1. The bilateral removal of the posterior neural folds of axolotl neurulae may lead to the development of larvae in which a greater or lesser length of the ventral fin of the tail between the cloaca and a point just short of the tail-tip is missing. In other cases regulation permits the development of a normal-looking tail.

  2. The later amputation of such ‘normal’ tails, or of defective ones where the plane of amputation lies anterior to the defective region, is followed by the formation of normal regenerates.

  3. The amputation of defective tails where the plane of amputation passes through a region lacking ventral fin is followed by the formation of a regenerate lacking ventral fin as far posteriorly as a point just short of the tail-tip.

  4. The presence or absence of a ventral tail-fin in the regenerate depends upon its presence or absence in the plane of amputation, and is not otherwise related to the degree of abnormality of primary urogenesis.

  5. Tails which have defective ventral fins and poorly developed somitic musculature (as a consequence of simultaneous removal of posterior neural folds and part of the posterior medullary plate) behave in regeneration as do those lacking ventral fin only, even where the plane of amputation passes through a region with deficient somites.

  6. The removal of the dorsal half of the tail-bud of embryos of R. temporaria may also be followed by almost normal urogenesis as a consequence of regulation. In most cases, however, the resulting tails are extensively defective lacking a region of the dorsal fin, parts of the spinal cord, and having poorly developed somitic musculature.

  7. In nearly every case the amputation of these tails was followed by the formation of a regenerate that was normal if the original tail was ‘normal’ or if the plane of amputation lay anterior to the region of the tail which lacked a dorsal fin. Where the plane of amputation passed through a region lacking dorsal fin the regenerate was formed without one.

  8. In a very small number of cases exceptional and defective regenerates were formed, either after a first or a subsequent amputation, where normal ones were to be expected. The occurrence and significance of these exceptions are discussed.

  9. It is concluded that the successful completion of normal primary urogenesis in amphibians is not a prerequisite of the formation of a normal regenerate. It remains possible that the formation of normal regenerates from abnormal tails involves regulative properties in the blastema.

  10. The need for representation of tail-fin at the plane of amputation if it is to be present in the regenerate is further confirmation of the departure of amphibian tail regeneration from the pattern established for the urodele limb.

When the loss of an animal member is followed by its regeneration the new structure is not necessarily a replica of the old. It may, for example, be an anatomically inadequate substitute, or hypomorph. In certain cases it is recognizably the equivalent of some other member characteristic of the species in question— a heteromorph. Systemic and environmental influences operating during the process of regeneration sometimes play an important part in deciding what is to be formed, but so, too, of course, do the properties of the cells of the blastema and stump (for review see A. E. Needham, 1952). These cells, and their forebears, have all had the experience of being part of an organism undergoing primary embryogenesis, but it is far from clear to what extent this experience has relevance for the successful completion of regeneration.

It is true that the very capacity to regenerate at all is a product of normal development and usually arises only after embryogenesis is accomplished, but this is another matter. Once given the ability to regenerate, is the information contained within the cells of the blastema that guides their morphogenesis inherited without change through the cell lineage which connects them to the fertilized egg, or is part of it acquired during embryogenesis?

So long as regeneration is studied only where it follows the removal of normal structures this question must remain unanswered. But structures whose primary development has been abnormal offer us the possibility of finding out, to put it in an extreme and simple form, how far regeneration is a process tending to the reproduction of what was there and how far to the production of what should have been there. It is unlikely that the answer will be the same in every case. Indeed, we know that it will not be, for a white axolotl (i.e. one homozygous for the gene d) is, in one sense, abnormal and will replace a lost hand by another white one, whereas transplanted limb disks may give rise to malformed limbs which, on amputation, can be replaced by normally formed ones (Harrison 1918; Swett, 1924).

It is clear that abnormalities of different kinds will have different meanings for this problem. Unfortunately no classification of them can be wholly satisfactory. Thus the following provisional scheme cannot claim to be based upon mutually exclusive categories with a fundamental status in teratology, since only in extreme cases will an abnormality lie almost entirely within one category alone. At best it can help to make possible an orderly approach to the relationship of regeneration to abnormal development.

