J. Embryol. exp. Morph. 77, 221-241 (1983)
Printed in Great Britain © The Company of Biologists Limited 1983
221
The form and structure of supernumerary hindlimbs
formed following skin grafting and nerve deviation
in the newt Triturus cristatus
By SUSAN REYNOLDS, NIGEL HOLDER 1 AND MANUEL
FERNANDES
From the Anatomy Department, King's College, London
SUMMARY
The results of a series of skin grafting experiments performed on the hindlimb of the newt
Triturus cristatus are described. In an attempt to limit the position of origin and the complexity
of supernumerary limb outgrowths four experimental features were varied. These were 1) size
of skin grafts; 2) their position of origin; 3) the position to which they were grafted; and 4) the
presence or absence of nerves of two different sizes at the graft site. The degree of distal
outgrowth and the pattern of the supernumerary structures in the anterior to posterior and
dorsal to ventral axes were assessed. The results are discussed in terms of the control of pattern
regulation in localized populations of limb blastema cells.
INTRODUCTION
One of the most striking responses to grafts whereby tissues are positionally
misplaced in regenerating or mature amphibian limbs is the formation of supernumerary structures (see reviews by Tank & Holder, 1981 and Wallace, 1981).
Numerous types of tissue-grafting procedures have been developed to analyse
the possible cellular interactions which underlie the formation of these extra
limbs. These include the rotation of blastemas upon the stumps from which they
derive (Bryant & Iten, 1976; Tank, 1978,1981a; Maden, 1980,1982; Maden &
Mustafa, 1982; Maden & Turner, 1978; Stock, Krasner, Holder & Bryant, 1980;
Papageorgiou & Holder, 1983; Wallace & Watson, 1979); the contralateral
exchange Of blastemas from One limb to another (Bryant & Iten, 1976; Tank,
1978; Maden, 1982); and the rotation or contralateral transplantation of mature
limb tissues (Bryant & Iten, 1977). These types of operation yield reliable
frequencies of supernumerary limbs in which the number, handedness and
general anatomical characteristics and position of origin have been assessed.
Other methods resulting in supernumerary formation tend to yield less-clearly
structured limbs which are more difficult to interpret in terms of these features.
1
Author's address: Anatomy Department, King's College, Strand, London, WC2R 2LS,
U.K.
EMB77
222
S. REYNOLDS, N. HOLDER AND M. FERNANDES
Such methods generally involve the positional misalignments necessary for induction of extra limb parts by grafting individual tissues within the limb, followed
by either amputation through the graft or nerve deviation to the graft site. Both
the skin (Carlson, 1974, 1975a; Droin, 1959; Rahamani, 1960; Settles, 1978;
Rollman-Dinsmore & Bryant, 1982; Lheureux, 1975) and muscles (Carlson,
19756) produce extra structures when axially misaligned in a limb which is
subsequently amputated. The production of supernumerary limbs also occurs if
nerves are deviated to the limb surface in association with specifically induced
trauma to underlying tissues (Bodemer, 1958,1960) or with positional mismatch
of skin grafts (Lheureux, 1977).
A number of features of supernumerary limbs have been gleaned from these
various methods of their production. Of particular interest are the recent
demonstrations of clear differences in the anatomical makeup of supernumerary
limbs formed following contralateral and ipsilateral blastema grafts.
Contralateral grafts result in the formation of one or two extra limbs in predictable circumferential limb positions and these limbs are always of normal anatomy with regard to the dorsal to ventral and anterior to posterior limb axes if
complete limbs are formed (Maden, 1982). In contrast, supernumerary limbs
formed following ipsilateral 180° blastemal rotation show a range of distinct
anatomies (Maden, 1980, 1982; Maden & Mustafa, 1982; Tank, 1981/);
Papageorgiou & Holder, 1983). This range includes normal limbs, partsymmetrical part-asymmetrical limbs, symmetrical limbs and limbs of mixed
handedness. The production of such supernumerary limbs by blastemal transplantation has revealed that the mechanisms underlying distal outgrowth and
spatial patterning may be complex and cannot be readily explained by the naive
interpretation of models for pattern specification based on continuity (see Lewis,
1981; Bryant, French & Bryant, 1981).
The experiments described in this paper attempt to define a simple in vivo
assay for analysing the cellular interactions which lead to new ordered outgrowth
in amphibian limbs. If a simple assay is devised the apparently complex interactions occurring after blastemal grafting or amputation of a whole limb may be
easier to analyse at the cell and tissue level. The grafts described involve manipulations of different-sized pieces of skin in the thigh region of the leg of Triturus
cristatus, coupled with deviation of a severed nerve to the graft site. The presence
of a nerve at any site of future outgrowth is required as a trophic influence. The
skin was chosen as the tissue for manipulation because it is easy to graft with little
damage to deep tissues and it is known to play a major role in positional control
of spatial patterning during limb regeneration (see also Carlson, 1975a; Tank,
19816).
The results presented reveal the difficulties which are encountered in trying to
establish a simple in vivo assay for pattern regulation. In order to define an assay
we considered it necessary to examine three major properties of amphibian limbs
which are known to be required for the production of supernumerary limbs.
