‘Bilateral innervation of a single hindlimb bud was induced by amputating the other limb bud and disrupting the barriers between the two sides. Though the routes of the crossed nerves were necessarily abnormal, the motor projections that developed subsequently were normal as determined by horseradish peroxidase tracing. The limb therefore appears to be innervated selectively, each region being invaded and/or synapsed with only by motoneurones at particular locations. The numbers of motoneurones surviving after metamorphosis were almost normal on both sides provided the operation was done before motor invasion of the limb bud begins. From this it is argued that the axons were probably guided actively to their correct destinations. Without such guidance, axons would probably not have been able to find their correct termination sites and motoneurone survival would therefore have been depressed. The normal motoneurone numbers also imply that the single limb was supporting twice its usual quota of motoneurones. The hypothesis that motoneurones compete in the limb for survival is therefore not supported.

There seem to be two schools of thought on how the motor projections to the limb develop in the embryo. The passive innervation school maintains that axons follow paths of least resistance like water down a gully. Complexity of surrounding anatomy divides the flow into branches and it is fortuitous which motoneurones invade given muscles. By implication, synaptogenesis is not selective, any motoneurone may connect with any muscle. Projections vary little because the anatomy creating the pathways is common to all (Horder, 1978). The opposing school claims that there must be some selectivity to ensure that muscles always receive innervation from motoneurones at particular locations in the spinal cord. Selectivity may take the form of active axonal guidance bringing the motoneurones into contact with their correct muscles and/or selective synaptogenesis such that their motoneurones cannot synapse with incorrect muscles but only with their own (Landmesser, 1980).

Both schools are supported by recent experiments in the chick embryo. In support of the passive, rotated limb buds and grafted supernumerary limb buds receive abnormal projections apparently because motor axons continue to follow their usual courses, finding foreign sites where their correct sites should be (Hollyday, Hamburger & Farris, 1977; Morris, 1978; Stirling & Summerbell, 1979, 1980). In support of selective innervation, spinal cords rotated in the rostrocaudal axis send their projections to correct sites, undeterred by having started from the wrong end (Lance-Jones & Landmesser, 1980 b).

In this paper I describe an experiment further supporting selective innervation. Tadpoles were reared with one hindlimb bud removed. The spinal nerves were induced to innervate the opposite limb bud. It was found that despite the abnormal routes taken by the spinal nerves the motor projections were normal.

The experiment also contains a test of a hypothesis about the control of motoneurone numbers. Up to 75 % of developing motoneurones die in normal embryos (Prestige, 1974; Hamburger, 1975) and it is hypothesized that they are the losers in a competition for an unknown but essential limb factor. Bilateral innervation effectively doubles the number of motoneurones competing in the one limb which ought to lower the proportion of survivors. Contrary to predictions, it was found in this study that the number surviving to maturity was close to normal on both sides so that the limb was supporting twice its normal complement of motoneurones. This result argues strongly against peripheral competition. Preliminary results have been reported previously (Lamb, 1980).

Operations were carried out on Xenopus laevis tadpoles in a solution containing 1 gram of sea salt, 4 grams of NaCl, and 9 grams of D-glucose per litre plus 1:5000 MS222 (Sandoz). The left hindlimb bud was removed from 10- to 15-day-old tadpoles at Stages 49 to 51. (Nieuwkoop & Faber, 1967), spanning the stage at which motor axon invasion of the limb bud begins (Stage 50; Lamb, 1976). The other limb bud was left undisturbed and not moved into the midline as described previously (Lamb, 1980). Midline barriers dorsal to the anus which normally confine the spinal nerves to their own sides were disrupted with a needle. Tadpoles recovered in saline (5 gm/1). During the next few stages of development some tadpoles regenerated new limb buds which were removed immediately. None were longer than 0·3 mm on removal. One to 6 weeks after completion of metamorphosis frogs were given injections of horseradish peroxidase solution (HRP. Sigma Type VI, 30 % in phosphate buffer, pH 7·0) into one of five different muscles to label motoneurone pools. Since the object was to determine the extent of a pool and not to label all the neurones within it, small injections were given in about half the animals (approx. 0·01 μl). Though few motoneurones were labelled, small injections had the advantage that any ectopic labelled motoneurones could be more confidently ascribed to abnormal projections rather than spread of HRP to other muscles. Other animals were given larger injections (up to 0·5 μl) to allow comparison with previous projection studies in juvenile frogs (Lamb, 1976). The frogs were killed 2 days later for motoneurone counts and mapping of HRP-labelled motoneurones from serial transverse sections. Diaminobenzidine was used for the HRP histochemistry. Full descriptions of the methods are given elsewhere (Lamb, 1976). Motoneurones were identified using the criteria of Prestige (1967) and those with nucleoli counted in every third section without correction for nucleolar splitting. Labelled motoneurones were mapped in the rostrocaudal and medio-lateral axes.

