ABSTRACT
I have used a novel quantitative electron microscopic method to determine the rate at which nerve fibres grow towards their targets during development. The rate of recruitment of nerve fibres to the maxillary nerve of the mouse embryo was determined by counting the number of axon profiles in the nerve sectioned close to its emergence from the trigeminal ganglion at closely staged intervals throughout its early development. The rate of change of fibre number with distance along this nerve was determined by counting the number of axon profiles at intervals along the nerve at stages during the period that fibres are growing to their targets. From these two parameters, both of which were linear functions during the mid-period of fibre recruitment to the nerve, it has been possible to calculate that embryonic sensory nerve fibres grow to their peripheral targets at the surprisingly slow rate of just over 20μmh−1.
Introduction
The difficulty of observing cells in the dynamic state within vertebrate embryos has precluded direct measurement of a basic parameter of the developing nervous system, namely the rate at which nerve fibres grow towards their targets. Estimation of the rate of regeneration of peripheral nerves following injury (for review see Sunderland, 1978) and measurements of the growth of nerve fibres observed directly within the tadpole tail (Speidel, 1941) may not be applicable to early embryonic development. Likewise, the growth rate of embryonic neurites in culture (Harrison, 1910; Hughes, 1953; Nakai, 1956; Bray, 1973; Luduena, 1973) gives little indication of the normal rate in vivo since it is greatly influenced by the choice of culture substratum (Luduena, 1973; Letourneau, 1975; Collins, 1978; Collins & Dawson, 1982).
Estimation of the rate at which normal embryonic nerve fibres grow towards their respective target fields in vivo is necessarily based on the study of static images. Attempts to ascertain this rate from measurements of nerve length in a series of staged embryos is unsatisfactory. First, this approach is restricted to a study of the short period of development during which the very earliest fibres of a nerve grow towards and reach their target field and hence is critically dependent upon the accuracy of staging embryos within this period. Second, it is difficult to locate with certainty the point to which the furthest nerve fibres of an early nerve have grown.
In normal development of the mouse (Davies & Lumsden, 1984) the earliest fibres of the maxillary nerve emerge from the trigeminal ganglion during the 9th embryonic day (E9, day of vaginal plug is E0) and reach the epithelium of the maxillary process by Ell. The fibre complement of the nerve increases until E13 after which there is a decline commensurate with neuronal death within the ganglion. Here I have studied the maxillary nerve from E10 to E13, the period during which fibres are recruited to the nerve and are growing towards the periphery.
* In addition to sensory nerve fibres, the maxillary nerve does contain a very small number of postganglionic para-sympathetic fibres from the pterygopalatine ganglion that briefly run with the nerve for a short part of its course.
Materials and methods
Embryos of known age were obtained from 8h overnight mating of CD1 albino mice (Charles River). Females were examined for vaginal plugs the following morning, which was considered to be day 0 of embryonic development (E0). The precise stage of development was determined by comparison of the external features of the embryos with the staging tables of Theiler (1972).
Tissue blocks consisting of the trigeminal ganglion and maxillary process of E10, Ell, E12 and E13 embryos of CD1 mice were fixed by immersion in 2% glutaraldehyde and 2·5% paraformaldehyde in 0·1 M-sodium cacodylate, postfixed in 1 % osmium tetroxide and embedded in Epon by standard procedures.
The maxillary nerve was sectioned in the transverse plane using a Reichert Ultracut ultramicrotome and ultrathin sections were examined and photographed on a Philips EM 300 electron microscope. In three or four embryos of each age the left nerve was sectioned close to its emergence from the trigeminal ganglion and the number of nerve fibres in this part of the nerve (n′) was determined by counting axon profiles in electron micrograph photomontages (see below). In addition, the left maxillary nerve in two El 1 and two E12 specimens was serially sectioned in the same plane and the numbers of axon profiles were counted from photomontages of ultrathin sections cut at 100 ;tm intervals along the nerve.
The early maxillary nerve is composed of several dozen fasciculi of unmyelinated nerve fibres (Fig. 1). The total numbers of axon profiles in fasciculi selected at random across each sectioned nerve were counted directly from electron micrograph photomontages at a final magnification of × 10000. At least one third of the total number of fibres in each nerve or at each level in a serially sectioned nerve was determined in this manner. The number of fibres in remaining fasciculi was estimated from surface area measurements as follows: fasciculus profiles (excluding accompanying Schwann cells) were traced from adjacent semithin sections with the aid of a drawing tube at a final magnification of ×1500 and the cross sectional areas of these profiles were accurately measured with a planimeter. The number of nerve fibres per unit area fasciculus was calculated for fasciculi in which fibres had been counted directly and the mean of these figures was used to calculate the number of nerve fibres in the remaining fasciculi. It was necessary to make these estimates separately at each level in serially sectioned nerves since fibre number per unit area fasciculus exhibited a gradual decrease with distance from the ganglion. Estimates of nerve fibre number by this method were subject to an error of less than ±5 %.
Transmission electron micrograph of a small fasciculus of unmyelinated axons in a section of the E10 maxillary nerve lying close to the ganglion. Bar, 1 μm.
