In montages of electron micrographs of the sciatic nerve of Eleutherodactylus martinicensis, the numbers of fibers of all classes have been counted, from the 8·5 day embryo, through the early juvenile to the adult. These counts have been compared with the total numbers of cells in the lumbar ventral horn plus those in the lumbar spinal ganglia 8, 9 and 10.

In the embryo, both sets of counts rise to a peak on the 13th day, and fall early in the 14th day. In the embryo, the cell count is 600–1000 more than the fiber count, while in the 6-day juvenile onwards, the fiber count is the greater.

These differences are held to arise from the play of two independent factors, namely production of axons in the embryo by only a minority of cells in ganglia and ventral horns, and secondly, the extent to which axons branch between spinal roots and sciatic nerve at all stages.

In the embryo, numbers of cells and fibers maintain a parallel course up and down the 13-day peak, indicating that many cells which are lost by degeneration had already sent axons into the nerve.

Myelinated fibers first appear in the limb nerves at 8-5 days, when limb motility is first seen. The course of formation of the earliest myelin in the sciatic nerve resembles that of fibers in the central nervous system.

In spinal roots there are both myelinated and unmyelinated fibers, the proportion of the latter in ventral roots being the greater.

Cell counts within nervous centers at successive stages in embryonic and larval tetrapods have shown that numbers often decrease during development. Most of these studies have been concerned with the lateral motor column of the cord, particularly in Anura where the ventral horns which innervate limb and girdle muscles originate as discrete and readily countable groups of cells. These findings are reviewed by Kollros (1968) and by Hughes (1968). They invite inquiry into the relevance of such data to the general problems of the establishment of relationships between periphery and center in neural development. Cells are lost by cell death; in Xenopus cell degeneration is particularly conspicuous in ventral horns (Hughes, 1961) and the related dorsal root ganglia (Prestige, 1965) at a time when the limbs begin to move, and there is some evidence for a selective mechanism by which the neural circuits through limbs and cord approach their precise final pattern by the elimination of elements acquired at random during earlier phases of development when contact between nerve and muscle fiber is less specific (Hughes & Prestige, 1967). The question rests however on whether a corresponding loss and replacement of axons can be demonstrated within the limb nerves. One way of examining this question is to compare at successive stages cell numbers with counts of fibers in the limb nerves, an inquiry which has already been attempted (Hughes, 1965a) for the embryonic anuran Eleutherodactylus. This however was based on counts of fibers under the light microscope in silvered preparations. It was not realized at that time how incomplete are such data in that large numbers of small fibers below the limits of resolution escape notice. The present study is a further attempt at numerical comparison between cells and fibers, by the use of the electron microscope in the development of the same animal. We also present cytological evidence for the loss of fibers in limb nerves during development, together with observations on the formation of myelin.

In those species of Eleutherodactylus on which we have worked, the juvenile leaves the egg capsule some 14 days after fertilization as a fully formed miniature frog. At a stage 24 h from hatching, counts of various parameters in the development of nerves and muscles in E. martinicensis rise steeply to a maximum, followed by an even sharper fall (Hughes, 1965a). In the ventral horns, this peak interrupts the general decline in cell numbers comparable to that of other Anura, but its significance is at present far from clear. However, these rapid events provide an opportunity for detailed comparison between changes in cells and fibers at this period of development. With the related E. ricordii (Hughes, 1959; Hughes & Reier, 1972) no peak in the number of ventral horn cells in the late embryo has been found, nor is it present when embryos of E. martinicensis are cultured in relatively large volumes of saline.

Clutches of eggs of Eleutherodactylus martinicensis were laid in a well-watered pit containing vegetable matter at the Department of Zoology, University of the West Indies, Jamaica.

Embryos were freed of their egg envelopes, and plunged into 4 % glutaraldehyde at 4 °C, and staged according to the data given in Hughes (1966a). They were eviscerated, and the skin removed over the whole surface. The body wall and somitic musculature was dissected away, leaving the limbs attached to the body axis in undisturbed relationships. After 1–2 h fixation in glutaraldehyde, these embryos were transferred to cold isotonic sucrose, in which condition the material was subsequently brought to Cleveland in a vacuum flask packed with ice. Embryos were then post-fixed in buffered osmic acid (Sabatini, Bensch & Barnett, 1963) for one hour, and then dehydrated and embedded in Araldite. So small are these embryos that adequate fixation and infiltration of plastic in both spinal cord and peripheral nerves was obtained when neural axis and limbs were fixed in one, with undisturbed anatomical relationships.

From the blocks of embedded material, thin sections were prepared and mounted on one-hole formvar or carbon-coated grids. Micrographs of transverse sections of sciatic, tibial and peroneal nerves were combined into montages of the entire cross-section at magnifications of 15000–24000, such as Egar & Singer (1971) have recently described for spinal nerves in Triturus. Such montages are of the order of a square meter in area. Eleutherodactylus is probably the smallest of extant tetrapods, and in others the problems of handling, study and storage of montages of whole major nerve trunks would be even greater.

One such montage was rephotographed and reduced to form Fig. 3 of the present work.

Fig. 1.

Tibial nerve of embryo at the stages indicated. Total fibers ( + ); scale to right in hundreds. Singly wrapped (◊) and myelinated fibers (□); scale to left, in hundreds.

Fig. 1.

