ABSTRACT
Aneurogenic arms of young axolotls were implanted into the flank as heterotopic autografts with reversed proximo-distal orientation. The formerly proximal ends of such arms regressed to a variable extent, and then either regenerated or could do so following a second amputation. The regenerate always contained a complete sequence of skeletal elements between the adjacent stump skeleton and terminal digits, being a mirror image of the implanted arms with identical transverse axes but an opposed proximo-distal axis. Many reversed arms also regenerated fingers from the implanted hand. Identical results were obtained from reversed arms of control larvae, confirming previous studies on reversed well-innervated arms. Nerves are not required for the establishment of a new proximo-distal axis, therefore, and probably have no influence on the determination of any limb axis. Morphogenesis of the regenerate is clearly related to the position along the proximo-distal axis where the blastema originates. Although this axis is reasonably envisaged in terms of a gradient, its polarity is ignored during regeneration.
INTRODUCTION
The technique of removing most of the central nervous system from an amphibian embryo maintained in parabiosis has been exploited for several studies of limb development and regeneration. Sparsely innervated and aneurogenic arms produced in this way develop normally at first and can regenerate (Yntema, 1959), although the arms of older larvae and adult urodeles apparently require a considerable innervation in order to regenerate (Singer, 1952). This contradiction has not been fully resolved but comparisons between aneurogenic and normally innervated larval arms should reveal what aspects of regeneration are influenced by nerves. Local irradiation, for instance, inhibits regeneration in both cases, implying a direct action of X rays which is not mediated by nerves (Wallace & Maden, 1976). I have applied the same technique to enquire whether or not nerves influence the proximo-distal (PD) polarity of regenerating arms.
Milojevic & Grbic (1925) first tested the permanence of the limb’s PD axis by implanting the distal ends of limb segments into the backs of adult newts. The free end them sometimes regenerated an almost complete limb with a normal PD polarity, opposed to that of the implanted stump. Subsequent experiments have confirmed this result while attempting to reduce wastage and resorption of the reversed limb by improving its blood supply and innervation (Efimov, 1933; Monroy, 1942). Butler (1955) adapted the operation for larval urodeles by allowing an implanted wrist to heal into the flank for several days before amputating the dually attached arm close to the shoulder. Even then, most of the upper arm was resorbed before healing or regeneration occurred near to the elbow (Deck, 1955). Deck & Riley (1958) also noticed considerable regression and a common failure of regeneration when repeating this operation on the hind limbs of larval urodeles. Dent (1954) increased the frequency of regeneration after the same operation on adult newts by deflecting brachial nerves to the implanted wrist and delaying the amputation for 2 weeks. Oberheim & Luther (1958) ingeniously grafted the arm of one salamander larva through the tail muscle of another, diverted sciatic nerves into the host’s wound region and kept the two larvae attached for three days under a light narcosis. Amputating the grafted arm through the humerus and wrist then sometimes caused ‘bipolar regeneration’, at both ends simultaneously.
The extent of the regenerated structures encountered in these experiments varied according to the level of amputation and amount of regression. From a base which conformed to the adjacent region of the limb stump, each regenerate contained the usual sequence of structures which normally lie distal to that region. This feature is commonly ascribed to the law of distal transformation governing normal limb development and regeneration (Stocum, 1978). All the regenerates apparently preserved the anterio-posterior and dorso-ventral axes of the limb from which they arose. Consequently, the formerly proximal end of a right-arm stump regenerated a left hand. Even though the PD polarity of such a regenerate is opposed to that of the implanted stump, it is perfectly normal in respect to the rest of the body and to the local innervation which has grown into the implant. The direction of nerve growth could dictate distal transformation, as several investigators have suggested (e.g. Needham, 1952). If that were so, a reversed aneurogenic arm might be incapable of regenerating distal structures or even show some form of proximal transformation. Precedents for both of these expectations have been described for reversed limb buds (Swett, 1927, 1928; Gräper, 1922), and it is of general interest to discover if the analogy between developing and regenerating limbs can be extended to this situation.
