40, 129-153 (1974)
in Rana catesbeiana
Tadpoles MICHAEL Department
S. LETINSKY of California, April
Los Angeles, California
The physiological properties of developing nerve-muscle junctions in Rana catesbeiana tadpoles are described. Developing neurons at different stages of ontogeny formed functional synaptic connections with a section of tail muscle implanted in place of the hind limb bud. Transmission is quanta1 in nature, sensitive in normal ways to calcium and magnesium concentrations, and conforms to a Poisson distribution. The quanta1 content is initially low and increases with development. Mepp’s occur randomly and have low frequencies which increase slightly with development. The size of a single quantum of transmitter does not change during development. The muscle fibers are multiply innervated, resulting in Epp’s with distinct peaks and complex skewed mepp amplitude histograms. No significant increases were observed in the level of differentiation of the developing motor neurons as a result of their having innervated a portion of mature tail muscle. The numbers of developing motor neurons increased in the experimental lateral motor column, and a lag in their maturity was observed relative to motor neurons in the control lateral motor column. INTRODUCTION
The morphological and physiological properties of developing synapses are of great importance to an understanding of how cells interact during development. The first movements seen in developing rat embryos correlate well with the appearance of neuromuscular junctions (Straus and Weddell, 1940). Endplates are first seen at 16 days’ gestation when the muscle fibers are in myotube stage. At this time there are synaptic vesicles and mitochondria in the developing nerve terminal, the postsynaptic membrane appears thickened, and an amorphous basement membrane is present. By 18 days the nerve terminal is covered by Schwann cells, a well-formed synaptic cleft is present, and the secondary postsynaptic substructure and folds first begin to appear by day 20 (Kelly and Zacks, 1969). Diamond and Miledi (1962) found that the frequency of spontaneous miniature endplate potentials from fetal rat diaphragm increases with development, and Redfern (1970) observed complex endplate potentials (Epp’s) with multiple components in neonate rat dia129 Copyright
0 1974 by Academic
phragm, indicating that at this stage each muscle fiber is innervated by several different nerves. By the second to third week after birth, these extra nervous connections are lost and simple, single-component Epp’s persist throughout life. The primary stages of synaptic development have not yet been analyzed, however, because of the small size and fragility of embryonic cells. Most of the detailed evidence available concerning the physiological properties of developing synapses has come from tissue culture studies. Functional connections and synaptic transmission have been shown to develop in culture from nerve and muscle explants (reviewed by Shimada and Fischman, 1973). The development of chemical transmission between embryonic spinal cord and muscle during the initial formation of synapses has only been studied in culture (Robbins and Yonezawa, 1971a,b; Fischbach, 1972). Transmission was shown to be quanta1 in nature, following a Poisson distribution, having initially small quanta1 contents which increased with further development of the culture. The main objective of the following
study was to investigate the properties of developing neurons during the formation of peripheral connections. Because the limbs of amphibians develop late in ontogeny, tadpoles present the distinct experimental advantage of having embryonic nerve and muscle tissue present in a relatively large, extensively studied animals. Tadpole tail muscle innervated by developing limb nerves was chosen for the experimental system in order to overcome the technical problems of working with tiny delicate developing muscles. The mature tail muscle provides a postsynaptic structure suitable for use with electrophysiological techniques. METHODS
In these experiments frog tadpole tail musculature was implanted in place of an excised hind limb bud at different developmental stages. The implant would potentially be innervated by the whole complement of developing hind limb nerves. In control animals, tail musculature was removed, then reimplanted into the tail and allowed to become innervated by normal regenerating tail nerves; or the tail was simply denervated and allowed to become reinnervated. Frog tadpoles (Rana catesbeiana) were selected as the experimental preparation because the limbs develop very late in ontogeny. Bullfrog tadpoles are relatively large, hearty animals, with functionally mature body and tail musculature and nervous innervation, yet having wholly embryonic limb musculature and nervous innervation at the same time. Sections of mature tail muscles were used as the target for the developing hind limb nerves because they have relatively large muscle fibers (range lo-100 pm) and could be removed intact and transplanted with a minimum of trauma to the animal.
Staging The developmental stage of the tadpoles was determined by an adaptation of the Taylor and Kollros (1946) tables of larval development of Rana pipiens tadpoles. Staging was based upon changes in the external morphology of limb buds with development. A representative sample of limb buds from each stage was examined histologically so that the normal degree of nerve and muscle development could be correlated with the nerve’s physiological properties. These developmental changes are summarized below. Stage II. The hind limb bud at this stage consists of a homogeneous, evenly distributed population of mesodermal cells enclosed within a thickened epidermal layer. Early developing nerve fibers from the three lumbral-sacral spinal roots intermingle at the base of the limb bud and form the eventual sciatic plexus (Fig. 1A). Stage III. The limb bud has increased slightly in size but there are few internal morphological changes. The mesenchyme remains relatively homogeneous, lacking the medial condensation of mesodermal nuclei that Taylor (1943) described at this stage in Rana pipiens. While nerve fibers are present in the bud, there are no indications of any major nerve branches or pattern formation (Fig. 1B). Stage IV. The external changes between stages III and IV are only slight, but internally there is a generalized condensation of mesenchymal nuclei. Clear regional concentrations corresponding to presumptive femur or thigh musculature have still not formed. The nerve has split into the main cruralis and sciatic trunks (Fig. 1C). Stage V. The anlagen of the femur and thigh musculature can be discerned at the base of the limb bud. Differentiation of presumptive thigh musculature has proceeded to myoblast stage in the most proximal muscular condensations. The cruralis and sciatic branches of the nerve
FIG. 1. (A) Horizontal section through a limb bud from a stage II tadpole. The mesoderm is evenly packed and the nerve is present in the proximal half of the bud. (B) Horizontal section of a stage III tadpole hind limb bud. The bud has elongated slightly and the nerve extends further into the limb bud. There are still no noticeable condensations of mesodermal cells. (C) Horizontal section of a stage IV hind limb bud. The first signs of a medial condensation of mesodermal cells are present, and the nerve has branched into the main cruralis and sciatic trunks. (D) Horizontal section of a stage V hind limb bud. The condensations of the femur and the thigh musculature can be discerned. (E) Horizontal section of a stage VI hind limb bud. The presumptive thigh musculature and femur are now more well developed and separated by channels of less dense tissue. (F) High power magnification of nerve entering presumptive muscle from a stage VII hind limb bud. Calibration is 75 pm for A through E, and 15 pm for F.