Tibial nerve of embryo at the stages indicated. Total fibers ( + ); scale to right in hundreds. Singly wrapped (◊) and myelinated fibers (□); scale to left, in hundreds.

Figure 2.

Part of sciatic at 8·5 days. Single axons with dark outlines are the first myelinated fibers (m); those with lighter outlines are at the promyelin stage (p). Smaller axons are wrapped in bundles (b) by processes of Schwann cells. Note abundant Schwann cells with nuclei. 1 μm equivalent as shown by bar.

Figure 2.

Part of sciatic at 8·5 days. Single axons with dark outlines are the first myelinated fibers (m); those with lighter outlines are at the promyelin stage (p). Smaller axons are wrapped in bundles (b) by processes of Schwann cells. Note abundant Schwann cells with nuclei. 1 μm equivalent as shown by bar.

Fig. 3.

Montage of sciatic at hatching stage, rephotographed and reduced. The nerve contains 1471 unmyelinated and 399 myelinated axons. 1 μm equivalent as shown by bar.

Fig. 3.

Montage of sciatic at hatching stage, rephotographed and reduced. The nerve contains 1471 unmyelinated and 399 myelinated axons. 1 μm equivalent as shown by bar.

For the adult sciatic nerve, a semi-thin plastic section was photographed under the phase microscope, and a montage prepared ; partial montages of the adult sciatic under the electron microscope were then compared with the lightmicroscope picture.

Embryonic stages

7–7·5 days

This stage is about halfway through embryonic life, and digits are present in both fore- and hindlimbs. Movement of the hindlimb is just beginning (Hughes, 1965b). In the tibial nerve there are already present nearly a thousand fibers (Fig. 1). Light-microscopical preparations have indicated that axons first enter the limb at 4·5 days (Hughes, 1962); the following three days are a specially active period of axonal growth and migration into the limb.

The limb nerves are surrounded by an investment made up of flattened expansions of perineurial cells, each of which extends over only part of the circumference of the nerve. Together they make up a system of parallel separate membranes, of which there may be as many as six at some points. Within the nerve, the most conspicuous bodies are the nuclei of Schwann cells, of which about twelve may lie in a plane of section through the tibial nerve.

Schwann cytoplasm is densely packed with ribosomes, which in broader areas are associated with irregular cisternae of endoplasmic reticulum. Round the inner surface of the nuclear membrane are dense irregular masses of chromatin. Centrally, the nucleoli appear similarly opaque while the background of the nucleoplasm is granular. During the rest of the embryonic period the Schwann nuclei are similar in appearance, a condition which indicates a continued high level of synthetic activity.

Round the nucleus of each Schwann cell is a thin layer of cytoplasm from which extended processes stretch to enclose bundles of nerve fibers, ten or more in number, the smallest of which are about 0·1 μm in diameter. A basement membrane or glycocalyx is recognizable over the surface of cell processes and nerve fibers which border the spaces within the nerve, within which fibrils of endoneurial collagen are sparsely distributed.

One fascicle of the nerve mainly consists of a number of larger fibers. Their axonal circumference is as much as 0·5 μm in diameter. Each of these is separately wrapped by a process of a Schwann cell in the form of an uneven sheet which extends in a loose spiral several times round the nerve fiber. More than one may wrap a single fiber. Several may provide the combined wrappings for two adjacent fibers. The wrapping may flatten in places to about 100 Å in thickness, though the spaces between them are generally distinct. They correspond to the ‘promyelin’ fibers described by Friede & Samorajski (1968) in the sciatic nerve of the infant rat.

8·5 days

At this stage the hind limb can retract after a light touch, and independently of trunk flexure. Random spontaneous movements of the limbs have already begun (Hughes, 1966b).

During the 8th day (Fig. 2), the appearance of the sciatic and tibial nerves under the electron microscope changes more rapidly than in any succeeding stage. Their cross-sectional area increases several times, though with little change in the number of fibers. There is now much free space within the nerve with abundant endoneurial collagen. More than three times the number of Schwann cells are seen. Fibroblasts with widely extending processes are evident, and with nucleus and cytoplasm as dense as in Schwann cells. Over the fibroblast surface there is no apparent basement membrane.

The smallest fibers are now from 0·15 to 0·2 μm in diameter; they are grouped in bundles, each with an envelope of Schwann processes. Most contain between 5 and 25 fibers (Fig. 4tz). Larger fibers, 0·6 μm or more in diameter, with wrappings similar to those of large fibers at the previous stage, occur singly, or in groups of two. Again, one large axon may be associated with a bundle of small fibers.

Fig. 4.

Numbers of unmyelinated fibers (▄, scale top left), fiber bundles (♦, scale to right) in sciatic nerve from 8·5-day embryo to 9-day juvenile, together with adult. Below (ae), distributions of bundle size at the stages indicated by arrows. Numbers of instances on scale to lower left, numbers of fibers in bundles, by successive groups of five fibers, on scales at lower margin of figure. Notice in (c) loss of bundles with 5–10 fibers during a fall in total numbers during 14th day, and recovery in 9-day juvenile (d).

Fig. 4.

Numbers of unmyelinated fibers (▄, scale top left), fiber bundles (♦, scale to right) in sciatic nerve from 8·5-day embryo to 9-day juvenile, together with adult. Below (ae), distributions of bundle size at the stages indicated by arrows. Numbers of instances on scale to lower left, numbers of fibers in bundles, by successive groups of five fibers, on scales at lower margin of figure. Notice in (c) loss of bundles with 5–10 fibers during a fall in total numbers during 14th day, and recovery in 9-day juvenile (d).