Supernumerary hindlimbs in Triturus
223
These are: 1. the degree of wounding at the site of future outgrowth with the
consequent formation of a wound epidermis; 2. the presence of a sufficient
nervous trophic influence and 3. the positional mismatching of limb tissues. A
reliable assay that would allow the future analysis of such variables as the degree
of positional mismatch needed for induction of outgrowth and the number of
graft cells needed to induce outgrowth following positional mismatching of
tissues, would require a consistently high frequency of induction of a predictable
type of outgrowth. The outgrowths would also need to be sufficiently well
developed to allow comprehensive structural analysis. The experimental results
presented here demonstrate that only some of these requirements have been
fulfilled. However, the system is sufficient to allow us to make some useful
deductions with regard to the structure of supernumerary limbs, and confirms the
dogma concerning the roles of nerves and wound epidermis in the process of limb
regeneration.
MATERIALS AND METHODS
Male and female adult Italian crested newts {Triturus cristatus) were obtained
from Gerrard Haig Ltd. They were kept in individual plastic containers in tap
water for the duration of the experiment and fed raw ox heart twice weekly.
Prior to surgery, animals were anaesthetized in MS222 (Sigma). All operations
were carried out on the upper hindlimbs of the newts. Skin grafts were found to
adhere to underlying muscle and connective tissue and remain in place without
the use of sutures.
Experiment 1
Experimental Skin from the anterior half of the left upper hindlimb was exchanged with skin from the posterior half of the right upper hindlimb. This procedure
resulted in either double-half anterior or double-half posterior skin cuffs overlying normally oriented deep tissues (Fig. 1). The grafts measured approximately
2-5 mm in the proximodistal axis. The sciatic nerve was severed proximal to the
knee and deviated to the proximal wound edge.
Controls Three control experiments were conducted.
Group 1: Anterior and posterior skin strips were exchanged as in the experimentals but without deviation of the sciatic nerve.
Group 2: Left anterior and right posterior skin strips were removed and then
replaced in their original positions (no exchange) (Fig. 1). The sciatic nerve was
deviated to the proximal wound edge.
Group 3: As group 2, but without deviation of the sciatic nerve.
Experiment 2
This experiment was designed to investigate whether supernumerary limbs
224
S. REYNOLDS, N . HOLDER AND M. FERNANDES
A
Fig. 1. Diagram of the operation performed in experiment 1. Skin from anterior and
posterior halves of the thigh region were exchanged (A), or removed and replaced
(B). In the first instance positional conflict between the skin and underlying tissues
is created. The drawings on the right represent end on views of the limb stumps, a,
anterior; p, posterior.
were still produced when the size of skin grafts used in experiment 1 was considerably reduced.
Experimentals A very narrow strip of skin from the anterior half of the left upper
hindlimb was exchanged with a similar strip from the posterior half of the right
upper hindlimb (Fig. 2). The grafts measured approximately 1 mm in the proximodistal axis. The sciatic nerve was again deviated to the proximal wound edge.
Controls Similar very narrow left anterior and right posterior skin strips were
removed and replaced in their original positions (no exchange) with deviation of
the sciatic nerve to the proximal wound edge in each case (Fig. 2).
Experiment 3
This experiment was designed to minimize wounding around the circumference of the limb and create instead a comparable degree of wounding in the
skin along the proximodistal axis of the leg.
Experimentals A narrow strip of skin measuring approximately 1 mm circumferentially and 3 mm in the proximodistal axis was removed from the ventral side
of the left upper hindlimb and exchanged with a similar strip from the dorsal
aspect of the right hindlimb, preserving the correct orientation in the
proximodistal axis (Fig. 3). The sciatic nerve was deviated to the posterior edge
of each ventral wound. However, to minimize internal damage a small dorsal
nerve (a branch of the extensor nerve) was deviated to the posterior edge of each
dorsal wound instead of the sciatic, which lies ventrally.
Supernumerary hindlimbs in Triturus
225
Fig. 2. Diagram of the operation performed in experiment 2. The design is the same
as in Fig. 1, but the size of the skin transplants is considerably smaller in the
proximodistal axis, a, anterior; p, posterior.
Fig. 3. Diagram of the operation performed in experiment 3. A thin strip of skin,
narrow in the a-p axis, was exchanged between dorsal and ventral thigh regions (A),
or removed and replaced (B). a, anterior; p, posterior; d, dorsal; v, ventral.
Controls Two control experiments were carried out.
Group 1: Left ventral and right dorsal skin strips were removed and then
replaced in their original positions (no exchange) (Fig. 3). Ventral control grafts
received sciatic innervation and dorsal control grafts extensor innervation.
Group 2: In this control experiment the smallest possible wound was made
commensurate with locating and deviating the sciatic nerve. This was a 2mm
226
S. REYNOLDS, N. HOLDER AND M. FERNANDES
long proximodistal cut on the ventral aspect of the upper hindlimb. The sciatic
nerve was severed and deviated to the wound.
Experiment 4
The results of experiment 3 indicate that there exists a difference in potential
to induce supernumerary limbs between dorsal and ventral graft sites, in both the
experimental and control groups. This experiment was designed to investigate
whether this difference is a consequence of the variation in nerve supply to the
two types of wound.
Experimentals A narrow strip of skin was removed from the dorsal aspect of the
right upper hindlimb and replaced with a similar strip from the ventral aspect of
the left upper hindlimb, as in experiment 3. In this experiment, however, the
sciatic nerve was deviated up to this dorsal wound instead of the extensor nerve.
Controls A narrow dorsal strip of skin was removed and then replaced in its
original position on both the left and right upper hindlimbs. The extensor nerve
was severed and deviated to the posterior edge of the wound on the left limb but
on the right limb the sciatic nerve was deviated up to the wound edge.
