Sequential double labelling with different fluorescent ...

Journal of Neuroscience Methods, 39 (1991) 9 - 1 7

9

© 1991 Elsevier Science Publishers B.V. All rights reserved 0165-0270/91/$03.50

NSM 01271

Sequential double labelling with different fluorescent dyes coupled to dextran amines as a tool to estimate the accuracy of tracer application and of regeneration B. Fritzsch and R. Sonntag UniL,ersity of Bielefeld, Faculty of Biology', Bielefeld (F.R.G.) (Received 30 May 1990) (Revised version received 12 February 1991) (Accepted 23 May 1991)

Key words: Double labelling; Motoneurons; Estimation of numbers; Regeneration We present a technique to estimate the accuracy of a given application procedure for neuronal tracers. In a second series of animals we used this technique for the estimation of successful regeneration of peripheral nerves. Dextran amine coupled to rhodamine was applied to the cut trochlar nerve in Xenopus tadpoles. To assess the accuracy of tracer application, experiments were done in which a second dye, dextran amine coupled to fluorescein, was applied after 1 day proximal to the first dye. More then 90% of all trochlear m o t o n e u r o n s were doubly labelled after this procedure. Their total n u m b e r s were not significantly different from n u m b e r s obtained after single labelling with H R P in a comparable age group, To assess success of regeneration after 5 and 8 days, the second application of fluorescein dextran amine was distal to the first application side. Statistically significant differences suggest incomplete regeneration of many neurons. After 42 days the numbers of singly and doubly labelled m o t o n e u r o n s was in the same proportion as before regeneration. This suggests that about 90% of the surviving motoneurons had successfully regenerated back to the periphery.

Introduction Counting neuronal numbers accurately is important for neurobiology in general (Williams and Herrup, 1988) and in particular for studies of numerical matching during development (Oppenheim, 1989) and evolution (Fritzsch and Sonntag, 1988). In addition to problems underlying the estimation of neuronal numbers from sectioned material (Coggeshall et al., 1990), the unambiguous identification of motoneurons is also critical. Three techniques are in use: (1) counting presumed motoneurons in Nissl-stained material

Correspondence: Dr. B. Fritzsch, Creighton University, Dept. of Biomedical Sciences, Division of Anatomy, Omaha, NE 68178, U.S.A. Tel.: 1-402-2802917. Fax: 1-402-2802690.

(Oppenheim, 1986); (2) counting the nerve fibres in a given nerve (Sohal and Weidmann, 1978; Harris and McCaig, 1984); and (3) counting motoneurons after retrograde labelling (Baulac and Meininger, 1983; Sonntag and Fritzsch, 1987; Madison et al., 1990). The data obtained by each technique are in general agreement with respect to over-production of neurons and subsequent loss through cell death during development. However, in systems where axons branch (Sohal and Weidmann, 1978), data obtained with different techniques disagree with respect to the magnitude of neuronal loss during ontogeny. In fact, each of these techniques can be criticized for the following reasons. First counting supposed motoneurons in Nissl-stained material neglects the potential heterogeneity of a set of neurons and is confounded by the ambigu-