In the sciatic nerve, the range of size among single promyelin fibers is similar for those within bundles, though larger fibers tend to be single. The size ranges of promyelin fibers and of myelinated axons do not overlap at this stage.

From the spirals of Schwann cell processes round the promyelin fibers, such as are described for the preceding stage, arise the first segments of myelin. In the first place they envelop the axon loosely, with empty spaces between. They then begin to compact together, first obliterating the spaces between. Next, over their surface, segments of myelin appear, and finally the Schwann protoplasm within is eliminated. Thus in Eleutherodactylus, wrappings undergo compaction in two stages. On these grounds, we do not regard the first lamellae as arising from inwardly growing cell membrane within the Schwann cell, as in the generally accepted account of peripheral myelination, first described by Geren (1954). Loose spirals at the promyelin stage have been described by Friede & Samorajski (1968) in the 6-day rat sciatic nerve. The early stages of myelination in Eleutherodactylus resemble the course of central myelination as first described by Peters (1960) in Xenopus, where a process of an oligodendrocyte extends as a ‘tongue’ towards and round a nerve fiber. Schwann cell processes at first behave in a similar way towards peripheral axons in Eleutherodactylus, but their form is often too complex to be called tongues, however forked. Accordingly we use the term ‘lappet’ to describe them. In the following stage of development their variety of form will call for further notice.

Where only two or three lappets are present at any point, compaction is restricted to half or even less of the circumference, but extends more widely in axons with four or more lamellae. None as yet are uniformly compacted. In some, a dark line about 1000 Å wide over part of the circumference can be resolved at higher magnification into some layers of myelin, often with Schwann cytoplasm still present within the lappets. Outside the myelin is an incomplete layer of Schwann cytoplasm, stretching to a variable extent round the circumference, usually about half way. Over the remainder, the cytoplasm thins out leaving the myelin apparently covered only by basement membrane, a feature characteristic of central fibers wrapped by oligodendroglial tongues.

11·5–12 days

Figure 5. In living embryos light touch now results in flexion followed by vigorous extension (the ‘kick’, Hughes, 1965b) which may be repeated, but not yet continuously maintained. Yet during the last three days wrapped fibers have been lost from both sciatic and tibial nerves, mainly through decrease in the category of promyelin fibers within bundles, without however a corresponding increase in myelinated fibers (Figs. 1 and 6). Among the single promyelin fibers there are signs of recent isolation. Thus there has been both loss and gain of wrapped axons, with the former predominating. In nearly all fibers in early stages of wrapping there are projections from the outer lappet, usually in a radial direction. These often stretch towards other fibers, or to groups, and may be continued by a tract of material apparently similar to the basement membrane which covers Schwann cell surfaces. This appearance may indicate a recent withdrawal of a connecting lappet. In such instances, it is possible that both fibers still derive their wrappings from a single Schwann cell.

In other instances, the Schwann lappet connecting two fibers may be interrupted by a local focus of degeneration, with empty vesicles inside a swollen membrane, such as we have recently described within the developing spinal cord (Hughes, Egar & Turner, 1969) where degeneration seems to proceed in a similar manner for both uncovered axons and for glial processes. A bag of vesicles may also be seen within the complex lappets round an axon, an appearance which may result from the degeneration of one of a pair of fibers.

Figure 5.

Part of sciatic of 11·5-day embryo, showing promyelin fibers with loose wrappings (p), myelinated fiber with four incomplete lamellae (m), and bundle of small axons (b). 1 μm equivalent as shown by bar.

Figure 5.

Part of sciatic of 11·5-day embryo, showing promyelin fibers with loose wrappings (p), myelinated fiber with four incomplete lamellae (m), and bundle of small axons (b). 1 μm equivalent as shown by bar.

Fig. 6.

(a) Counts of lumbar ventral horn cells together with cells in ganglia 8, 9 and 10 (▄) to compare with total numbers of fibers in sciatic nerve ⊡, at the stages of development shown, (b) Counts of singly wrapped fibers (upper) and of myelinated fibers (lower) in the embryo. Scale of abscissa in hundreds of cells or fibers as appropriate.

Fig. 6.

(a) Counts of lumbar ventral horn cells together with cells in ganglia 8, 9 and 10 (▄) to compare with total numbers of fibers in sciatic nerve ⊡, at the stages of development shown, (b) Counts of singly wrapped fibers (upper) and of myelinated fibers (lower) in the embryo. Scale of abscissa in hundreds of cells or fibers as appropriate.

The appearance of many of the wrapped fibers differs from that seen at 8·5 days, to an extent which suggests that many are not the same individuals which were seen at the earlier stage. They are now even more irregular in transverse section. At the promyelin stage, the gaps between the Schwann lappets are more conspicuous. Their arrangement is now particularly clear, and has been studied in some detail. Most wrapped fibers are invested by a single lappet (Fig. 7, ag), but in a few the axon is surrounded by more than one incomplete and separate wrappings (Fig. 7 A). Sometimes a lappet may split and branch during its course round a fiber (Fig. 7 i), an appearance which has already been noticed at earlier stages.

Fig. 7.