Analysis of results
The animals were examined twice weekly during the first two weeks after
surgery and once a week thereafter. Any limb from which the skin graft was lost
was discounted from the experiment. Two to three months after the initial surgery the animals were anaesthetized and their hindlimbs amputated through the
proximal end of the femur. The limbs were fixed in Bouin's fixative for 24 h and
then decalcified in neutral EDTA for three weeks. Following this treatment they
were stained for cartilage with Victoria Blue B using the technique described by
Bryant & Iten (1974). The limbs were examined and the skeletal pattern of any
supernumerary outgrowths documented. A selection of well-developed supernumerary limbs were embedded in wax together with the region of the hindlimb
from which they originated. Transverse sections 10 fjm thick of the supernumeraries were then cut and stained with haematoxylin and eosin. The pattern
of muscles in each of these supernumeraries was documented.
RESULTS
The first sign of a supernumerary outgrowth on a limb was in general the
appearance of a small swelling over the cut end of the deviated nerve. In some
instances this swelling developed no further or declined. Swellings first appeared
at variable times after surgery from two weeks onwards and the supernumeraries
which developed from these swellings had attained their final structure and
completed their growth within two months of the initial surgery. The supernumerary outgrowths were divided into three categories:-
Supernumerary hindlimbs in Triturus
227
Spike - a single conical, finger-like projection containing cartilage elements.
Indeterminate - a large outgrowth containing cartilage with several digit-like
projections protruding at odd angles, but with no clear overall foot-like structure.
Organized structures with digits - an outgrowth containing cartilage elements,
foot like in appearance with between two and five clearly separated digits.
It is interesting that only the most distal limb structures ever developed and no
elements proximal to what might be the distal ends of the tibia and fibula were
ever seen. The cartilage elements in the supernumerary limb never articulated
with the femur of the host limb.
Experiment 1
The surgical creation of a symmetrical skin cuff overlying normal deep tissues
only elicits supernumerary production if the sciatic nerve is deviated to the
wound: in the presence of the severed nerve end about 40 % of the limbs in each
group (AA and PP) produced outgrowths, (Table 1). Three limbs in the AA skin
cuff group were very well structured and foot like in appearance. One such
supernumerary limb is shown in Fig. 5. If no nerve was deviated no extra outgrowths formed (12 cases AA and 13 cases PP).
If one half of the circumference of skin was merely removed and replaced with
no nerve deviation then supernumeraries were not produced (18 cases A control,
18 cases P control). If, however, the sciatic nerve was deviated to the wound edge
of such a control then a proportion of limbs developed supernumerary outgrowths (Table 1). In the case of the anterior control grafts these were confined
to mere cartilaginous spikes. In contrast, six of the 17 posterior control grafts
produced well-structured outgrowths with two to five digits. An example of a
five-digit supernumerary limb from this group is shown in Fig. 6.
In Table 1 the last two columns show data on the skeletal elements developed
in the supernumerary limbs. The final column gives the average number of
Table 1.
Experiment 1
Operation
Cases
2
none spike unsure digits
AA
PP
A cont.
P cont.
13
13
16
15
Cartilage
elements
Supernumerary outgrowths
11
7
3
4
digits digits
0
0
0
1
1
0
0
0
5
aver/
digits total super.
58
54
5
81
11-6
10-6
1
10-1
When the sciatic nerve was not deviated no extra outgrowths were formed (see text).
228
S. REYNOLDS, N. HOLDER AND M. FERNANDES
1
-- 4
<*;
Fig. 4. Dorsal view of a normal foot stained with Victoria blue, showing the number
and form of the skeletal components. Digits are numbered from anterior to posterior, a, anterior; p, posterior. Magnification xl4.
elements per limb for those limbs developing supernumeraries. The normal newt
ankle and foot contains 23 elements (Fig. 4), and thus those supernumeraries
which develop comprise an average of one third to one half of a foot. The
exception is the anterior skin control group which produce markedly fewer
supernumerary outgrowths.
Experiment 2
The results of the second experiment are given in Table 2.
The use of narrow instead of wide strips of skin, either to create doubleanterior and double-posterior cuffs or in control operations, was also efective in
Table 2.
Experiment 2
Operation
Cases
Cartilage
elements
Supernumerary outgrowths
A
t
none spike
AA
PP
Acont.
P cont.
11
13
10
10
2
6
7
3
7
5
3
4
unsure
2
digits
3
digits
4
digits
1
1
0
0
1
0
0
0
0
0
0
1
0
0
0
1
A
(
aver/
5
digits total super.
0
1
0
1
26
31
8
51
2-9
4-4
2-7
7-3
Supernumerary hindlimbs in Triturus
229
inducing supernumerary outgrowths. With the exception of the anterior control
grafts, those supernumeraries which developed exhibited rather fewer cartilage
elements per limb (last column) than were seen following the use of wider skin
grafts.
Fig. 7 shows a five-digit supernumerary produced by making a doubleposterior narrow skin cuff, while Fig. 8 shows a similar five-digit supernumerary
produced by merely removing and replacing a narrow posterior skin strip.
Experiment 3
Even the smallest possible wound, a 2 mm ventral cut running proximodistally
on the ventral side of the limb, caused a small proportion of limbs to produce
outgrowths, although these were confined to mere spikes containing some cartilage (Table 3). When dorsal skin replaced ventral skin (sciatic innervation) a
high percentage of limbs developed outgrowths with well-defined cartilage elements (Table 3).
Following removal and replacement of ventral control skin strips (sciatic innervation) half of the limbs produced supernumeraries, but the total number of
cartilage elements was less than in the experimental group in which dorsal skin
replaced ventral skin.
Fig. 9 shows a five-digit supernumerary limb which resulted from an operation
in which dorsal skin replaced ventral skin, and Fig. 10 shows a spike produced
in response to a 2 mm long control cut.
