indicating that the correlation on the dorsal side is coincidental and that the dermis and the avascu­lar zone are defined by different means.

The embryonic cutaneous nerves, like the blood vessels, are excluded from the most superficial mesen­chyme and form a plexus immediately beneath their exclusion zone. This plexus begins to develop as soon as the first cutaneous nerve branches reach the skin (stage 29, E6), and can be revealed in whole mounts by silver staining. In cross-sections through the limb at stage 36 (Fig. 4), it can be seen that the nerve plexus is at exactly the same level as the blood vessel plexus, even on the ventral side of the limb, where both are located in the loose connective tissue deep to the dermis. The corre­lation is clearer still in flat whole-mount preparations of skin dissected from the surface of a limb whose blood vessels have been marked with Indian ink and whose nerves have been silver stained. By racking up and down with the fine focus of the microscope, it can be seen that the nerves and blood vessels weave their way over and under one another (Fig. 5): they are every­where confined to the same stratum, give or take a few micrometres. Experiments in which the ectoderm is destroyed suggest, however, that the extent of the nerve plexus is not simply defined by the extent of the vascular plexus but depends also on some neurotrophic influ­ence from the ectoderm.

Fig. 5.

A small region of a preparation of stage 36 dorsal chick wing skin with both nerves (arrows) and blood vessels simultaneously revealed by injection of Indian ink combined with silver staining. The nerves weave a path over some blood vessels and under others. Scale bar: 100 μm.

Fig. 5.

A small region of a preparation of stage 36 dorsal chick wing skin with both nerves (arrows) and blood vessels simultaneously revealed by injection of Indian ink combined with silver staining. The nerves weave a path over some blood vessels and under others. Scale bar: 100 μm.

u. v. irradiation can be used to create a patch of limb surface permanently denuded of ectoderm before cutaneous nerve branches have begun to form

In a previous paper, we showed that it is possible to denude a part of the limb bud of ectoderm by exposing the surface to short-wavelength u.v. irradiation: unlike a surgical lesion, the u.v. lesion does not heal, and the irradiated patch remains denuded for at least 7 days '(Martin & Lewis, 1986). In the earlier experiments, we were concerned with effects on the development of the skeleton, and we irradiated the limb bud very early, at about stage 18 (E3). This resulted in loss not only of ectoderm but also of most of the soft tissues on the irradiated side of the limb.

In the experiments to be presented here, we ir­radiated one day later, at stage 23 or 24, as described in Materials and methods. A total of 4$ embryos subjected to this treatment were analysed in detail in the various ways to be discussed below. Ectoderm was permanently lost from the irradiated patch, within 24 h, as before (Fig. 6). However, the loss of soft tissue did not go so deep as in the specimens irradiated at stage 18. As illustrated in Fig. 7A & B, lesions created by irradiation at stage 23/24 and examined 6 days later (stage 35/36) appear, at least at the periphery of the exposed region, to lack only the layer of dense dermal mesenchyme which would normally lie directly beneath the ecto­derm. This dermal tissue, which normally comprises closely packed, rounded cells, comes to an abrupt halt at the wound margin. Although the mesenchyme cells at the exposed surface are flattened to form a single­layered epithelioid sheet, the cells immediately beneath them do not resemble dermal fibroblasts; rather, they look more like the elongated loose connective tissue fibroblasts of the hypodermis. Dorsal muscles located beneath regions denuded by late u.v. irradiation are usually normal in pattern, as are all the other deep dorsal soft tissues of the limb, as well as the skeleton and mixed nerve trunks; this could be seen both from serial sections and from examination of appropriately stained whole mounts. Occasionally the damage ap­peared to have gone a little deeper and one or two of the dorsal muscles were found to be missing.

Fig. 6.

(A) Scanning electron micrograph of a stage 26+ wing bud irradiated 24 h previously at stage 24. The denuded patch (arrows) lies in the prospective elbow region. (B) High power detail from (A) showing the wound margin and the exposed mesenchymal cells. Scale bars: (A) lmm; (B) 20μm.

Fig. 6.

