The effect of replacing different regions of limb skin with head skin on regeneration in the axolotl

A technique enabling different regions of the limb to be isolated using pieces of skin from the head has been used previously to show major differences in the regenerative capacity of posterior and anterior tissue in the upper arm (Wigmore & Holder, 1985). Specifically, stumps containing posterior tissue were able to produce normal 4-digit limbs while stumps containing only anterior tissue regenerated only single digits. In the lower arm this difference between anterior and posterior tissue was less marked and no half of the limb could produce the complete pattern. All lower arm halves produced between 2 and 3 digits with the pattern of skeletal structures in some cases ending abruptly part way across the anteroposterior axis. In addition to the skeletal pattern the dorsoventral axis was assayed using the pattern of muscles (Maden, 1980, 1982; Maden & Mustafa, 1982a; Holder & Weekes, 1984). The presence of dorsal or ventral tissue in the regenerate appeared to be dependent on the presence of dorsal or ventral tissues in the stump. Stumps lacking either of these tissues, whether in the upper or lower parts of the limb, produced regenerates deficient in dorsal or ventral muscle. These results left several questions unanswered. In particular, which components of the stump enable the regeneration of the normal pattern of skeletal and muscular tissues.

It has been known for some time that the dermis of the skin exerts a powerful influence on limb regeneration. Rotation of skin cuffs (Carlson, 1974,1975) or the movement or implantation of limb dermis to different positions on the limb (Lheureux, 1972, 1975; Settles, 1978; Tank, 1981; Rollman-Dinsmore & Bryant, 1984; Maden & Holder, 1984) causes the production of additional structures in the regenerate while stumps composed of double posterior skin regenerate double posterior limbs (Slack, 1980,1983; Maden & Mustafa, 1982b). Recently flank skin has been used to replace different regions of limb skin in the upper arm of the newt (Tank, 1985); the results showed regional variations in the effect of the graft, with grafts to the ventral side of the stump producing the highest proportion of hypomorphic limbs. Less well studied is the effect of skin on the underlying pattern of muscles. However, transplantation of dorsal or ventral skin has been shown to cause the production of abnormal patterns of dorsal and ventral musculature (Maden & Mustafa, 1982b).

Autografts of head skin heal rapidly onto the limb and several experiments have demonstrated that skin from this region has no effect on pattern formation during limb regeneration. Unlike limb skin, head skin does not restore the regenerative ability of irradiated limbs (Umaski, 1938; Lazard, 1967) and implants of head skin into limb stumps do not affect the pattern regenerated (Tank, 1981). The technique of using head skin to isolate different parts of the limb and prevent the incorporation of tissues from different circumferential regions can be extended to examine the role of different regions of the skin in limb regeneration. In particular this paper deals with the role of posterior skin in producing normal regenerates from upper arm posterior half stumps. Additionally, the role of skin in regeneration from upper and lower arm amputation planes is compared together with the influence of skin on the muscle pattern in the regenerate. The results demonstrate a clear anteroposterior asymmetry in the ability of skin regions to control pattern regulation in the upper arm, which is absent in the lower arm. In the upper arm dorsal and ventral skin is seen to affect the muscle pattern although again this effect is not seen in the lower arm.

All experiments were carried out using axolotls (Ambystoma mexicanum), 10–15 cm in length, spawned in the colony at King’s College. Animals were kept in tap water in individual containers and fed on raw heart.

Animals were operated on whilst anaesthetized with MS222 (Sigma) and allowed to recover in the dark at 10°C for 2 days. Operations consisted of the removal of half the circumference of limb skin (either posterior, dorsal, ventral or anterior) for a distance of 3–4 mm and its replacement with a similar-sized strip of skin from the forehead. The wound in the forehead was found to be covered by epidermis within 24h and so no attempt was made to treat it. The head skin was sutured into place and the limb amputated immediately through the middle third of the humerus for upper arm operations and the middle third of the radius and ulna for lower arm operations (Fig. 1A). The stumps were viewed end on under a dissecting microscope to check the position of the graft. Controls consisted of carrying out the same procedure as described above but without grafting any head skin onto the limb.