In both operations where the small extensor nerve was deviated to the graft
site (ventral skin to dorsal site, 10 cases, and dorsal controls, 15 cases) no extra
structures were formed.
Experiment 4
When ventral skin replaced dorsal skin and the sciatic nerve was deviated up
Table 3.
Experiments 3 and 4
Operation
Supernumerary outgrowths
Cases
Cartilage
elements
A
spike
DtoV
VtoD*
V cont.
Vcut
13
10
21
19
1
6
11
16
2
digits
3
digits
4
digits
5
aver/
digits total super.
82
43
48
7
* Category from experiment 4 (see text).
All grafts with extensor nerve deviation formed no extra outgrowths (see text).
6-8
10-8
4-8
2-3
230
S. REYNOLDS, N. HOLDER AND M. FERNANDES
to the wound instead of the extensor nerve, four out of ten limbs produced supernumerary outgrowths (Table 3). In contrast, the presence of the cut end of the
sciatic nerve at the wound edge of a dorsal control graft completely failed to
stimulate any such supernumerary growth (10 cases). As was shown in experiment
X
V
r
8
\
9
10
Supernumerary
hindlimbs
in Triturus
231
3, no supernumerary limbs were developed when the smaller extensor nerve was
used to innervate either an experimental or a control graft.
General characteristics
1. The role of the nerve
Three basic types of variations with respect to nerve involvement in supernumerary limb formation were performed. In the first variation no nerve was
deviated. Of a total of 61 cases in this group no supernumerary outgrowths of any
kind were formed. In the second category the small dorsally situated extensor
nerve was cut and deviated to the graft edge. Again, in 44 cases no extra outgrowths were stimulated. In contrast, when the large sciatic nerve was deviated
to the graft edge (in the remaining 174 cases - Tables 1, 2 and 3) supernumerary
outgrowths were often formed. The frequency of extra structure formation in
this group is discussed below.
2. Frequency of supernumerary outgrowth
Of a total of 279 operated limbs 105 cases involved either no nerve deviation
or deviation of the small extensor nerve to the graft edge. Therefore 174 cases
involved the sciatic nerve, which was of sufficient size to allow some degree of
outgrowth. In these 174 cases 78 extra structures containing cartilage were
found. This represents a frequency of 44%. However of the supernumerary
outgrowths only 27 (15 %) had clearcut patterns with identifiable digits. The
remaining structures were almost all spikes (46 cases, 25 %) and just five (4 %)
of the outgrowths were too disorganized to allow any further analysis.
Fig. 5. Supernumerary foot withfivedigits stained with Victoria blue, formed after
a skin graft creating symmetrical double anterior skin overlying normal deep tissue,
coupled with sciatic nerve deviation (experiment 1). Magnification x20.
Fig. 6. Supernumerary foot with five digits stained with Victoria blue, formed
following a control posterior skin removal and replacement operation coupled with
sciatic nerve deviation, (experiment 1). Magnification xl8.
Fig. 7. Supernumerary foot with five digits stained with Victoria blue, formed
following the grafting of a narrow posterior skin strip to the anterior thigh region,
coupled with sciatic nerve deviation, (experiment 2). Magnification x20.
Fig. 8. Supernumerary foot withfivedigits stained with Victoria blue formed following posterior skin removal and replacement, coupled with sciatic nerve deviation,
(experiment 2). Magnification x20.
Fig. 9. Supernumerary foot withfivedigits stained with Victoria blue formed following an exchange of dorsal skin to a ventral limb position, (experiment 3). Magnification x20.
Fig. 10. Supernumerary spike outgrowth (arrow) stained with Victoria blue formed
following a 2 mm ventral control cut, coupled with sciatic nerve deviation, (experiment 3). Magnification xl6.
232
S. REYNOLDS, N. HOLDER AND M. FERNANDES
3. Symmetry properties of the supernumerary outgrowths
Of the 27 supernumerary limbs with clear-cut patterns, 26 were serially sectioned to analyse the muscle patterns in order to establish the symmetry relations
of the outgrowths with respect to the dorsal to ventral axis. The normal muscle
pattern of the metatarsal region of the foot is shown in Fig. 11. Of the 26 sectioned supernumerary outgrowths, four cases were normal in this respect (three
with 3 digits and one with 4 digits); four cases were part symmetrical and part
dm
dm
dm
dm
dm
Supernumerary hindlimbs in Triturus
233
Table 4. Axial symmetry of supernumerary outgrowths
Total supers. ,
sectioned
normal*
26
6
double Vt
14
Types of symmetry
^
^
double DJ part sym/part asym.§ a-psym.**
1
4
1
* All involved positional mismatching of tissue.
t All occurred in a ventral limb position.
X Occurred in a dorsal limb position.
§ All involved positional mismatching of skin.
