/. Embryol, exp. Morph.. Vol. 11, Part 2, pp. 431-444, June 1963 Printed in Great Britain Functional Specificity of Spinal Cord Segments in the Control of Limb Movements by GEORGE SZEKELY 1 From the Department of Anatomy, University of Pecs WITH ONE PLATE INTRODUCTION THERE is ample evidence that limbs innervated by spinal segments which normally do not supply limbs, do not exhibit co-ordinated movements (Detwiler, 1920). Nerve formation under such circumstances seems to be fairly normal, i.e. essentially characteristic for the innervated limb (Detwiler, 1920; Piatt, 1956), so that failure of nerve formation and of re-innervation of the muscles cannot be blamed for the result. It is more probable that the limbinnervating segments of the spinal cord: the brachial (segments 3, 4, 5 in the newt) and the lumbo-sacral (segments 16, 17, 18) might alone possess a central apparatus determined in early embryonic life with the capacity to innervate and move limbs in a co-ordinated manner (Detwiler, 1936; Weiss, 1955; Rogers, 1934). Although some experimental approaches (Moyer, 1943, Piatt, 1957) were unsuccessful, it is obvious that this problem could best be investigated by transplanting brachial or lumbo-sacral segments into the place of the thoracic segments, and by additionally implanting at the same level supernumerary limbs to be innervated by the grafted cord segments that might contain the postulated specific apparatus for co-ordinated limb movement. By suitable variation of the experimental conditions it could not only be decided (1) whether there exists any specific capacity of the limb-bearing segments to move limbs in a co-ordinated manner even if they are transplanted to heterotopic levels of the cord but also (2) whether the character and rhythm of movement of transplanted limbs depends on the nature of the limb (fore- or hindlimbs), or on that of the cord segments (brachial or lumbo-sacral), and (3) whether the position of the transplanted cord segments (rostro-caudal) influences the timing of the motor rhythm. The result of an experimental approach to these questions is presented in the following pages. 1 Author's address. The Department of Anatomy, University Medical School, Dischka u. 5, Pecs, Hungary.
432 GEORGE SZEKELY METHODS Larvae of Pleurodeles waltlii and of Triturus vulgaris were used at stages between 24 and 38. To determine the exact age of the larvae, stage series of Triturus vulgaris from Sato (1933) and, for older stages, from Glucksohn (1932) have been adapted to Pleurodeles. Donors were usually somewhat younger than hosts. In younger stages the embryos were motionless; application of narcotic drugs was, however, necessary for operation on the older embryos. The general procedure of operation was as follows. The brachial or lumbosacral segments of the medullary tube were excised from the donor, carefully cleaned from the underlying chorda and from myotome material. In older stages the covering ectoderm was also removed. The prepared segments were then fitted into the place of the previously removed medullary tube segments either at the thoracic or at one of the limb levels of the hosts. After the onset of feeding a pair of fore- or hindlimbs was grafted into the myoseptal region close to the transplanted cord segments at the thoracic level. Hosts and donors always belonged to the same species. Four groups of experimental animals were distinguished according to where the heterotopic segments were placed. Group 1. The brachial segments were grafted into the thoracic region and in eleven cases a pair of forehmbs was attached, while in four cases hindlimbs were attached. Group 2. The grafted brachial segments were situated in the lumbo-sacral region of eight animals. Group 3. The lumbo-sacral segments were grafted into the thoracic region and a pair of forelimbs was attached in fourteen cases, hindlimbs in seven cases. Group 4. The grafted lumbo-sacral segments were situated in the brachial region of seven animals. In Groups 2 and 4 no limb transplantation was, of course, necessary, the normal hind- and forelimbs indicated the function of the heterotopic cord segments. For the purpose of control, in the fifth group, limbs were grafted at the level of the thoracic segments in four intact animals. The total of fifty-one experimental animals of the four groups, which included only the successful cases with complete healing and functioning supernumerary limbs, were selected from several hundred operated embryos. The locomotion of the animals was observed during the larval period underwater with a lowpower dissecting microscope. A more detailed analysis of ambulatory movements was made after metamorphosis. The majority of the animals were recorded on cinematographic film at the rate of 16 and 64 frames per second. Normal projection of the latter rendered possible a considerable reduction of the speed and hence a detailed analysis of the movements on the screen. After careful observation the animals were sacrificed and subjected to histological examination by serial sections stained either with Bodian's or with Holmes' silver methods in order to determine the exact innervation of the supernumerary limbs.
