ABSTRACT
Denervated forelimbs and contralateral innervated forelimbs of Ambystoma larvae were injured internally distal to the elbow by compression with watchmaker’s forceps. Innervated controls completely repaired the crush injury within one week; denervated limbs failed to repair the injury and exhibited varying degrees of limb regression. Histological examination revealed that the process of tissue dedifferentiation initiated by injury was more extensive in denervated, regressing limbs than in controls. In innervated limbs, both the DNA labelling index and the mitotic index peaked approximately 4-6 days after the injury and returned to baseline levels by 10 days. In denervated limbs, the DNA labelling index also increased and remained at an elevated level for at least 2 weeks after the injury, but significant mitotic activity was not observed. The data indicate that intact nerves are not needed for cellular dedifferentiation, cell cycle re-entry, and DNA synthesis in injured limbs, but are required for the cells to proliferate and repair the injury. These results are discussed together with those of similar experiments on the role of nerves during the initiation of epimorphic regeneration in amputated limbs.
INTRODUCTION
The process of limb regeneration in amphibians depends directly on at least three factors: a stimulus produced by the tissue injury which initiates the ‘dedifferentiative’ cellular events leading to regeneration; an agent released from nerves which promotes mitotic activity in the cells affected by the injury; and an inductive influence from an apical wound epithelium which causes the underlying cells to remain in the proliferative state and give rise to a blastema out of which the regenerate develops (Thornton, 1968; Tassava & Mescher, 1975; Wallace, 1981). The importance of nerves in promoting cell division in the injured limb tissues is demonstrated more dramatically in larval urodeles than in adults. The latter fail to regenerate denervated and amputated limbs, but do heal the stump with a fibrocellular scar and cartilage over the bone end (Schotté, 1926). If larval limbs are denervated before or shortly after amputation, however, the consequent absence of normal mitotic activity in the stump tissues results not only in the failure of the limb to establish a blastema and regenerate, but also in the abrogation of the tissue repair process (Butler & Schotté, 1941). Barring re-innervation, dedifferentiation continues unchecked and the resultant cells are eliminated, causing the structural regression of the limb as far as the shoulder (Schotté & Butler, 1941 ; Schotté & Karczmar, 1944; Karczmar, 1946). Thornton (Thornton, 1953; Thornton & Kraemer, 1951) showed that amputation trauma was not necessary to cause regression of denervated larval limbs, but that simple compression of the limb with watchmaker’s forceps so as to injure the muscle and cartilage without breaking the skin was sufficient to induce the histological changes leading to limb regression. Thus in larval limbs the neural influence is required even for the relatively simple process of tissue repair.
Recent histological and autoradiographic studies have attempted to elucidate the role of nerves in amphibian limb regeneration by examining the effects of denervation on cell cycle parameters during tissue dedifferentiation or in the blastema. It has been learned that the neurotrophic influence is not needed for the initial events of dedifferentiation which follow amputation, including reentry of muscle and connective tissue cells into the proliferative cycle and replication of their DNA (Tassava, Bennett & Zitnik, 1974; Mescher & Tassava, 1975; Tassava & Mescher, 1976; Maden, 1978). Although the percentage of cells incorporating tritiated thymidine in denervated limbs increases in parallel with that of control limbs, the reports by Tassava and co-workers indicate that the denervated cells do not divide, i.e. the percentage of cells in mitosis does not increase significantly in the denervated limb stumps. This finding led to the suggestion that the neurotrophic effect on the dedifferentiating cells is exerted during the G2 phase of the cell cycle (Tassava & Mescher, 1975; Tassava & McCullough, 1978). Maden’s (1978) cell cycle studies were similar to those of Tassava et al. (1974) but yielded results which suggested other, less specific, controls of cell cycle events by the nerves in the amputated limb. The methods of cell cycle analysis have been applied here to injured larval limbs with and without nerves, like those originally studied by Thornton (1953), in an attempt to clarify the neurotrophic control of cellular events following injury to the urodele limb.