Seventeen frogs were studied, three of which were used for cell counts in an earlier study (Lamb, 1980). All developed only one hind limb; there was no trace of the other. The hind limbs were normal in every respect including size and function (Fig. 1). Movements in particular were normal as judged by eye, and the animals, once accustomed to having only one hind leg, were able to swim and feed with alacrity.

Fig. 1.

Monopodal frog compared to a normal frog. The remaining limb seems normal in all respects. Movements of the limb were visually observed, unassisted, and were indistinguishable from normal. Though subtle abnormalities may not be detected without electromyography, it is most improbable that such abnormalities could affect motoneurone survival should they be present. Increased survival is associated only with gross motor impairment bordering on paralysis (Pittman & Oppenheim, 1979; Olek, 1980).

Fig. 1.

Monopodal frog compared to a normal frog. The remaining limb seems normal in all respects. Movements of the limb were visually observed, unassisted, and were indistinguishable from normal. Though subtle abnormalities may not be detected without electromyography, it is most improbable that such abnormalities could affect motoneurone survival should they be present. Increased survival is associated only with gross motor impairment bordering on paralysis (Pittman & Oppenheim, 1979; Olek, 1980).

Motoneurone numbers were almost normal on the operated sides in the majority of cases (Table 1). In tadpoles operated at later maturity the numbers tended to be lower.

Table 1.

Motoneurone counts in monopodal frogs: left limb bud removed

Motoneurone counts in monopodal frogs: left limb bud removed
Motoneurone counts in monopodal frogs: left limb bud removed

On the amputated side the spinal nerves which normally innervate the limb (principally segmental nerves 9 and 10 with a variable contribution from 8) fused into a single well-defined nerve trunk which crossed dorsal to the anus and joined the opposite sciatic nerve at the base of the limb (Fig. 2d). At the junction, the two nerve trunks intermingled with fibres from each interweaving to all parts of the common sciatic nerve. However, there was a tendency for fibres from the amputated side to lie more peripherally in the common nerve (Fig. 2 b). Stripping apart of the two trunks inevitably caused tearing of many of the axons but nevertheless fascicles from each trunk could be followed into both the anterior and posterior tibial nerves. No detectable abnormalities were found of the routes of the main branches of the sciatic nerve or tibial nerves. A formal study of the paths of axons from each side during development and at maturity is in progress and will be reported later.

Distributions of HRP-labelled motoneurones were all normal (Figs. 38). Labelled motoneurone pools on both sides were correctly situated in the rostrocaudal and mediolateral axes. The correctness of the mediolateral representation is especially significant since the limb and the contralateral ventral horn must have been 180° out of register from their normal axial alignment. To conclude from these results that motor projections were normal it must be assumed that any motoneurones projecting inappropriately would have been labelled. This seems safe since several experiments have shown that inappropriately projecting motoneurones can be labelled (e.g. Lamb, 1976; Hollyday et al. 1977; Morris, 1978; Stirling & Summerbell, 1980), while no evidence has been found to show that such motoneurones cannot be labelled.

Fig. 2.