trunk extend the length of the developing limb bud, but there are no prominent nerve branches entering presumptive muscle, and no evidence of nerve-muscle connections (Fig. 1D). Stage VI. The primordia of the thigh musculature and the femur are more prominent, and elongated spindle-shaped myoblasts are present. In addition, the nerve’s main branches are easily recognizable, and often smaller branches are seen entering differentiating muscle masses in the thigh (Fig. 1E). The general pattern is the same as that previously observed by Taylor (1943) for Rana pipiens. The major nerve branches are formed and the nerve’s distribution is essentially complete by stage V to VI in Rana pipiens and by the end of stage VI for Rana catesbeiana. At stage VI in Rana pipiens, Taylor (1943) observed many motor fibers whose tips were closely associated with newly striated myoblasts in a manner which he characterized as being quite typical of mature amphibian motor endings. In the present experiments, at the same developmental stage neither cross striations nor close association of nerves with myoblasts were observed. Both direct electrical stimulation of the limb bud and indirect stimulation of the spinal roots failed to produce any contractions or noticeable movement when observed under a dissecting microscope. From evidence to be presented later, it is known that the nerve is capable of functioning physiologically at this stage of development; however, it is not known whether functional neuromuscular connections have yet formed. The inability to elicit movement by direct and indirect electrical stimulation plus the lack of visible striations suggest immature muscle differentiation and possibly, although not nonfunctional or immature necessarily, nerve-muscle connections. Stage VII. Externally, stage VII is marked by the beginning of digit formation, and internally by increased differenti-
ation of the skeletal and muscular elements. Multinucleated myotubes and young myofibers as well as myoblasts are present in the thigh. Cross-striations are seen for the first time in the extreme proximal portion of the limb bud. Small nerves can be seen entering muscle bundles and occasionally presumptive nerve terminals can be discerned. At this stage, indirect stimulation of the limb bud through the spinal roots produces visible tremorlike contractions. This is also the first stage at which localized acetylcholinesterase activity occurred. It should be noted that although functional nerve-muscle connections have formed, there is no reflexive movement of the limb until stages IX to X. Stage VIII. All major muscle masses of the limb bud have differentiated. Small striated muscle fibers can be seen in the thigh, and small neuromuscular junctions can be observed. Correspondingly, the limb bud responds to direct and indirect electrical stimulation with larger and more forceful contractions. However, still no reflexive limb movements can be elicited. Stages IX-X. In these later stages there is a gradual maturation of the limb’s musculature and nervous innervation. The nerve pattern first established by stage VI fills out as more nerve fibers grow in and the adult innervation becomes realized. The first reflexive movements of the developing limb are seen, consisting of a simple leg extension with passive relaxation. Because of the protracted larval development of Rana catesbeiana tadpoles, there was generally no change in developmental stage during the time between the operation and the animal’s sacrifice. In a few cases there were slight developmental changes-however, never more than one stage. For consistency, these and all other experimental animals were staged according to their developmental stage at the time of sacrifice. Since there is ongoing outgrowth of new axons throughout most of development (Taylor, 1943; Hughes and
like connective tissue. The hind limb bud was removed, along with a small amount of the surrounding tissue to prevent regeneration (Goss, 1969). The implant was positioned inner side down, directly over the severed nerve stump, in place of the excised limb bud. One end was secured with 7-O sutures to the intact tail musculature, stretched, and the other end sutured to the abdominal musculature. Both incisions were closed with 6-O silk sutures. The positioning of the implanted tail muscle is critical, and must be such that the severed sciatic stump is completely covered by the implanted muscle. If this is done accurately, the cut nerve, which originally entered the limb bud, consistently grows into the implanted tail muscle and forms functional synaptic connections. The operated tadpoles were stored in Surgical Procedure pairs in plastic containers with 500 ml of Bullfrog tadpoles (Rana catesbeiana) be- aerated water, to which 0.1 mg of tetracytween stages II and X were selected. The cline per milliliter and one drop of 0.35% animals were anesthetized with 1/000 malachite green were added daily for the (0.025%) MS 222 (tricaine methanesulfofirst 3-4 postoperative days to prevent nate, Finquel, Ayerst) and kept continuinfection. Every day the animals were ally moist during the operation by being examined, and the water in each tank was covered with gauze saturated with icechanged. They were fed boiled lettuce chilled Ringer’s. Three intact myotomes of daily. tail muscle, containing 200-300 muscle Forty-seven limb bud replacement operfibers, were removed from the dorsal half of ations were performed; of these, 29 surthe tail, approximately mid-way down its vived and were examined 21-66 days later. length, their inner surface was cleaned and any connective tissue was carefully re- Physiological Procedures moved. Care was taken not to injure the For physiological experiments the imspinal cord during removal of the muscle. planted muscle and the underlying abdomVery fine threads from unraveled 6-O su- inal musculature were removed together, tures were then tied, one to each corner, to and mounted in a small plastic chamber on the lateral two myotomes, leaving the cenSylgard resin. In initial experiments, the tral myotome intact. The excised and rectus abdominus musculature was discleaned muscle was then stored in icesected away and microelectrode recordings chilled Ringer’s prior to implantation. were made from the exposed superficial Next, an L-shaped incision was made in fibers on the implant’s inner surface. Howthe abdominal skin just above the hind ever, this procedure apparently damaged limb bud extending down the midline and the fragile nerves growing over the muscle’s then laterally. This skin flap was reflected inner surface, for many fibers that had iniforward toward the head, and the abdomitially contracted could no longer be driven. nal musculature was cleaned of any jellyTherefore, in the majority of the later
Egar, 1972), it must be remembered that the developmental stage marks the maximum age of any nerves present, but younger, less mature nerves could be present as well. That is, the implanted muscle could become innervated by both regenerating severed axons as well as intact, but later maturing, developing axons that were not severed at the time of the operation. In these experiments, the numbers assigned to the spinal roots were the same as those used by Ecker (1889) for adult frogs, in which the three main spinal nerves of the hind limb were numbered 7, 8, and 9. This was done to avoid any confusion with alternative numbering systems found in the literature (e.g., Taylor, 1943, who numbered the corresponding roots to the hind limb bud 8, 9, and 10).
experiments, the rectus abdominus musculature was left attached and recordings were made from the superficial fibers of the outer surface, which had faced the skin. Although the connective tissue buildup on the outer surface was much denser than that on the inner surface, making electrode penetration difficult, recording from the outer surface presented the advantage of not disturbing the muscle’s innervation pattern. Microelectrode penetrations were usually made in the central region of the muscle where it was easier to distinguish the outlines of individual fibers. Physiological experiments were performed in Ringer’s of the following composition: 115 mM sodium chloride, 2 mM potassium chloride, 1.8 mA4 calcium chloride, and phosphate buffer (pH 7.2). When it was desirable to reduce the endplate potential below threshold for action potential initiation, modified Ringer’s with either 0.5 to 1 mM calcium chloride and 2 to 8 mM magnesium chloride or 1 to 10 x 10mB M d-tubocurarine chloride (Abbott) were used. For stimulation the nerve was sucked up into a fine capillary electrode with some Ringer’s, and current was passed between the inside and outside. The temperature was constantly monitored and maintained between 18” and 19°C. Electrical activity was recorded with very fine glass microelectrodes filled with either 3 M potassium chloride or 4 M potassium acetate and having resistances of 50-100 megohms. The responses were amplified with a WPI negative capacity preamplifier, and displayed on a Tektronix 561B oscilloscope or a Brush Mark 220 chart recorder. Sharp, high-resistance electrodes were necessary since tail muscle fibers appear to have very fragile membranes and to be easily susceptible to damage, in spite of their relatively large fiber diameter. Current for electrical measurements was supplied by an operational amplifier current pump driven by a Devices Digitimer stimulator. A second oper-
ational amplifier was used to monitor current and keep the bath at virtual ground. Histological
Cross sections of hind limb buds were stained for developing nerve and muscle, as in the earlier literature (Taylor, 1943), by a combination of Bodian silver impregnation and the Mallory-Heidenhain trichrome method. The presence of acetylcholinesterase activity at the endplate in normal and implanted tail muscle innervated by developing hind limb nerves was determined by Karnovsky’s thiocholine technique (Karnovsky and Roots, 1964). Spinal cords were removed and fixed at least 3-5 days in Bouin’s fixative, then 10 pm serial sections were cut and mounted on albuminized slides. These were stained by Mayer’s Acid Alum Hematoxylin and Mallory Eosin to visualize the motor neurons in the developing ventral horn. The developing nerves were stained using an adaptation of the Gros-Bielchowsky silver impregnation technique (Letinsky, 1972). RESULTS
Hind Limb Bud Development
The 7th, 8th, and 9th spinal nerves approach the developing hind limb bud from a mediodorsal direction and join near its base to form the sciatic plexus. These nerves consist of two distinct fiber types: a small number of myelinated primary fibers, and a large number of extremely fine, unmyelinated secondary fibers which stain only faintly (Taylor, 1943; Letinsky, unpublished results). The mature primary fibers can be easily followed and seen to branch off the main nerve root at intervals to innervate the already well-differentiated body musculature and the most anterior portions of tail muscle and skin. The fine secondary nerve fibers appear later in development, forming a bundle adjacent to the large fibers, which tapers peripherally as it approaches and enters the sciatic
plexus at the base of the limb bud (Fig. 2). A few primary fibers can sometimes be seen among the fine fibers at the base of the sciatic plexus. With careful dissection these fibers can be followed and observed to branch off from the sciatic plexus before it enters the limb bud. They were never observed to grow into the hind limb bud itself. Taylor (1943) demonstrated that in the larva of Rana pipiens the primary fibers innervated only myotomes and skin, end organs which are normally functionally innervated during the embryonic period, and that the innervation of the developing limb is supplied exclusively by the fine secondary fibers. This was confirmed to be
true of Rana catesbeiana by physiological evidence to be presented later. An analysis of hind limb bud development was performed (see Methods; Letinsky, 1972). This was done to provide a baseline for comparing the developing nerve’s physiological properties to the level of morphological development of the normal hind limb bud musculature and its nervous innervation. The changes that occur during normal hind limb development are summarized below. The first developing nerves (secondary fibers) are present in the limb bud at stage II, at the time well before any muscle differentiation. Myoblasts first appear at stage VI and striated myotubes are found in the presumptive thigh musculature by stage VII, coinciding with the first appearance of functional nerve-muscle connections and localizations of AChE activity. Normal reflex limb movements do not occur, however, until two to three stages later (IX-X). Evidence for Innervation of Transplanted Tail Muscle by Developing Secondary Nerves
FIG. 2. Schematic representation of a spinal root (SpR) leading to the hind limb bud (HLB). The spinal root consists of two fiber types: primary fibers (PF) and secondary fibers (SF). Note that only secondary fibers enter the limb bud. The primary fibers branch off from the main spinal root to innervate somatic body muscle (EM) and proximal tail muscle (‘I’M).