In order to investigate further the role of posterior skin in upper arm regeneration, anterior half upper arm stumps were constructed as described in Wigmore & Holder (1985). This operation involved the removal of all posterior soft tissues and the suturing of a strip of head skin to the medial wound surface. Posterior skin was then separated from the removed posterior half of the limb and sutured to the medial surface of the anterior half stump, distal to the head skin. The limb was then immediately amputated through the posterior skin graft and the middle third of the humerus. This produced a stump with a complete circumference of limb skin together with all the soft tissues on the anterior side of the limb and the humerus but deficient in the subdermal tissues of the posterior side (Fig. 1B). It has been shown previously that anterior half stumps constructed in this way but without posterior skin consistently regenerate only single digits (Wigmore & Holder, 1985).

All experimental animals were checked for retention of the graft every few days for the first few weeks after the operation. Operated animals were left for 60 days after which time they were reanaesthetized and the regenerates amputated, fixed in Bouin’s solution, dehydrated and stained with Victoria Blue (Bryant & Iten, 1974) to show the skeletal pattern. Camera-lucida drawings were made of all regenerates. Limbs were then wax embedded and serial sections were cut at 10 μm and stained with haematoxylin and eosin. These sections were used to check the position of the graft and to analyse the muscle pattern.

The muscle pattern was examined at two levels, the proximal metacarpals and the midforearm. In normal limbs at the level of the proximal metacarpals, ventral muscle is continuous across all digits while dorsal muscle, the extensores digitorum breves, occurs as discrete crescents over each metacarpal (Maden, 1980, 1982; Maden & Mustafa, 1982a; Burton, Holder & Jesani, 1986). There is also a greater concentration of Leydig cells in the dorsal epidermis at the metacarpal level. At the midforearm level, the dorsoventral axis can be determined from the overall muscle pattern and the presence of the pronator quadratus (pq) muscle which runs from a spine on the ventral side of the ulna towards the radius (Holder & Weekes, 1984; Burton et al. 1986).

Eight upper arm stumps with this operation were constructed (see Table 1). The majority of regenerates (five) produced only a single digit and had a proximodistal skeletal pattern consisting of a single forearm bone, two carpals, a metacarpal and two phalanges (Fig. 2). One stump failed to regenerate and the remaining two regenerates were 3- and 4-digit limbs. Neither of these limbs were normal, both showing reduction or loss of the most posterior digit or carpals.

Sections of these last two limbs showed the 4-digit limb to have a normal muscle pattern while the 3-digit limb lacked dorsal muscle at both metacarpal and forearm regions (see Table 2). Single-digit limbs cannot be analysed for their muscle patterns.

In contrast to the posterior side of the limb, all stumps where anterior skin was replaced regenerated normal 4-digit limbs with both radius and ulna. Sections of these limbs showed them to have a normal muscle pattern.

Eleven out of twelve limbs with this type of operation had either 3 or 4 digits (see Table 1) but several limbs showed fusion between digits (four cases) and both 3-digit limbs lacked the most posterior digit (Fig. 3). All regenerates, with the exception of that with a single digit, had both radius and ulna. Sections of the regenerates showed that three limbs (one with 4 digits, and two with 3 digits) lacked any muscle on the dorsal side of the metacarpals (Fig. 4; Table 2). At the forearm level, one of these lacked dorsal muscle, and of the other two, one was double ventral (having ventral muscle on both sides of the limb) and the other was normal.

This group of twelve limbs showed the greatest spread of results and the lowest mean digit number after those with posterior skin replaced (see Table 1). The three limbs with 2 or 3 digits showed loss of posterior skeletal structures (Fig. 5) and one 4-digit limb lacked the most posterior carpals. All limbs, including those with only a single digit, had both radius and ulna. Four limbs (a 4-digit, two 3-digit and a 2-digit limb) lacked ventral muscle in the metacarpal region (dorsal half limbs, Fig. 6; Table 2) but only one of these lacked ventral muscle at the forearm level. The remaining limbs had a normal muscle pattern (the single-digit limbs could only be scored at the forearm level).

