Forearms of juvenile axolotls contain about 5000 axons, of which only 25% are myelinated and visible by light microscopy. Virtually all the axons degenerate after transection of the brachial plexus, but repeated operations fail to keep the arm completely denervated. Regrown nerve fibres were detected by electron microscopy after 6 weeks of attempted denervation and related to the quantity usually considered necessary for limb regeneration. Such arms regenerated quite normally, provided their innervation had been depleted for several weeks before amputation. Among other ways of reconciling these observations to the neurotrophic theory of limb regeneration, it is suggested that tissues can adapt to deprivation of their nerve supply.

According to the neurotrophic theory of limb regeneration, the establishment and early growth of a regeneration blastema are dependent upon an adequate nerve supply but older regenerates are capable of differentiation, morpho-genesis and some further growth in the absence of nerves. Both sensory and motor nerves contribute to the total nerve requirement in adult urodeles, giving the theory an explicit quantitative basis which is conducive to the argument that higher vertebrates may be incapable of limb regeneration because of an inadequate innervation (Singer, 19521974). The requirement for amphibian limb regeneration has been estimated by partial denervation at about the time of amputation and expressed either as the fibre density at the amputation surface, the fraction of the surface area occupied by axoplasm or, more simply, as a percentage of the normal innervation. The frequency of regeneration exhibited by such partially innervated arm stumps is strongly correlated to their residual innervation in larval salamanders (Karczmar, 1946), or correlated within a ‘threshold’ of about 30 to 50% of the normal nerve supply in newt upper arms (Singer, 1946b).

There are two clear exceptions to the quantitative and threshold aspects of the neurotrophic theory. Transplanted arms can regenerate when very poorly innervated (Singer & Mutterperl, 1963); aneurogenic arms of young larvae regularly regenerate in the virtual or complete absence of nerves (Yntema, 1959; Egar, Yntema & Singer, 1973; Wallace, 1980). Singer (1965) sought to incorporate these exceptions into an addictive version of the trophic theory by postulating either that tissue sensitivity to the trophic factor depended on its prior experience of nerves, or that other tissues could also produce the trophic factor when stimulated by excessive operational trauma. These modi-fications seemed to cover all eventualities and were vindicated by an ingenious test involving the transplantation, innervation and later denervation of aneurogenic larval arms (Thornton & Thornton, 1970). The test may be unnecessarily complicated, however, for a tedious argument presented elsewhere (Wallace, 1981) contends it should be possible to demonstrate the regeneration of denervated arm stumps without resorting to either trans-plantation or the embryonic operations needed to produce aneurogenic arms. A preliminary experiment established that juvenile axolotl arms could regenerate while being denervated at weekly intervals, but only if amputation was delayed for several weeks after the initial denervation. Such arms with early regenerates were judged to be sparsely innervated or devoid of nerves when examined by routine histological staining of paraffin sections (Watson, unpublished). Small unmyelinated axons would have escaped detection by light microscopy (cf. Egar & Singer, 1971), so we have repeated and refined the experiment in an attempt to meet this criticism. We find our results are in better accord with the addictive version of the neurotrophic theory than with its quantitative or threshold aspects.

Ten 55–70 mm long axolotls (Ambystoma mexicanum) were selected as an experimental series. Their right arms were denervated by severing the brachial plexus close to the head of the humerus. This operation was repeated 2 weeks later and after a month had elapsed, cutting a more proximal part of the plexus on each occasion. Both arms were amputated above the wrist at the time of the third denervation, when the specimens had reached lengths of 66-88 mm. A final resection of the right brachial plexus was performed 10 days later and both forearms were preserved 7 days after that, 17 days after amputation.

Ten 60–80 mm long specimens were used as a control series, being subjected to bilateral amputation above the wrist at the same time as the experimental series. Their right arms were denervated 10 days later and both forearms were preserved 17 days after amputation.

All 40 forearms were processed identically by a standard technique for electron microscopy: fixation for 2 h in 2·5% glutaraldehyde in 0·1 M-cacodylate buffer pH 7·4; 1·5 h in 1% osmium tetroxide in the same buffer; dehydration through a series of ethanol steps (70–100%) and two changes in propylene oxide; 1 h in equal parts of propylene oxide and epon, before embedding overnight at 60°C in epon 812. One micron sections were stained with toluidine blue and azur II, and examined by light microscopy to obtain counts of myelinated axons. Thin sections were placed on coated single slot grids, stained with lead citrate and examined with a Phillips 300 electron microscope. A series of overlapping photographs from two major nerves were printed (ca. × 7000) and assembled into montages, from which the density of small unmyelinated axons was estimated relative to the numbers of Schwann cell nuclei. The diameters of 300 axons in each montage were measured to obtain estimates of the mean cross-sectional area and thus of the total axoplasmic area at an amputation surface.