(A) Scanning electron micrograph of a stage 26+ wing bud irradiated 24 h previously at stage 24. The denuded patch (arrows) lies in the prospective elbow region. (B) High power detail from (A) showing the wound margin and the exposed mesenchymal cells. Scale bars: (A) lmm; (B) 20μm.

Fig. 7.

(A) Transverse Araldite section, showing a denuded region of the dorsal forearm region of a stage 35 wing after irradiation 5 days previously at stage 23. Note that blood vessels of the cutaneous vascular plexus (arrows) continue beyond the wound margin, beneath the denuded region, while a layer of dense dermal mesenchyme (asterisks), normally present beneath the epidermis, is absent where the epidermis is missing. (B) A high power view of the margin of a lesion similar to that shown in (A). Scale bars: 100 μm.

Fig. 7.

(A) Transverse Araldite section, showing a denuded region of the dorsal forearm region of a stage 35 wing after irradiation 5 days previously at stage 23. Note that blood vessels of the cutaneous vascular plexus (arrows) continue beyond the wound margin, beneath the denuded region, while a layer of dense dermal mesenchyme (asterisks), normally present beneath the epidermis, is absent where the epidermis is missing. (B) A high power view of the margin of a lesion similar to that shown in (A). Scale bars: 100 μm.

The cutaneous vascular plexus is present in a large part of the denuded patch, although absent in its centre

It was remarkable that a cutaneous vascular plexus was present even in a large portion of the region denuded of epidermis. Thus in cross-sections through lesions created at stage 23/24 and left to develop for 6 days before fixation (Fig. 7A & B), this cutaneous vascular plexus could be seen to extend beneath the exposed mesenchyme for distances of the order of 300 μιη beyond the lesion margin; yet, as described above, the mesenchyme superficial to the vascular plexus in such regions had nothing of the character of dermis. Closer to the centre of the lesion, a surface vascular plexus often seems to be absent (as judged from histological sections); it is quite likely that the depth of lesion is slightly greater in the centre, and therefore the level at which the vascular plexus would normally form has been lost in that region. However, from our obser­vations at the wound periphery, it would appear that ectoderm per se is not required for maintenance of the cutaneous vascular plexus.

Where ectoderm has been removed, the cutaneous nerve plexus does not develop

In all of our present series of specimens, the denuded areas lie within a region of skin overlying the dorsal forearm and normally innervated by DC Int, DC Elb and DCs prox and dist Uln (see Fig. 3), but the size and precise location of the lesion are somewhat variable. The range is roughly indicated in Fig. 8. In all our specimens, the cutaneous nerve plexus was absent throughout the denuded region. The homologous re­gion in the unirradiated contralateral limb was seen to be well innervated at the stages examined. By superim­posing camera lucida drawings of (a) the uncleared limb with its wound margin clearly visible upon (b) the cleared limb with its cutaneous neural plexus silver- stained, it was clear that small twigs from branches outside the wound region that approached the margin of the epidermis came to an end or changed direction at that margin or very close to it (Figs 9 & 10). The correspondence was checked more closely for selected specimens by examining the wound front and the neural plexus simultaneously, either in partly cleared limbs with oblique illumination (Fig. 10C) or in cleared limbs using interference contrast optics. The limit of inner­vation coincided with the margin of the epidermis to within about ±10μm.

Fig. 8.

Diagram to show the range of sizes of denuded regions seen in stage 35/36 limbs after irradiation at stage 23/24. Cross-hatching lines in the two different orientations represent maximal and minimal areas of denudation.

Fig. 8.

Diagram to show the range of sizes of denuded regions seen in stage 35/36 limbs after irradiation at stage 23/24. Cross-hatching lines in the two different orientations represent maximal and minimal areas of denudation.

Fig. 9.