** Formed following ventral skin control in experiment 3, (see text).
normal (one with 2 digits, one with 4 digits and two with 5 digits, Fig. 12); one
case was a symmetrical double dorsal (3 digits, Fig. 13) and eleven cases were
symmetrical double ventral (six with 2 digits, one with 3 digits, and four with 5
digits, Fig. 14). In five of the six remaining cases, one digit in the pattern could
not be assessed due to lack of muscle or plane of section. Two of these had
normal symmetry (in one case a 3 digit and in one case a 5 digit) and three had
double ventral symmetry (in one case a 3 digit and in two cases 4 digits) (see
Table 4). The one remaining case, which formed following a ventral control graft
in experiment 3, deserves special attention because it showed symmetry about
the anterior to posterior axis. It had three digits and a clearly symmetrical muscle
pattern (Fig. 15). The presence of this limb at least demonstrates that a-p
symmetry can occur in these supernumerary limbs. The extremely low incidence
of such supernumeraries may be evidence for the conclusion that at least a
proportion of the spikes are in fact duplications of the a-p axis. It is interesting
in this regard that the spikes that were sectioned also appeared to be symmetrical
Fig. 11. Camera-lucida drawing of a transverse section cut through the midmetatarsal level of a normal foot showing the normal muscle pattern. The dorsal (d)
muscles (dm) are discrete and lie directly above the metatarsals (c). The ventral (v)
muscles (dotted) are more complex and are continuous across the a-p axis. Magnification x32.
Fig. 12. Camera-lucida drawing of a transverse section cut through the midmetatarsal level of a five-digit supernumerary outgrowth formed after grafting a
narrow posterior skin strip to the anterior thigh region, (experiment 2). Of the five
digits three are asymmetric and two are symmetrical double ventral (muscle dotted).
The dorsal discontinuity between dorsal and ventral muscles lies to one side of the
central digit. The Victoria blue whole mount of this limb is shown in Fig. 7. d, dorsal;
v, ventral; c, metatarsals. Dorsal muscles are arrowed. Magnification x34.
Fig. 13. Camera-lucida drawing of a transverse section cut through the midmetatarsal level of a three-digit supernumerary formed after grafting a ventral skin
strip to the dorsal thigh region (experiment 3), coupled with sciatic nerve deviation.
Clear dorsal masses (dm) are located above and below the metatarsals (c). Magnification x38.
234
S. REYNOLDS, N. HOLDER AND M. FERNANDES
Supernumerary hindlimbs in Triturus
235
(Fig. 16), although, due to the small number of pattern elements, it was not
possible to assess which axis was affected in these cases.
The distribution of these supernumeraries with different classes of anatomy
was also determined with respect to site of origin on the host limb (Table 4). The
greatest number of cases occurred in the double ventral group where a total of
fourteen (including the three cases where all but a single digit could be determined) cases were found. Similarly, of the four limbs where part of the pattern
was symmetrical and part normal three of the cases showed double-ventral symmetry. This bias toward ventral symmetry may be expected as in the great
majority of the experiments the sciatic nerve was deviated to the ventral edge of
the grafted skin even though the grafts themselves may have involved anterior
and posterior halves of skin. The only experimental group in which the sciatic
nerve was deviated to the dorsal limb region was when a thin ventral skin strip
was grafted to the dorsal limb surface. This is the only group in which any dorsally
symmetrical digits were seen (Table 4, column 1). Of the three supernumerary
limbs sectioned from this group, one was double dorsal (3 digit), one part normal
and part symmetrical, where the single symmetrical digit in a 4-digit outgrowth
was double dorsal, and one had normal dorsal to ventral asymmetry (a 3-digit
outgrowth). No double ventral symmetry was seen in any digits in this group.
It seems therefore, that double-ventral digits only occur in supernumerary
limbs formed in a ventral host limb position and double-dorsal digits in those
formed in a dorsal host limb position.
DISCUSSION
The experiments presented here were initially designed to attempt to establish
an in vivo assay with which to examine the cellular interactions underlying pattern regulation in the amphibian limb. The results of the extensive series of skin
Fig. 14. Camera-lucida drawing of a transverse section cut through the midmetatarsal level of afive-digitoutgrowth formed after posterior skin removal and
replacement, coupled with sciatic nerve deviation to the ventral skin edge, (experiment 2). All five metat'arsals (c) are surrounded by ventral muscles (dotted). The
Victoria blue whole mount of this limb is shown in Fig. 8. Magnification x36.
Fig. 15. Camera-lucida drawing of a transverse section cut through the mjdmetatarsal level of a three-digit outgrowth formed after removing and replacing a
narrow strip of skin from the ventral thigh region, coupled with sciatic nerve deviation. The muscles (dotted) and skeletal elements (c) appear to be symmetrically
arranged about the a-p axis although anterior and posterior poles cannot be individually identified. The dotted line marks the apparent line of symmetry which runs
from dorsal to ventral. Magnification x40.
Fig. 16. Light micrograph of a transverse section of the spike shown in Fig. 10. Note
the apparent symmetry of the structure about the plane represented by the dashed
line. The horse-shoe-shaped muscle tissue is also outlined by a dotted line indicating
the symmetrical appearance of the structure, m, muscle; c, cartilage. Magnification
x50.
236
S. REYNOLDS, N. HOLDER AND M. FERNANDES
grafting operations have enabled us only partially to achieve this goal. The roles
of wounding, nerve trophic influences and positional mismatches of skin have
been clarified to some extent but the frequency of formation of clearly developed
and extensively patterned outgrowths is not high enough to allow reproducibility. However, the structure of the supernumerary limbs which did form allow
several conclusions to be drawn which are highly relevant to current opinion. Of
particular relevance are the observations that contralateral skin grafts can induce
supernumerary outgrowths which are symmetrical about the dorsal to ventral
axis; and that digits which are asymmetrical in the dorsal to ventral axis only form
following positional mismatching of tissues. Furthermore, control operations in
which no positional mismatching of skin occurs produce digits which are always
symmetrical in the dorsal to ventral axis. Whether these digits reveal dorsal or
ventral symmetry appears to be determined by their position of origin on the host
limb. Extra digits appearing in a ventral position show ventral symmetry whereas
extra digits appearing in a dorsal position show dorsal symmetry.