These limbs differed from those described previously in that the deeper soft tissues had been removed from the posterior side of the upper arm and the stump was made into an anterior half limb by sealing the medial surface with head skin. Posterior upper arm skin was then added to the medial surface, distal to the head skin (Fig. 1B). All but two of the nine regenerates from these stumps had a normal skeletal pattern with 4 digits (see Table 1; Fig. 7). The 5-digit limb included a serial duplication of digit 3 while the 3-digit limb lacked digit 4 and had only a single forearm element.

Sections of these limbs showed a normal pattern of muscles at both metacarpal and forearm levels.

In contrast to the upper arm, all regenerates from lower arm operations had at least 3 digits (see Table 1) and the majority, irrespective of the position of the head skin, had 4. Minor abnormalities were noted in regenerates from stumps where posterior skin had been replaced. These showed some loss of posterior skeletal structures, for example loss of digit 4 resulting in a 3-digit limb or reduction in size of posterior digits and carpals. In one case digit 4 had three phalanges while the 5-digit regenerate appeared to include a serial duplication of digit 3. This was also the case in both 5-digit regenerates from stumps with dorsal skin replaced. This type of operation also produced two limbs with digit 4 having three phalanges. Several limbs showed branching at the phalangeal level.

The muscle pattern of all lower arm regenerates was normal at both metacarpal and forearm levels with the exception of one regenerate from a stump with posterior skin replaced; this limb lacked ventral muscle at both levels.

Five control limbs were constructed for each type of operation in both upper and lower arms (40 limbs in all). All regenerates from control limbs produced 4 digits with a basically normal skeletal pattern. Six regenerates had an extra basal carpal between the fused basal carpal one and two and basal carpal three. Sections of control regenerates at both metacarpal and forearm levels showed a normal pattern of muscles.

The results presented here support the conclusions drawn from previous work on regeneration from half limb stumps that demonstrated that the regenerative ability of the limb shows regional variations (Wigmore & Holder, 1985; Wigmore, 1986). In these studies normal limbs regenerated from upper arm half stumps containing posterior tissue but not from stumps composed of only anterior tissues. This difference between posterior and anterior parts of the limb was much less marked in the lower arm, but posterior halves still produced the highest mean digit number of all four quadrants of the limb. The present work was carried out to investigate these differences further not only by regions of the limb but also by eliminating the influence of a single tissue, the limb skin, from particular parts of the limb circumference.

The greatest effect after replacement of half the limb skin with head skin was in the posterior half of the upper arm. Removal of this skin resulted in the regeneration of a high proportion of single digits. This occurred despite the fact that it was only 50 % of the skin that had been replaced and the rest of the limb remained intact. The single digits obtained from these stumps are virtually identical to those regenerating when the whole of the posterior half was removed to produce anterior half stumps (Wigmore & Holder, 1985) and suggests that the cause of the hypomorphic regenerates in those experiments was due solely to the absence of posterior skin from the stump.

The importance of the posterior skin was demonstrated directly by its ability to rescue anterior half stumps which normally only regenerate single digits (Wigmore & Holder, 1985). This result indicated that, in the upper arm, the only essential component of the stump for producing more than a single digit is the posterior skin. This conclusion is consistent with the construction of symmetrical limbs composed of both skin and deep tissues (Stocum, 1978; Tank & Holder, 1978; Holder, Tank & Bryant, 1980). In these experiments double anterior stumps produced single digits and double posterior stumps regenerated double posterior multidigit limbs. When only the skin was made symmetrical and the deep tissues were left intact (Maden & Mustafa, 1982b), stumps bearing double posterior skin again regenerated multidigit, double posterior limbs. However, limbs symmetrical for anterior skin regenerated normal 4-digit limbs in an apparent contradiction of the principle that posterior skin is necessary for the formation of a normal anteroposterior axis in the upper arm. This result is puzzling as all of the tissues present have been shown to be inessential to pattern formation (present work and Wigmore & Holder, 1985). The only unique feature of this type of stump is a discontinuity between the deep posterior tissues and the overlying anterior skin and it must be concluded that this enables the pattern-forming mechanism to function. The role of tissues that appear to be inessential to regeneration is unclear but certainly in the case of anterior tissues they do contribute large numbers of cells, mainly to anterior structures of the regenerate, when normal stumps are amputated (Rollman-Dinsmore & Bryant, 1984; Tank, Connelly & Bookstein, 1985). The present results and those of the authors mentioned above are shown in diagrammatic form in Fig. 8. As can be seen with the one exeption mentioned previously, only stumps bearing posterior skin show consistent regeneration of more than a single digit.