10

ity to delineate a nucleus. For example, only 55% of the neurons in the abducens motor nucleus of the rat are motoneurons, the rest are interneurons (Cabrera et al., 1988). Likewise, during regeneration, some motoneurons may atrophy and hence may not be readily recognizable (Murphy et al., 1990). Second, counting of nerve fibres can lead to unambiguous numbers. However, most nerves contain fibres from more than one population of neurons and each fibre may in addition branch (Braekvelt et al., 1986; Dunlop and Beazley, 1987), may have loops (Dunlop et al., 1990), or one neuron may have two axons (Leonard and Willis, 1979; Holt, 1989). This is particularly obvious during regeneration when many neurons degenerate or atrophy whereas the number of nerve fibres remains constant suggesting extensive branching within the nerve (Murphy et al., 1990). Third, differences in transport at different developmental stages or difficulties in identifying labelled motoneurons has produced some suspicion over counting of motoneurons labelled with tracers (Oppenheim, 1986). Because most of the criticism against the latter approach appears invalid in our opinion, we have designed an experiment to find out whether retrograde tracing techniques are reliable enough to be used for quantitative studies. We employed the recently developed dextran amines coupled to different fluorescent dyes (Glover et al., 1986; Fritzsch and Wilm, 1990) in our studies for several reasons. These tracers are not detrimental to cells (Gimlich and Braun, 1985), provide a long-lasting label, do not leak, and their transport properties are essentially identical. When applied for a short time, dextrans will be taken up only by severed axons and, therefore, the accuracy of the surgery determines the selectivity of the application. We are aware that different combinations of other fluorescent dyes such as fast blue, fluorogold, lucifer yellow, rhodamine or rhodaminelabelled beads, diamidino yellow and DiI can and have been used to assess success of regeneration. One should, however, keep in mind that some of these dyes reportedly leak out of cells over time (Fritzsch and Wilm, 1990; Johnson et al., 1990) and some have been shown to be toxic (Illert et al., 1982). Moreover, some of these tracers cannot

be combined easily or may require readjustments of the protocol that can result in reduced sensitivity of either tracer technique as a compromise (Wigston and Kennedy, 1987; Duffy et al., 1990). Nevertheless, the carbocyanine dye DiI (Honig and Hume, 1989), if used in combination with fast blue (Vidal-Sanz et al., 1988) or fluorogold (Madison et al., 1990), can achieve over 90% double labelling. Compared to the fluorescent dyes, the use of H R P in conjunction with other substances (Stewart, 1981; Basbaum and Menetrey, 1987) can be tedious and time consuming.

Material and methods

For this study we used 44 larval Xenopus laevis (stages 48, 49, 50, 56; Nieuwkoop and Faber, 1967). The trochlear nerve was chosen for labelling because it has a number of unique features not shared with other nerves. In Xenopus it contains only nerve fibres from a distinct group of motoneurons in the contralateral midbrain (Fritzsch and Sonntag, 1987). A gap separate trochlear motoneurons at the stages relevant for this study from the only other contralateral motoneuron group, the oculomotor population projecting to the superior rectus (Fig. 1). Thus, even if we may have hit unnoticed an oculomotor nerve branch in addition to the trochlear nerve, there is no risk of misidentification of the motoneurons. In addition, the trochlear nerve courses rather isolated immediately below the skin, allowing easy operation and assessment of successful regeneration. The trochlear nerve was exposed in tadpoles (stages 48-56) anesthetized with tricaine methanesulfonate (1:3000). The nerve was cut with fine scissors close to its sole target, the superior oblique muscle. Care was taken to avoid teasing of the nerve fibres. H R P (Boehringer, Mannheim) was applied in the first group of animals as crystals to the cut nerve for 1 min, a sufficient time for complete labelling with little or no additional unwanted labelling (Fritzsch et al., 1990). In the second to fifth group of animals rhodamine dextran amine (RDA, 10000 MW, lysine fixable; Molecular Probes, Eugene, OR) was applied in

11

Fig. 1. The appearance of trochlear (bottom) and oculomotor (top) motoneurons is shown after application of rhodamine dextran amine to the cut oculomotor and trochlear nerves in the left orbit and fluorescein dextran amine in the right orbit. Note the distinct gap between the contralateral oculomotor (green, top) and the contralateral trochlear (green, bottom) neurons at this stage. Dendrites are clearly labelled after this short survival time and green- and red-labelled cells are distinct in this double exposure. Flat mount of a stage-48 animal: transport time was 6 h; caudal is to the bottom; lateral is to the left. Bar = 10/zm. Photographed on Kodak Ektachrome 200 ASA.