In the present experiments, the nerves which normally innervate the hind limb bud were induced to innervate implanted tail muscle. These spinal roots consist of both primary and secondary nerve fibers. In the normal animal, the primary nerve fibers branch off from the spinal roots proximal to the sciatic plexus and thus do not enter the hind limb bud. However, since primary nerves normally innervate tail and body musculature, and since the experimental muscle was a section of mature tail muscle, it was conceivable that primary fibers, rather than the intended secondary fibers, had grown into and innervated the transplanted muscle. First, studies were conducted on the experimental animals to test for any possible sprouting of primary nerve fibers. In animals where the implanted tail muscle
became successfully innervated, various primary rootlets near the sciatic plexus were stimulated in situ after they had left the main spinal root (see Fig. 2). If branching had occurred, it was assumed that the nerve impulse would be conducted antidromically to the branch point, invade the newly formed sprout and cause contractions in the implanted muscle. No such response was ever seen in the experimental muscles, although stimulation of the main spinal root (secondary fibers) did produce contractions. As a further test, the sciatic trunk was stimulated in situ at the point where it passed through the body wall to innervate the implant. None of the primary innervation had been disturbed. As before, it was assumed that if sprouting had occurred, the nerve impulse would be conducted antidromically to the branch point, invade the original primary neuron and cause movements in the body and tail muscles. Again, no evidence of primary nerve sprouting and stray innervation was observed. Additional evidence that the implanted muscle became innervated selectively by secondary fibers was obtained from animals at advanced developmental stages (stages IX-X). It is at this point in ontogeny that normal reflexive connections between the developing hind limb and the spinal cord are first formed. In normal unoperated animals at about stages IX to X, when one limb was moved or prodded, the reflexively connected contralateral limb moved simultaneously. At a comparable stage in experimental tadpoles, a similar result was observed in the control limb and the implanted muscle. When the control limb was moved, the implanted muscle also contracted, implying that those nerves which would have normally innervated the hind limb bud (developing secondary fibers) were now innervating the implanted muscle. Furthermore, in every experiment in which the implanted muscle was found to
be functionally innervated, there was no regeneration of the excised limb bud. Limb bud regeneration did occur in two experimental animals whose implants did not become functionally innervated, and in control experiments where the limb bud was removed but no muscle implanted. Because nerves are required for limb bud regeneration (Singer, 1952), these results suggest that the nerves in the experimental animals which would normally have reinnervated the limb bud and supported its regeneration were, in fact, innervating the implanted muscle. Finally, the observed physiological responses were characteristic of what one would expect from small, immature, developing motor neurons, differing markedly from responses normally seen in muscle innervated by primary nerves or in adult frog nerve-muscle preparations. For example, the threshold for nerve excitation was considerably higher for the experimental nerve (3.5-5.0 V) than for that of primary nerves (0.2-0.5 V). The nerve conduction velocity, as measured from the latencies of the synaptic events, was 0.068 * 0.036 (mean i SD) meters per second, as compared to 0.54 * 0.09 (mean * SD) meters per second in the controls. Moreover, the synaptic physiology, as described below, also differed noticeably in several respects from control and adult preparations. General Ph.vsiology of Tail Muscle Innervated by Developing Secondar?/ Nerve Fibers Muscle contraction evoked by developing secondary nerves. Although functional synapses in normal developing limb musculature were first seen at approximately stage VII, the developing sciatic nerve was capable of driving a tail muscle implant in a host of only stage II development, the youngest tested. In stage II-IV animals, indirect stimulation generally elicited contractions in only a small percentage of the fibers in the implant. Usually, the first
evoked response was the largest; with repeated stimulation the muscle twitches became progressively weaker. This was assumed to be due to nerve fatigue. In order to overcome this and to prevent complete nerve failure, only a few muscle contractions were elicited, always at low repetition rates of 0.1 to 0.2 impulses per second. Sometimes, even at these low stimulus rates, there were intermittent contraction failures, where some contracting fibers ceased twitching completely while the remaining fibers continued to contract. In the more advanced animals (stages V-X), indirect stimulation produced muscle contractions in a larger percentage of the implant; however, sections of the muscle were still inactive and apparently uninnervated. It is possible that these apparently uninnervated fibers were, in fact, innervated but that the synaptic potentials were subthreshold for spike initiation so no muscle contractions occurred. This was the case in several experiments, for if twin pulses were used, summation or facilitation of the synaptic potentials caused more fibers to contract, although never the entire muscle. Using small gradations in stimulus intensity it was possible to distinguish motor units having different thresholds. Two or three such discrete motor units were observed in immature experimental animals (stages II-IV) and up to five at more mature developmental stages. Thus there was a definite trend toward an increasing number of functionally innervated fibers with advancing stages of development. The stage seemed much more important than the length of time following the actual muscle implantation. Long-term experimental muscles occasionally seemed to be better innervated, having a greater number of functionally innervated muscle fibers compared to shortterm but identically staged experimental muscles. However, these differences were not as obvious or as consistent as those attributable to different stages of ontog-
eny. I have no convincing evidence to support the hypothesis that these smaller and less obvious differences were the result of the length of time of innervation; they could equally be attributed to random variables resulting from the experimental procedures. Resting membrane potentials. Once it was determined that the muscle fibers were functionally innervated, electrophysiological studies were carried out. In these experimental muscles the resting potentials were often unstable and decreased toward zero within minutes, although in several cases stable resting potentials were held for periods of up to 1 hr or more. The instability of these resting potentials was due, at least in part, to damage to the extremely fragile tail muscle fiber membrane during microelectrode penetration. Relatively stable recorded resting potentials were generally low in the experimental animals, between -45 to -70 mV, although occasionally fibers were penetrated with membrane potentials as large as -90 mV. These values were below those found in normal unoperated tadpole tail muscle fibers which, like adult frog skeletal muscle fibers, have resting potentials of approximately -90 to -95 mV. In the present experiments, even though the implanted tail muscle fibers were innervated by developing motor neurons, there was no correlation between the resting potential and either the tadpoles’ developmental stage or the chronological age of the implant. It seems probable that the low values resulted from damage during electrode penetration rather than from any trophic influence exerted by the developing nerve. Spontaneous electrical activity. Small spontaneous potentials (Fig. 3) resembling miniature endplate potentials recorded at adult frog nerve-muscle junctions (Fatt and Katz, 1952) were observed in the experimental muscles at each developmental stage studied from II to X. These potentials were presumably due to quanta1
FIG. 3. Spontaneous mepp’s from a stage IX experiment. Three different classes of mepp’s can be distinguished. Calibration is 1 mV and 100 msec.