In contrast to the upper arm, replacement of posterior lower arm skin did not prevent the formation of normal 4-digit limbs. At this level, posterior skin does not appear to be essential for pattern regulation. Posterior lower arm skin is, however, known to have morphogenetic properties as stumps bearing double posterior skin regenerate double posterior limbs (Slack, 1980,1983; Maden & Mustafa, 1982b). Isolation of half lower arms (Wigmore, 1986) showed that no half could regenerate the complete pattern of the limb. It can be concluded therefore that while the information for pattern formation is concentrated in the posterior half of the upper arm, at distal levels the information is more spatially separated. This conclusion is further supported by the observation that, in contrast to the upper arm, lower arm double posterior and double anterior stumps produce approximately equal amounts of symmetrical outgrowth (Krasner & Bryant, 1980).

In previous experiments where dorsal and ventral halves of the limb were isolated (Wigmore & Holder, 1985; Wigmore, 1986) or symmetrical double dorsal or double ventral stumps were constructed (Burton et al. 1986), limbs deficient in dorsal or ventral muscle often regenerated. In the light of these results the present replacement of dorsal or ventral skin was designed to examine the role of skin on pattern regulation in the dorsoventral axis. Replacement of dorsal or ventral skin in the upper arm resulted in the formation of limbs deficient in muscle on the side of the limb where the limb skin had been replaced. This effect is probably not due to direct inhibition of muscle formation by head skin because, as described below, similar operations on the lower arm produced limbs with a normal muscle pattern. In the upper arm the production of the underlying muscle pattern appears to be dictated by the overlying dermis and in the absence of dorsal or ventral skin, muscle on that side of the limb fails to form. This conclusion is supported by the results of Maden & Mustafa (1982b) where stumps symmetrical for dorsal or ventral skin produced, in a proportion of cases, regenerates with double dorsal or double ventral muscle. The mechanism by which the muscle pattern is controlled in the axolotl is at present unknown but in the chick limb (Chevallier, Kieny & Mauger, 1977; Chevallier & Kieny, 1982), it is the connective tissue of the limb that dictates the pattern of the different muscles. It is possible that fibroblasts from the dermis of the axolotl skin enter the blastema during regeneration and exert a similar control.

The majority of stumps with dorsal or ventral skin replaced regenerated 4-digit limbs. Since these stumps always contained some posterior skin, this is the expected result if the presence of posterior skin enables the regeneration of multidigit limbs. Similarly removal of posterior or anterior skin leaves approximately 50 % of dorsal and ventral skin in situ which would appear to be sufficient for the regeneration of a normal muscle pattern.

In the lower arm the results of dorsal and ventral skin replacement were different in that all regenerates had normal muscle patterns and, from the distribution of Leydig cells, the overlying skin also appeared normal. This result was unexpected as removal of dorsal or ventral halves of the lower arm (Wigmore, 1986) produced limbs deficient in dorsal or ventral muscle similar to those formed in the upper arm. The results obtained here are, however, consistent with the construction of lower arm stumps bearing symmetrical dorsal or ventral skin (Maden & Mustafa, 1982) where, despite the symmetry in the skin, normal asymmetric muscle patterns were regenerated. It is unclear whether this result indicates a difference in the patterning mechanism in upper and lower parts of the limb or is due to a difference in their response to grafted tissue.

It is a pleasure to thank Stephen Wilson, Gavin Swanson and Paul Martin for many useful comments on this work and Liz Delaney for her dedicated technical assistance. This work was supported by an SERC project grant to Nigel Holder.