12 TABLE I DEXTRAN AMINE AND HRP-LABELLED MOTONEURONS IN STAGE 56 XENOPUS TADPOLES One group of animals (column 1) received a single application of H R P to the cut trochlear nerve. A second group of animals (columns 1-3), matched in age and size, first received R D A to the cut trochlear nerve followed 1 day later by F D A to the cut proximal stump. Ninety-seven percent of all R D A and 95% of FDA-labelled cells were also double labelled (columns 1-3). The maximal number of single-labelled cells in all dextranamine applications is close to the number of HRP-labelled cells in the control batch. Group 1 Labelling (HRP) 30 33 40 40 40 46 53 56 59 -

*

44 ± 10

Group 2 First (RDA)

Second labelling (FDA)

30 34 35 36 39 41 41 42 56 60

35 31 37 36 38 36 44 40 61 61

41 ±9

42 ± 10

Double (RDA/FDA) 30 3(/ 34 35 38 35 41 39 56 59 40Mean ± 10 SD.

* One animal died.

the same way as HRP in the first group. Animals were subsequently reanimated in water. The HRP group (10 animals operated at stage 56: group 1) was processed after 2 days postoperative survival time for HRP histochemistry as described elsewhere (Fritzsch and Sonntag, 1987). In the second group (10 animals, RDA application at stage 56), the trochlear nerve was re-exposed after 1 day, sectioned proximal to the previous cut and dextran amine (FDA) was applied for 1 min. After an additional survival time of 2 days, the animals were anesthetized and fixed with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). In the third to fifth groups (group 3: RDA application at stage 48, 8 animals; group 4: RDA application at stage 49, 10 animals; group 5: RDA application at stage 50, 4 animals), the regenerated trochlear nerve was cut again after 5 days (group 3), 8 days (group 4) or 6 weeks (group 5) distal to the first cut and FDA was applied. Alter additional 18 h the animals were anesthetized and fixed by immersion in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). Brains of animals in groups 1, 2 and 5 were frozen sectioned at 50/xm and mounted on gela-

TABLE II D E X T R A N - A M I N E LABELLED M O T O N E U R O N S A F T E R V A R I O U S TIMES O F R E G E N E R A T I O N This table shows that close to 90% of the trochlear neurons that survive axotomy may eventually regenerate back to the periphery with time. This process is not yet completed 5 or 8 days after surgery. Group 3

Group 4

Group 5

5 days of regeneration

8 days of regeneration

6 weeks of regeneration

Stage of operation

48

49

50

N of first-labelled cells

4, 6, 6, 9, 9, 14, 15

20, 28, 30, 32

Mean

9 ± 3.8

10, 16, 17, 17, 18 24, 25, 26, 28, 29 21 ± 5.9

N of second-labelled cells

1, 4, 5, 9, 12, 7, 14

21, 29, 26, 38

Mean

6.7 ± 3,4

- , 2, 8, 12, 13, 17, 18, 24, 30, 35 16.7 ± 9.9

N of doubly labelled cells

1, 4, 4, 9, 9, 5, 7

15, 26, 24, 32

Mean

5.6 ± 2.7

- , 2, 8, 12, 13, 17, 18, 21, 23, 24 15.3 ± 6.9

24.3 ± 6.1

Doubly labelled cells compared to first labelling (%)

62

73

88

27.5 _+ 4.5

28.5 ± 6.2

13 tinized slides. Sections from group ~ were reacted for H R P detection using diamino benzidine (Fritzsch and Sonntag, 1987). Fluorescent labelling was studied in the sections from groups 2 and 5, using an epifluorescence microscope with the appropriate filter sets for fluorescein isothiocyanate (FITC) and rhodamine isothiocyanate (RITC) and photographed (T-Max 100, Tri-X 400 or Ektachrome 200, Kodak). In groups 3 and 4 the brains were split in the dorsal midline (e.g., through the tectum and diencephalon) and mounted flat with the ventral side up on a slide (Aitken, 1987). These brains were coverslipped with Elvanol and viewed in an epifluorescence microscope. In addition, the head with the trochlear nerve of groups 3-5 was mounted flat on a slide and the trochlear nerve was examined for double labelling with both fluorescent dyes and for regeneration. Numbers of double-labelled and single-labelled cells were counted with a 40 x oil immersion objective (Tables I and II). There was no problem in distinguishing the two colors by eye or in color photographs (Fig. 1) but it sometimes was difficult in the black and white documentation. No