The brachial plexus is easily exposed in these large specimens, so that the initial operation can be guaranteed to transect all nerves to the forearm except for a minor sympathetic supply. Each successive operation becomes more difficult as the amount of scar tissue increases and the nerve trunks become more transparent. The first three denervations were judged to be successful by the reduced growth or atrophy and the permanent immobility of the operated arms throughout the experiment, in contrast to the recovery of movement expected about 3 weeks after a single denervation (Thornton, 1960; Maden, 1977). The success of the final denervation could only be assessed retrospectively from sections.

Ten days after amputation, both arms of all the experimental and control specimens had formed distinct conical blastemata. The control right arms were denervated then, for a cone blastema is still dependent upon the integrity of its nerve supply. All the regenerates were examined 1 week later in order to determine whether or not they had progressed beyond the cone stage (Fig. 1). Confirming the preliminary test, most of the experimental right arms had progressed to palette, notch or early digit regenerates under conditions designed to eliminate their nerve supply. More surprisingly, almost all of the experimental specimens now showed a more advanced regenerate on the denervated right arm than on the innervated left arm (Table 1). We can explain this difference by invoking the Tweedle effect. Tweedle (1971) demonstrated that denervating or amputating one arm impedes the regeneration of the contralateral arm by disturbing transneuronal contacts and causing some chromatolysis of contralateral neurons. Maden (1977) also reported that denervation causes the degeneration of some contralateral axons. The left arms of these experimental specimens would be sensitive to the last two operations on the right brachial plexus, so that their regeneration should be appreciably delayed. The experimental right arms need not be affected, however, assuming their innervation could not be reduced any further. The right arms of the control specimens showed a typical response to denervation 10 days after amputation. None of their cones grew at all in the following week and most of them shrivelled to a smaller blastema. That would also be expected of any experimental right arm which contained nerves when it was amputated but lost them subsequently, as may have happened in a single case (Table 1). The left arms of the control specimens which were also subject to the Tweedle effect consistently reached the palette stage after 17 days, duplicating the rate of regeneration shown by the innervated left arms of the experimental series.

Table 1

Stages of regeneration scored at 17 days after amputation

Stages of regeneration scored at 17 days after amputation
Stages of regeneration scored at 17 days after amputation
Fig. 1

Arms of an experimental (A, B) and a control (C, D) specimen when scored 17 days after amputation. A, C, the innervated left arms both have palette stage regenerates. B, the experimental right arm has regenerated a 3-digit hand. D, the control right arm has regressed from a cone stage, 1 week earlier when it was denervated, to a small blastema.

Fig. 1

Arms of an experimental (A, B) and a control (C, D) specimen when scored 17 days after amputation. A, C, the innervated left arms both have palette stage regenerates. B, the experimental right arm has regenerated a 3-digit hand. D, the control right arm has regressed from a cone stage, 1 week earlier when it was denervated, to a small blastema.

Transverse sections from all 40 forearms were examined by light microscopy, although a few were too poorly orientated or too close to the elbow to give quantitative results. The remaining innervated arms of both series provided a fairly consistent standard for comparison (Table 2). They contained about 1300 large myelinated axons, mostly packed close together in three major central trunks and two radial nerves with up to 12 minor nerves in the dermis and others scattered through the muscle, and about 250 Schwann cell nuclei (Fig. 2). The control right arms contained less than 30 apparently undamaged myelinated fibres, interspersed among empty myelin sheaths and debris (Fig. 3). Many of the Schwann cell nuclei had enlarged at 7 days after denervation, but very few were dividing and the number in each section had not increased appreciably. The experimental right arms contain 0·24 relatively small and lightly myelinated axons, scattered among enlarged intercellular spaces but without much debris, suggesting that material had been lost earlier during the prolonged denervation (Fig. 4). These sections contained significantly fewer than the normal number of Schwann cell nuclei but nuclear divisions were relatively common (Table 2). The counts of myelinated fibres shown in Table 2 reveal that at least 97·5% of the major axons are degenerating 7 days after denervation, while the apparently intact residue may only be resistant internodes of severed axons. The presence of a few lightly myelinated axons in about half of the experimental right arms indicates that repeated denervations were not uniformly successful in excluding regrown fibres or those invading the limb from adjacent spinal nerves. These fibres never amounted to more than 2% of the counts in normal arms, so the experimental right arms must be considered as sparsely innervated unless they contain an excessive number of unmyelinated axons.