J. Embryol. exp. Morph. Vol. 11, Part 2 EXPLANATION OF PLATE Moving picture series illustrating the locomotion of an animal (Pleurodeles waltlii) in which the 8th, 9th and 10th spinal cord segments are replaced by the 3rd, 4th and 5th (brachial) segments. The supernumerary limbs (middle pair) are moving synchronously with the normal forelimbs on the same side. GEORGE SZfiKELY (Facing page 433)
SPINAL CORD SEGMENTS AND CONTROL OF LIMB MOVEMENTS 433 RESULTS The transplanted limbs soon received a blood supply and the first movements appeared 6-8 days after transplantation. Since locomotion during the larval period, was by quick swimming movements, detailed analysis of ambulation was done at and after metamorphosis. In the first experimental group, however, the labyrinthine reflexes enabled good observation of the movements of grafted limbs even during larval life. If a dish containing the larva resting on its bottom is tilted, the elbow on the side of the downwards slope is extended and that on the upward slope is flexed to prevent tilting of the animal around its longitudinal axis. These responses occurred in the same manner and time in the grafted limbs innervated by heterotopic brachial segments, indicating a synchronous function with the normal forelimbs. Another phenomenon of larval life worth mentioning was found in animals of Groups 3 and 4: at the time of transplantation of the supernumerary limbs the host's own hindlimbs were not yet developed. No movement in the transplanted limbs or in the normal forelimbs, innervated by heterotopic lumbo-sacral segments, was noticed before the onset of movement in the host's own hindlimbs, although the grafted limbs had shown movement before transplantation, the circulation had developed normally and considerable growth had occurred during the immobile period after transplantation. This phenomenon may reveal that functional maturation of heterotopic lumbo-sacral segments occurs simultaneously with that of the normal segments. After metamorphosis the following patterns of co-ordination and rhythm were found. Group 1 Supernumerary limbs innervated by heterotopic brachial segments performed co-ordinated ambulatory movements in perfect synchronization with the normal forelimbs (Plate). An analysis of movement with slow-motion film revealed only a slight delay in the rhythm of the transplants when they were situated near the hindlimbs. No difference was noticed in rhythm and co-ordination when hindlimbs were transplanted save for the poor motility of the knee joint, which even in the normal hindlimb, apparently because of its gross anatomical structure, showed only slight extension and flexion during normal locomotion. Some disturbance, however, could be seen both in the co-ordination and rhythm of supernumerary limbs situated too caudal, when the animals walked on uneven ground, or came across an obstacle and tried to surmount it. During such attempts the co-ordination even between the four normal limbs also became disturbed: e.g. forelimb protraction was not always followed by hindlimb protraction on the contralateral side; or the movement of the hindlimb proceeded or became independent of that of the forelimb. Supernumerary limbs close to the forelimbs maintained synchronous movement even under such circumstances, but when situated caudally they appeared to be overruled by the
434 GEORGE SZEKELY hindlimb's rhythm, or even showed movement with quite irregular timing. In undisturbed locomotion the usual pattern of co-ordination was soon resumed. Group 2 As limb transplantation was not necessary, the movements of the host's own hindlimbs were studied when the lumbo-sacral segments were replaced by brachial segments. Such hindlimbs showed regular co-ordinated stepping movements, as had been already shown by Holtzer (1950). Orthotopic hindlimbs TEXT-FIG. 1. One complete locomotor cycle redrawn after a cinematographic film demonstrating the time delay and rhythm shift of the hindlimbs innervated by grafted brachial cord segments. The movement of the left hindlimb starts in the third phase (arrow) lagging two phases behind the forelimb movement and reaches its maximum in the fifth phase. The right forelimb starts moving in the seventh phase and is accompanied three phases later (arrow) by the hindlimb which reaches the maximum in the last phase. In the following cycle these delays become even greater and the parallel co-ordination gradually shifts into diagonal