packets of transmitter spontaneously released from developing secondary nerve terminals. They will be referred to here as miniature endplate potentials (mepp’s). Mepp’s were not observed in every muscle fiber, but generally only from the relatively few fibers which later proved to be functionally innervated. However, mepp’s were occasionally present in muscle fibers in which no Epp’s could be elicited. It was possible that these mepp’s were from very immature nerve terminals. This is consistent with the finding that mepp’s are present without Epp’s during the early stages of regeneration in adult frogs (Miledi, 1960). The mepp frequency observed in functionally innervated experimental muscles was always extremely low, and sometimes no mepp’s appeared at all. As shown in Fig. 4, the average mepp frequency at stage II was the lowest (0.43 mepp’s per minute). The frequency increased slightly with further development to 2.45 mepp’s per min-
ute at stage X. These values were significantly less than those found at normal primary nerve reinnervated tail muscle endplates (range: 0.4 to 5 mepp’s per second), or those found at adult nervemuscle junctions (Fatt and Katz, 1952). Furthermore, at immature developmental stages there were several examples of functionally innervated experimental muscle fibers in which no spontaneous mepp’s could be recorded, but in which a few mepp’s could be observed both on the falling phase of Epp’s and following Epp failures, representing delayed release of transmitter following nerve stimulation (Rahamimoff and Yaari, 1973). These latter fibers were considered to have zero mepp frequencies. The length of time that the muscle had been innervated had no effect on the mepp frequency; long-term experimental animals were just as likely to have low or zero mepp frequencies as similarly staged but shorter-termed ones. The frequency of experimental mepp’s, like that of normal adult mepp’s, was sharply affected by hypertonic solutions. When the tonicity of the Ringer’s was doubled by the addition of either sucrose or NaCl, the experimental mepp frequency increased markedly; within minutes the accelerated rate was 50-100 times normal.
FIG. 4. Average mepp frequency for each developmental stage between II and X. The vertical bar represents + SD.
For example, in one experiment (stage III, 42 days) the mepp frequency increased over 100 times, from 0.72 to 81 mepp’s per minute; and in another more advanced animal (stage VII, 41 days), the frequency increased from 0.94 to 46 mepp’s per minute, a 48-fold increase. These increased mepp frequencies were still quite low compared to the accelerated mepp frequencies of similarly treated normal tadpole tail muscle (approximately 100 mepp’s per second) or of adult motor endplates [between 90-150 mepp’s per second (Fatt and Katz, 1952) 1. Furthermore, in the experimental muscles these accelerated release rates could not be sustained and eventually the frequency declined noticeably, sometimes even to zero. This decrease was most evident and had an earlier onset in the more immature animals, where there was often a complete cessation of both spontaneous activity and of nerve-evoked Epp’s. This precipitous drop in mepp frequency can be attributed to depletion of transmitter stores in the presynaptic terminal. At every stage, when recording from the central region of single experimental muscle fibers, numerous mepp’s were observed with different amplitudes and time courses (Fig. 3). The presence of these different classes of mepp’s was suggestive of multiterminal or multiple innervation, i.e., packets of transmitter being released at numerous spots along the fiber length, either from different branches of the same nerve or (as is likely the case in these experiments) from separate, individual nerve fibers. It is also possible that these were degeneration mepp’s due to transmitter released from Schwann cells at the sites of the old endplates (Birks et al., 1960). This latter explanation seems unlikely because the time courses of the experimental mepp’s generally corresponded well with the time courses of evoked Epp’s and, unlike degeneration mepp’s, the experimental mepp’s responded to hypertonic solutions with markedly increased frequen-
cies. Histograms of the experimental mepp amplitude distributions, as shown in Fig. 5, were complex and often skewed. The largest potentials could have resulted from spontaneous action potentials in the developing terminals. These differed from the relatively simple amplitude histograms constructed from primary nerve-reinnervated tail muscle fibers (Fig. 5D) and from normal tail or adult endplates. There were no obvious changes in the amount of variation in mepp amplitudes and time courses per fiber during either increasing age or development. Although these experimental fibers were multiply innervated, all the mepp’s recorded should still be independent and subject to statistical analysis. To test for random release, a mepp interval histogram was constructed and compared with the expected theoretical curve generated by the equation: n = N t/T exp (~ t/T) (Fatt and Katz, 1952). Figure 6 shows that this is indeed the case.
Size of an individual quantum of transmitter. There was large variability in the amplitude of mepp’s recorded from different experimental muscle fibers. It was of interest to ascertain whether a similar variability existed in quanta1 size, the absolute size of the individual packets of transmitter which produce the mepp’s, and whether there were any changes in quanta1 size during development. Katz and Thesleff (1957) demonstrated that at the frog neuromuscular junction, despite the wide range of mepp amplitudes, the absolut,e size of the individual quanta of transmitter was relatively constant. The same type of analysis was applied to t,he present experimental mepp’s. Figure 7 contains pooled results from several different experimental muscles between stages II and X. These experimental results must be considered in the light of the recording procedure and the physiology of these developing nerve-muscle junctions. First, focal recordings were not made because the
FIG. 5. Amplitude histograms of mepp’s recorded from three experimental muscles and one primary-reinnervated control muscle. The experimental histograms (A, B, and C) are complex and skewed, while the control histogram (D) is relatively simple.
20 24 INTERVAL(SEC)
FIG. 6. Interval histogram for spontaneously occurring mepp’s from an experimental innervated muscle fiber. The curved line represents the expected exponential decay for a random event, with a mean interval of 4.95 msec.
exact location of the developing endplates was unknown, and also because repeated penetrations injured the muscle fibers. Therefore, due to spatial decay, the recorded mepp amphtudes were smaller than
at their sites of origin. Second, because the muscle fibers were generally multiply innervated and because the mepp’s occurred very infrequently, there was considerable overlap in the individual classes of mepp’s. It was thus difficult to obtain a true value for the mepp amplitude. Third, the resting potentials varied and were generally low, between -45 and -70 mV. Consequently, the electrochemical driving force responsible for the miniature potentials differed, resulting in variations in the recorded mepp amplitude. Fourth, the resistance measurements were made at the recording electrode and not at the origin of the mepp’s. In order to reduce these inaccuracies, only those mepp’s with the fastest rise times and only muscle fibers with relatively large resting potentials were considered. To compensate for the low resting
FIG. 7. The relationship between the mepp amplitude (mV) and the input resistance (megohms) of pooled results from seven experimental animals between stages II and X. The solid line has a slope of 1, which depicts the idea1 linear relationship.