MATERIALS AND METHODS
Reversed aneurogenic arms were produced by combining the techniques of Yntema and Butler outlined in the introduction. That entailed subjecting axolotls (Ambystoma mexicanum) to four successive operations, each of which involved some wastage or mortality. Pairs of embryos were joined in parabiosis before the tailbud stage. The hindbrain and trunk spinal cord were extirpated from the right embryo of each pair on the following day. Further details of these operations are provided by Wallace & Maden (1976). Nerve fibres tend to spread from the nurse embryo to the defective one, so only the right arm of the latter could be treated as sparsely innervated or sometimes completely free from nerves. All such right arms are termed aneurogenic in this report.
Reversal was performed on well grown arms of 25 – 30 mm specimens. The digits were removed from aneurogenic arms (or the right arms of contiol larvae) to create a wounded hand which was implanted into the peritoneal cavity. The specimens were kept virtually immobile in dilute MS 222 overnight, as movement of either the arm or body tends to dislodge the implanted hand. The right arms of a second control series were denervated immediately before implanting the hand. Implanted arms were amputated close to the shoulder one week later, and displaced repeatedly to ensure a complete separation.
Regular inspection during the next 4 weeks provided records of regeneration from the shoulder, from the free end of the arm and from the implanted hand. Reversed arms which showed no regenerate on the free end after five weeks were reamputated and observed for a further period. The orientation of each regenerate was assessed before fixation and preparation of whole mounts stained with methylene blue. Aneurogenic arms were finally sectioned longitudinally and impregnated with silver according to Blest (1976) to display nerve fibres.
A few additional reversals were performed on non-parabiotic aneurogenic larvae which are completely free of nerves (Popiela, 1976), using Oberheim and Luther’s technique outlined in the introduction. One such arm was processed for electron microscopy (cf. Egar, Yntema & Singer, 1973; Popiela, 1976).
RESULTS
Less than half of the parabiotic twins (Fig. 1) survived to feed and grow normally. Subsequent operational failures, fungal infection, and a novel recessive gene which impedes limb development all reduced the number of useful twins to about 20% of the starting material. Aneurogenic arms grew more slowly than those of nurse or control larvae but reached a relatively slender 3- to 4-digit stage. Implanting the hand into the flank only interrupted the brachial blood flow for a few days. The subsequent shoulder amputation eliminated the normal blood supply to the arm but circulation resumed in the implanted wrist two days later. This amputation provoked regeneration from the shoulder, resulting in new 3-digit right hands in as many weeks. The implants regressed initially to lose at least the free end of the upper arm. Some were completely resorbed, most healed near the elbow as a permanent stump, and the rest regenerated as left hands from about the same region (Fig. 2). Differences in healing of the implanted hand affected those responses to a considerable extent, by Dent (1954). These arms could not be actively flexed and so might become implanted more securely. The implanted hand might also attract flank nerves earlier because its own innervation has been destroyed. Table 1 offers mild support for these predictions, in that denervated control arms were less liable to regenerate from the implantation site but regenerated at their free ends more frequently than innervated control arms. In fact, the denervated arms resemble aneurogenic arms more closely in this respect. The expected frequency of bipolar regenerates can be obtained by treating regeneration at the free end and implanted wrist as independent events and calculating their coincidence. The observed frequency of bipolar regenerates is remarkably close to this expectation in each series (Table 1). More importantly, the two control series included cases of regeneration at either or both ends of the reversed arm. Such regenerates were identical to those obtained from reversed aneurogenic arms in position, structure and orientation (Fig. 3).
The static reversed arms, which had merely healed after the first amputation, were reamputated close to the elbow after 5 weeks. Except for one aneurogenic arm, each regenerated a complete wrist and hand from the new wound surface. All these regenerates, like those resulting from the first amputation, clearly preserved the antero-posterior and dorso-ventral axes of the limb stump but showed distal transformation (Fig. 4). The stained skeletal cartilages seen in cleared whole mounts leave no doubt that the free end of the implanted arm regenerated a mirror image of itself. Many of these regenerates contained a surprisingly complete skeleton, often including part of the humerus, but others had regressed further and regenerated correspondingly fewer skeletal parts (Figs. 3, 4). Since regeneration invariably involved distal transformation, these results confirm previous observations and extend them to aneurogenic arms, suggesting that the amount of innervation has no appreciable influence on the structure and orientation of a regenerate.