With regard to the roles of nerves and wounding in this experimental system,
several points can be made. The initial results of Bodemer (1958, 1959, 1960)
have been confirmed with respect to the requirement for a large nerve at the site
of outgrowth. The lack of a nerve or the presence of a small nerve, in this case
the dorsal extensor nerve of the thigh, do not allow formation of any supernumerary outgrowths. The sciatic nerve, however, supplies sufficient trophic
influence for outgrowth to occur. The wounding requirement is a more complex
issue. The grafts performed all involved the removal and replacement of a piece
of hindlimb skin. The pieces were either returned to their normal site or placed
in an abnormal limb position. In each case the skin piece was fitted neatly into
its eventual position. The degree of free wound epidermis was therefore not
varied. However, the size of the piece of skin was varied in three sizes. In the
most extreme case the skin was merely opened and the sciatic nerve drawn to the
surface (experiment 3, Table 3). In these cases a transparent thin wound epidermis was evident along the line of the cut for up to two weeks but no outgrowths
other than spikes were formed. In the second case a thin strip of skin about 1 mm
wide (p-d axis) was cut out and either returned to its original site or exchanged
with a comparably sized piece from the opposite limb position. In these cases
some outgrowths were induced. When a larger piece of skin was utilized (one
whole half of the circumference from hip to knee) supernumerary limbs were
again induced. Thus, the act of removing and grafting the skin appeared to be
necessary for the induction of patterned supernumerary outgrowths. It would be
useful to establish how many skin cells need to be removed and replaced before
sufficient localized dedifferentiation is stimulated to allow such outgrowths to
begin.
The analysis of the structure of the outgrowths also revealed several points. In
terms of overall pattern and structure, the most numerous type of extra outgrowth was a spike. Such structures appear to be tapered rods of cartilage which
Supernumerary hindlimbs in Triturus
237
may contain one or several elements. The significance of these structures is
difficult to determine, but it is interesting to note that exactly such tapered,
incomplete outgrowths are predicted by the polar-coordinate model as the result
of localized duplications of positional values which form as small symmetrical
circumferential sets (see Bryant et al. 1981; Bryant, Holder & Tank, 1982;
Holder, Tank & Bryant, 1980). It is possible therefore that these structures are
the result of very localized interactions of cells with locally non-adjacent
positional values beneath the grafted skin in the region of the deviated nerve.
Dedifferentiated cells with locally disparate positional values may make contact
through the extensive cell processes which characterize them (Geraudie &
Singer, 1981), or may interact with the perineurial connective tissue or Schwann
cells which are present at the site of outgrowth as a result of nerve deviation.
Apart from a low percentage of disorganized outgrowths the other types of
supernumerary structures are clearly patterned foot structures which had between two and five digits. In the well-formed cases the four or five digits were
associated with tarsal elements although a complete tarsus was not seen (see Figs
5-9). Although the frequency of such structures was low, sufficient were formed
to allow an analysis of their muscle patterns.
A striking feature of these outgrowths was the absence of the femur and tibia
and fibula, despite the fact that all experiments were performed in the thigh
region of the leg. This observation may be explained in one of two ways. It is
possible that these structures reflect the efforts of the outgrowth to form a pattern
within a limited population of blastemal cells. The number of cells may be limited
for any number of reasons but a limitation on nerve trophic factors may be
important. In any event, this feature of the structure of the limbs was very
consistent. It may reflect the proposal of Maden (1977) and Stocum (1978) that
the distal regions of the limb pattern are established first during distal outgrowth.
However, a similar pattern of distally more complete structure is equally well
achieved by a mechanism which produces the pattern in a proximal to distal
sequence, such as the progress zone model for the establishment of proximal to
distal positional values (Summerbell, Lewis & Wolpert, 1973; Smith, Lewis,
Crawley & Wolpert, 1974). For example, if the population of cells within the
progress zone is initially small, then the number of cells leaving the zone with
proximal values will be small. As cell division adds to the number of cells within
the progress zone the number of cells leaving the zone with successively more
distal values will progressively increase resulting in more complete distal parts
(see Wolpert, Tickle & Sampford, 1979).
A second feature of the anatomy of the extra outgrowths is the symmetry of
the dorsal to ventral axis. It is clear that the supernumerary outgrowths can be
divided into three groups. In the first category normal symmetry is found in all
digits formed. Such limbs occur only infrequently (four cases out of 25 sectioned). The second category of limbs had part of the pattern normal and part
symmetrical. Of the two 5-digit examples where such patterns were seen both
238
S. REYNOLDS, N. HOLDER AND M. FERNANDES
symmetrical regions were double ventral. In one 3 digits were symmetrical and
2 normal, in the other 2 were symmetrical and 3 normal. In the remaining
category, all digits in the pattern were either symmetrical double ventral or
double dorsal. These cases were clearly identifiable and are similar to those
formed in the forelimb of Triturus following 180 ° blastemal rotation (Papageorgiou & Holder, 1983) and in the axolotl (Maden, 1980,1982; Maden & Mustafa,
1982; Tank, 1981a). The only anatomical type not yet identified in the newt that
has been found in the axolotl is the mixed-handed outgrowth where part of the
a-p pattern is reversed with respect to the remaining part in the d-v axis. As yet
the interactions leading to the formation of these different types of limbs are
unclear. However, the formation of extra structures containing ventral symmetry only occurred in outgrowths formed on the ventral limb region. Similarly,
although only two cases are relevant, dorsal symmetry only occurred in structures formed on the dorsal limb side.