Junction of right and left sciatic nerves at the base of the limb (L). Each sciatic nerve is itself formed by the fusion of spinal nerves 8, 9 and 10. In this animal spinal nerve 11 also made a small contribution, (a) Dorsal view after removal of the verebral column which normally overlies the entire plexus. The stumps of spinal nerves visible in this photograph are numbered according to Gaupp (1896). The crural nerve (CN) on the unoperated side leaves the plexus proximal to the junctions of spinal nerves 9 and 10 and after the junction of 8 and 9 as in normal animals. No crural nerve was ever seen on the operated side × 25. (b) The common sciatic nerve dissected free to the division into anterior and posterior tibial nerves. The two contributing sciatic nerves have been pulled apart to show more clearly the manner of their fasciculation. The uncrossed sciatic nerve (U) was partly ensheathed by the crossed nerve (C) though some fibres from each were intermingled. Both tibial branches received fascicles from each sciatic nerve, × 50.

Fig. 2.

Junction of right and left sciatic nerves at the base of the limb (L). Each sciatic nerve is itself formed by the fusion of spinal nerves 8, 9 and 10. In this animal spinal nerve 11 also made a small contribution, (a) Dorsal view after removal of the verebral column which normally overlies the entire plexus. The stumps of spinal nerves visible in this photograph are numbered according to Gaupp (1896). The crural nerve (CN) on the unoperated side leaves the plexus proximal to the junctions of spinal nerves 9 and 10 and after the junction of 8 and 9 as in normal animals. No crural nerve was ever seen on the operated side × 25. (b) The common sciatic nerve dissected free to the division into anterior and posterior tibial nerves. The two contributing sciatic nerves have been pulled apart to show more clearly the manner of their fasciculation. The uncrossed sciatic nerve (U) was partly ensheathed by the crossed nerve (C) though some fibres from each were intermingled. Both tibial branches received fascicles from each sciatic nerve, × 50.

Fig. 3.

Horseradish-peroxidase-labelled motoneurones (arrows) in a transverse section of spinal cord. Low power × 130 high power × 340, phase contrast.

Fig. 3.

Horseradish-peroxidase-labelled motoneurones (arrows) in a transverse section of spinal cord. Low power × 130 high power × 340, phase contrast.

Fig. 4.

Distributions of motoneurones projecting to semimembranosus in mono-podal frogs. Each pair of rectangles represents right and left ventral horns of one animal. Rostral is to the top and medial to the right of each rectangle. The left limb hud was removed at the stage of development shown above each pair. Motoneurones were labelled by retrograde axonal transport of HRP injected into semimembranosus 2 days before the frogs were killed. The volume injected is given above each pair. Scale gives the numbers of labelled motoneurones. Total labelled motoneurones in each ventral horn is given below as is the total motoneurone count (labelled plus unlabelled).

Fig. 4.

Distributions of motoneurones projecting to semimembranosus in mono-podal frogs. Each pair of rectangles represents right and left ventral horns of one animal. Rostral is to the top and medial to the right of each rectangle. The left limb hud was removed at the stage of development shown above each pair. Motoneurones were labelled by retrograde axonal transport of HRP injected into semimembranosus 2 days before the frogs were killed. The volume injected is given above each pair. Scale gives the numbers of labelled motoneurones. Total labelled motoneurones in each ventral horn is given below as is the total motoneurone count (labelled plus unlabelled).

Fig. 5.

Motoneurones projecting to knee extensors.

Fig. 5.

Motoneurones projecting to knee extensors.

Fig. 6.

Motoneurones projecting to gastrocnemius.

Fig. 6.

Motoneurones projecting to gastrocnemius.

Fig. 7.

Motoneurones projecting to toe dorsiflexors.

Fig. 7.

Motoneurones projecting to toe dorsiflexors.

Fig. 8.

Motoneurones projecting to intrinsic plantar flexors of the toes.

Fig. 8.

Motoneurones projecting to intrinsic plantar flexors of the toes.