Table 2

Light microscope counts of nerve constituents (mean ± standard deviation) from forearm sections

Light microscope counts of nerve constituents (mean ± standard deviation) from forearm sections
Light microscope counts of nerve constituents (mean ± standard deviation) from forearm sections
Fig. 2

Light microscope sections of major nerves in the forearm. 2, normal myelinated axons in experimental left arm EL 10; 3, degenerating axons and myelin debris, one week after denervation in control right arm CR 8; 4, absence of myelinated axons after chronic denervation in experimental right arm ER 1.

Fig. 2

Light microscope sections of major nerves in the forearm. 2, normal myelinated axons in experimental left arm EL 10; 3, degenerating axons and myelin debris, one week after denervation in control right arm CR 8; 4, absence of myelinated axons after chronic denervation in experimental right arm ER 1.

Fig. 3

Light microscope sections of major nerves in the forearm. 2, normal myelinated axons in experimental left arm EL 10; 3, degenerating axons and myelin debris, one week after denervation in control right arm CR 8; 4, absence of myelinated axons after chronic denervation in experimental right arm ER 1.

Fig. 3

Light microscope sections of major nerves in the forearm. 2, normal myelinated axons in experimental left arm EL 10; 3, degenerating axons and myelin debris, one week after denervation in control right arm CR 8; 4, absence of myelinated axons after chronic denervation in experimental right arm ER 1.

Fig. 4

Light microscope sections of major nerves in the forearm. 2, normal myelinated axons in experimental left arm EL 10; 3, degenerating axons and myelin debris, one week after denervation in control right arm CR 8; 4, absence of myelinated axons after chronic denervation in experimental right arm ER 1.

Fig. 4

Light microscope sections of major nerves in the forearm. 2, normal myelinated axons in experimental left arm EL 10; 3, degenerating axons and myelin debris, one week after denervation in control right arm CR 8; 4, absence of myelinated axons after chronic denervation in experimental right arm ER 1.

Electron microscopy of innervated and recently denervated control arms confirmed the preceding observations and extended them to unmyelinated fibres. All the major nerves normally contained bundles of small axons embedded in the folds of scattered Schwann cells. The vast majority of all axons showed signs of degeneration 7 days after denervation, and this applied to all the nerves examined of control right arms. The six experimental right arms examined by electron microscopy presented a more confusing picture, for their major nerve trunks contained large spaces with fine debris between the Schwann cells whose processes enveloped typical axonal profiles (Fig. 5). Although the Schwann cell processes in severed nerves can fold round each other to give a very similar appearance (Payer, 1979), the elongated profiles seen in oblique and longitudinal sections convince us that most of them are genuine axons. A montage of prints from a major nerve of the smallest experimental right arm yielded an average ratio of 16 axons to each Schwann cell nucleus in the area. Averaging two perpendicular diameters of 300 axons, their mean diameter was 0-96 μm (range 0·3–3 μm) and the mean cross-sectional area was calculated to be 0·885 μm2. A similar montage from a major nerve of the smallest innervated arm yielded a lower density of unmyelinated axons, 13 per Schwann cell nucleus, of much the same size range with mean values of 1 μm diameter and 0-995 μm2 area. The myelinated fibres had a mean diameter of 3-7μm (range 2-9 μm) and area of 11·8 μm2. These average values misrepresent the irregular distribution of unmyelinated axons, which occurred in bundles in both montages, but they can be combined with light microscope data to calculate the relative innervation of the experimental right arm. As shown in Table 3, this arm contained a substantial number of axons, amounting to almost half the normal supply and nearly 90% of the normal density of axons at an amputation surface. That degree of innervation might be exceeded slightly in other experimental right arms which contained a few remyelinated axons. These arms can still be considered as sparsely innervated in the restricted sense that the total mass of axoplasm (or axoplasmic area in a section) is much less than that present in a normally innervated arm.

Table 3

Characteristics of the two smallest arms from the experimental series, combining data from thick sections and EM montages

Characteristics of the two smallest arms from the experimental series, combining data from thick sections and EM montages
Characteristics of the two smallest arms from the experimental series, combining data from thick sections and EM montages
Fig. 5

Detailed structure from a major nerve in experimental right arm ER1, showing unmyelinated axons (A) enveloped in a process (P) of Schwann cell cytoplasm; Schwann cell nucleus (N).