coordination. The animal has moved in shallow water. Pleurodeles waltlii. innervated by transplanted brachial segments were not only capable of coordinated movement, but showed, in addition, a definite inclination to move in parallel with the forelimbs on the same side. This parallel co-ordination, however, was not as obvious as in the former group. Even during smooth locomotion the parallel co-ordination turned occasionally into diagonal (i.e. normal) co-ordination for a short while, and the smallest disturbances in locomotion caused quite irregular, sometimes simultaneous movements in both hindlimbs. Cinematographic records revealed a considerable delay in the rhythm of the hindlimbs (Text-fig. 1), becoming greater with every step and thus eventually coming into phase with the rhythm of normally innervated hindlimbs. After a
SPINAL CORD SEGMENTS AND CONTROL OF LIMB MOVEMENTS 435 few steps the parallel co-ordination usually resumed suddenly and the two limbs on the same side moved in parallel, in camel-gait manner from which the rhythm gradually shifted again towards a diagonal co-ordination. It must be admitted that the reconstruction of the vertebral canal, owing presumably to the spatial incongruity between the large brachial segments and the narrow lumbo-sacral canal, was not always perfect in these cases. This might also have some effect on the co-ordination of the four limbs. From the observation of several cases we could, however, determine a tendency for transplanted brachial segments to maintain a forelimb rhythm in the hindlimbs. Group 3 In supernumerary limbs innervated by heterotopic lumbo-sacral segments two types of patterns of limb function could generally be distinguished. In three k TEXT-FIG. 2. Ambulation of a Triturus vulgaris over solid ground. The 11th, 12th and 13th segments have been replaced by heterotopic lumbo-sacral segments in the 38th (Gliicksohn, 1932) embryonic stage. The supernumerary hindlimbs are moving synchronously with the normal hindlimbs on the same side. Redrawn from a cinematographic film. animals the supernumerary limbs yielded completely synchronized steps with the corresponding hindlimbs (Text-fig. 2), the same behaviour which was shown by animals of the first group. The grafted limbs of the remaining eighteen animals showed greater or smaller delay relatively to the normal hindlimbs. This delay was either so small that it could only have been detected on the slow-motion cine pictures, or else it was so large that the movements of the animals were obviously quite irregular. In such cases there was, in fact, a regular hindlimb rhythm, though with such considerable delay that the regularity was only apparent when the film was analysed. The movement was even more obscured by extra steps, usually of the left-grafted limb (Text-fig. 3). This extra motion occurred simultaneously with the next step of the other grafted limb and was either very small in excursion or sometimes nearly as large as a normal step. It was difficult to find any relation between the function of supernumerary limbs and the type of operation. One got, however, the impression that these 28
436 GEORGE SZfiKELY delays were larger the younger the stages at which the lumbo-sacral segments were transplanted, and the nearer they were situated to the brachial segments. The transplanted limbs of one of the animals innervated by lumbo-sacral segments transplanted into the place of 6-8 segments at stage 30, showed such delay that their movements preceeded the hindlimbs during the next cycle of stepping and, therefore, lagged immediately behind the ipsilateral forelimbs. TEXT-FIG. 3. Locomotor cycle of a Pleurodeles waltlii in which the 8th, 9th and 10th spinal cord segments have been replaced by lumbo-sacral segments in the 30th (Sato, 1933) embryonic stage. The normal hindlimbs are paralysed by cord transsection which resulted in greater excursion of the supernumerary hindlimbs without altering their rhythm. This series illustrates the delay and the eventual synergic motion of the supernumerary hindlimbs. The left supernumerary limb is moving in diagonal co-ordination with the right forelimb, but its rhythm lags behind two phases (arrow). The right supernumerary limb, similarly, starts moving two phases later than the left forelimb (arrow) and at the same moment the left supernumerary limb produces an extra step (double arrow). All cycles are showing the same pattern. Movements performed in shallow water. Redrawn after a cinematographic film. This movement could easily be mistaken for synchronous movement with the ipsilateral foreleg. This was, however, an extreme case. When the operation was performed at the same stage but the transplanted cord was placed caudally, the delay in the rhythm was shorter. In each of the three animals showing synchronization between the supernumerary limbs and the hindlimbs, the operation was done at stage 38 and the heterotopic lumbo-sacral segments were placed beyond the ninth segment. No difference was found in rhythm and co-ordination between forelimb and hindlimb grafts. However, the movement