Selective innervation

Motor axons from the left ventral horn were induced to innervate the right limb bud, the left having been amputated earlier. The resulting crossed innervation is functional; stimulation of right or left spinal nerves causes equivalent movements of the limb (Denton, C., Lamb, A. and Wilson, P., unpublished results). To reach the limb bud the axons needed to take very abnormal routes, crossing the midline dorsal to the anus before meeting up with ipsilateral axons at the base of the limb bud. Despite this, and the 180° misalignment that must exist between the transverse axes of the ventral horn and the contralateral limb, all motor projections tested were normal. The results do not support the hypothesis that motoneurones grow into and synapse with the limb in a non-selective manner or that the characteristic adult motor projections are fortuitous. Selectivity must have operated at some point to give the normal projections. Either the axons must have been actively guided, or synaptogenesis must have been selective, or both. Guidance of axons is found in normal tadpoles (Lamb, 1976), but whether it is active or passive is not clear. Selective synaptogenesis is strongly supported by experiments in tadpoles with partially ablated limb buds. Proximal or distal segments of the limb bud were removed before being innervated. Using HRP the corresponding motoneurones were subsequently found to terminate in the remaining inappropriate part of the limb. Later still all these cells died leaving alive only motoneurones appropriate for the remaining limb segment (Lamb, 1981 a).

Though selective synaptogenesis on its own would seem sufficient to account for the normal projections in this study, the numbers of surviving motoneurones suggest that active guidance was also partly if not entirely responsible. Before showing how this is so, it should be noted that contralateral motoneurone numbers were almost normal in most of the tadpoles operated at Stage 49. Poorer survival rates were strongly correlated with greater maturity at operation. The most probable reason is that axons have greater difficulty finding the contralateral limb bud as distances increase. In addition, many axons must have been severed in older tadpoles since proximal limb bud becomes innervated during stage 50. Although motor axons severed before the onset of cell death can regenerate, axotomy nevertheless depresses survival rates (Lamb, 1981b).

The fact that motoneurone survival was approximately normal when the operations were done early enough means it is unlikely that axons grew passively or randomly into the limb bud. This is because without active guidance, axons would first have had difficulty finding the remaining limb bud and then, those that succeeded, finding their appropriate regions within it. As a result few if any motoneurones would have survived in contrast to the normal numbers found. It is therefore reasonable to interpret the results in favour of active guidance. A theoretical alternative to active guidance is that axons could successfully find their correct sites by branching non-selectively to all sites and then eliminating incorrect branches. However, developing motor axons appear not to branch to any significant degree except for their terminal ramifications (Lamb, 1976; Landmesser, 1978; Lance-Jones & Landmesser, 1980a). Studies of projections at intermediate stages are needed to resolve these points. It is important to note that even if active guidance is confirmed, passive guidance is not denied. Passive forces such as contact guidance undoubtedly play an important role, but active forces seem necessary to give at least minimal directives to the growth cone.

A question arises whether contralateral axons used ipsilateral axons as guides. That is not to suggest that the contralateral axons were passively guided for they would still have had to choose which ipsilateral axons to follow. Though this may have involved fasciculation, usually considered passive, the fasciculation would have to have been selective requiring active recognition between compatible axons. Selective fasciculation was first proposed by Weiss 40 years ago (e.g. Weiss, 1941), but it has received little attention because it is difficult to examine experimentally. However, the concept is attractive since all that is required is for axons to respond selectively to information transmitted by predecessors already at the target. Retrograde chemical or electrical signals are not hard to envisage. The difficulty is to devise tests which can show unambiguously that axons make choices on the basis of that information and not in response to non-axonal cues. There is hope that monopodal tadpoles may provide a suitable model and this is to be investigated.

Other theories of axonal guidance have been discussed in recent reviews (Constantine-Paton, 1979; Landmesser, 1980).

The results of this experiment complement two recent studies. In the first (Beazley & Lamb, 1979), optic nerves induced to take abnormal routes in Xenopus tadpoles, were found to make normal retinotopic projections onto the tectum. In contrast to the present study, neuronal loss (in the retina) was probably increased judging from the small size of the mature optic nerves, and the authors were unable to decide whether the axons were actively guided. However, the experiment showed that non-selective synaptogenesis could not be induced simply by making axons enter the wrong part of the tectum. In the other study, in chick embryo, spinal cords reversed in the rostrocaudal axis were nevertheless able to establish their normal motor projections, providing excellent evidence for active guidance (Lance-Jones & Landmesser, 1980 b).