A. Inherited abnormality

A.l. Chromosomal defects whose origin may lie a few or many generations before the fertilization of the zygote in which they become manifest. Usually assumed to be represented in each somatic cell.

A.2. Non-chromosomal: all other abnormalities due to the constitution of the gametes, including those which are products of the parental genotypes.

B. Abnormality due to environmental insult

B.l. In post-embryonic life: abnormalities resulting from interference with a structure after the completion of its primary morphogenesis, but which are not made good by regenerative or other regulative processes.

B.2. In early development: abnormalities resulting from the failure of a developing system to regulate after suffering interference.

C. Experimentally produced artefacts

Structures which never, or exceedingly rarely, occur in nature. Species or organ chimaerae, induced supernumerary structures, &c.

It must be stressed, however, that an adult organism may owe its normal appearance to regulatory processes (e.g. in cases of incomplete expression of certain genotypes or of embryonic regulation after surgical interference). Here the course of primary morphogenesis has not been normal although its outcome is. We cannot exclude the possibility that the amputation of a normal-looking structure might be followed by the formation of a regenerate showing features of the non-regulated condition. This possibility is perhaps especially important in cases of poor penetrance or minimum expression of genetic defects, but, as will be seen later, has also to be considered in a case where regulation has made good extensive loss of material in early embryonic stages.

Such information as we already have on regeneration after the amputation of abnormal structures, and it includes cases in each of these categories, will be fully discussed in a later paper. All that need be emphasized at this stage is the impossibility of predicting, a priori, that a constant pattern of behaviour will emerge within each category. Thus in the cases of A.l, abnormalities, even though we may assume that the genetic constitution of the blastema cells is identical with that of the fertilized egg from which they are descended, it does not follow that the morphogenesis of a regenerate is sufficiently similar to that of primary embryogenesis for replication of the abnormality to occur. We may well find that the action of certain genes is limited to specific phases of development and that the amputation of a genetically abnormal structure will be followed by the formation of a normal one.

In the previous literature there are two important accounts of work on abnormal amphibian tails. The first, that of Vogt (1931), has received considerable attention. By removing the ectoderm lying laterally to the blastopore of urodele neurulae he obtained larvae in which the ventral fin was missing over part of the length of the tail. When such defective tails were transected in the region of the defect the regenerates subsequently formed lacked a ventral fin. Although he makes only a brief reference to this result it has been taken to show that animals cannot regenerate that which they have never possessed (Huxley & de Beer, 1934), with which view it is, of course, consistent.

Later, Woronzowa & Liosner, of whose work I have only been able to learn from the account in Woronzowa (1949), claimed more surprising results from work with Rana temporaria. They produced tadpoles with severe defects of the dorsal part of the tail by removing the dorsal half of the tail-bud of young embryos. Not all their mutilated animals had obviously defective tails, in many cases regulation during urogenesis succeeded in producing an apparently normal tail. Subsequently amputation was performed, the plane of transection being about half-way along those tails which were of normal appearance and anterior to the defective region in the abnormal ones. The majority of all regenerates were of normal appearance, but a substantial proportion showed defects which were generally similar to those appearing in the original tails. Such defective regenerates were formed both in replacement of abnormal tails and of normal ones. A second, and later a third, amputation gave a second and then a final crop of regenerates in which the same general result obtained. Thus even in the group which had normal original tails, and normal first and second regenerates, the third regenerates were in some cases defective.

These results would be remarkable indeed if it could be established that the amputations were actually performed at a level anterior to the defects caused by the primary removal of tail-bud tissue. Unfortunately it is likely that the tails were more extensively defective than would appear from examination of their profile only. Thus the plane of amputation might be chosen to be anterior to a region in which the dorsal fin was missing but still leave a stump in which somites, neural tube, and notochord were defective.

In view of the provisional nature of Vogt’s report and the difficulty in assessing Woronzowa & Liosner’s results, it was decided to repeat and extend their work. In particular it seemed important to discover whether the regenerative replication of defective tails could take place where the plane of amputation passed through the tail at a level where its cross-section was normal so far as visible structure was concerned.