Fig. 5

Detailed structure from a major nerve in experimental right arm ER1, showing unmyelinated axons (A) enveloped in a process (P) of Schwann cell cytoplasm; Schwann cell nucleus (N).

Prolonged denervation of an axolotl arm prior to amputation clearly allows it to regenerate through all the stages which are usually dependent upon a nerve supply. If we had been content to assess the residual innervation of these arms by light microscopy, we should have been satisfied they were virtually devoid of nerves and thus concluded nerves were dispensible agents in re-generation. The substantial numbers of fine axons in these arms 17 days after amputation compel us to moderate that conclusion, but not to abandon it altogether.

Observations on axolotl arms subjected to a single denervation (Egar, unpublished) indicate that axons regrow into the upper arm within a week and increasing numbers of them gain a thin myelin sheath by 11–14 days. The scarcity of myelinated fibres in the repeatedly denervated arms considered here thus implies that most of the detected axons only entered the arm after amputation, perhaps during the final week. Others might have been present earlier and degenerated after a subsequent denervation, of course, but our experience of the increasing difficulty in finding nerves to resection convinces us the arms were more sparsely innervated during the early stages of regeneration than at the end of the experiment. In that case, Table 3 probably over-estimates the residual nerve supply at the actual amputation surface 17 days previously.

Assuming axons of all sizes dispense equal amounts of trophic factor, then its concentration should be proportional to the density of axons at an amputation surface. Partly due to the cessation of growth and muscular atrophy during prolonged denervation, regrown axons amounting to only half the normal innervation achieved about 90% of the normal density (Table 3). We cannot pretend that estimate is accurate to within 10% or that it is valid for all the experimental right arms, but the kind of effect predicted by the neurotrophic theory here is a greater delay of regeneration than that ascribed to the Tweedle effect on the contralateral innervated arms. Karczmar (1946) noted a reduced rate and frequency of regeneration by arms of A. maculatum larvae when he reduced their innervation to 90% at the time of amputation, although axolotls do not respond so clearly to partial denervation (Egar, unpublished). We conclude there is a discrepancy between the present results and previous ones which is probably related to the relative timing of denervation and amputation.

The main alternative version of the neurotrophic theory assumes large axons secrete more trophic factor than small ones and consequently relates the frequency of regeneration to the fraction of the amputation surface occupied by axoplasm (Singer, Rzehak & Maier, 1967; Singer, 1974). Table 3 reveals a serious discrepancy from this assumption, for arms which had regenerated rapidly and perfectly only contained about 11% of the normal axoplasmic area or 21% when normalized to equal sized arm cross sections. If the calibre of nerve fibres is an important parameter of the trophic factor then prolonged denervation before amputation certainly disturbs its expected relationship to regeneration.

Curiously enough, some of the most persuasive evidence adduced for the quantitative neurotrophic theory also comes from experiments involving delayed amputation. A regenerated motor supply provides sufficient innervation for about 50% of the tested newt arms to regenerate (Singer, 1946a; Sidman & Singer, 1960), or for regeneration in all the larval salamander arms tested by Thornton (1960). It is usually argued that a hyperplastic reinnervation by motor nerves must have boosted the supply of trophic factor above a threshold value in order for regeneration to occur several weeks or months later. No fibre counts were reported for these experiments, however, and no attempt was made to determine if the timing of amputation influenced the results. The present results supply this missing control, suggesting the known delay of amputation provides a more valid explanation of regeneration than the presumed degree of innervation.

The data considered here and reviewed in more detail by Wallace (1981) oblige us to reformulate the neutrotrophic theory in a more flexible way. Nerves undoubtedly have a trophic influence on amputated limbs and young regenerates, for an initial denervation then reduces a variety of synthetic activities, cell division and growth (e.g. Singer & Caston, 1972). The regeneration of aneurogenic arms and the present results, however, suggest there can be no fixed demand for a particular limiting threshold quantity or density of nerve fibres. Apparently, an arm can become accustomed in a matter of weeks to a depleted residual innervation, or to none at all, and then regenerates perfectly normally as long as that degree of innervation is maintained. The simplest interpretation of the nervous control over regeneration, therefore, is that tissues are sensitive (or addicted) to their current supply of neutrotrophic factor. A marked reduction of innervation, causing an abrupt step-down of the trophic supply, results in temporary withdrawal symptoms which either prevent regeneration or delay it until the tissues have adapted to the new level. Adaptation could be achieved by a local production of trophic factor to compensate for that normally supplied by the nerves. It is not related to operational trauma, however, for Karczmar (1946) often prevented or delayed regeneration by a sequence of repeated denervations like those employed here. There is no evidence for any alternative source of the trophic factor at present, but compensatory production may offer one way of reconciling the present results to the quantitative aspects of the neurotrophic theory.