Drawings from electron micrographs, from montage of transverse section of sciatic nerve at 11·5 days, to show relationships between lappets of Schwann cells (hatched), round axons (dotted), (ag) Progressive wrapping of single lappets; (h) double lappet; (i) divided single lappet. 1 μm equivalent as shown by bar.

Fig. 7.

Drawings from electron micrographs, from montage of transverse section of sciatic nerve at 11·5 days, to show relationships between lappets of Schwann cells (hatched), round axons (dotted), (ag) Progressive wrapping of single lappets; (h) double lappet; (i) divided single lappet. 1 μm equivalent as shown by bar.

Before the wrapped layers of investment become compacted together, it can be seen how the attentuation of a lappet and the approximation of its walls leads to the formation of a dense line of myelin. Often this process occurs unevenly along the course of a lappet, and a stretch with recognizable cytoplasm may give place on either side to a single line. Partial compaction is mainly restricted to axons with small numbers of wrappings.

A sector with a relatively thick layer of Schwann cytoplasm and with uncompacted lamellae in regular layers may show a clear boundary between the overlapping edges of the outer lappet, the oblique line of apposition between them representing an uncompacted outer mesaxon. The recognition of an internal mesaxon within an uncompacted system of lappets is more obscure.

By the 12th day, myelinated axons have not increased further in circumference, or acquired further wrappings, but smaller fibers from 2·5 μm in circumference have begun to myelinate. The lower limit of size among myelinated fibers overlaps with that of promyelin axons. The maximum number of layers round any axon is eight, which represents little progress in the general course of myelination during the last three days.

13–13·5 days

By the 13th day, the behaviour of the living embryo has developed further. It is now able to maintain continuous bilateral action of the hindlimbs, and hence to swim.

Rapid advance in the development of peripheral nerves is resumed during the thirteenth day. The total number of fibers in both sciatic and tibial nerves has increased by about half the previous levels. The number of bundles of unmyelinated fibers has nearly doubled since the previous stage, with a marked peak in number of bundles with 5 to 10 fibers (Fig. 4). In the sciatic nerve trunk the number of wrapped fibers has also doubled, mainly by increase in myelinated fibers. In the tibial and peroneal nerves, however, this peak in the count of wrapped fibers is not seen (Fig. 1) ; this difference may be related to rapid development of fibers within the Rami profundi which are given off early from the sciatic, and supply the large muscles of the thigh (Hughes, 1965 a) to which the extensor thrust of the leg is mainly due. Promyelin fibers, both single and within bundles, are still present, and their sizes continue to overlap those of myelinated fibers. Fibers round which myelination has apparently only just begun, with two or three compacted lamellae extending round part of the circumference, are still found, but are increasingly restricted to smaller axons.

The free spaces between Schwann cell lappets of the previous stages are no longer seen, and there is no reason for regarding the formation of the myelin at this stage onwards as in any way different from that in peripheral nerves of other animals. We are unable to decide whether fibers which now myelinate in the manner usual in peripheral nerves represent new generations of axons. In them a distinction is still seen between a less compacted inner zone, and the fully compacted lamellae outside. Such differences have also been seen by Webster (1971) in the sciatic nerve of the infant rat. In Eleutherodactylus at 13 days, many sheaths show uncompacted areas even where as many as 12 lamellae are present. In a few of the largest fibers there is continuous myelin over the whole circumference. The larger fibers are less angular in appearance than they were at the previous stage, but they are still far from circular in cross-section. Some develop short branches, which can be shown by comparison of adjacent sections to be blindly ending diverticula, within the confines of a single Schwann cell.

At 13 days is first seen a relationship between axonal circumference and number of lamellae (Fig. 8c) such as Friede and Samorajski (1968) have described in the post-fetal rat. In Eleutherodactylus, fibers of the central processes which enter the cord from the dorsal poles of the spinal ganglia are less heavily myelinated than are sciatic axons (Fig. 8f).

Fig. 8.

Relationships between axonal circumference (horizontal scales in μm) to numbers of lamellae in myelin sheaths at embryonic stages indicated, (ae) in sciatic nerve (partially compacted lamellae, • ; largely compacted lamellae, ▄). (f) shows both myelinated fibers in sciatic nerve (▄) and in lumbar central process (+).

Fig. 8.

Relationships between axonal circumference (horizontal scales in μm) to numbers of lamellae in myelin sheaths at embryonic stages indicated, (ae) in sciatic nerve (partially compacted lamellae, • ; largely compacted lamellae, ▄). (f) shows both myelinated fibers in sciatic nerve (▄) and in lumbar central process (+).

At this stage samples of the lumbar spinal roots were studied in horizontal sections at three levels, most dorsally through a central process (Fig. 9), then through a sensory root at the ventral pole of a ganglion, and finally through part of a ventral root. It would not be possible within a single grid space to include the whole of the lumbar spinal roots, either dorsal or ventral, nor were we able to identify precisely the transverse level of these sections, which could have belonged either to S8 or S9. In the central process, the ratio of unmyelinated to myelinated fibers was 1·14:1, and in the dorsal root 1·50:1. These figures may suggest some branching of unmyelinated fibers within the ganglion. Both of the values are however much lower than that of the sciatic trunk at this stage, namely 4·7:1.

Figure 9.