SPINAL CORD SEGMENTS AND CONTROL OF LIMB MOVEMENTS 437 of the transplanted forelimbs innervated by lumbo-sacral segments displayed some sort of 'hindlimb character', i.e. there was only slight extension and flexion in the elbow although the anatomical structure would have allowed greater excursions. The forelimbs thus showed a motility characteristic of their central innervation apparatus rather than of their anatomical structure. Group 4 After replacing the brachial segments by lumbo-sacral segments, the behaviour of the host's own forelimbs was very similar to that of the hindlimbs of TEXT-FIG. 4. Characteristic 'camel-gait' walking cycle of a Pleurodeles waltlii in which the brachial segments have been replaced by lumbo-sacral segments. The movements in the elbows are restricted and the left forelimb is somewhat rotated, so that its lower surface is directed medio-caudally. There is a delay of two phases in the motion of the forelimbs (arrows), they reach, however, the maximum of protraction simultaneously with the hindlimbs on the same side. Such pattern of co-ordination could be seen in several consecutive cycles. Movements performed in shallow water. Redrawn after a cinematographic film. the animals in Group 2. Beautiful parallel co-ordination developed with the hindlimbs on the same side, so that during smooth locomotion the camel-gait walking was obvious (Text-fig. 4). Obstacles to locomotion caused irregularities in co-ordination within which even the normal diagonal pattern could be well recognized. Several times a series of quick oar-stroke-like movements appeared simultaneously in both forelegs, especially underwater at the beginning of
438 GEORGE SZEKELY swimming. The movement in the elbow during walking was just as poor as was observed in supernumerary forelimbs innervated by heterotopic lumbosacral segments. The posture of the forelimbs in the resting animal was also very characteristic. While a normal animal stands still, the lower arm is bent at a right-angle (or even more acutely) to the abducted upper arm and the palms rest against the bottom of the dish beside the head. In forelimbs with lumbo-sacral innervation the upper arm was somewhat rotated, the elbow was only slightly flexed and the palm directed somewhat caudally. This posture is very characteristic of the hindlimbs at rest. Orthotopic forelimbs, therefore, innervated by lumbo-sacral segments, yielded the same 'hindlimb character' both in movement and posture. Control group Supernumerary limbs grafted into mid-thoracic levels of intact hosts remained motionless. Some rather vague movement in the shoulder could be observed, occurring simultaneously with stronger movements of the trunk. This movement never developed into co-ordinated stepping, but gradually disappeared and the limbs became progressively retarded in their growth. This is in good agreement with several findings obtained both from amphibia and birds (Detwiler, 1936; Weiss, 1955; Piatt, 1956; Szekely & Szentagothai, 1962) that non-limb segments of the spinal cord are unable to control co-ordinated limb movements. In adults such limbs were about two-thirds of the size of normals, the muscle tissues were atrophied and the joints became ankylotic. It is interesting to note that in contrast to these, limbs innervated by heterotopic limb segments even although without any visible movement and therefore discarded from the experimental series maintained their normal development. Histological findings Histological examination showed that the nerves supplying the transplanted limbs might emerge from one, two or three segments, forming a regular plexus before entering the limbs. Limbs with single segmental innervation showed as complete motility as others with multisegmental innervation. This supports Weiss' (1937) earlier finding that a single spinal