(A,B) Camera-lucida drawings of the cutaneous nerve branch patterns of an irradiated wing (A) and its control (B), from a stage 36 embryo after irradiation of a patch of dorsal wing ectoderm at stage 23. In (A) a drawing of the uncleared limb with the margin of the denuded patch visible (dark shading) has been superimposed on the drawing of the cleared silver-stained limb with nerves visible. Only nerves innervating the dorsal forearm skin are illustrated (solid lines: A = DC Al; B = DC Int; C = DC Uln; D = DC Elb) along with the proximal part of the dorsal mixed nerve trunk, which lies deeper (broken lines). The region of the humerus (H) around which the dorsal mixed trunk courses, and over which the cutaneous nerve branches normally diverge from the trunk, is lightly shaded. (C,D) Photographs of the specimens depicted in (A) and (B) respectively. The silver-stained whole mounts, cleared in methyl salicylate, are viewed using dark-field illumination. Nerves to muscles as well as skin can be seen. (The bony collars of the developing radius, ulna and humerus also stain black.) Only the forearms and elbows of the two limbs are shown. Extremes of contrast make it difficult to display the entire pattern of innervation clearly in a single photograph. Scale bars: 1mm.

Fig. 9.

(A,B) Camera-lucida drawings of the cutaneous nerve branch patterns of an irradiated wing (A) and its control (B), from a stage 36 embryo after irradiation of a patch of dorsal wing ectoderm at stage 23. In (A) a drawing of the uncleared limb with the margin of the denuded patch visible (dark shading) has been superimposed on the drawing of the cleared silver-stained limb with nerves visible. Only nerves innervating the dorsal forearm skin are illustrated (solid lines: A = DC Al; B = DC Int; C = DC Uln; D = DC Elb) along with the proximal part of the dorsal mixed nerve trunk, which lies deeper (broken lines). The region of the humerus (H) around which the dorsal mixed trunk courses, and over which the cutaneous nerve branches normally diverge from the trunk, is lightly shaded. (C,D) Photographs of the specimens depicted in (A) and (B) respectively. The silver-stained whole mounts, cleared in methyl salicylate, are viewed using dark-field illumination. Nerves to muscles as well as skin can be seen. (The bony collars of the developing radius, ulna and humerus also stain black.) Only the forearms and elbows of the two limbs are shown. Extremes of contrast make it difficult to display the entire pattern of innervation clearly in a single photograph. Scale bars: 1mm.

Fig. 10.

(A) Camera-lucida drawing of another representative specimen, irradiated at stage 23+ and fixed and silver-stained at stage 35 + . (B) Contralateral control for (A). Shading and labelling conventions are as in Fig. 9. (C) Photograph of a detail from the specimen depicted in (A) (boxed region), illuminated obliquely to show the finest cutaneous nerve twigs. The black line has been drawn with a camera lucida to show the ectodermal margin which was accurately discernible under the microscope but not easy to capture in the photograph. The broad shadowy diagonal bar (R) deep to the nerves is the radius. Scale bars: (B) 1 mm; (C) 100 μm.

Fig. 10.

(A) Camera-lucida drawing of another representative specimen, irradiated at stage 23+ and fixed and silver-stained at stage 35 + . (B) Contralateral control for (A). Shading and labelling conventions are as in Fig. 9. (C) Photograph of a detail from the specimen depicted in (A) (boxed region), illuminated obliquely to show the finest cutaneous nerve twigs. The black line has been drawn with a camera lucida to show the ectodermal margin which was accurately discernible under the microscope but not easy to capture in the photograph. The broad shadowy diagonal bar (R) deep to the nerves is the radius. Scale bars: (B) 1 mm; (C) 100 μm.

Where a target skin territory is largely destroyed, the corresponding cutaneous nerve branch is completely absent

Although the specimens in our experimental series were denuded over large parts of the territories nor­mally innervated by DC Int, DC Elb, and DC Uln, some outlying parts of these territories generally had remained undamaged. As a rule, in such specimens there was no nerve branch to be seen at the normal site of origin of DC Int (22 of 28 specimens), DC Elb (23 of 28 specimens) or DC Uln (24 of 28 specimens) from the parent deep mixed nerve trunk (Figs 9 & 10). The surviving undamaged parts of the territories of these nerves were therefore deprived of innervation by the norma! route. These deprived territories were invaded by ramifications from remaining cutaneous nerves, the clearest examples being where vacant DC Int territory was invaded by DC Al (Figs 9 & 10). In the most extreme example of invasion by DC Al into vacant DC Int territory, the resulting DC Al territory was at least 50% larger than the corresponding DC Al territory' in the unirradiated contralateral wing (Fig. 9).