Part of horizontal section through 13-day embryo to show central process of lumbar spinal ganglion (left) with adjacent part of dorsal funiculus of cord. The central process contains abundant Schwann cells with large myelinated fibers (m), and bundles of smaller unwrapped axons (u). Some of each group have just entered the cord, as indicated. Central process fibers are less heavily myelinated than are sciatic axons. 1 μm equivalent as shown by bar.

Figure 9.

Part of horizontal section through 13-day embryo to show central process of lumbar spinal ganglion (left) with adjacent part of dorsal funiculus of cord. The central process contains abundant Schwann cells with large myelinated fibers (m), and bundles of smaller unwrapped axons (u). Some of each group have just entered the cord, as indicated. Central process fibers are less heavily myelinated than are sciatic axons. 1 μm equivalent as shown by bar.

In the ventral root, the proportion of unmyelinated fibers is very much higher, and was at least 12:1. Many occur loose, without enclosure in Schwann processes. Some unknown proportion of these small fibers must be pre-ganglionic sympathetic axons, which do not enter the sciatic nerve. However, it is clear that whatever allowance is made for this factor, the proportion of unmyelinated fibers is higher in ventral than in dorsal roots at this stage.

14 days. (Fig. 3)

During the last day of embryonic life, resorption of the tail is largely completed. At this time in embryos freed from the egg capsule, spontaneous behavior is largely concerned with diagonal ambulatory movements (Hughes, 1966b). Within the limb nerves, the total number of fibers in both sciatic and tibial sharply decreases. The loss is most marked in the former where there is a corresponding fall in the number of myelinated fibers (Figs. 1, 6). Thus a sharp peak is demarcated by the rise during the 13th, and the fall during the 14th day.

In the sciatic nerve, the most drastic loss is in the number of unmyelinated fibers which, together with their bundles, fall to a level below that of the 12th day. These losses are concentrated among the smaller bundles (Fig. 4c). At this time unmyelinated fibers are being lost from the sciatic nerve at a rate of about fifty an hour. Our electron micrographs give little indication of this depletion. Within Schwann cells surrounding myelinated fibers, we have seen irregular masses of unorganized material at this stage, and at no other (Fig. 10A). These may represent phagocytosed axonal remnants. Round the smaller bundles of unmyelinated fibers, among which losses are especially severe, Schwann cells with nuclei are only occasionally seen in a single transverse section, and the chances of seeing any engaged in resorption may well be low.

Figure 10.

(A) From sciatic of 13·5-day embryo. Schwann cell enclosing myelinated fiber (left) with globular inclusions (right), possibly phagocytosed remnants. 1 μm equivalent as shown by bar.

(B) From sciatic of 4-day juvenile, to show bundle of unmyelinated fibers now being subdivided by Schwann processes (S). Wrapped fibers, below, and to either side showing incomplete compaction of myelin lamellae (i). 1 μm equivalent as shown by bar.

Figure 10.

(A) From sciatic of 13·5-day embryo. Schwann cell enclosing myelinated fiber (left) with globular inclusions (right), possibly phagocytosed remnants. 1 μm equivalent as shown by bar.

(B) From sciatic of 4-day juvenile, to show bundle of unmyelinated fibers now being subdivided by Schwann processes (S). Wrapped fibers, below, and to either side showing incomplete compaction of myelin lamellae (i). 1 μm equivalent as shown by bar.

There is little further advance in myelination during the last day of embryonic life. No more than 15 lamellae of myelin envelop the largest axons (Fig. 8e), and evident Schwann cytoplasm still extends round most of the circumference of the sheath.

Post-embryonic stages

During post-embryonic life several changes can be traced within the sciatic nerve. By the fourth day after hatching, Schwann cytoplasm round myelinated fibers is already less abundant than during embryonic life, and the rough endoplasmic reticulum of the embryonic Schwann cell studded with polysomes becomes much less conspicuous. Only thin arcs of Schwann cytoplasm are visible in transverse sections of the adult nerve, while Schwann nuclei are then very infrequent.

The promyelin stage of wrapping becomes increasingly rare, and fibers with less than four compacted lamellae are rarely found in the juvenile, and none with less than six in the adult. Yet myelinated fibers steadily increase in number after hatching; presumably therefore the earliest stages of myelination are then passed through more swiftly. In the smaller myelinated fibers of the 4-day juvenile, a sector with incompletely compacted lamellae is frequently seen, but only rarely at later stages. The maximum number of lamellae continues to increase in juvenile life, and reaches eighty in the adult nerve.

The number of unmyelinated fibers changes little in post-embryonic life, but the number of bundles steadily increases by subdivision consequent on further penetration of Schwann lappets between the fibers (Fig. 10 B). This change hastens about six days after hatching, and restores the peak in numbers of smaller bundles with 5 to 10 fibers, a condition which then persists into adult life (Fig. 4d, e). In the adult nerve each unmyelinated fiber is separated from its neighbours by a single thin lamina of Schwann cytoplasm. Unmyelinated fibers slightly outnumber myelinated axons, a condition little different from the 1:1 ratio found by Egar & Singer (1971) in spinal nerves of Triturus.

The data presented above are summarized in Table 1, together with information on cell counts, yet to be described. It is clear that the development of these limb nerves does not follow a continuous course, but that successive distinct stages are recognizable. There are two active periods of increase in fiber number and in structural advance, namely from 4·5 to 8·5, and from 11·5 to 13·5 days. They are separated by an interval when total fiber numbers remain stationary, wrapped fibers decrease, and there is no general progress in myelination. The second period of advance is again followed by a decline in fiber number.