cord segment from limb level can effectively innervate the limb. The central branches of the innervating plexuses were painstakingly traced back to their origin (see Text-fig. 5) in order to rule out cases in which the transplanted limbs might have received nerves from normal limb plexuses, as according to the investigations of Detwiler & Carpenter (1929) even a single branch from limb nerves would be sufficient to establish co-ordinated stepping movements in the transplanted limb innervated by thoracic segments. Connexions between the nerves of transplanted and normal limb segments occurred in none of our cases, with the exception of a single animal which has not been included in the foregoing description. In this
SPINAL CORD SEGMENTS AND CONTROL OF LIMB MOVEMENTS 439 case hindlimb segments were transplanted into the place of the segments 6-8, and a thick branch from the fifth segment contributed to the innervation of the transplanted limbs. Unfortunately, only some ill-defined twitches were seen in the hands, thus not allowing a minute study of the function of these limbs. However, the possibility of a double innervation calls for a more detailed investigation of this phenomenon. TEXT-FIG. 5. Graphic reconstruction showing the innervation of supernumerary limbs by implanted brachial cord segments (originally 3, 4, 5). From all cases investigated, the cord and limb grafts in this animal were situated most cranial positions. Two slight narrowings indicate the cranial and caudal ends of the grafted segments. The left supernumerary limb is innervated by the 7th spinal nerve to which the 6th nerve contributes with a thin branch. The right supernumerary limb is innervated by the 7th and 8th spinal nerves which before entering the limb form a plexus. A thin branch from the 6th nerve could be followed to the girdle. The supernumerary limbs showed beautiful stepping movements synchronously with the forelimbs. DISCUSSION The experiments reported gave decisive answers to two of the three questions raised in the Introduction. 1. There undoubtedly exists a specific apparatus in the brachial (3-5) and lumbo-sacral (16-18) segments of the cord, determined early in embryonic life,
440 GEORGE SZEKELY capable of innervating and bringing to co-ordinated function implanted supernumerary limbs. Transplantation of the prospective segments, that under normal conditions innervate the limbs, to any other part of the cord, does not alter the differentiation of this specific apparatus if the transplantation is made after stage 24. Complete histological fusion of the grafted segments with the neighbouring parts of the host's cord and normal development of its tissue is, of course, a necessary condition for orderly function. The negative result of Moyer (1943) and of Piatt (1957) can be fully explained from their histological descriptions showing in many cases defective development of the graft. As seen in the second section of this paper dealing with methods, from a great many operated animals only relatively few about 10 per cent. could be accepted as successful. These experiments do not give any exact information about the time at which the supposed central apparatus is developed. A detailed investigation into this question is being carried out at present in this department. 2. Apart from minor discrepancies, arising probably from differences in the anatomical structure of joints and muscles, the character of movement depends largely on the nature of the segmental apparatus. While the proximal (shoulder and hip) and distal (wrist and ankle) joints disclose equally good motility whether innervated by fore- or hindlimb segments, the brachial segments can effectively move the elbow but only to a limited extent the knee. Lumbo-sacral segments give equally poor movement both of the elbow and knee. Thus the motility of a forelimb with lumbo-sacral innervation is similar to that of a hindlimb. The extent of movement in each joint is therefore determined by the nature of the innervating segments. Likewise the posture of a resting forelimb with lumbosacral innervation is similar to that of a hindlimb,-so that a 'hindlimb character' is established both in motility and posture. The development of a movement of 'forelimb character' in hindlimbs with brachial innervation is checked by the anatomical structure of the hindlimb. A much more important body of evidence is furnished by the observation concerning timing and participation of the