In the absence of the target patch of skin, cutaneous nerves apparently do not even begin to form

It is important to know whether absent cutaneous nerve branches had been absent from the outset or had started to form and then retracted on finding no target skin to innervate. We have attempted to answer this question by irradiating some specimens as before and fixing and silver-staining them at stage 29-30, when the cutaneous nerves of the forearm would normally have just reached the skin. We analysed 8 such specimens. In all of these, the lesion centered on the territory of DC Int and probably extended beyond it also. Judging from equiv­alent specimens fixed at stage 35/36, we should expect that the DC Int nerve would be found to be absent in most such cases at stage 35/36. In 8 out of 8 of the specimens fixed at stage 29/30, DC Int was in fact totally absent. In the majority of cases, DC Uln and DC Elb were absent also. It is difficult to draw definite conclusions from these observations but they suggest that in the absence of the target patch of skin the cutaneous nerve branch does not even begin to form (see Discussion below).

Where cutaneous nerve branches are present, their deep portions follow essentially normal routes, even if their peripheral territory is enlarged or reduced

As noted above, among our specimens there were many examples where absence of a particular cutaneous nerve branch allowed a neighbouring surviving branch to extend its territory' abnormally. There were also a few cases where outlying regions of a nerve’s territory were deleted but a significant part of the central region remained; the corresponding nerve branch then formed but ramified over a restricted region. If the target patch of skin controlled the route of outgrowth of the cu­taneous nerve fibres towards it (directing the fibres, for example, towards the centre of the patch), one should expect that a change in the extent of the target patch should entail a change in the route taken by the cutaneous nerve branch from the deep mixed nerve trunk towards the skin. This was not what happened: each surviving branch was clearly identifiable by the route that it took towards the skin - a route that was only slightly displaced from that taken by the corre­sponding nerve branch in the contralateral control limb (Figs 9 & 10). Only after reaching the skin did the nerve begin to show a grossly abnormal pattern of ramifi­cation.

In cases where DC Al had widened its ramifications to innervate an enlarged area of skin, there appeared to be a correlated increase in thickness of the main DC Al nerve branch. We saw no evidence to suggest that where DC Al had extensively invaded vacant DC Int territory, it had been at the expense of some of its own territory. However, there is some limitation, other than competition with neighbouring nerve branches, that restricts the distance over which neurites from any given nerve branch will ramify: in some specimens it was clear that some territory had remained vacant and had not been invaded by ramifications from neighbour­ing nerve branches.

A signal from the skin may trigger divergence of cutaneous nerve branches from the mixed nerve trunks

The present findings clearly demonstrate the absence of cutaneous nerve branches whose target field has been partly or wholly lost following irradiation. It seems that these results neatly parallel those shown for limbs devoid of muscle (Lewis et al. 1981), in that absence of a target tissue results in the absence of just those nerve branches that would have innervated it, while the main nerve trunks from which those branches derive remain unaffected. There is, however, at least one important difference between the two systems: most muscle nerve branches are initially formed by axons that turn aside from the main nerve trunk as it passes close by a muscle rudiment, perhaps even within filopodial grasping dis­tance (Lewis et al. 1981; Tosney & Landmesser, 1985), whereas the point where the growth cones of the developing cutaneous nerves diverge from the nerve trunk is often a considerable distance from the skin. The muscle, therefore, may not need to emit any long- range chemotactic signal to attract its innervation: it would be enough for the growth cones to recognize their target by an early contact interaction. For the skin, on the other hand, contact recognition of the target is not an adequate mechanism for initiating formation of the nerve branch, since the main nerve trunk does not bring the growth cones close enough: either the skin itself must emit some relatively long-range chemotactic signal, or the tissue between the deep nerve trunk and the skin must provide a favoured branch pathway for nerve outgrowth, or both these mechanisms must oper­ate together.

Can our experiments distinguish between these possi­bilities? If the formation and location of the cutaneous nerve branch were controlled simply by a chemotactic influence from the skin, we should indeed expect the branch to be missing when the whole target patch of skin is missing; but when only a part of that patch is missing, one should expect a cutaneous nerve branch (of reduced size) to form in an unusual position. In fact we observed that cutaneous nerve branches were either absent or, if present, followed an approximately normal route from the deep nerve trunk to the skin. This suggests that the routes taken by the cutaneous nerve branches are defined not by the skin but by the intervening connective tissues. The role of the skin, it seems, is to trigger growth cones to diverge from the deep nerve trunk where the option exists and/or to guarantee their survival when they reach the target itself.