We are conscious of how much this study depends upon investigations conducted by Marcus Singer over the past 40 years, even if our conclusions make this seem a rather back-handed compliment. M. Egar was supported by N1H Grant NS-07403-19 to Dr Singer, and thus feels doubly grateful. A. Watson acknowledges receipt of an SRC Studentship.

Egar
,
M.
&
Singer
,
M.
(
1971
).
A quantitative electron microscope analysis of peripheral nerve in the urodele amphibian in relation to limb regenerative capacity
.
J. Morph
.
133
,
387
397
.
Egar
,
M.
,
Yntema
,
C. L.
&
Singer
,
M.
(
1973
).
The nerve fiber content of Amblystoma aneurogenic limbs
.
J. exp. Zool
.
186
,
91
95
.
Karczmar
,
A. G.
(
1946
).
The role of amputation and nerve resection in the regressing limbs of urodele larvae
.
J. exp. Zool
.
103
,
401
427
.
Maden
,
M.
(
1977
).
The role of Schwann cells in paradoxical regeneration in the axolotl
.
J. Embryol. exp. Morph
.
41
,
1
13
.
Payer
,
A. F.
(
1979
).
An ultrastructural study of Schwann cell response to axonal degeneration
.
J. comp. Neurol
.
183
,
365
383
.
Sidman
,
R. L.
&
Singer
,
M.
(
1960
).
Limb regeneration without innervation of the apical epidermis in the adult newt, Triturus
.
J. exp. Zool
.
144
,
105
110
.
Singer
,
M.
(
1946a
).
The nervous system and regeneration of the forelimb of adult Triturus. IV The stimulating action of a regenerated motor supply
.
J. exp. Zool
.
101
,
221
239
.
Singer
,
M.
(
1946b
).
The nervous system and regeneration of the forelimb of adult Triturus. V. The influence of number of nerve fibers, including a quantitative study of limb innervation
.
J. exp. Zool
.
101
,
299
337
.
Singer
,
M.
(
1952
).
The influence of the nerve in regeneration of the amphibian extremity
.
Q. Rev. Biol
.
27
,
169
200
.
Singer
,
M.
(
1965
).
A theory of the trophic nervous control of amphibian limb regeneration, including a re-evaluation of quantitative nerve requirements
.
In Regeneration in Animals and Related Problems
(ed.
V.
Kiortsis
&
H. A. L.
Trampusch
), pp.
20
30
.
Amsterdam
:
North-Holland
.
Singer
,
M.
(
1974
).
Neutrotrophic control of limb regeneration in the newt
.
Ann. N.Y. Acad. Sci
.
228
,
308
321
.
Singer
,
M.
&
Caston
,
J. D.
(
1972
).
Neurotrophic dependence of macromolecular synthesis in the early limb regenerate of the newt, Triturus
.
J. Embryol. exp. Morph
.
28
,
1
11
.
Singer
,
M.
&
Mutterperl
,
E.
(
1963
).
Nerve fiber requirements for regeneration in forelimb transplants of the newt Triturus
.
Devi Biol
.
7
,
180
191
.
Singer
,
M.
,
Rzehak
,
K.
&
Maier
,
C. S.
(
1967
).
The relation between the caliber of the axon and the trophic activity of nerves in limb regeneration
.
J. exp. Zool
.
166
,
89
98
.
Thornton
,
C. S.
(
1960
).
Regeneration of asensory limbs of Ambystoma larvae
.
Copeia
4
,
371
373
.
Thornton
,
C. S.
&
Thornton
,
M. T.
(
1970
).
Recuperation of regeneration in denervated limbs of Ambystoma larvae
.
J. exp. Zool
.
173
,
293
301
.
Tweedle
,
C.
(
1971
).
Transneuronal effects on amphibian limb regeneration
.
J. exp. Zool
.
177
,
13
29
.
Wallace
,
H.
(
1980
).
Regeneration of reversed aneurogenic arms of the axolotl
.
J. Embryol. exp. Morph
.
56
,
309
317
.
Wallace
,
H.
(
1981
).
Vertebrate Limb Regeneration
.
Chichester
:
Wiley
(in the press).
Yntema
,
C. L.
(
1959
).
Regeneration in sparsely innervated and aneurogenic forelimbs of Amblystoma larvae
.
J. exp. Zool
.
140
,
101
123
.