supernumerary limbs in the pattern of stepping. This, again, entirely depends on the nature of the segmental apparatus which innervates the supernumerary limbs, and is completely independent of the nature of the limbs. In particular, brachial segments implanted to any site between the two limb-innervating regions elicit movements of the supernumerary limbs strictly synchronous with the movements of the host's ipsilateral foreleg. The timing of stepping movements in supernumerary limbs innervated by grafted lumbo-sacral segments was diagonal in character, i.e. the movements corresponded more or less to the movement of the contralateral forelimb. It was thus, in principle, a hindlimb stepping pattern, irrespective of whether the limb actually was a hindlimb or forelimb. This became especially clear if the host's own hindlimbs were paralysed by transecting the cord below the grafted segments, or amputation of both hindlimbs. 3. Besides these results there are findings which do not allow an unequivocal
SPINAL CORD SEGMENTS AND CONTROL OF LIMB MOVEMENTS 441 interpretation. While the time lag between the movements of the ipsilateral and of the supernumerary limbs innervated by transplanted brachial segments was just discernible with the methods available to us, there was a considerable delay in the timing of supernumerary limbs innervated by heterotopic lumbo-sacral segments in most cases. The inadequacy of the analytical method and the relatively small number of cases of this type do not permit an exact statistical evaluation of the relation between the time lag and the type of operation; one gets, however, the impression that it depends on the age of transplantation and on the position of the graft in the spinal cord. The fact that the time lag in the movement of the supernumerary limbs relative to the movement of the ipsilateral hindieg of the host was greater the larger the distance of the lumbo-sacral graft from the host's lumbo-sacral region i.e. the more cranial the graft was situated immediately rules out the possibility that conduction distance of nervous impulses travelling downward through the spinal cord might account for the time lag. This assumption could be used, at best, for the barely appreciable time lag of the limbs innervated by heterotopic brachial segments. The next obvious possibility, that the movement of the implanted limb is initiated by impulses ascending from the host's lumbo-sacral segments towards the lumbosacral graft, is ruled out by the fact that cord transection immediately below the graft does not influence the timing of the movement (see Text-fig. 3). The finding that the time lag was usually greater when the animals received the cord graft in younger embryonic stages suggests a further possibility: since the specific capacity of different spinal cord segments for the control either of limb or trunk musculature is determined during embryonic development, the specificity of rhythm in the function of fore- or hindlimb segments may similarly be the result of some secondary determination which follows the former in time, and provides the fore- or hindlimb character of the timing. If the functional determination of the timing mechanism is not completed at the time of transplantation, the heterotopic hindlimb segments may adapt to their more cranial position, resulting in a shift in the timing, approaching the rhythm of the forelimb segments. This is only a theoretical possibility in the interpretation of the observed time lag and requires further investigation. Difficulties arise in the interpretation of the function of heterotopic brachial segments in lumbo-sacral positions, and of heterotopic lumbo-sacral segments in brachial positions. As already described, the characteristic function of these segmental apparatus is only temporarily obvious even during smooth walking. Disturbances in locomotion cause irregularities as great as if the fore- and hindlimbs were moving independently of each other. This, again, suggests determination problems, i.e. the influence of the surroundings may be strongest in the brachial and lumbo-sacral regions, disturbing the fore-.or hindlimb character of the grafted cord segments. As an alternative there is the possibility that central impulses addressed to brachial segments may be more effective in controlling the function of heterotopic lumbo-sacral segments in brachial