To decide between these latter possibilities is diffi­cult, since growth cones that travel out initially towards a nonexistent target might then retract and disappear very rapidly. Failing to catch them in the act, one might conclude that they had never made the exploratory journey. We have looked at denuded specimens pre­cisely when cutaneous nerve branches should first be appearing (stage 29) and find no trace of either DC Int, DC Elb or DC Uln diverging from the mixed nerve trunk except in a few cases, which probably correspond to cases where a branch would be seen in any case in the more mature partially denuded limb. Tosney (1987) reported a similar absence of even transient nerve outgrowth following removal of dermamyotomal target tissues, as did Lewis et al. (1981) in the case of muscle nerve branches in muscleless limbs. It is tempting to conclude, therefore, that a long-range stimulus from the skin is required to evoke the initial formation of a cutaneous nerve branch. To be definite about this, however, one needs to show that retraction is too slow a process to be missed if it occurs. Our preliminary experiments to test this by ablating skin at later stages, after cutaneous nerves have begun to form, suggest that retraction takes several days.

Lumsden & Davies (1983) have shown in vitro that mouse trigeminal ganglion neurites are guided to their whisker-pad targets by diffusible factors synthesized by the skin itself. Moreover, this chemotactic influence is not blocked by adding antibodies against nerve growth factor to the medium and therefore presumably is not mediated by NGF. In the same system, regionally inappropriate target tissues do not evoke directed outgrowth of trigeminal ganglion neurites, suggesting that development of the trigeminal system may depend on regionally specific chemotactic factors (Lumsden & Davies, 1986). Whether or not there are regionally specific factors in the limb causing sensory neurites to diverge from the mixed nerve trunks and grow towards the skin is less clear; but it is clear that muscle and cutaneous nerve branches are triggered to diverge from mixed nerve trunks by independent cues because re­moval of one target tissue does not affect the pattern of innervation to the other (Lewis et al. 1981; data reported in this paper). Moreover, the skin sensory nerve fibres do not appear to require guidance from motor nerve fibres (Swanson & Lewis, 1986; Scott, 1988).

The skin has two antagonistic short-range actions on cutaneous nerves

In birds, as in mammals, the initial innervation of the skin consists of a nerve plexus confined to a layer just beneath the dermis. This state of affairs persists for some time; then, in a second phase of cutaneous nerve development, axons change direction and sprout up­wards towards the surface, forming specialized sensory corpuscles in the superficial dermis and penetrating the epidermis also (Ram0n y Cajal, 1929; Saxod, 1978). TTie timing of this invasion of the superficial tissues is not accurately known for the chick, but from our own observations must be later than embryonic day 10. Thus, throughout the stages discussed in this paper, cutaneous nerves do not penetrate the epidermis, and they even refrain from entering the zone of connective tissue immediately beneath it. The epidermis, directly or indirectly, seems to exert some repulsive influence on the nerves - an idea that is supported by Verna’s (1985) work in vitro. Yet just beneath the resulting exclusion zone the nerves ramify profusely to form a plexus; and where ectoderm (together with dermis) is missing, no nerve plexus develops. These latter two observations imply that the cutaneous nerves depend on a stimula­tory or trophic influence from the skin to grow. In short, the cutaneous nerves seem to have a love-hate relation­ship with the epidermis and/or dermis. Their behaviour in this respect is like that of the cutaneous blood vessels.