Results and discussion
From E10 to E13 there was a steady increase in the fibre complement of the maxillary nerve (Table 1; Fig. 2). During this period of development there was a highly significant linear correlation between n′ and age (r = 0·987, P < 0·001). This suggests that throughout this period nerve fibres are recruited to the nerve at a constant rate of 325 fibres per hour (dn′/dr obtained by linear regression). It is likely, however, that the relationship of n′ to t is not strictly linear but logistic (sigmoidal) with a point of inflexion lying between Ell and E12. This is supported by the finding that the n′ means at E10 and E13 are respectively above and below the straight line passing through the n′ means at Ell and E12. Also, the line of best fit obtained from the data by linear regression intersects the abscissa at E9·6 whereas it is known that the earliest nerve fibres emerge from the ganglion between E9 and E9·5 (Davies & Lumsden, 1984). The implication of a logistic function is that the rate of recruitment of fibres to the nerve increases to reach a maximum between Ell and E12 and de-creases after this time. Since, however, there is no significant difference between the n′ means at each age and the corresponding n′ values interpolated from the straight line of best fit (t-test for related samples, P< 0 · 001) and because the estimated limits of values for dn′/dt at Ell and E12 for logistic functions lie within 5% of the value of 325 μmh−1 obtained by linear regression, the latter figure was used in the calculation of growth rate.
Number of nerve fibres in the maxillary nerve sectioned close to its emergence from the trigeminal ganglion at ages E10 to E13

Graph of the number of nerve fibres in the maxillary nerve close to its emergence from the trigeminal ganglion at E10, Ell, E12 and E13. The points indicate the number of fibres counted in individual specimens. The line passing in relation to these points is the line of best fit determined by linear regression.
Graph of the number of nerve fibres in the maxillary nerve close to its emergence from the trigeminal ganglion at E10, Ell, E12 and E13. The points indicate the number of fibres counted in individual specimens. The line passing in relation to these points is the line of best fit determined by linear regression.
The fibre complement of the nerve at intervals along its length at both Ell and E12 (Table 2; Fig. 3) appears to be a linear function of distance from the trigeminal ganglion; the correlation coefficients are very nearly —1 at Ell (r = –0·9996 and –0·9994) and at E12 (r=–0·9993 and –0·9989) and are highly significant in all cases (P< 0·001). Given the almost constant rate of recruitment of nerve fibres to the maxillary nerve during this period of development, the linear relationship of fibre number to distance implies that all nerve fibres grow towards the periphery at the same rate. If nerve fibres grew with a range of different growth rates then the relationship of fibre number (n) to distance from the ganglion (s) would approximate to the following form: n = ABs —C, where A, B and C are constants and B has a value of less than one.
Number of nerve fibres in the maxillary nerve at 100 pm intervals from the ventral pole of the trigeminal ganglion at Ell and E12

Graph of the number of nerve fibres in the maxillary nerve at 100 μm intervals from the ventral pole of the trigeminal ganglion in two Ell and two E12 specimens. The lines passing in relation to the points are the lines of best fit determined by linear regression.
The occurrence of either degeneration or collateral branching along the course of the maxillary nerve would confound the estimation of growth rate by the present method since either of these processes would influence the rate of change of fibre number with distance (dzi/ds). Degeneration of nerve fibres is, however, not evident until E13 (Davies & Lumsden, 1984) and collateral branching is first apparent at E12 (Davies & Lumsden, 1986). Since neither degeneration nor collateral branching occur in the Ell nerve, estimates of dn/ds from specimens of this age were used in the final calculation of growth rate. By linear regression dn/ds was markedly similar in both Ell specimens (–15·2 and –15·6). Estimates of dn/ds from the E12 specimens were slightly less negative (–14·6 and –14·2) than Ell figures which is the predicted consequence of collateral branching along the course of the nerve at E12.
Substituting the figure of 325 for dn′/dt and the means of the Ell values for dn/ds in the formula set out in the Introduction gives a value of 21 μm h−1 for the growth rate of sensory nerve fibres towards their peripheral target field in development. This growth rate is from one tenth to one half of the rates of regeneration of mammalian peripheral nerves (Sunderland, 1978). It is one half of the growth rate of cutaneous nerve fibres observed directly in the tail fin of frog tadpoles (Speidel, 1941) and is from one quarter to two thirds of the growth rates of a variety of pioneer fibres in the frog embryo determined by measurement of fibre lengths at different stages of development (Jacobson & Huang, 1985). It should be pointed out, however, that the method of calculating growth rate employed here specifically applies to the rate at which embryonic nerve fibres grow along their predecessors rather than the growth rate of pioneer fibres since the analysis is based on changes in the fibre composition of a portion of the nerve rather than its absolute length. It is not possible to ascertain using the present method whether the growth rate of the relatively small number of pioneer fibres is significantly greater than that of fasciculating nerve fibres.
It will be of interest to investigate, using the current method, whether there are differences in the growth rates of fibres from populations of sensory neurones whose target fields are near or distant (e.g. different cranial sensory ganglia). Also, it will be of interest to investigate whether there are differences in the growth rates of fibres from different classes of embryonic neurones (e.g. somatic motor, automonic, etc.).
ACKNOWLEDGEMENTS
This work was supported by a grant from the Nuffield Foundation.