Eggs of the axolotl (Siredon mexicanum) were used for the urodele experiments and of R. temporaria for the anuran ones. Operation procedures were standard. Animals were communally cultured between the infliction of the primary defect and the time of amputation. From then on they were maintained individually in order that each amputate could be compared with its appropriate regenerate.

(a) The urodele tail

Two experiments were performed to test the regenerative properties of defective urodele tails. In the first 109 axolotl neurulae were deprived of their posterior neural folds and some of the adjacent juxta-blastoporal ectoderm (see Text-fig. 1). Most of them subsequently became larvae in which the ventral fin was lacking over a greater or lesser length of the tail, but in some regulation during urogenesis permitted the formation of a normal-looking tail. The ventral fin could be affected from its anterior margin along the anal tube posteriorly to a level just short of the tail-tip. The tail thus never lacked a ventral fin over the last 5 per cent, of its length.

Text-fig. 1.

The regeneration of axolotl tails in animals from which the posterior neural folds had earlier been removed, with numbers of animals surviving at each stage.

Text-fig. 1.

The regeneration of axolotl tails in animals from which the posterior neural folds had earlier been removed, with numbers of animals surviving at each stage.

Shortly after the surviving larvae had started to feed all but 12 of them suffered amputation of the tail. In the case of those possessing normal-looking tails the level of transection was approximately half-way between the cloaca and the tail-tip. Defective tails were transected either immediately anterior to the defective region or in the middle of it. Some selection was involved at this point since any larva in which the fin was missing so far anteriorly as the anal tube was automatically placed in the group suffering mid-defect amputation. This was done because it was known that amputation in front of the cloaca is not followed by regeneration in normal larvae. The 12 animals which were left intact had defective tails. They were kept as controls for the stable nature of their defects. In fact, as was expected, they showed no tendency to restore the missing part of their fin. Three months after the other animals had completed regeneration the tails of these 12 too were amputated and they are consequently included in the series as a whole.

The major results of the experiment are shown in Text-fig. 1. Only 71 animals survived until the process of regeneration was complete. Each of the 13 of them which had had normal-looking tails produced a normal regenerate. Each of the 25 regenerates formed after amputating defective tails immediately anterior to their defects was normal in form, though in two cases the ventral fin of the regenerate was remarkable for the complete absence of melanophores. Each of the 33 regenerates formed after the amputation of defective tails at a level where they lacked ventral fin, was abnormal. The abnormality of these regenerates took a constant form, even though the original defects affected differing lengths of the tail. In each case the regenerate as a whole was inclined sharply upwards at the level of amputation, so that its axis lay at an angle of between 40° and 80° to the axis of the stump. In each case the regenerate lacked a ventral fin posteriorly to a point little short of the tail-tip.

Histological examination of stumps, regenerates, and amputates revealed no differences between the experimental animals and normal larvae whose tails had been amputated, other than the absence of tail-fin material in the appropriate experimental animals.

The second experiment performed on axolotls differed from the first in one respect only. In 35 neurulae part of the posterior neural plate was removed as well as the presumptive fin material (see Text-fig. 2). This was done in order to produce larvae with tails which had both an externally visible defect in the ventral fin and an internal deficiency of the tail somites. It was to be expected that the somites would be deficient over a region extending farther anteriorly than the fin defect (see, for example, the fate maps of Chuang, 1947). This would make possible the amputation of the tail at a level anterior to the fin defect but within the region of defective musculature. To do this seemed desirable in view of the possibility that the results reported by Woronzowa & Liosner for Rana would prove to be explicable in terms of structural deficiency of internal tissues, including somites, at the plane of amputation.

Text-fig. 2.

The regeneration of axolotl tails in animals from which the posterior neural folds and part of the medullary plate had earlier been removed, with numbers of animals surviving at each stage.

Text-fig. 2.

The regeneration of axolotl tails in animals from which the posterior neural folds and part of the medullary plate had earlier been removed, with numbers of animals surviving at each stage.

The results, as shown in Text-fig. 2, were precisely parallel to those of the first experiment. Histological examination showed that in many cases the somites were abnormally small in the stump, though in none were they missing altogether.