A serious discrepancy with several earlier experiments is raised. Limb buds rotated or grafted as supernumeraries alongside normal limb buds have been found to receive systematically abnormal projections (Hollyday et al. 1977; Morris, 1978; Stirling & Summerbell, 1979, 1980). In particular, rotated wing buds receive the projections that would be predicted from passive axonal guidance and non-selective synaptogenesis (Stirling & Summerbell, 1979, 1980). The implications for specificity of neuromuscular synaptogenesis have been discussed elsewhere (Lamb, 1981a). Lance-Jones & Landmesser (1980,b) have tried to reconcile the results in terms of active guidance. (See also Landmesser, 1980, review). They suggested that normal projections result after cord rotation, but not limb rotation because axons leaving the rotated cord immediately encounter tissues of incongruous polarity and have time to adjust before reaching the limb. Axons approaching a rotated limb have no such opportunity and find themselves caught irretrievably in the wrong part of the plexus. Unfortunately, this very plausible explanation is weakened by the present study. Assuming the motor axons were actively guided, those on the contralateral side could have had no direct knowledge of abnormality until near the amputation site and no axial confusion until they had crossed behind the anus. Unless knowledge about the limb field was obtained indirectly, directional adjustments must have been made at the last moment either at the base of the limb bud or within the limb bud itself. For the moment it seems that the discrepancies cannot be resolved.

Competition hypothesis

An important result of this study is the confirmation of an earlier report that the limb is able to support twice its usual complement of motoneurones (Lamb, 1980). It was argued that the result refutes the hypothesis that motoneurone survival is determined by competition for limited target factors. However, Purves (1980) has speculated that the initial bilateral innervation may somehow induce a greater supporting capacity within the limb (as opposed to the operation itself doing so; see Lamb, 1980). Since there appears to be no way of convincingly disproving the suggestion, peripheral competition among motoneurones remains a possibility.

A more recent study reinforces this need for caution. Removal of parts of the motoneurone pools to certain muscles of the chick hindlimb appeared to reduce the number of dying motoneurones in the remaining portions (Lance-Jones & Landmesser, 1980,a). This evidence is by far the best in support of competition. It is of interest that cell rescue seemed possible for motoneurone pools of only some muscles. If it is shown that these are in the minority it may explain the rather puzzling observation that supernumerary limbs are able to rescue some cells from death but not the majority (Hollyday & Hamburger, 1976; Lamb, 1979a).

A similar experiment in Xenopus produced no evidence for competition. Part of the ventral horn supplying the knee flexors was removed on one side only in young tadpoles and the animals were examined well after cessation of cell death. Total cell numbers in the remaining segments were not increased. It was also found that at least some motoneurone deaths are definitely not due to competition (Lamb, 1979b).

It seems the evidence at hand cannot decide whether competition controls motoneurone survival. Though it should be borne in mind that competition probably controls survival elsewhere in the nervous system (e.g. Pilar, Land-messer & Burstein, 1980), that does not constitute an argument for competition among motoneurones. In any case, whether they compete or not, it is almost certain that competition on its own is insufficient to account for all naturally occurring motoneurone deaths. Searches for additional causes therefore seem justified.

I thank Dr. L. D. Beazley for her comments and Mr. S. B. Baker for technical assistance. Support was provided by the National Health and Medical Research Council of Australia and the Muscular Dystrophy Research Association of Western Australia.