Table 1.

graphic
graphic

While these periods of fiber loss in themselves indicate that some proportion of the original fibers of the limb nerves do not survive into post-embryonic life, the evidence for neuronal turnover is strengthened when we consider the largely parallel changes in numbers of cell bodies now to be described.

Cell counts

The counts here presented (Figs. 6, 11) relate to normal development within the egg capsule on material which has been collected over a number of years. In all counts in both ventral horns and ganglia only cells are scored in each section where the main nucleolus is seen within the outline of the nucleus, a procedure which obviates the need for any correction due to partition of cells between adjacent sections (Jones, 1937 ; Ebbeson & Tang, 1965 ; Hughes, 1969). In general such counts have been found to be reproducible to within 10 or 15 %. The major changes in development which are here described greatly exceed these limits.

Fig. 11.

Counts at the stages shown of cell totals (a) in dorsal root ganglia 8,9 and 10; and (b) in lumbar ventral horn.

Fig. 11.

Counts at the stages shown of cell totals (a) in dorsal root ganglia 8,9 and 10; and (b) in lumbar ventral horn.

Ventral horn cells

The ventral horn contains both interneurons and motor cells, though unlike the mammal, in the frog there is no special category of fusimotor or gamma cells or fibers (Katz, 1961). The same criteria were used in counting ventral horn cells as in previous studies on Eleutherodactylus (Hughes, 1959, 1962, 1965a, 1966a). Our aim was to count only those cells which are distinct in size and development from those of the more medial mantle layer. They have a visible layer of perikaryal cytoplasm, with some traces of Nissl substance, They are clearly bipolar. These criteria are met by nearly all cells in the mid-region of the ventral horn, while towards the caudal pole the proportion of ‘differentiated’ cells decreases. The adjacent cells which are not counted will either be immature ventral horn cells or interneurons. The question of what proportions of these cell categories are present will be further discussed below in comparing numbers of cells and fibers. The data on numbers of ventral horn cells show that in E. martinicensis the peak at 13 days temporarily restore the total to that seen at 8 to 9 days, only to fall steeply during the 14th day (Fig. 11 b).

Dorsal root ganglia

The anatomy of the lumbar plexus in Eleutherodactylus (Hughes, 1965a, text-figure 4) shows that the 7th spinal root gives rise mainly to the crural nerve, with a small anastomotic branch to nerve 8. Sensory fibers of the sciatic nerve thus come very largely from S8, S9, and S10, to which counts here presented are confined.

In the counts of cell numbers in spinal ganglia published earlier in these studies (Hughes, 1959) only the larger and more mature cells were scored, a procedure which was then thought to correspond to the criteria used for ventral horn cells. Renewed study of the spinal ganglia shows that in the course of differentiation there are no recognizably distinct stages between the early neuroblast and the mature neuron with its round nucleolus, large nucleus and wide margin of Nissl-laden cytoplasm, though at the stages studied in the present work the distinction is clear between nerve cells and non-neural elements, the Schwann and satellite cells. Accordingly, all neuronal cells have here been counted in the ganglia, and the figures given here for late embryos (Fig. 11a) are higher than those previously reported. An argument for this difference in treatment between ventral horns and ganglia is that within the latter are no cells which correspond to interneurons. Figure 11a shows that the total cell numbers in the lumbar ganglia rise to a peak on the 13th day, from which they fall sharply during the 14th day to values similar to those at earlier stages of development. Counts on the brachial ganglion (S2) show that the peak does not relate only to the hindlimb. In both ventral horns and lumbar ganglia the decline slowly continues during the first few days of juvenile life, and is followed by a subsequent differentiation of fresh neurons.

Cell death

Counting dying cells is more difficult than for the living. The earliest stage of pycnosis seen in Xenopus where the nuclear membrane is still intact, and with chromatinic aggregations within (Hughes, 1961), occurs also in Eleutherodactylus. Under the light microscope the later stages appear as scattered blobs of deeply stained material. Electron micrographs show that these consist of phagocytosed cell debris. Where several deaths occur in a small area, it is not clear whether these ‘degeneration sites’ represent the remains of one or of several cells. This difficulty mainly affects counts at peak periods of degeneration.

Cell death is common within the embryonic ventral horns from the 9th day until the hatching stage, with larger numbers early in this period (Fig. 12). As in Xenopus at the corresponding phases of the development of limb movement (Hughes & Prestige, 1967) it is likely that there is an active turnover of ventral horn neurons with newly differentiating cells continuously replacing those lost by degeneration. The present results show that this period is also one when numbers of wrapped fibers in the sciatic nerve are declining. What significance can be attached to the second phase of cell loss in the ventral horns during the 14th day (Fig. 11 b) is more obscure. It is, however, related both to fiber loss and to reduction in volume of some thigh muscles (Hughes, 1965a, 1966a). In the spinal ganglia of Eleutherodactylus, cell death is much rarer than in the ventral horns, and only during the 14th day when cell numbers are rapidly falling are degenerations common (Fig. 12). It seems that in the ganglia of Eleutherodactylus, elimination of the first generation of cells is largely confined to this brief period.

Fig. 12.

Counts of degeneration sites in ventral horn (•) and in lumbar ganglia (○) from 9 days onwards.