MATERIALS AND METHODS
Ambystoma maculatum embryos, collected locally, were allowed to hatch in the laboratory and the larvae were raised in separate dishes to a length of 25-30 mm with daily feedings of freshly hatched brine shrimp (Artemia salina). Fifty animals were anaesthetized lightly with MS 222 (Sigma) and the left forelimbs were denervated by transecting the third, fourth, and fifth brachial nerves distal to the brachial plexus. At the same time both forelimbs were injured according to the method of Thornton & Kraemer (1951) by compressing 0·5 mm of the limb between the elbow and wrist with watchmaker’s forceps, taking care not to break the skin. Left forelimbs were re-denervated every six days. Every other day for 2 weeks following the initial operations, three larvae were selected at random and each was injected intraperitoneally with 5 μCi [3H]thymidine (ICN Pharmaceuticals, specific activity 73·5 Ci/mmole). After 3 h of incorporation, each larva was fixed in Carnoy’s fluid. Three uninjured, undenervated animals were labelled similarly to establish ‘day 0’ levels of cell labelling. Experimental larvae not used for DNA labelling were maintained for six weeks to observe the morphological changes accompanying regression of the denervated limbs.
Forelimbs of the labelled animals were embedded in paraffin and sectioned longitudinally at 10 μm. Sections from each limb were processed either for autoradiography or by the Feulgen method as described previously (Mescher & Tassava, 1975). Mesodermal tissues in the area between the elbow and wrist, i.e. the radius, ulna, adjacent muscles and loose connective tissue, were examined quantitatively with the aid of a grid in the microscope’s eyepiece. For each limb a DNA labelling index (percentage of cells labelled with at least a dozen silver grains) was determined from the haematoxylin- and eosin-stained auto-radiographs and a mitotic index (percentage of cells with mitotic figures) was determined from both the Feulgen-stained sections and the autoradiographs. Approximately 2500 cells were counted per limb, using non-adjacent sections to give representative areas throughout the entire tissue sample. Statistical significance of the data was determined by means of a t test.
RESULTS
Morphology and histology
The innervated limbs appeared to recover quickly from the injury, and by one week after the operations no visible indication of the tissue crushing could be observed. However, none of the denervated limbs ever recovered completely, and the deleterious effects of the injury gradually became more pronounced in the continued absence of nerves. Within three days of the operations, four denervated limbs detached at the point of compression and the stumps regressed to various levels in the upper arm region over the next five weeks. All of the denervated, injured limbs showed morphologically some signs of the regression described for such limbs by Thornton & Kraemer (1951). Reduction in the overall size of the still-compressed forearm and shrinkage of the digits were apparent within 10-20 days of the injury and in many cases gradual disappearance of the forearm, digits and elbow had occurred by the end of the 6-week observation period.
The histological changes accompanying the onset of regression in denervated, injured limbs have been detailed by Thornton (1953) and resemble the de-differentiative events that occur in denervated, amputated larval limbs (Butler & Schotté, 1941). Only the first two weeks after denervation and crushing have been studied histologically here. Sections of the forelimbs from one animal 12 days after injury are shown in Fig. 1. In the denervated limbs, the cartilage of the radius and ulna at the site of injury became progressively vacuolated and devoid of cells, while adjacent muscle detached and dedifferentiated. Wrist elements and the epiphyseal cartilage of the radius and ulna were more resistant to vacuolation and remained intact with little histological change during the 14 days of observation. Tissues of the innervated limbs, although injured in a similar manner, showed much less extensive dedifferentiation. Cells lost from the crushed cartilage were quickly replaced and muscle remained present. By 10-14 days after the injury, innervated limbs were completely repaired and the presence of new, less deeply stained cartilage at the level of limb compression was the only histological evidence that an injury had occurred.
Autoradiographs of sections through the forearm region of the denervated (a) and innervated (b) limbs of an Ambystoma larva 12 days after crush injury. The approximate level of the crush is indicated by the large arrows. Note the overall reduction in size and amount of soft tissue, as well as the higher percentage of nuclei labelled with [3H]thymidine (small arrows), in the denervated limb. Both photographs were taken at 100 x magnification.
Autoradiographs of sections through the forearm region of the denervated (a) and innervated (b) limbs of an Ambystoma larva 12 days after crush injury. The approximate level of the crush is indicated by the large arrows. Note the overall reduction in size and amount of soft tissue, as well as the higher percentage of nuclei labelled with [3H]thymidine (small arrows), in the denervated limb. Both photographs were taken at 100 x magnification.