In a normal limb, the nerve plexus lies at precisely the same depth beneath the epidermis as the vascular plexus in every region, even though the depth varies from one region to another. It seems likely, therefore, that the two are either interdependent, or are both responding to the same depth-control (or exclusion) cue relative to the epidermis. Their depth is apparently not determined by the depth of dense dermal mesenchyme which is in some places shallower than the plexuses. We have shown that at the margin of a denuded patch the blood vessel plexus remains even though nerves are absent; and in limb buds deprived of innervation, a blood vessel plexus still develops at the normal depth (Martin & Lewis, unpublished data). Conversely, other observations of a lack of detailed correlation between nerves and blood vessels make it unlikely that nerves are guided by blood vessels (Tosney & Landmesser, 1985; Martin & Lewis, unpublished data). Therefore it seems that nerves and blood vessels are responding independently to the same environment. But what might the relevant environmental cues be? Cutaneous nerves are known to be sensitive to neural growth factors and in particular NGF, and endothelial cells are known to be responsive to fibroblast growth factor (FGF) and certain other factors (Davies, 1988; Folk-man & Klagsbrun, 1987). But if these stimulatory factors were the whole story, there would be no reason to expect the plexuses to form at exactly the same level. However, that phenomenon can be explained if we suppose that nerves and blood vessels, while depending on different stimulatory factors from the epidermis, both face the same barrier blocking entry into the exclusion zone just beneath it. There is good evidence that the exclusion zone for blood vessels is due to hyaluronic acid that is present in this region (Feinberg & Beebe, 1983). There is some in vitro evidence that hyaluronic acid is an ineffective substratum for nerve outgrowth (Carbonetto et al. 1983). Moreover, Knud-son & Toole (1985) have shown that connective-tissue cells secreting hyaluronic acid surround themselves with a jacket of this material, hindering cell-cell contact; and according to Al-Ghaith & Lewis (1982), the ad­vance of growth cones through limb mesenchyme in­volves the formation of close contacts with the mes­enchymal cells. Thus hyaluronic acid might well be expected to inhibit growth cone movement.

There is one obvious objection that might be raised to our interpretation of the finding that where there is no epidermis no cutaneous neural plexus forms: it might be suggested that the absence of the cutaneous nerve plexus is merely the consequence of destruction of the mesenchyme through which it would normally grow, rather than loss of a neurotrophic signal from the epidermis or dermis. While this cannot be absolutely excluded, it seems unlikely in the light of our obser­vation that the cutaneous vascular plexus is still present beneath much of the denuded region.

Competition between fibres of neighbouring cutaneous nerve branches shapes their eventual territory boundaries

Selective removal of particular cutaneous nerve branches has allowed us to see how surviving cutaneous branches respond to territory that is left undamaged but vacant. Our observations support the view that territory boundaries between different cutaneous nerve branches are governed by competition. A complemen­tary experiment to ours, reported by Scott (1984), involved depriving the chick of a number of dorsal root ganglia that normally innervate defined overlapping territories (dermatomes) of the hind limb skin. Scott’s electrophysiological studies show that neurites from neighbouring ganglia now innervate the vacant derma­tomes, although it is not clear whether this invasion is via the normal pattern of cutaneous nerve branches or if particular nerve branches expand their territories whilst others do not form at all, as in our experiment.

Competition has also been shown to be important in the development of the cutaneous nerve patterns in the frog limb: when dorsal root ganglia that normally supply the limb are removed, axons of nonlimb dorsal root ganglia sprout into the limb territories (Miner, 1956). Furthermore, removal of the trigeminal ganglia from one side of a Xenopus embryo prior to cutaneous innervation results in the invasion of that ganglion’s territory by neurites from the contralateral side (Kitson & Roberts, 1983).

It seems likely that the competition for skin territory may reflect a competition for a trophic factor derived from the epidermis or dermis, for in regions deprived of these tissues, no cutaneous neural plexus forms. In situ hybridization data from the mouse whisker pad system (Davies et al. 1987) suggest that NGF is being syn­thesized at the right time and by the right cells in the embryonic skin to be acting as such a trophic factor. They also show that trigeminal ganglion neurites begin to express receptors to NGF when they reach the whisker pad skin. More recently Rohrer et al. (1988) have shown that similar levels of NGF message are being synthesized by chick wing skin at the stages when cutaneous nerve branches reach the skin.