Fig. 12.

Counts of degeneration sites in ventral horn (•) and in lumbar ganglia (○) from 9 days onwards.

Our main concern in the interpretation of the present data is to examine the relationships between the numbers of cells in ventral horns and lumbar ganglia and the number of fibers in the sciatic nerve. This comparison is beset by various complications, some of which vary in importance during development. One factor which affects our data to a small extent at all stages is the neglect of the small dorsal rami of the lumbar nerves.

A major aim is to determine at various stages what proportions of cells in cord and ganglia are not represented by axons in the sciatic nerve. In the ventral horn these include both immature motor cells and also interneurons. In the previous paper (Hughes, 1969) the range of numbers of large and mature cells within the lumbar ventral horn of the later embryo overlapped those of the lumbar ventral roots counted in silvered sections. These, we can now surmise, included larger axons as well as bundles of smaller axons which when coated with silver appeared as single fibers. This correspondence was then held to support a rough identification of the cells counted with motor neurons. In the present data, the number of ventral horn cells counted in late embryos are somewhat greater than in the previous paper. At these stages, the numbers of the larger and singly wrapped fibers in the sciatic nerve are about one hundred more than that of silver-stained axons previously counted in ventral roots. Inspection of Fig. 9 suggests that at least this number of wrapped fibers in the sciatic nerve must belong to dorsal roots. In the late embryo, and even more so at earlier stages, these considerations indicate that a considerable proportion of cells in the ventral horn are not represented by large axons in the sciatic nerve. Some will have small unmyelinated axons, other motor cells will not yet have formed axons, and some cells counted will be interneurons. When, however, we add to the cells of the ventral horn those of the lumbar dorsal root ganglia, and compare these totals with fibers of all calibers in the sciatic nerve, a clearer conclusion emerges. There is an excess of cells over fibers of about 600 at 13–14 days (Fig. 6). Factors shortly to be discussed suggest that the numbers of cells without sciatic axons at these stages may be even higher.

In the juvenile after six days, as in the adult, the position is reversed, and sciatic fibers then exceed lumbar cells by about 200. In the adult there are presumably no immature cells in ganglia or ventral horns. In counting the latter, we again neglected the smaller cells, which it can be argued will largely consist of interneurons. The apparent deficit of cells at these stages may well be due to the presence of post-ganglionic sympathetic axons in the sciatic nerve, a factor which is less likely to affect the embryo, where cells of the sympathetic chain are in an early stage of differentiation.

A further factor which affects comparison of cells and fibers is the extent of axonal branching between spinal cord and sciatic nerve, the effect of which on the present data is to add to the fiber count, and thus to underestimate the proportion of cells without sciatic axons. Evidence for axonal branching in the embryo comes from a higher proportion of unmyelinated fibers in dorsal roots than in central processes (p. 401 above) and histological demonstration of branching of ventral root axons within the cord (Hughes, 1965 a).

Furthermore, it was argued in previous papers (Hughes, 1964a, b) that the withdrawal of motor innervation from supernumerary grafts in the late embryo was evidence for reduction in the extent of branching at that time. We suggest then that in the late embryo axonal branching is decreasing, and axons are growing from the remaining undifferentiated cells. Also at these stages cells are degenerating, more in the dorsal root ganglia than in the ventral horns. An important question which relates to the interpretation of cell death in the nervous system at this time is whether the cells which degenerate had already sent axons to the periphery before their death. Here the present data, despite the uncertainties of interpretation which we have just discussed, seem to provide an answer. In Fig. 6 we see that numbers of cells and fibers follow a parallel course up and down the 13-day peak. This means that during the brief period of the descending phase, as many fibers are lost as are cells, a fact which suggests that many cells which are then lost were represented by sciatic axons. However, during this 12-h period axonal branching is decreasing, which may be responsible for some of this decline in fiber numbers. If so, some cells lost at this time were without sciatic axons, though it is highly unlikely that the steep descents in both cell and fiber numbers are wholly unrelated events.

However incomplete the glimpse of the changes within the peripheral nervous system which the present data afford, the conclusion can hardly be avoided that profound changes in axonal distribution do occur even after functional relationships have been established.

In the dorsal root ganglia, cell death (Fig. 12) and cell loss (Fig. 11) are largely confined to the day before hatching, when fiber loss (Fig. 6) is mainly among unmyelinated axons (Fig. 4). At this time there must be large-scale changes in the pattern of somatic sensory innervation, due both to loss of neurons and to the production of fresh axons during the first few days of juvenile life. If in larval Anura there is a corresponding replacement of sensory cells near the metamorphic climax, here may lie the explanation of the observations of Jacobson & Baker (1968) and Jacobson (1969) on the change in reaction to touch of juveniles bearing reversed skin grafts. They found that reactions at first were normal, but were later replaced by a maladaptive response which resulted in a misdirected attempt of a limb to wipe the stimulated area (Miner, 1956). Modulation of sensory cells by the reversed graft may affect only cells which replace an earlier generation.

Our sincere thanks are due to Dr T. Turner for providing the living material used in this study, to Mrs Elinor Koo for her diligent technical assistance, and to Dr Paul Reier for much help and advice.