potentials each mepp was multiplied by the factor 75/(E-15) (Katz and Thesleff, 1957)) whereby E, the measured membrane potential, was normalized to a standard resting potential of 90 mV. These experimental limitations made rigorous quantitative studies impossible. Nevertheless, a roughly linear relationship was apparent in the experimental plot (Fig. 7); there were no aberrantly large or small values. The muscle fibers with the largest input resistances (smallest diameters) had the largest mepp’s. These data suggest that the size of the individual packets of transmitter were not significantly different and did not change markedly during the developmental period studies. Synaptic events-endplate potential fluctuations. Electrophysiological correlates of synaptic transmission were present at every developmental stage examined. The recorded synaptic potentials were made up of multiple Epp components, each of which fluctuated widely and sometimes failed even when no synaptic blocking agent was applied. This was consistent with the fluctuations observed during whole muscle contractions. Samples of
these composite synaptic potentials can be seen in Figs. 8 and 9. In Fig. 8, a curareblocked preparation (stage V, 53 days), the stimulus intensity was increased between each trace, bringing in successively one, two and then three individual Epp components, each with a different threshold and latency. In the magnesium-blocked experiment shown in Fig. 9 (stage VII, 41 days), the individual Epp’s fluctuated and failed independently of each other. No obvious correlation existed between the latency of the separate components and either their threshold or time course. Often Epp’s with relatively slow time courses occurred with shorter latencies than Epp’s with faster rise and fall times. These differences in time course and the different classes of mepp’s suggest large distances between different components, probably the result of innervation by individual nerves having different thresholds. In many instances it was
FIG. 8. Illustration of the response to slowly increasing stimulus intensity. The stimulus intensity was gradually increased between each trace. In the top there was no response. In the second trace there was a single-component Epp. When the stimulus intensity was further increased, a second, later component appeared in the third trace. At a still higher threshold, a third Epp was evoked. Calibration is 2 mV and 40 msec.
FIG. 9. Superimposed traces of a Mg-blocked endplate. Individual Epp’s fluctuated and failed independently of each other. Note the different latencies and time courses. Calibration is 2 mV and 40 msec.
not possible to distinguish separate components by fine gradations in stimulus intensity, but the various components could still be seen to fluctuate independently. These nonseparable composite Epp’s could result from innervation by different nerves with similar thresholds or different branches of the same nerve. The latencies of the individual components within a given set of Epp’s were very constant (e.g., 25.6 * 0.4 msec, corresponding to a conduction velocity of approximately 0.039 meters/second). Thus, these components could not be accounted for by nonsynchronous release of transmitter from a single terminal or by spontaneous mepp’s. No correspondence was found between the amount of multiple innervation and the developmental stage of the nerve or the length of time following the operation. Epp amplitudes varied in a steplike fashion and sometimes failed completely during these experiments. It was important to determine, if possible, whether these failures were the result of transmission
VOLUME 40, 1974
failure in the presynaptic terminal or the result of nerve conduction block. Abrupt, long-lasting presynaptic nerve failures have been demonstrated in adult rat (Krnjevic and Miledi, 1959) and in neonate rat (Redfern, 1970), during high frequency stimulation (20-50/set). This was not the case in the present experiments. The stimulus frequency was kept very low (0.2/set) and no extended periods of repeated failures occurred. Still, the possibility to conduction failures had to be ruled out. Mepp’s were occasionally observed both on the falling phase of evoked Epp’s and following Epp failures with a frequency much higher than expected from their low frequency without stimulation (less than one per minute). The slight increase in mepp frequency during stimulation may be correlated with the depolarization which results when the nerve action potential invades the presynaptic terminal. The fact that mepp’s were observed following failures as well as Epp’s implies that action potentials entered the terminal in both cases; t,herefore, zero responses may, in at least some cases, be attributed to failures in the transmitter release mechanism, rather than to invasion of the terminal. In such cases, the quanta1 contents were not inconsistent with what one would expect from endplates with occasional transmission failures. Quanta1 content can be determined by two independent methods derived from the Poisson distribution: (1) failure analysis, and (2) calculating the ratio of the average Epp amplitude to the average mepp amplitude (de1 Castillo and Katz, 1954). If the observed failures were the result of nerve conduction block rather than the release process, the number of failures would be overestimated and the two methods of determining m would give different results. As will be discussed in detail below, when these Epp’s and failures were subjected to Poisson statistics, the resulting distribution and estimation of quanta1 contents
conformed well with the predicted Poisson distribution. The experiment illustrated in Fig. 10A was especially interesting because it was from one of the most immature stages (stage II-III, 27 days) and was recorded in normal Ringer’s without any blocking agents. It represents, therefore, the true synaptic output of the nerve at this stage. For the Poisson theorem to apply in this context, there must be a relatively large store of possibly reacting units (n), and the probability (p) of any one unit being released following a single nerve stimulus must be small. Ideally, there should be no interaction between the effects of successive nerve impulses, each Epp being independent of what occurred before it. However, there is evidence (Letinsky, 1974) that these young terminals show marked
depletion of transmitter (nerve fatigue) when stimulated at frequencies greater than 0.33lsecond. In this experiment, the nerve was stimulated once every 5 set, so that enough responses could be obtained for statistical studies while, hopefully, significant depletion did not occur. Conventionally, the average quanta1 content (m) is first calculated from the average Epp size (Epp)/average mepp size (mepp) and then the theoretical curve expected from a Poisson distribution is generated (Martin, 1966). However, in this experiment, for the construction of a Poism was calculated in the son distribution, same manner as that used by Dude1 and Kuffler (1961). The mean quanta1 content for the Poisson distribution was determined from the number of failures (In (N/N,) = m) rather than from the average
FIG. 10. Epp amplitude histograms. A is from an immaturely staged experimental muscle (stage II-III) recorded in normal Ringer’s without any blocking agents. B and C are amplitude histograms from muscles recorded in Mg-Ringer’s (B is stage VI, C is stage VII). The number of failures is indicated by the bar at the origin. The curved line was theoretically determined by application of the Poisson distribution using the predicted number of quanta1 events derived from the experimentally determined mean quanta1 content. Roman numerals represent multiples of a single quanta1 unit.
mepp size. This was necessary because, as with externally recorded mepp’s in crayfish, the mepp frequency was extremely low and determination of the average mepp amplitude was inaccurate. However, the few mepp’s that were recorded can be used later as a further independent check for the accuracy of the Poisson distribution. Next, the theoretical Poisson curve was constructed. The single unit size (9) normally equals the mepp size, but in this experiment it was derived from the average Epp size and m:
q = Epplm By using this calculated value for q (0.64) and an average statistical variance, a theoretical Poisson distribution curve was constructed (Fig. 10A) according to the general procedure outlined in Martin (1966). Further confirmation that the observed results fit a Poisson distribution was obtained from the spontaneous mepp’s. Three mepp’s which had the same time courses as the Epp’s were recorded: 0.65 mV, 0.65 mV, and 0.60 mV. These values correspond well with the size of the calculated unit potential (0.64) and further support the hypothesis that transmission was quanta1 in nature. Another method of testing for quanta1 release was to compare the mean quanta1 content calculated using various methods derived from the Poisson theorem. In addition to determining m from In (N/N,,) or e/rnepp, the mean quanta1 content was also estimated by the quantity 1/CV2, where CV is the coefficient of variation of the Epp distribution (de1 Castillo and Katz, 1954; Martin, 1966). A sampling of m from several different developmental stages calculated by three different methods is presented in Table 1. These independent estimates were in agreement at every stage studied. Quanta1 content. Quanta1 contents were calculated for each developmental stage using both unblocked or curare-blocked
preparations. Table 2 shows that there was a general increase in the quanta1 content during advancing development. The lowest average values were found at stage II (m = 1.32 * 0.74 SD), the quanta1 content increasing to 11.45 * 4.46 SD by stage X. While the average quanta1 content tended to increase during development, the values at any given stage covered a wide range and there was much overlap between different stages. In fact, at every stage examined, including the most advanced, there were still nerve fibers with extremely low quanta1 contents, 1.62 at stage IX and 3.78 at stage X. These low values could be due to very immature, newly developing secondary nerves which had just grown out
COMPARISON OF QUANTAL CONTENTS ESTIMATED BY MORE THAN ONE METHODS -l/CVZ In (N/N,) Qwlmepp 2.32 0.82 3.42 2.37 0.99 1.82 2.42 1.20 1.14
2.40 0.79 3.38 2.41 0.83 1.73 2.40 1.29 1.46
3.20 0.68 1.95 2.57 1.14 1.19
“Values were taken from animals developmental stages.