442 GEORGE SZfiKELY positions. Competition between central control of a brachial character and the heterotopic cord segments of lumbo-sacral character may result in the observed irregularities. The same may be true for heterotopic brachial segments in lumbosacral positions. From the present experiments we cannot decide how far these assumptions approach reality. Although the present experiments have raised many unsolved questions, further analysis of such experimental models may provide the necessary data for an explanation of the mechanisms by which the specific segmental apparatus for limb movements are linked together and fitted into the motor system as a whole. SUMMARY 1. Spinal cord segments from limb levels (both brachial and lumbo-sacral) were homoplastically grafted into the place of previously removed thoracic segments in newt embryos. In other embryos the brachial segments were replaced by lumbo-sacral segments or vice versa. In early larval stages a pair of supernumerary limbs were transplanted at the level of the grafted segments in the thoracic region. The function of normal and supernumerary limbs innervated by heterotopic spinal cord segments was studied. 2. When brachial segments were substituted for thoracic segments the supernumerary limbs yielded co-ordinated stepping movements, perfectly synchronized with the normal forelimbs. Apart from minor discrepancies arising from differences in gross anatomical structure, the same rhythm and co-ordination of stepping movements were found whether fore- or hindlimbs were transplanted. 3. Brachial segments in lumbo-sacral positions were able to bring the normal hindlimbs into co-ordinated movement. Although there was no correlation in the movements between the four limbs, the hindlimbs showed a definite tendency to move in parallel with the forelimbs on the same side. 4. Supernumerary limbs, innervated by heterotopic lumbo-sacral segments grafted into thoracic regions, showed co-ordinated movements which accompanied the movements of the normal hindlimbs either synchronously or, in most cases, with some delay. Essentially the same movement patterns were given both by transplanted fore- and hindlimbs except that the motility of the elbow was poorer than that of a normal forelimb. This was regarded as a 'hindlimb character' in the movement of forelimbs innervated by lumbo-sacral segments. 5. Orthotopic forelimbs innervated by heterotopic lumbo-sacral segments grafted into brachial regions also disclosed co-ordinated movements with a tendency to move in parallel with the hindlimbs on the same side. Well-defined 'hindlimb character' was verified in both the motility and the posture of such limbs. 6. The following conclusions were drawn, (i) A special segmental apparatus, already determined in early embryonic stages, exists at limb levels of the spinal
SPINAL CORD SEGMENTS AND CONTROL OF LIMB MOVEMENTS 443 cord for controlling co-ordinated limb movements, (ii) The rhythm and character of limb movements executed by these apparatus corresponded always to the nature of the innervating segments irrespective of the nature of the innervated limbs, (iii) A secondary process of determination was assumed to provide the fore- or hindlimb character of the differentiation of these segmental apparatus. The incomplete establishment of this determination might explain the observed time lag and irregularities in function of heterotopic lumbo-sacral segments under certain experimental conditions. RESUME Specificite fonctionnelle de segments de la moelle epiniere dans le controle des mouvements des membres 1. On a pratique le greffe homoplastique de segments de moelle epiniere du niveau des membres (a la fois brachiaux et lombo-sacres) a Femplacement de segments thoraciques prealablement excises, chez des embryons de triton. Chez d'autres embryons, les segments brachiaux ont ete remplaces par les segments lombo-sacres ou vice versa. Aux premiers stades larvaires, on a transplants une paire de membres surnumeraires au niveau des segments greffes dans la region thoracique. On a etudie le fonctionnement des membres normaux et surnumerai res innerves par des segments heterotopiques de moelle epiniere. 2. Quand des segments brachiaux ont ete substitues a des segments thoraciques, les membres surnumeraires ont presente des mouvements de marche coordonnes, parfaitement synchronises avec les membres anterieurs normaux. A part quelques ecarts mineurs provenant de differences dans la structure anatomique generate, on a observe le meme rythme et la meme coordination des mouvements de marche apres transplantation de membres anterieurs ou posterieurs. 