Beazley
,
L. D.
&
Lamb
,
A. H.
(
1979
).
Rerouted axons in Xenopus tadpoles form normal visuotectal projections
.
Brain Research
179
,
373
378
.
Constantine-Paton
,
M.
(
1979
).
Axonal navigation
.
Bioscience
29
,
526
532
.
Gaupp
,
E.
(
1896
).
Anatomie des Frosches, Abt
.
3
.
Braunschweig
:
Vieweg und Sohn
.
Hamburger
,
V.
(
1975
).
Cell death in the development of the lateral motor column of the chick embryo
.
J. comp. Neurol
.
160
,
535
546
.
Hollyday
,
M.
&
Hamburger
,
V.
(
1976
).
Reduction of naturally occurring motor neuron loss by enlargement of the periphery
.
J. comp. Neurol
.
170
,
311
320
.
Hollyday
,
M.
,
Hamburger
,
V.
&
Farris
,
J. M. G.
(
1977
).
Localization of motor neuron pools supplying identified muscles in normal and supernumerary legs of chick embryo
.
Proc. natn. Acad. Sci., U.S.A
.
74
,
3582
3586
.
Horder
,
T. J.
(
1978
).
Functional adaptability and morphogenetic opportunism, the only rules for limb development?
Zoon
6
,
181
192
.
Lamb
,
A. H.
(
1976
).
The projection patterns of the ventral horn to the hind limb during development
.
Devl Biol
.
54
,
82
99
.
Lamb
,
A. H.
(
1979a
).
Ventral horn cell counts in a Xenopus with naturally occurring supernumerary hind limbs
.
J. Embryol. exp. morph
.
49
,
13
16
.
Lamb
,
A. H.
(
1979b
).
Evidence that some developing limb motorneurones die for reasons other than peripheral competition
.
Devl Biol
.
71
,
8
21
.
Lamb
,
A. H.
(
1980
).
Motoneurone counts in Xenopus frogs reared with one bilaterally innervated hindlimb
.
Nature
284
,
347
350
.
Lamb
,
A. H.
(
1981a
).
Target dependency of developing motoneurones in Xenopus laevis
.
J. comp. Neurol. (In the Press)
.
Lamb
,
A. H.
(
1981b
).
Axon regeneration by developing limb motoneurones in Xenopus laevis
.
Brain Research
,
204
,
315
323
.
Lance-Jones
,
C.
&
Landmesser
,
L.
(
1980a
).
Motoneurone projection patterns in embryonic chick limbs following partial deletions of the spinal cord
.
J. Physiol
.
302
,
559
580
.
Lance-Jones
,
C.
&
Landmesser
,
L.
(
1980b
).
Motoneurone projection patterns in the chick hind limb following early partial reversals of the spinal cord
.
J. Physiol
.
302
,
581
602
.
Landmesser
,
L.
(
1978
).
The development of motor projection patterns in the chick hindlimb
.
J. Physiol
.
284
,
391
414
.
Landmesser
,
L.
(
1980
).
The generation of neuromuscular specificity
.
Ann. Rev. Neurosci
.
3
,
279
302
.
Morris
,
D. G.
(
1978
).
The development of functional motor innervation in supernumerary hind limbs of the chick embryo
.
J. Neurophysiol
.
41
,
1450
1465
.
Nieuwkoop
,
P. D.
&
Faber
,
J.
(Eds.) (
1967
).
Normal Table of Xenopus laevis (Daudin)
.
Amsterdam: North-Holland
.
Olek
,
A. J.
(
1980
).
Effects of alpha and beta bungarotoxin on motoneurone loss in Xenopus larvae
.
Neuroscience
5
,
1557
1563
.
Pilar
,
G.
,
Landmesser
,
L.
&
Burstein
,
L.
(
1980
).
Competition for survival among developing ciliary ganglion cells
.
J. Neurophysiol
.
43
,
233
254
.
Pittman
,
R. H.
&
Oppenheim
,
R. W.
(
1979
).
Cell death of motoneurones in the chick embryo spinal cord. IV. Evidence that a functional neuromuscular interaction is involved in the regulation of naturally occurring cell death and the stabilization of synapses
.
J. comp. Neurol
.
187
,
425
446
.
Prestige
,
M. C.
(
1967
).
The control of cell number in the lumbar ventral horns during the development of Xenopus laevis tadpoles
.
J. Embryol. exp. Morph
.
18
,
359
387
.
Prestige
,
M. C.
(
1974
).
Axon and cell numbers in the developing nervous system
.
Brit. Med. Bull
.
30
,
107
111
.
Purves
,
D.
(
1980
).
Neuronal competition
.
Nature
278
,
585
586
.
Stirling
,
R. V.
&
Summerbell
,
D.
(
1979
).
The segmentation of axons from the segmental nerve roots to the chick wing
.
Nature
278
,
640
642
.
Stirling
,
R. V.
&
Summerbell
,
D.
(
1980
).
Evidence for non-selective motor innervation of rotated chick limbs
.
J. Physiol
.
300
, 7P.
Weiss
,
P.
(
1941
).
Nerve patterns: The mechanics of nerve growth
.
Growth
5
,
163
203
.