Our data imply that the epidermis and/or dermis control the development of the cutaneous innervation in several distinct ways. Target patches of skin appar­ently exert a long-range influence to make cutaneous nerve branches diverge from the deep mixed nerve trunks. On reaching the skin, cutaneous neurites be­have as though responding to a trophic factor from the epidermis and/or dermis that acts locally to encourage the formation of a cutaneous plexus; and, as though in competition for this skin-derived factor, neighbouring cutaneous nerve branches ramify competitively to oc­cupy the available territory. Apparently antagonistic to the trophic influence is an inhibitory cue from the epidermis that prevents neurites from coming too close, thereby confining the plexus to a shallow stratum some distance beneath the surface. Our findings emphasize once again that there is no single unitary cue guiding nerves to their targets (Dodd & Jessell, 1988). A series of influences of different types operate at different points along the pathway.

We thank the MRC and ICRF for financial support and Drs Alun Davies, Gavin Swanson, Peter Wigmore and Raymond Saxod and colleagues at both King’s College London and the ICRF Developmental Biology Unit, Oxford, for discussion and comments.

Al-Ghaith
,
L. K.
&
Lewis
,
J. H
. (
1982
).
Pioneer growth cones in virgin mesenchyme: an electron-microscope study in the developing chick wing
.
J. Embryol. exp. Morph
.
68
,
149
160
.
Bennett
,
M. R.
,
Davey
,
D. F.
&
Uebel
,
Κ. E.
(
1980
).
The growth of segmental nerves from the brachial myotomes into the proximal muscles of the chick forelimb during development
.
J. comp. Neurol
.
189
,
335
357
.
Caplan
,
A. I.
&
Koutroupas
,
S
. (
1973
).
The control of muscle and cartilage development in the chick limb: the role of differential vascularization
.
J. Embryol. exp. Morph
.
29
,
571
583
.
Carbonetto
,
S.
,
Gruver
,
Μ. M.
&
Turner
,
D. C.
(
1983
).
Nerve fiber growth in culture on fibronectin, collagen, and glycosaminoglycan substrates
.
J. Neurosci
.
3
,
2324
2335
.
Davies
,
A. M
. (
1988
).
Role of neurotrophic factors in development
.
Trends Genet
.
4
,
139
144
.
Davies
,
A. M.
,
Bandlow
,
C.
,
Heumann
,
R.
,
Korsching
,
S.
,
Rohrer
,
Η.
&
Thoenen
,
H.
(
1987
).
Timing and site of nerve growth factor synthesis in the developing skin in relation to innervation and expression of the receptor. Nature
,
Lond
.
326
,
353
358
.
Dodd
,
J.
&
Jessel
,
T. M
. (
1988
).
Axon guidance and the patterning of neuronal projections in vertebrates
.
Science
242
,
692
699
.
Feinberg
,
R. N.
&
Beebe
,
D. C
. (
1983
).
Hyaluronate in vasculogenesis
.
Science
220
,
1171
1179
.
Feinberg
,
R. N.
,
Repo
,
M. A.
&
Saunders
,
J. W
. (
1983
).
Ectodermal control of the avascular zone of the peripheral mesoderm in the chick embryo
.
J. exp. Zool
.
226
,
391
398
.
Folkman
,
J.
&
Klagsbrun
,
M
. (
1987
).
Angiogenic factors
.
Science
235
,
442
447
.
Hamburger
,
V.
&
Hamilton
,
H. L
. (
1951
).
A series of normal stages in the development of the chick embryo
.
J. Morph
.
88
,
49
92
.
Hamburger
,
V.
&
Yip
,
J. W
. (
1984
).
Reduction of experimentally induced neuronal death in spinal ganglia of the chick embryo by nerve growth factor
.
J. Neurosci
.
4
,
767
774
.
Karnovsky
,
M. J
. (
1965
).
A formaldehyde-gluteraldehyde fixative of high osmolarity for use in electron microscopy
.
J. Cell Biol
.
27
,
137
138
.
Kjtson
,
D. L.
&
Roberts
,
A
. (
1983
).