This consistent observation suggests that symmetry around the d—v axis may
result from localized interactions in the blastema regions forming symmetrical
structures. It is possible that the range of structures from a spike to a 5-digit
symmetrical outgrowth reflects the size of the initial blastema population and the
extent of circumferential positional values which are available for interaction at
the outset. A blastemal population which forms a 5-digit double-ventral outgrowth, for example, must be able to generate the a and p extremes of the
circumferential set of values, yet only has sufficient values present to allow
pattern duplication in the dorsal to ventral axis (see Bryant et al. 1981). Such
limited interactions must, in a few cases, also be sufficient to allow regeneration
of at least part of the pattern. Even so, normal asymmetrical digits are less
frequent than symmetrical ones. Of a total of 82 digits analysed 29 were
asymmetrical and 53 were symmetrical double ventral or double dorsal.
The question of interaction between cells with sufficiently disparate positional
values to establish regeneration of a complete or partial set of asymmetrical
digits within a pattern must be examined with respect to the exact skin grafts
which were performed. Three basic types of graft were performed. In the first
category anterior or posterior skin pieces were exchanged from left to right limbs
to create symmetrical skin regions overlying asymmetric deep tissues. Although
the size of the grafted skin piece varied, a range of structures was formed. Only
in such cases where extremes of positional mismatch between skin and the underlying connective tissue occurred did asymmetric digits form as the whole or part
of the digit pattern. The second category involved control grafts where anterior
or posterior skin was removed and replaced. In all outgrowths where digits
formed in such grafts (see Tables 1 and 2) they were symmetrical. Thus, when
local dedifferentiation of cells occurs in the absence of positional discrepancy the
number of circumferential positional values present are only sufficient to result
in pattern duplication. At present it appears that this duplication only manifests
itself predominently with respect to the dorsal to ventral axis. However, it is
Supernumerary hindlimbs in Triturus
239
possible that duplication in the a-p axis results in the formation of spikes, (see
Holder et al. 1980), and one supernumerary limb was clearly duplicated in the
a-p axis in this study. The third category of skin grafts involved the positional
mismatch of dorsal and ventral tissues (Table 3). Again, in this group, the
formation of asymmetrical digits comprising the whole or part of the pattern only
occurred when dorsal or ventral skin pieces were exchanged. In the control grafts
of this group any supernumerary digits formed were symmetrical (Table 4).
One final point is raised with regard to positional mismatching effects on
pattern regulation. Up to now all patterns where symmetry or mixed handedness
has been noted in the d-v axis have occurred following ipsilateral transplantation
of blastemas. In these experiments the creation of symmetrical skin regions
involves the transposition of skin from contralateral limbs. It is likely, therefore,
that the creation of symmetry in the d-v axis is not a sole property of ipsilateral
operations, as appears to be the case with blastema grafts, but rather is a function
of the circumferential range over which cell interactions occur during their
formation. In the skin grafts reported here the number of blastemal cells formed
is governed by the limited degree of trauma induced by the graft and the extent
of the nerve trophic influence which will allow further division of these cells
following dedifferentiation. In contrast, following contralateral blastema transplantation dedifferentiation is extensive and all limb nerves are severed to
produce a maximal trophic influence. Therefore no limitation is placed upon the
number of dedifferentiated cells available for outgrowth. However, the formation of anatomically perverse limbs in 180° blastemal rotations is thought to
result from fusion of initially discrete blastema populations formed in the grafthost junction in the axolotl (Maden & Mustafa, 1982). As many blastema populations may be formed around the graft-host junction following ipsilateral rotation it is possible that numerous limited sized interactive blastemal units begin
to grow out and subsequently fuse to form the range of anatomies that are seen.
In conclusion, the attempt to define an assay for pattern regulation in the absence of amputation and its consequential production of large numbers of
blastemal cells has revealed several interesting points. The structure, position of
origin and form of supernumerary outgrowths show specific and consistent
features. The local nature of cell interactions leading to distal outgrowth and pattern regulation which is governed by skin grafting and site of nerve deviation may
be responsible for the formation of three distinct classes of supernumerary limbs.
It is a pleasure to thank Rosie Burton, Richard Glade and Malcolm Maden for comments
on the manuscript and Charleston Weeks for technical assistance. This research was supported
financially by the SERC and the Wellcome Foundation.
REFERENCES
V., FRENCH, V. & BRYANT, P. J. (1981). Distal regeneration and symmetry. Science
212, 993-1002.
BRYANT, S.
240
BRYANT, S.
S. REYNOLDS, N. HOLDER AND M. FERNANDES
V., HOLDER, N. & TANK, P. W. (1982). Cell-cell interactions and distal outgrowth
in amphibian limbs. Amer. Zool. 22, 143-151.
BRYANT, S. V. & ITEN, L. E. (1974). The regulative ability of the limb regeneration blastema
of Notophthalmus viridescens: experiments in situ. Wilhelm Roux Archiv. EntwMech. Org.
174, 90-101.
BRYANT, S. V. & ITEN, L. E. (1976). Supernumerary limbs in amphibians: experimental
production in Notophthalmus viridescens and a new interpretation of their formation. Devi
Biol. 50, 212-234.
BRYANT, S. V. & ITEN, L. E. (1977). Intercalary and supernumerary regeneration in regenerating and mature limbs of Notophthalmus viridescens. J. exp. Zool. 202, 1-16.
BODEMER, C. W. (1958). The development of nerve-induced supernumerary limbs in the adult
newt, Triturus viridescens. J. Morph. 102, 555-581.
BODEMER, C. W. (1959). Observations on the mechanism of induction of supernumerary limbs
in adult Triturus viridescens. J. exp. Zool. 140, 79-99.