As expected from previous studies (cf. Egar et al. 1973), most of the reversed aneurogenic arms were sparsely innervated by one or two small bundles of axons which were easily detected in longitudinal sections. Only one regenerate lacked any visible nerve fibres, an isolated case which clearly required more rigorous confirmation. The reamputated aneurogenic arms failed to provide this confirmation, for they had all become innervated in the nine weeks since implantation. Aneurogenic axolotls do not survive long enough to support this experiment unless they are maintained in parabiosis, when they are inevitably invaded by nerves from the nurse larva. Arms of advanced single aneurogenic larvae can be implanted into younger ones, however, and then have time to regenerate. Two advanced regenerates obtained in this way showed the typical features of distal transformation. Only one of them was successfully processed for electron microscopy, but it was certainly devoid of nerves.
DISCUSSION
The regenerated structures noted in this study correspond perfectly to the categories described by Butler (1955) and Deck (1955) for reversed arms of A. maculatum larvae, and provide a means of reconciling their results to those obtained from embryos of the same species. When Swett (1927, 1928) reversed limb buds after Harrison stage 37, the implanted tip often managed to grow out from the flank but the formerly proximal end invariably healed over without regenerating. That was also the commonest result in the present control series. Butler (1955) only described his successful operations on relatively advanced larvae, presumably meaning well-embedded arms which did not grow any fingers at the implantation site. He noted that the free ends of such reversed arms could grow a new forearm and hand but usually did so only after a second amputation, as the present results confirm. Reversed arms commonly regenerate digits at the implantation site and perhaps do so whenever the formerly distal end is partly exposed. The formerly proximal free end of the arm sometimes regenerates promptly and almost always does so following a later amputation. There is clearly no distal dominance of regenerative power, for both ends of the arm can regenerate simultaneously (cf. Monroy, 1942; Oberheim & Luther, 1958). In an analogous experiment performed by Eiland (1975), regeneration occurred at a better innervated site in preference to a more distal one. Better innervation of the implanted hand of a reversed arm could also explain why regeneration occurs more frequently there than at the free end, especially as the reduced innervation of aneurogenic arms tend to reduce this disparity (Table 1).
Since the control reversed arms, like other heterotopically transplanted arms, only gain a subnormal innervation (Deck, 1955; Singer & Mutterperl, 1963), it is tempting to suppose those which did not regenerate promptly had not yet acquired an adequate nerve supply. That is, the innervation of control arms fell within the threshold level of the quantitative neurotrophic theory (Singer, 1952). The identical behaviour of the aneurogenic arms casts doubt on this explanation for they would have to respect a lower threshold level, being more sparsely innervated both before and after implantation. Perhaps a limb can adapt to any amount of innervation, but fails to regenerate only when its nerve supply has been reduced recently. That seems preferable to the threshold concept as a means of explaining both the present results and experiments designed to test the neurotrophic theory (Thornton & Thornton, 1970).
All the regenerates considered here faithfully perpetuated the antero-posterior and dorso-ventral axes of the limb from which they arose, and all showed distal transformation. The former two axes evidently remain fixed after their determination during early development, as Harrison (1921) and others have demonstrated. Contrary to Swett’s (1927) contention, however, the proximo-distal axis is reversible by regeneration. Since this also applies to aneurogenic arms, distal transformation is not determined by nerves which do not seem to have any polarising influence on the regenerate. It is difficult to avoid the conclusion that distal transformation is an autonomous property of the regeneration blastema and one shared by the developing limb bud (Gräper, 1922). Admittedly, the base of the blastema redifferentiates in conformity with the region of the stump from which it originated. That conformity is still commonly ascribed to a stump influence or field affect, but it can equally well be interpreted as a predetermined property of the local cells which form the blastema (Wallace & Watson, 1979). If the original blastemal cells are predetermined in this way, their determination is most easily envisaged in terms of a PD gradient along the intact limb (ignoring the transverse axes). The direction of that gradient, however, is evidently unimportant and probably meaningless once regeneration begins. Perhaps that is implicit in the numerical gradient proposed by Maden (1977) and elaborated by Stocum (1978).
In conclusion, blastemal morphogenesis corresponds closely to that of a limb bud in being an autonomous but internally regulative process. Such a view is reinforced by finding that the establishment of limb axes, which anticipates innervation during development, also occurs without reference to nerves during regeneration.