(b) The anuran tail

In repetition of the experiment of Woronzowa & Liosner, the dorsal half of the tail-bud was removed from 550 embryos of R. temporaria. The great majority of them became tadpoles with defective tails, though some regulated successfully to produce normal-looking ones. The defects were usually gross and affected not only the dorsal tail-fin but also internal tissues. The tails were frequently much distorted, being bent dorsally in the plane of bilateral symmetry.

Of the animals which survived until feeding began, 40 with clearly defined defects were set aside as controls for the stability of their defects. They were kept alive until the onset of metamorphosis but failed to show any tendency to repair their defects spontaneously. A number of other animals were discarded as having tails too malformed for effective amputation. The remainder suffered tail amputation as had the axolotls (see Text-fig. 3). On the completion of the ensuing regeneration most of the tails were re-amputated and a second crop of regenerates produced. By the time these were fully formed the tadpoles were approaching metamorphosis and it was judged impossible to repeat the Russian workers’ feat of obtaining three regeneration generations.

Text-fig. 3.

The regeneration of tails in Rana tadpoles which had earlier lost the dorsal half of their tail-buds. The numbers are of animals surviving to the stages indicated.

Text-fig. 3.

The regeneration of tails in Rana tadpoles which had earlier lost the dorsal half of their tail-buds. The numbers are of animals surviving to the stages indicated.

The results are shown in Text-fig. 3. The numbers of animals surviving at each stage are shown there. It will be noted that a second amputation was not performed on tails which had been transected in the middle of their defective region.

Clearly the great majority of the experimental animals behaved as did the axolotls, and this despite the fact that in this case the plane of amputation frequently passed through the tail at a level at which the internal tissues were very abnormal. Indeed, subsequent histological examination showed that effectively normal regenerates were obtained even when the stump tissues were defective in the extreme. Poorly represented notochord, somites reduced to a fraction of their normal cross-sectional area and even a spinal cord reduced to a barely perceptible strand were sufficient to support the formation of a regenerate not markedly inferior to those formed by normal tadpoles.

Six of the regenerates were, however, remarkable in their departure from the general pattern. They all occurred in the series in which tails had been amputated anterior to the defective fin. Five of them were first generation regenerates which also showed a defect in the dorsal fin. The remaining one was formed after the amputation of a normal-looking regenerate.

These six cases are the only ones in which the more unexpected results of Woronzowa & Liosner were repeated. They represent a smaller proportion of a smaller total material than the corresponding cases obtained by these workers. These authors also obtained, it will be remembered, some defective regenerates after the first and second amputations of regulated tails. This phenomenon was not observed in the present experiments.

Histological examination of these six regenerates and of the stumps and amputates of the animals to which they belonged failed to reveal the explanation of the behaviour of the group of five. The cross-section of the stump was, indeed, abnormal in each case, but no more so than in animals which provided normal regenerates. The single abnormal regenerate from the second generation Was exceptional in that the spinal cord was not represented at the plane of transection of the first amputation, though it was at the second. In other words the second amputation took place somewhat farther forward than the first. It is difficult to see any connexion between this fact and the aberrant nature of the second regenerate.

It must be recorded that all six abnormal regenerates lacked the dorsal fin over a short region which did not approach the tail-tip at all closely. They were thus quite distinct in appearance from the regenerates formed after mid-defect amputation. On the other hand, though little weight probably attaches to the point, they did not replicate the original defective tails they were replacing in detail.

The results of the experiments with axolotls confirm Vogt (1931) in showing that, in contrast to experience with the urodele limb, the structural integrity of the plane of amputation in the tail is necessary for the formation of a normal regenerate—at least so far as the ventral fin is concerned. This is consistent also with the conclusions drawn by Holtzer et al. (1955) and Holtzer (1956) from their experiments. They recognize that the cartilages and connective tissue of a urodele tail regenerate are formed by blastema cells analogous to those from which a limb is regenerated, but find that muscle and spinal cord spring more directly from their equivalents in the stump. The ventral fin may now, with some confidence, be added to the latter two.