3. Les segments brachiaux en position lombo-sacree ont pu induire des mouvements coordonnes chez les membres posterieurs normaux. Bien qu'il n'y ait pas eu de correlation des mouvements entre les quatre membres, les membres posterieurs ont montre une tendance nette a se mouvoir parallelement aux membres anterieurs du meme cote. 4. Les membres surnumeraires innerves par les segments lombo-sacres heterotopiques greffes dans les regions thoraciques ont presente des mouvements coordonnes qui accompagnaient ceux des membres posterieurs normaux soit synchroniquement, soit, dans la plupart des cas, avec quelque retard. Les membres anterieurs et posterieurs transplantes ont presente dans l'ensemble les memes types de mouvements, sauf que la motilite du coude etait moindre que celle d'un membre anterieur anormal. On a considere ce fait comme un 'caractere de membre posterieur' du mouvement des membres anterieurs innerves par des segments lombo-sacres. 5. Les membres anterieurs orthotopiques innerves par des segments lombosacres heterotopiques greffes dans les regions brachiales ont manifeste aussi des 29
444 GEORGE SZEKELY mouvements coordonnes avec une tendance a se mouvoir parallelement aux membres posterieurs du meme cote. Le 'caractere de membre posterieur' bien defini a ete verifie a la fois pour la motilite et la posture de tels membres. 6. On a tire de ces experiences les conclusions suivantes: (i) II existe dans la moelle epiniere au niveau des membres un appareil segmentaire special deja determine aux stades embryonnaires precoces, pour le controle des mouvements coordonnes des membres. (ii) Le rythme et le caractere des mouvements de membres executes par ces appareils correspondent toujours a la nature des segments innervateurs, quelle que soit la nature des membres innerves. (iii) On suppose qu'un processus secondaire de determination donne le caractere de membre anterieur ou posterieur a la differentiation de ces appareils segmentaires. On pourrait expliquer par une determination incomplete le retard temporel et les irregularites observes dans le fonctionnement des segments lombo-sacres heterotopiques, dans certaines conditions experimentales. REFERENCES DETWILER, S. R. (1920). Experiments on transplantation of limbs in Amblystoma. The formation of nerve plexuses and the function of the limbs. /. exp. Zool. 31, 117-69. DETWILER, S. R. (1936). Neuroembryology: An Experimental Study. New York: The Macmillan Co. DETWILER, S. R. & CARPENTER, R. L. (1929). An experimental study of the mechanism of co-ordinated movements in heterotopic limbs. /. comp. Neur. 4,1, A21-A1. GLUCKSOHN, S. (1932). Aussere Entwicklung der Extremitaten und Stadieneinteilung der Larvenperiode von Triton taeniatus Leyd. und Triton cristatus Laur. Arch. EntwMech. Org. 125, 344^05. HOLTZER, H. (1950). Differentiation of the regional action systems in the urodele spinal cord. Anat. Rec. 108, 127-8. MOYER, E. K. (1943). Innervation of supernumerary limbs by heterotopically grafted brachial cords in A. punctatum. J. exp. Zool. 94, 97-114. PIATT, J. (1956). Studies on the problem of nerve pattern. I. Transplantation of the forelimb primordium to ectopic sites in Amblystoma. J. exp. Zool. 131, 173-202. PIATT, J. (1957). Studies on the problem of nerve pattern. II. Innervation of the intact forelimb by different parts of the central nervous system in Amblystoma. J. exp. Zool. 134, 103-25. ROGERS, W. M. (1934). Heterotopic spinal cord grafts in salamander embryos. Proc. nat. Acad. Sci., Wash. 20, 247-9. SATO, T. (1933). Uber die Determination des fetalen Augenspalts bei Triton taeniatus. Arch. EntwMech. Org. 128, 342-77. SZEKELY, G. & SZENTAGOTHAI, J. (1962). Reflex and behaviour patterns elicited from implanted supernumerary limbs in the chick. /. Embryol. exp. Morph. 10, 140-51. WEISS, P. (1937). Further experimental investigations on the phenomenon of homologous response in transplanted amphibian limbs. II. Nerve regeneration and innervation of the transplanted limbs. J. comp. Neur. 66, 481-535. WEISS, P. (1952). Central versus peripheral factors in the development of coordination. Res. Publ. Ass. nerv. ment. Dis. 30, 3-23. WEISS, P. (1955). Nervous system (neurogenesis). In Analysis of Development (ed. B. H. Willier, P. A. Weiss & V. Hamburger), pp. 346-401. Philadelphia and London: W. B. Saunders. {Manuscript received 17 th December, 1962)