at all different
AVERAGE QUANTAL CONTENT AT EACH DEVELOPMENTAL STAGE Stage II III IV V VI VII VIII IX X
Average * SD 1.32 1.34 4.08 5.38 6.95 10.26 6.24 7.18 11.47
zt 0.74 f 0.93 zt 1.95 i 1.71 + 2.87 i 5.71 i 3.38 + 4.60 f 4.46
MICHAEL S. LETINSKY
and formed connections of small terminal area. These values for the quanta1 content may not accurately reflect the transmitter output capabilities of the developing nerve terminals because of the marked depletion which occurs during repetitive stimulation (Letinsky, 1974). This would be especially apparent at the most immature developmental stages. Epp time course. The rise time of the experimental Epp’s ranged between 2.4 msec and 11.7 msec, but it was difficult to interpret these results because focal recordings of the developing endplates were not made. Nevertheless, a pattern did emerge. The fastest rising Epp’s were from the more advanced animals (e.g., 2.4 msec at stage VII, and 2.5 msec at stage IX). At the more immature stages (II to IV) no Epp’s were ever observed with rise times faster than 4.5 msec with an average of 7.02 h 2.15 SD msec. If the developing endplates were distributed randomly along the muscle fiber, one could expect at least some recordings to be very close to individual endplates. The fastest observed values may have corresponded to just this type of recording situation, and, thus, these values may have been close to the actual Epp rise time.
The experimental half-fall times were consistently very long compared to adult Epp’s, ranging from 5.2 msec to 36.4 msec with an average of 12.67 * 6.12 SD msec. It was not possible to show a correlation between the half-fall time and either developmental stage or age of the implant because the composite nature of the synaptic potential made measurements unreliable. The slow half-fall times of the experimental Epp’s suggest that there was no acetylcholinesterase (AChE) present at newly formed developing nerve-muscle junctions. Histological and physiological tests show this to be the case. There were no sharply localized deposits of AChE activity along the fiber length, although there was activity at the old degenerating endplates. It is known that AChE activity can persist for over 2 months at degenerating endplates (Feng and Ting, 1938; Filogamo and Gabella, 1966). Consistent with a lack of localized AChE activity, there was no change in either the time course or amplitude of developing mepp’s or Epp’s when 5 x 10e6 M prostigmine was added to the bathing solution (Fig. 11). During normal limb bud development, AChE localization first appears at stage VII; however, there was no physiological or histological evidence of localized AChE activity at
FIG. 11. Epp’s recorded in prostigmine. A is from a stage III preparation after 42 days. B is from a stage IX preparation after 64 days. In (a) the Epp’s are shown in Ringer’s without prostigmine, and in (b) in Ringer’s with 5 x 1O-B M prostigmine added. Notice the increased mepp release following the Epp in B. Calibration is 1 mV, 100 msec in A and 2 mV, 100 msec in B.
developing endplates in transplanted tail muscle from any stage examined between II and X. Epp latency and nerve conduction uelocity. The experimental and control conduction velocities were calculated from the latencies of the synaptic potentials. In the experimental muscles, only that portion of the developing nerve distal to the sciatic plexus was used. Consequently, the length of nerve from the stimulating electrode to the muscle was always very short, between approximately 670 and 1800 pm. Although the control nerves were longer, for purposes of comparison only short lengths of nerve were used corresponding to those of the experimentals. Even with these very short lengths of nerve the Epp latencies were relatively long; in the experimental muscles the Epp latencies ranged from 6 to 34 msec with an average latency of 18.8 & 6.3 SD msec. The average latency for the control Epp’s was 2.8 * 0.1 SD msec (range from 2.2 to 4 msec). Examples of these different latencies can be seen in Fig. 12. Approximate conduction velocities were calculated by using the synaptic latency and the distance from the stimulating electrode to the recording electrode. Conduction velocity values computed for the experimental nerves ranged between 0.034 and 0.13 meters per second with an average value of 0.068 + 0.036 SD meters per second. Compared to the experimental results the calculated control conduction velocities were nearly one order of magnitude greater; their average was 0.54 +=0.09 SD meters per second, with a range of 0.42 to 0.67 meters per second. There was no apparent correlation between the developmental stage and either the latency or the conduction velocity. Epp’s recorded from different fibers of the same muscle often had widely different latencies. For example, in one experiment (stage VI, 52 days), Epp’s were recorded with latencies of 14 and 34 msec, which
FIG. 12. A shows Epp’s with different synaptic latencies from three experimental muscle fibers. B is an example of a primary reinnervated control Epp. The control has a much shorter synaptic latency. Calibration is 1 mV and 40 msec in A, and 2 mV, 40 msec in B.
corresponded with conduction velocities of 0.12 and 0.039 meters per second. In another experiment (stage V, 53 days) one fiber had three separate Epp’s with latenties of 17, 22, and 30 msec. One possible explanation for these results is that the conduction velocity within the terminal may be the limiting factor and may predominate. The age of the terminal may be the important variable rather than the stage. Spinal Cord Analysis It was apparent from the physiological results that developing secondary nerves were able to function at very early developmental stages (as early as stage II), well before there normally was any observable
MICHAEL S. LETINSKY
nervous activity (stages IX-X). Two possible hypotheses may account for this: (1) At these early stages the developing nerves were functional, but ordinarily there is no differentiated target muscle, or (2) The developing nerves at these early stages were nonfunctional but had been induced to differentiate to a more mature functional state by innervating a mature muscle instead of the usual developing limb bud. In an attempt to distinguish between these two possibilities, serial sections were made of the lumbar-sacral region of the spinal cord, and the level of development and the numbers of developing secondary motor neurons were examined at each stage between II and X. Developing motor neurons were recognizable by their position and appearance. During normal development of the lumbar spinal cord, there is a gradient in the level of differentiation of these motor neurons from the cranial to the caudal ends of the lateral motor column. The cells at the cranial end always appear to be more mature than cells at the more caudal portions of the column. It was assumed that the more mature cells could be distinguished by their larger, paler, and more ovoid nuclei (Hughes, 1961). Similar gradations in motor neuron maturity have been observed in the lateral motor column of Xenopus tadpoles (Prestige, 1967). Because it was difficult to distinguish the actual longitudinal extent of the column, total cell counts were unreliable. Therefore, cell counts were made in the central portion of the lateral motor column to ensure that only developing motor neurons were included. As in earlier studies (Beaudoin, 1955; Race and Terry, 1965), the mean cell counts were based upon counts of 40 consecutive sections. Representative values of the mean number of cells per section can be seen in the control column of Table 3. The mean numbers of developing motor cells per section were initially high (38.6 per section), with a slight decrease
CELL COUNTS IN THE LATERAL MOTOR COLUMNS Stage v VI VII VIII IX X “Average error.