We must admit that the material described provides no evidence for a general dependence of regeneration upon primary urogenesis, except in so far as this is necessary to provide a plane of amputation of normal cross-section. Those authors who have seen in Vogt’s results a demonstration that organisms cannot regenerate what they have never possessed are not really justified. It is, of course, true that a larva which lacked the whole of its ventral fin could not be given one by amputating the tail, but this is because the fin normally extends anteriorly to the cloaca and amputation in front of this point is not followed by regeneration at all. It is also true that one can devise an experiment in which the regenerate mimics the original tail in point of ventral fin deficiency. But this can only be done where the original deficiency extends posteriorly to a level just short of the tail-tip. The correspondence is in this case fortuitous, however, since what determines whether the regenerate should have a complete ventral fin or none at all is the presence or absence of fin at the plane of amputation. The same animal could be made to regenerate a normal tail if the level of amputation were shifted forwards into a region with the fin intact.

However, it is important to recognize the possibility, in principle, that the process by which an abnormal tail is replaced by a normal one may differ from normal regeneration although its outcome does not. That is to say, it remains possible that where primary urogenesis is abnormal, regeneration to form a normal tail calls upon regulative processes that are not usually involved. This possibility would have to be taken seriously if, in a proportion of cases, regulation were apparently incompletely effective and the regenerate showed some sign of abnormality.

This, at first sight, might appear to be so in the experiments of Woronzowa & Liosner and in the experiments on Rana reported above which repeat them. However, the assessment of these results is, unfortunately, not a simple matter. In the first place it must be admitted that the abnormal regenerates which do not conform to the general pattern are so few that they might be regarded as the products of contingent events. Indeed, Woronzowa states that she obtained a higher proportion of abnormal regenerates among animals whose culture conditions were poor than among those which were really healthy. On the other hand, I have not obtained similar abnormalities in cultures of normal tadpoles whose tails were amputated. Far more important is the fact that in the experiments on Rana there was no question of the cross-section of the tail at the level of amputation being normal. It was in all cases very clearly defective. It is therefore simpler to view the occasional abnormal regenerate as the product of failure to regulate in a situation rendered abnormal not by the more remote events of unusual urogenesis but by the immediate consequences of a defective stump.

We may conclude that amphibian tails bearing witness to defects incurred during the mosaic phase of development do not necessarily produce abnormal regenerates after transection in front of the defective part, and that there is no firm evidence for regarding the regeneration that takes place as differing in any respect from the normal.

Chuang
,
H. H.
(
1947
).
Defekt- und Vitalfärbungsversuche zur Analyse der Entwicklung der kaudalen Rumpfabschnitte und des Schwanzes bei Urodelen
.
Roux Arch. EntwMech. Organ
.
143
,
19
-
125
. ‘
Harrison
,
R. G.
(
1918
).
Experiments on the development of the forelimb of Amblystoma, a selfdifferentiating equipotential system
J. exp. Zool
.
25
,
413
62
.
Holtzer
,
H.
,
Holtzer
,
S.
, &
Avery
,
G.
(
1955
).
An experimental analysis of the development of the spinal column. IV. Morphogenesis of tail vertebrae during regeneration
.
J. Morph
.
96
,
145
68
.
Holtzer
,
S.
(
1956
).
The inductive activity of the spinal cord in urodele tail regeneration
.
J. Morph
.
99
,
1
39
.
Huxley
,
J. S.
, &
De Beer
,
G. R.
(
1934
).
The Elements of Experimental Embryology
.
Cambridge
:
University Press
.
Needham
,
A. E.
(
1952
).
Regeneration and Wound Healing
.
London
:
Methuen
.
Swett
,
F. H.
(
1924
).
Regeneration after amputation of abnormal limbs in Amblystoma
.
Anat. Rec
.
27
,
273
88
.
Vogt
,
W.
(
1931
).
Uber regeneratives und regulatives Wachstum. (Nach Defektversuchen an Schwanz und Schwanzknospe von Amphibienkeimen
.)
Verh. anat. Ges
.
71
,
141
4
.
Woronzowa
,
M. A.
(
1949
).
Regeneratsia organov i zhivotnich
.
Moscow
:
Sovietskaya Nauka
.