Competition during innervation of embryonic amphibian head skin
.
Proc. R. Soc. Lond. B
218
,
49
59
.
Knudson
,
C. B.
&
Toole
,
B. P
. (
1985
).
Changes in the pericellular matrix during differentiation of limb bud mesoderm
.
Devi Biol
.
112
,
308
318
.
Lance-Jones
,
C.
&
Landmesser
,
L
. (
1981
).
Pathway selection by chick lumbosacral motoneurons during normal development
.
Proc. R. Soc. Lond. B
214
,
1
18
.
Landmesser
,
L.
&
Honig
,
M. G
. (
1986
).
Altered sensory projections in the chick hind limb following early removal of motoneurons
.
Devi Biol
.
118
,
511
531
.
Lewis
,
J.
,
Chevaluer
,
A.
,
Kieny
,
M.
&
Wolpert
,
L
. (
1981
).
Muscle nerve branches do not develop in limbs devoid of muscle
.
J. Embryol. exp. Morph
.
64
, 211-232.
Lumsden
,
A. G. S.
&
Davies
,
A. M
. (
1983
).
Earliest sensory nerve fibres are guided to peripheral targets by attractants other than nerve growth factor. Nature
,
Lond
.
306
,
786
788
.
Lumsden
,
A. G. S.
&
Dames
,
A. M
. (
1986
).
Chemotropic effect of specific target epithelium in the developing mammalian nervous system. Nature
,
Lond
.
323
,
538
539
.
Martin
,
P.
&
Lewis
,
J
. (
1986
).
Normal development of the skeleton in chick limb buds devoid of dorsal ectoderm
.
Devi Biol
.
118
,
233
246
.
Miner
,
N
. (
1956
).
Integumental specification of sensory fibres in the development of cutaneous local sign
.
J. comp. Neurol
.
105
,
161
170
⋅.
Ramon y
Cajal
, S. (
1929
). litudes sur la Neurogenese de Quelques Vertέbrέs. Madrid. Reprinted in English (tr. L. Guth, 1960). Studies on Vertebrate Neurogenesis. Illinois: Charles C. Thomas.
Rohrer
,
H.
,
Heumann
,
R.
&
Thoenen
,
H
. (
1988
).
The synthesis of nerve growth factor (NGF) in developing skin is independent of innervation
.
Dev! Biol
.
128
, 240-244.
Roncau
,
L
. (
1970
).
The brachial plexus and the wing nerve pattern during early develomental phases in chicken embryos
.
Monit. Zool. Ital. (N
.
5
.) 4, 81-98.
Saxod
,
R.
(
1978
). Development of cutaneous sensory receptors in birds. In Handbook of Sensory Physiology, vol. IX: Development of Sensory Systems (ed. M. Jacobson), pp. 337-417.
New York
:
Springer-Verlag
.
Scott
,
S. A
. (
1984
).
The effects of neural crest deletions on the development of sensory innervation patterns in embryonic chick hind limb
.
J. Physiol
.
352
,
285
304
.
Scott
,
S. A
. (
1988
).
Skin sensory innervation patterns in embryonic chick hindlimbs deprived of motoneurons
.
Devl Biol
.
126
,
362
374
.
Sengel
,
P.
(
1976
). Morphogenesis of Skin. Cambridge University Press, Cambridge.
Swanson
,
G. J
. (
1985
).
Paths taken by sensory nerve fibres in aneural chick wing buds
.
J. Embryol. exp. Morph
.
86
,
109
124
.
Swanson
,
G. J.
&
Lewis
,
J.
(
1982
).
The timetable of innervation and its control in the chick wing bud
.
J. Embryol. exp. Morph
.
71
,
121
137
.
Swanson
,
G. J.
&
Lewis
,
J
. (
1986
).
Sensory nerve routes in chick wing buds deprived of motor innervation
.
J. Embryol. exp. Morph
.
95
,
37
52
.
Tosney
,
K. W
. (
1987
).
Proximal tissues and patterned neurite outgrowth at the lumbosacral level of the chick embryo: Deletion of the dermamyotome
.
Devi Biol
.
122
,
540
558
.
Tosney
,
K. W.
&
Landmesser
,
L. T
. (
1985
).
Development of the major pathways for neurite outgrowth in the chick hindlimb
.
Devi Biol
.
109
,
193
214
.
Verna
,
J. M
. (
1985
).
In vitro analysis of interactions between sensory neurons and skin: evidence for selective innervation of dermis and epidermis
.
J. Embryol. exp. Morph
.
86
, 53-70.