Control lateral motor column 38.6 36.4 34.3 31.5 24.2 16.5 number
* i * i * i
1.92 1.71 0.93 0.88 0.63 0.77
Experimental lateral motor column 39.2 36.8 36.5 33.6 29.3 20.2
of cells per section
* * + + * i
1.97 1.63 0.81 1.07 0.74 0.68
between stages V and VIII; subsequent to stage VIII there was a decrease, and by stage X there were approximately 50% fewer developing neurons (16.5 cells per section) in the lateral motor column. This same pattern was seen during normal development in Rana pipiens (Beaudoin, 1955) and in Rana temporaria (Race and Terry, 1965). The entire lateral motor column was examined for any significant differences in the maturity of developing motor neurons on the control and experimental sides of the spinal cord. There were no cells observed at any stage which were noticeably more advanced developmentally. Thus, the implanted muscles apparently did not cause any large-scale change in the morphological maturity of the experimental motor neurons. There were differences in the number of cells in the experimental and control lateral motor columns, but only in animals staged higher than VII. As in the controls, at stages II to IV, no lateral motor column was present. The experimental column first became distinct at stage V (Fig. 13). Table 3 shows there were no differences in the average number of developing motor neurons per section at any stage up to VII, where a slight increase in numbers begins to appear on the experimental side. Hughes and Tschumi (1958) demonstrated in Xenopus tadpoles that excision of the limb bud before stages 52-53 (correspond-
FIG. 13. Differentiation of motor neurons in the lateral motor column. A shows a stage III spinal cord with no differentiated lateral motor column. B shows the first distinct lateral motor column at stage V. C is an example of control motor neurons and D shows experimental motor neurons from the contralateral side (stage VII). The motor neurons are slightly less well developed and more densely packed on the experimental side. E is the control lateral motor column, and F is the experimental lateral motor column from a stage IX spinal cord. Notice the larger number of less well differentiated neurons on the experimental side. Calibration is 40 Frn in A and B, and 15 pm in C through F.
ing to approximately stages VII-VIII in Rana catesbeiana tadpoles) had no effect, on the lateral motor column. This same trend was observed in control spinal cords where the limb bud was ablated and no
muscle was implanted. Up to stage VII there was no difference in numbers; then at stage VIII a small increase was seen on the operated side (operated: 23.9 cells/section; unoperated, 21.4 cells/section). This differ-
ence in numbers became larger by stage X (operated: 18.9 cells/section; unoperated: 9.4 cells/section). These results were in agreement with similar data for unilateral hind limb ablations in Rana pipiens (Beaudoin, 1955) and in Pseudacris nigrita tadpoles (Pearson and Kollros, 1961). In addition to these differences in number, there were subtle differences in the level of maturity of the developing secondary neurons in the experimental and control lateral motor columns. There were no observable differences through stage V, but at more advanced stages there were variations between the control and experimental motor neurons (Fig. 13). At stages VI-X, the cells on the experimental side appeared to be more densely packed and slightly less well developed. Nuclei of the experimental motor neurons were less pale and more heavily granulated than the corresponding control motor nuclei. This increased packing may account for the increased number of cells present in the lateral motor column. This difference in maturity was more apparent at the later stages, where the experimental motor neurons were about one stage less developed than motor neurons on the control side. Similar results were seen after unilateral limb bud excision in the control spinal cords and by Beaudoin (1955) in Rana pipiens. These results are consistent with the events initiated by removal of the developing hind limb bud at early developmental stages. The fact that no markedly advanced motor neurons were observed in the lateral motor column, and that there were either no differences or that the differences were toward less well developed cells, favors the hypothesis that the implanted tail muscle failed to induce the developing motor neurons to a more advanced level of morphological maturity. Developing
A Gros-Bielchowsky silver strain was used to stain the axis cylinder and terminal ends of developing secondary nerve fibers.
Examples of developing axons in the implanted muscles are shown in Fig. 14. Localization of the terminal ends was extremely difficult in this preparation. The fine terminals in these developing nerves did not take up silver strongly and were often lightly stained. Moreover, what appeared to be the fiber’s end may not have been its actual termination, for the growing tip may have consisted of very fine extensions and pseudopodia which failed to become silver impregnated or were too fine to observe. A possible developing terminal is shown in Fig. 14D. From histological preparations it was not possible to determine whether these synaptic connections were functional. However, the physiological data were consistent with what one would expect from extremely fine, immature developing nerves such as these. DISCUSSION
These experiments demonstrate that developing nerves have the ability to form viable synaptic connections and to function physiologically at developmental stages well before they would ordinarily form synapses and mediate reflex activity. In general terms, the physiological properties of these synapses were similar to those found in adult frog nerve-muscle junctions: spontaneous mepp’s were present and transmission was quanta1 in nature. The major discrepancies were the very low mepp frequency and the small quanta1 content. These differences between the developing and adult neuromuscular junctions were most likely a result of the immature state of the developing terminals, as with advancing development they became less pronounced. Hughes and Prestige (1967) have described the normal development of Xenopus tadpoles. They observed that the first reflexive limb movements appeared at Nieuwkoop and Faber (1956) stage 53, when the hind limb bud had developed to palette stage. Similar reflexive limb movements were first observed during normal
F'kc. 14. Gros-Bielchowsky silver stain of developing nerve terminals. In A, numerous fine nerves are shol Iyn grov ving out from the severed end of the sciatic trunk after muscle implantation. B and C contain examples of fine developing nerves crossing over and growing along the muscle fiber. D shows a possible developing ner ve tern linal from a stage IV preparation. Notice the small varicosities and fine interconnectives. Calibration is 40 pm in A and 10 pm-in B through D.
Rana catesbeiana tadpole development at stages IX to X, which corresponds to approximately the same level of development as in Xenopus. However, functional nervemuscle connections were formed in normal limbs by stage VII. Thus, during normal development the motor nerves are capable of functioning at least two or three stages before they are normally required to do so. Similar results were observed in young rat embryos, where nerve-evoked muscle contractions (Angulo y Gonzales, 1932; Straus and Weddell, 1940) and functional nervemuscle junctions (Diamond and Miledi, 1962) were observed at 16-17 days in utero. In the present experiments the developing secondary nerves were observed to be
capable of functioning and releasing transmitter while still very immature, at stages as young as stage II. Very immature motor neurons have also been observed to function in tissue culture preparations of embryonic spinal cord and muscle from Xenopus larvae (Cohen, 1972), embryonic chicks (Fischbach, 1972), and embryonic rats (Robbins and Yonezawa, 1971a,b). It is possible that more rapid development resulted when the cells were removed from their normal environment and artificially cultured. Accelerated development may also account for the ability of the developing motor neurons to function in the present experiments. It was possible that when these nerves grew into and innervated the
MICHAEL S. LETINSKY
implanted tail muscle they were induced to higher levels of maturity. In the very immature cells this would require very significant maturational alterations; stage II motor neurons would have to develop to approximately stage VII. Such dramatic changes were not observed in the external morphology of the developing motor neurons. It was possible, however, that accelerated biochemical maturation did occur in these cells but was not observed. Nevertheless, the present results imply that developing nerves have the capacity to function physiologically at extremely young developmental stages, and that this pattern may be true in the normally developing animal. The physiological properties of these very young developing synapses were markedly different from those of normal adult nerve-muscle junctions. The observed mepp frequencies and quanta1 contents were consistently extremely low initially compared to adult values, then increased slightly with further development. Low mepp frequencies which increased during development were also observed in developing rat embryos by Diamond and Miledi (1962). Nerve-evoked synaptic transmitter release from embryonic nerves has also been observed in tissue culture and shown to be very low initially, increasing with development (Robbins and Yonezawa, 1971a,b). Kuno et al. (1971) demonstrated in adult frogs that the quanta1 content and mepp frequency were positively correlated with terminal size. The low quanta1 contents and mepp frequencies found during development and the apparent depletion of available transmitter stores during repetitive stimulation (Letinsky, 1974) suggest that a similar relationship could exist for transmitter release at developing endplates. The present results suggest that during early stages of development the synaptic potentials have relatively slow time courses similar to those seen in developing rat
(Diamond and Miledi, 1962), and in cultures of embryonic chick (Fischbach, 1972), embryonic rat (Robbins and Yonezawa, 1971a,b), and in embryonic toad, Xenopus laevis (Cohen, 1972). However, the Epp rise times seem to decrease with further development. The slow time course may be due to the protracted action of transmitter resulting from the lack of AChE activity at the developing endplate, as first suggested by Diamond and Miledi (1962) and later by Fischbach (1972). This is consistent with the fact that the AChE concentration in fetal rat does not reach adult levels until after birth (Kupfer and Koelle, 1951; Beckett and Bourne, 1957), and that no AChE activity was observed in chick tissue cultures (Kano and Shimada, 1971; Fischbach, 1972). Similarly, no histological or physiological evidence of AChE activity was observed in the present experiments up to stage X. Prolonged synaptic potentials at these early stages of development could also result if the distance between the transmitter release site, the presynaptic terminal and the postsynaptic membrane was large. A gradual narrowing of the synaptic cleft may account for the reduction in Epp rise-time. Electron microscopy of newly forming nerve-muscle junctions between developing secondary fibers and transplanted tail muscle is now in progress to determine whether this is the case.