Ebbeson
,
S. O.
&
Tang
,
D.
(
1965
).
A method for estimating the number of cells in histological sections
.
JI, R. microsc. Soc
.
84
,
449
464
.
Egar
,
M. W.
&
Singer
,
M.
(
1971
).
A quantitative electron microscope analysis of peripheral nerve in the urodele amphibian in relation to limb regenerative capacity
.
J. Morph
.
133
,
387
398
.
Friede
,
R. L.
&
Samorajski
,
T.
(
1968
).
Myelin formation in the sciatic nerve of the rat
.
J. Neuropath, exp. Neurol
.
27
,
546
570
.
Geren
,
B.
(
1954
).
The formation from the Schwann cell surface of myelin in the peripheral nerves of chick embryos
.
Expl Cell Res
.
7
,
558
562
.
Hughes
,
A. F.
(
1959
).
Studies on embryonic and larval development in Amphibia. I. The embryology of Eleutherodactylus ricordii, with special reference to the spinal cord
.
J. Embyol. exp. Morph
.
7
,
22
38
.
Hughes
,
A. F.
(
1961
).
Cell degeneration in the larval ventral horn of Xenopus laevis (Daudin)
.
J. Embryol. exp. Morph
.
9
,
269
284
.
Hughes
,
A. F.
(
1962
).
An experimental study of the relationships between limb and spinal cord in the embryo of Eleutherodactylus martinicensis
.
J. Embryol. exp. Morph
.
10
,
575
601
.
Hughes
,
A. F.
(
1964a
).
The innervation of supernumerary and replacing grafts of limbs in Eleutherodactylus martinicensis
.
J. Embryol. exp. Morph
.
12
,
27
41
.
Hughes
,
A. F.
(
1964b
).
Further experiments on the innervation and function of grafted supernumerary limbs in the embryo of Eleutherodactylus martinicensis
.
J. Embryol. exp. Morph
.
12
,
229
246
.
Hughes
,
A. F.
(
1965a
).
A quantitative study of the development of the nerves in the hind limb of Eleutherodactylus martinicensis
.
J. Embryol. exp. Morph
.
13
,
9
34
.
Hughes
,
A. F.
(
1965b
).
The development of behaviour in Eleutherodactylus martinicensis (Amphibia, Anura)
.
Proc. zool. Soc. Lond
.
144
,
153
161
.
Hughes
,
A. F.
(
1966a
).
The thyroid and the development of the nervous system in Eleutherodactylus martinicensis’, an experimental study
.
J. Embryol. exp. Morph
.
16
,
401
420
.
Hughes
,
A. F.
(
1966b
).
Spontaneous movements in the embryo of Eleutherodactylus martinicensis
.
Nature, Lond
.
211
,
51
53
.
Hughes
,
A. F.
(
1968
).
Aspects of Neural Development
.
Logos
:
London
.
Hughes
,
A. F.
(
1969
).
Quantitative aspects of neurohistology
.
Int. Rev. Gen. exp. Zool
.
4
,
169
217
.
Hughes
,
A. F.
,
Egar
,
M. W.
&
Turner
,
T.
(
1969
).
Degeneration of nerve fibers within the embryonic spinal cord
.
Nature, Lond
.
221
,
579
581
.
Hughes
,
A. F.
&
Prestige
,
M. C.
(
1967
).
Development of behaviour in the hind limb of Xenopus laevis
.
J. Zool. Lond
.
152
,
347
349
.
Hughes
,
A. F.
&
Reier
,
P.
(
1972
)
A preliminary study on the effects of bovine prolactin on embryos of Eleutherodactylus ricordii
.
Gen. comp. Endocrinol, (in the Press)
.
Jacobson
,
M.
(
1969
).
Development of specific neuronal connections
.
Science, N.Y
.
163
,
543547
.
Jacobson
,
M.
&
Baker
,
R. E.
(
1968
).
Neuronal specification of cutaneous nerves through connections with skin grafts in the frog
.
Science, N.Y
.
160
,
543
545
.
Jones
,
R. L.
(
1937
).
Split nucleoli as a source of error in nerve cell counts
.
Stain Technol
.
12
,
91
95
.
Katz
,
B.
(
1961
).
The termination of the afferent nerve fibre in the muscle spindle of the frog
.
Phil. Trans. R. Soc. B
243
,
221
240
.
Kollros
,
J. J.
(
1968
).
Order and control of neurogenesis
.
Devi Biol. Suppl
.
2
,
274
305
.
Miner
,
M.
(
1956
).
Integumental specification of sensory fibers in the development of cutaneous local sign
.
J. comp. Neurol
.
105
,
161
170
.
Peters
,
A.
(
1960
).
The structure of myelin sheaths in the central nervous system of Xenopus laevis (Daudin)
.
J. biophys. biochem. Cytol
.
7
,
121
126
.
Prestige
,
M. C.
(
1965
).
Cell turnover in the spinal ganglia of Xenopus laevis tadpoles
.
J. Embryol. exp. Morph
.
13
,
63
72
.
Sabatini
,
D. D.
,
Bensch
,
K.
&
Barnett
,
R. J.
(
1963
).
Cytochemistry and electron microscopy. The preservation of cellular ultrastructure and enzymatic activity by aldehyde fixation
.
J. Cell Biol
.
17
,
19
58
.
Webster
,
H.
(
1971
).
The geometry of peripheral myelin sheaths during their formation and growth in rat sciatic nerves
.
J. Cell Biol
.
48
,
348
367
.