DNA labelling and mitotic indices
Autoradiography revealed significant increases in the number of cells under-going DNA synthesis in the mesodermal tissues of the larval forearms in the first few days following the injury. This was true both for the innervated limbs, involved in the process of tissue repair, and the denervated limbs, which underwent the phenomenon of regression rather than repair. Moreover, while the DNA labelling index of the innervated forearms peaked on day 4 and returned to the normal level by day 10, that of the denervated limbs remained at a significantly (P < 0·025) elevated level throughout the two-week period of measurement (Fig. 2). The mean labelling index and standard error for each day are given in Table i. Figure 3 shows the high percentage of labelled cells still found in the denervated, regressing limbs 14 days after the injury.
DNA labelling indices of innervated (•) and denervated (○) larval limbs during the two weeks following crush injury. Each point indicates the mean of three limbs. Standard errors are given in Table 1.
DNA labelling indices of innervated (•) and denervated (○) larval limbs during the two weeks following crush injury. Each point indicates the mean of three limbs. Standard errors are given in Table 1.
Autoradiograph of a section through the injured region of a denervated forelimb 14 days after crush injury. Note the large number of nuclei labelled with [3H]thymidine (arrows). 160 x magnification.
Thus cellular dedifferentiation in the denervated, injured larval limbs resembles that of amputated limbs, with or without nerves, in being characterized by cell cycle re-entry or acceleration and increased levels of DNA replication.
DNA labelling and mitotic indices in injured denervated and innervated forelimbs of Ambystoma larvae

Mitotic activity in the mesodermal cells of the innervated forearms was significantly increased for eight days after the injury, but by ten days had returned to the level shown by uninjured larval limbs (Fig. 4 and Table 1). In marked contrast to the innervated limbs, however, the denervated limbs were found to contain very few or no mitotic cells despite searches through every section at 400 x magnification. The mitotic index of the denervated fore-arms remained at or below that of control, uninjured limbs. It appears there-fore that the failure of these limbs to heal the injury and the loss of tissue which results in limb regression are due to failure of the dedifferentiated cells to divide normally in the absence of nerves.
Mitotic indices of innervated (•) and denervated (○) larval limbs during two weeks following crush injury. Each point indicates the mean of three limbs. Standard errors are given in Table 1.
Mitotic indices of innervated (•) and denervated (○) larval limbs during two weeks following crush injury. Each point indicates the mean of three limbs. Standard errors are given in Table 1.
DISCUSSION
The results presented here confirm the original observations of Thornton (Thornton, 1953; Thornton & Kraemer, 1951) with denervated, injured forelimbs of Ambystoma larvae and extend those findings in terms of the cell cycle. The tissue dedifferentiation which characterizes the onset of limb regression in these limbs has been found to resemble that which follows amputation of denervated larval limbs, not only histologically (Thornton, 1953), but also in the increased percentage of mesodermal cells replicating DNA. Like amputation (Tassava et al. 1974; Tassava & Mescher, 1976; Maden, 1978), the crush injury causes renewed or accelerated cell cycling and an increased DNA labelling index in the absence of intact nerves. The dedifferentiation/cell cycling stimulus elicited by injury apparently involves only the internal limb tissues, since open skin wounds affecting only the superficial tissues do not promote dedifferentiation and regression of denervated limbs (Thornton & Kraemer, 1951). Renewed cell cycling in response to tissue injury probably involves agents released during inflammation (see, e.g., Greenburg & Hunt, 1978). However, from an analysis of the regression process in denervated, amputated larval Ambystoma forelimbs, Karczmar (1946) suggested that nerves, degenerating as a result of the proximal transection, release a factor promoting the cellular events leading to regression. This was based on his finding that denervation of the limbs 4 days after amputation greatly accelerated dedifferentiation and regression, but that this process was retarded if denervation was performed 10 days prior to amputation. Thus one can speculate that nerves may participate in modulating the tissue dedifferentiation following limb amputation either by axoplasmic trophic factor(s) (Maden, 1978) or by factors released from degenerating myelin (Abercrombie & Johnson, 1946; Gospodarowicz & Mescher, 1980).