BODEMER, C. W. (1960). The importance of quantity of nerve fibres in development of nerveinduced supernumerary limbs in Triturus and enhancement of the nervous influence by
tissue implants. /. Morph. 107, 47-59.
CARLSON, B. M. (1974). Morphogenetic interactions between rotated skin cuffs and underlying stump tissues in regenerating axolotl forelimbs. Devi Biol. 39, 263-285.
CARLSON, B. M. (1975a). The effects of rotation and positional change of stump tissues upon
morphogenesis of the regenerating axolotl forelimb. Devi Biol. 47, 269-291.
CARLSON, B. M. (1975ft). Multiple regeneration from axolotl limb stumps bearing crosstransplanted minced muscle regenerates. Devi Biol. 45, 203-208.
DROIN, A. (1959). Potentialites et morphogenes dans la peau du Triton en regeneration. Rev.
Suisse. Zool. 66, 641-709.
GERAUDIE, J. & SINGER, M. (1981). Scanning electron microscopy of the normal and denervated limb regenerate in the newt, Notophthalmus, including observations on embryonic
amphibian limb-bud mesenchyme and blastemas offish-fin regenerates. Am. J. Anat. 162,
73-87.
HOLDER, N., TANK, P. W. & BRYANT, S. V. (1980). Regeneration of symmetrical forelimbs
in the axolotl, Ambystoma mexicanum. Devi Biol. 74, 302-314.
LEWIS, J. H. (1981). Simpler rules for epimorphic regeneration: The polar-coordinate model
without polar-coordinates. /. theoret. Biol. 88, 371-392.
LHEUREUX, E. (1975). Nouvelles donnees sur les roles de la peau et des tissus internes dans
la regeneration du membre du triton Pleurodeles waltlii Michah (Amphibien Urodele).
Wilhelm Roux' Archiv. devl Biol. 176, 285-301.
LHEUREUX, E. (1977). Importance des associations de tissus du membre dans le developpement des membres surnumeraires induits par deviation de ferf chez le Triton Pleurodeles
waltlii Michah. J. Embryol. exp. Morph. 38, 151-173.
MADEN, M. (1977). The regeneration of positional information in the amphibian limb. /.
theoret. Biol. 69, 735-753.
MADEN, M. (1980). Structure of supernumerary limbs. Nature 287, 803-805.
MADEN, M. (1982). Supernumerary limbs in amphibians. Amer. Zool. 22, 131-142.
MADEN, M. & MUSTAFA, K. (1982). The structure of 180 degree supernumerary limbs and a
hypothesis of their formation. Devi Biol. 93, 257-265.
MADEN, M. & TURNER, R. N. (1978). Supernumerary limbs in the axolotl. Nature 273,
232-235.
PAPAGEORGIOU, S. & HOLDER, N. (1983). The structure of supernumerary limbs formed after
180° blastemal rotation in the newt Triturus cristatus. J. Embryol. exp. Morph. 74,143-158.
RAHAMANI, T. (1960). Conflit de potentialites morphogenes et duplicature. Rev. Suisse Zool.
67, 589-675.
ROLLMAN-DINSMORE, C. & BRYANT, S. V. (1982). Pattern regulation between hind and
forelimbs after blastema exchanges and skin grafts in Notophthalmus viridescens. J. exp.
Zool. 223, 51-56.
SETTLES, H. E. (1978). Supernumerary regeneration caused by ninety degree turning of limb
skin in adult Notophthalmus. Growth 42, 297-307.
Supernumerary hindlimbs in Triturus
241
A. R., LEWIS, J. H., CRAWLEY, A. & WOLPERT, L. (1974). A quantitative study of
blastemal growth and bone regression during limb regeneration in Triturus cristatus. J.
Embryol. exp. Morph. 32, 375-390.
STOCUM, D. L. (1978). Organisation of the morphogenetic field in regenerating amphibian
limbs. Amer. Zool. 18, 883-896.
STOCK, G. B., KRASNER, G. N., HOLDER, N. & BRYANT, S. V. (1980). Frequency of supernumerary limbs following blastemal rotations in the newt. /. exp. Zool. 214, 123-126.
SUMMERBELL, D., LEWIS, J. H. & WOLPERT, L. (1973). Positional information in chick limb
morphogenesis. Nature 244, 492-496.
TANK, P. W. (1978). The occurrence of supernumerary limbs following blastemal transplantation in the regenerating forelimb of the axolotl, Ambystoma mexicanum. Devi Biol. 62,
143-161.
TANK, P. W. (1981a). Pattern formation following 180degree rotation of regeneration
blastemas in the axolotl, Ambystoma mexicanum. J. exp. Zool. 217, 377-387.
TANK, P. W. (1981/)). The ability of localised implants of whole or minced dermis to disrupt
pattern formation in the regenerating forelimb of the axolotl. Am. J. Anat. 162, 315-326.
TANK, P. W. & HOLDER, N. (1981). Pattern regulation in the regenerating limbs of urodele
amphibians. Q. Rev. Biol. 56,113-142.
WALLACE, H. (1981). Vertebrate Limb Regeneration. Wiley.
WALLACE, H. & WATSON, A. (1979). Duplicated axolotl regenerates. /. Embryol. exp. Morph.
49, 243-258.
WOLPERT, L., TICKLE, C. & SAMPFORD, M. (1979). The effect of cell killing by x-irradiation in
pattern formation in the chick limb. /. Embryol. exp. Morph. 50, 175-198.
SMITH,
(Accepted 3 May 1983)