I would like to thank Dr. Alan Grinnell for his helpful advice during the course of this research. Thanks also to Janet Seeley and Jane Chen for their expert technical assistance. This research was supported by USPHS Pre-Doctoral Training Grant 5 TO1 GM 00448-10.
REFERENCES ANGUI,O y GONZALEZ, A. W. (1932). The prenatal development of behavior in the albino rat. J. Camp. Neural. 55, 395-442. BEAUDOIN, A. R. (1955). The development of lateral motor column cells in the lumbro-sacral cord of
Rana pipiens. I. Normal development following unilateral limb ablation. Anat. Rec. 121, 81-95. BECKETT, E. B., and BOURNE, G. H. (1957). Cholinesterase in normal and abnormal human skeletal muscle. J. Neurol. Neurosurg. Psychiat. 20, 191-197. BIRKS, R., KATZ, B., and MILEDI, R. (1960). Physiological and structural changes at the amphibian myoneural junction, in the course of nerve degeneration. J. Ph.vsiol. (London) 150, 1455168. COHEN, M. W. (1972). The development of neuromuscular connections in the presence of D-tubocurarine. Brain Res. 41, 457-463. DEL CASTILLO, J., and KATZ, B. (1954). Quanta1 components of the end-plate potential. J. Physiol. (London) 124, 560-573. DIAMOND, J., and MILEDI, R. (1962). A study of foetal and new-born rat muscle fibers. J. Physiol. (Lendon) 162, 393-408. DUDEL, J., and KUFFLER, S. W. (1961). The quanta1 nature of transmission and spontaneous miniature potentials at the crayfish neuromuscular junction. J. Phvsiol. (London) 155, 514-529. ECKER, A. C. (1889). “The Anatomy of the Frog” (Haslam George, trans.), Asher, Amsterdam 1971. FATS, P., and KATZ, B. (1952). Spontaneous subthreshold activity at motor nerve endings. J. Ph,vsiol. (London) 117, 109-128. FENG, T. P., and TING, Y. (1938). Studies on the neuromuscular junction. XI. A note on the local concentration of cholinesterase at motor-nerve endings. Chin. J. Physiol. 13, 141-144. FILOGAMO, G., and GABELLA, G. (1966). Cholinesterase behavior in the denervated and re-innervated muscles. Acta Anat. 63, 199-214. FISCHBACH, G. D. (1972). Synapse formation between dissociated nerve and muscle cells in low density cell cultures. Deuelop. Riot. 28, 407-429. Goss, R. J. (1969). “Principles of Regeneration.” Academic Press, New York. HUGHES, A. (1961). Cell degeneration in the larval ventral horn of Xenopus laeuis (Daudin). J. Embryol. Exp. Morphol. 9,269-284. HUGHES, A., and EGAR, M. (1972). The innervation of the hind limb of Eleutherodactylus martinicensis: further comparison of cell and fiber numbers during development. J. Embryol. Exp. Morphol. 27, 389-412. HUGHES, A., and PRESTIGE, M. C. (1967). Development of behavior in the hind limb of Xenopus laeuis. J. 2001. 152, 347-359. HUGHES, A., and TSCHUMI, P. A. (1958). The factors controlling the development of the dorsal root ganglia and ventral horn in Xenopus laeuis (Daudin). J. Anat. 92, 498-527. KANO, M., and SHIMADA, Y. (1971). Innervation of skeletal muscle cells differentiated in vitro from chick embryo. Brain Res. 27, 402405. KARNOVSKY, M. J., and ROOTS, L. (1964). A “direct
coloring” thiocholine method for cholinesterases. J. Histochem. Cytochem. 12, 219-221. KATZ, B., and THESLEFF, S. (1957). On the factors which determine the amplitude of the “miniature endplate potential.” J. Physiol. (London) 137, 267-278. KELLY, A. M., and ZACKS, S. I. (1969). The fine structure of motor endplate morphogenesis. J. Cell Biol. 42, 154-169. KRNJEVIC, K., and MILEDI, R. (1959). Presynaptic failure of neuromuscular propagation in rats. J. Phvsiol. (London) 149, l-22. KUNO, M., TURKANIS, S. A., and WEAKLY, J. N. (1971). Correlation between nerve terminal size and transmitter release at the neuromuscular junction of the frog. J. Phvsiol. (London) 213, 545-556. KUPFER, C., and KOELLE, G. B. (1951). A histochemical study of cholinesterase during formation of the motor endplate of the albino rat. J. Exp. 2001. 116, 397-407. LETINSKY, M. S. (1972). The development of nervemuscle junctions in Rana catesbeiana tadpoles. Ph.D. thesis, University of California, Los Angeles. LETINSKY, M. S. (1974). Physiological properties of developing frog tadpole nerve-muscle junctions during repetitive stimulation. Deuelop. Riot. 40, 154-161. MARTIN, A. R. (1966). Quanta1 nature of synaptic transmission. Phvsiol. Reu. 46, 51-66. MILEDI, R. (1960). Properties of regenerating neuromuscular synapses in the frog. J. Ph.ysiol. (London) 154, 190-205. NIEUWKOOP, P. D., and FABER, J. (1956). Normal table of Xenopus laeuis (Daudin). North-Holland Publ., Amsterdam. PEARSON, C., and KOLLROS, J. J. (1961). Spinal cord response to amputation of the hind limb in larval Pseudacris. Amer. Zool. 1, 466. PRESTIGE, M. C. (1967). The control of cell number in the lumbar ventral horns during development of Xenopus laeuis tadpoles. J. Embryol. Exp. Morphol. 13, 359-387. RACE, J., JR., and TERRY, R. J. (1965). Further studies on the development of the lateral motor column in anuran larvae. I. Normal development in Rana temporaria. Anat. Rec. 152, 99-106. RAHAMIMOFF, R., and YAARI, Y. (1973). Delayed release of transmitter at the frog neuromuscular junction. J. Phvsiol. (London) 228, 241-257. REDFERN, P. A. (1970). Neuromuscular transmission in new-born rats. J. Phvsiol. (London) 209,701-709. ROBBINS, N., and YONEZAWA, T. (1971a). Developing neuro-muscular junctions: First signs of chemical transmission during formation in tissue culture. Science 172, 395. ROBBINS, N., and YONEZAWA,T. (1971b). Physiological studies during formation and development of rat neuromuscular junctions in tissue culture. J. Gen. Phvsiol. 58, 467-481.
Developing Nerue-Muscle Junctions
Y., and FISCHMAN, D. A. (1973). Morphological and physiological evidence for the development of functional neuromuscular junctions in oitro.
M. (1952). The influence of the nerve in Quart. regeneration of the amphibian extremity. Reu. Rio/. 27, 169-200. STRACS, W. L., and WEDDELL, G. (1940). Nature of the
first visible contractions of the forelimb musculature in rat foetuses. J. Neurophysiol. 3, 358-369. TAYLOR, A. C. (1943). Development of the innervation pattern in the limb bud of the frog. Amt. Rec. 87, 379-413. TAYLOR, A. C., and KOLLROS, J. J. (1946). Stages in the normal development of Rana pipiens larvae.
Amt. Rec. 94, 7-23.