While any sort of neural involvement in the onset of dedifferentiation is uncertain, there is clear evidence for a neurotrophic influence being required to effect repair of injured larval limbs (Thornton, 1953) and, in the case of amputated urodele limbs, to act together with the wound epithelium to bring about formation of a regeneration blastema (Singer, 1952, 1978). In the events leading to blastema formation, the principal role of the nerves seems to be to promote mitotic activity in the dedifferentiating cells (Singer, 1952; Thornton, 1968). Tassava & Mescher (1975), discussing the factors which initiate limb regeneration, suggested that nerves act during the G2 phase of the cell Oyele to allow mitosis. This hypothesis was based primarily on the observation that the early events of dedifferentiation and the initial increase in DNA labelling index were found to occur normally, but without subsequent cell division in denervated, amputated limbs of both larval axolotls (Tassava et al. 1974) and adult newts (Mescher & Tassava, 1975). The finding reported here, that during the initiation of regression in denervated, injured larval limbs there is an increase in the percentage of cells replicating DNA but no concomitant increase in the percentage of cells in mitosis, emphasizes once again that intact nerves are not needed for cell cycle re-entry and the S phase, but are necessary for the mitotic activity which leads to repair of the injury. The present results also support the explanation suggested by Tassava & Mescher (1975) for regression of denervated, injured larval limbs: that the dedifferentiated, non-dividing cells of denervated limbs are not viable and are removed from the limb. According to that hypothesis, gradual resorption of the limb would occur because the injury effect persists, causing more and more cells to dedifferentiate and replicate DNA, and without neural support the cells do not complete a normal cell cycle, become non-viable and are eliminated. This view predicts continued incorporation of [3H]thymidine by cells of denervated, injured limbs as reported here (Fig. 2). The elevated DNA labelling index and barely detectable level of cell division are consistent with the possibility of a G2 block to the dedifferentiating cells’ proliferative cycle in the absence of nerves. However, these data do not rule out other possibilities, such as a general neurotrophic effect on the rate of cell cycling so that in denervated limbs levels of mitosis leading to regeneration or repair would be precluded (see Mescher & Tassava, 1975; Tassava & McCullough, 1978; Globus, 1978).
Maden (1978) has repeated (with a variety of procedural differences) the experiment of Tassava et al. (1974) and, unlike those authors, found a significant number of mitotic cells in denervated limb stumps of larval axolotls. The reason for the difference in results is not clear, but the possibility of incomplete elimination of all nerve fibres by the denervation procedure of crushing the brachial plexus (Maden, 1978) should be considered. The presence of nerve branches outside the primitive, anastomosing plexus of amphibian limbs and the generalized distribution of nerves in this region of premetamorphic urodele limbs in particular are discussed in reviews by Detwiler (1933), Strauss (1946) and Hughes (1968). Despite this, Maden did not verify the absence of nerves in the limbs by nerve staining. His suggestion that in denervated limbs fewer cells than normal enter the cell cycle and divide appears to be contradicted by his own data, which show identical labelling indices in both innervated and denervated limbs until the onset of increased mitosis four days after amputation. In a related study Maden (1979) used Feulgen microdensitometry to show that cells of ‘cone stage’ blastemas accumulate in G1 after denervation, i.e. have a 2C DNA content, and cited this finding as evidence against the hypothesis of Tassava & Mescher (1975). However, the result he obtained could have been predicted, since it is well known that larval Ambystoma blastemas denervated at such a late stage of growth almost always continue to develop (Schotté & Butler, 1944; Butler & Schotté, 1949; Singer, 1952). Denervated cone-stage blastemas of newts show continued mitotic activity for at least two weeks (Singer & Craven, 1948) and normal tissue differentiation (Powell, 1969). The cells of these nerve-independent blastemas would be expected to accumulate in Gi/Go with a 2C DNA content as they begin to differentiate. The finding of a ‘Gi block’ (Maden, 1979) in such blastemas is, therefore, not surprising and is irrelevant to neurotrophic control of the cell cycle during earlier preblastemic or nerve-dependent blastema phases of regeneration when nerves are of major importance.
Thus the question of whether nerves control a specific phase of the proliferative cycle of mesodermal cells during the initiation of amphibian limb regeneration or repair remains open. The nerve-dependent repair of injured limbs in Ambystoma larvae is a system well suited for studies designed to answer that question, since it is not complicated by the influence of a wound epithelium converting the process of tissue repair into one of epimorphic regeneration (Tassava & Mescher, 1975; Mescher, 1976). Finally, as the discussion abov$ emphasizes, it is important for future work on the neurotrophic control of cell cycle events during limb regeneration to be based on what is already known about the role of nerves in this process.
ACKNOWLEDGEMENT
This work was supported by NIH grant GM 27735 to the author and a NIH Biomedical Research Support Grant to the George Washington University.