correction for double counting was employed in the sectioned material because this error is small in thick sections (Ebbesson and Tang, 1965) and because we are comparing relative numbers of similar aged animals prepared in a similar way and cut at the same thickness. Finally, in two additional animals (stage 48), we applied R D A to the left and F D A to the right orbit to fill all ocular motoneurons to show the distribution of contralateral and ipsilateral projecting motoneutons and the quality of the labelling with dextran amines (Fig. 1). Results

Qualitatice data In the H R P cases (group 1) the trochlear motoneurons were usually stained in a Golgi-like fashion with distinct dendrites. Labelling with H R P at this early stage revealed an extent of the dendrites not noticed in previous studies on adult frogs (Fritzsch and Sonntag, 1987). The intensity of labelling with fluorescent dyes varied between cells in a given animal, between animals and with survival time. However, most

Fig. 2. Sequential double labelling without regeneration. RDA (A) was applied l day prior to FDA (B) on a more proximal part of the stump. Note that all neurons are doubly labelled except one (white asterisk in A). 50 #m section of a stage-50 tadpole. Bar indicates 10 #m. Photographed on Tri-X 400 ASA.

14

cells were filled in a Golgi-like fashion as with H R P (Fig. 1). As with H R P (Fritzsch et al., 1984), the nucleus of these cells was diffusely labelled whereas the cytoplasm had both a diffuse background and a granular labelling. After long survival times (e.g., 5-42 days; Fig. 3) only a granular labelling of the cytoplasm remained. Label persisted in the proximal stump of the axon for at least 8 days while the distal stump degenerated within a few days. Anterograde filling of the axons with FDA showed that many motoneurons had successfully regenerated to their target after 5 and more days. Weakly labelled cells surrounded by strongly labelled cells were sometimes difficult to identify (Figs. 1-3). We counted only cells with distinctly labelled perikarya and may, therefore, have slightly underestimated the real population size. In contrast to the H R P reaction product, fluorescent dyes bleach with exposure time and weakly labelled neurons may in addition be bleached before they can be detected.

Quantitati~'e data Our data (Table I) showed that the numbers of double-labelled ceils (30-59) was comparable to those of single-labelled cells. Comparison of the mean of neurons labelled with both RDA and FDA showed no significant difference (t test, P = 0.05) to cells labelled only with FDA or RDA, or with HRP. The dye applied after I day (FDA) showed neither consistently stronger labelling (Fig. 2) nor consistently different numbers (Table I). This was true when pooled data as well as when individual animals were compared. Sequential second labelling after regeneration of cut fibres (groups 3-5) gave a rather different picture. After short-term regeneration (e.g., 5 days) many neurons were only labelled with the first applied tracer and only 62% of the neurons were doubly labelled. Results obtained after a somewhat longer regeneration of 8 days suggest a similar rate of regeneration (73% doubly labelled, Table II). In contrast, after long-term regeneration, the differences between first and second

Fig. 3. Sequential double labelling with regeneration. RDA (A) was applied to the cut trochlear nerve and FDA (B) to the regenerated nerve 42 days later. All cells are double labelled except one (white asterisk), 5[) g m section of a stage-51 tadpole. Bar indicates 10/.Lm. Photographed on Tri-X 400 Asa.

15 label were comparable to those of group 1, e.g., somewhat more neurons were labelled with the second fluorescent dye applied. About 88% of the cells were double labelled suggesting a similar success rate of regeneration.

Discussion The data show that counting of motoneurons labelled through their cut axons is a reliable procedure to identify motoneurons in as much as similar data in same aged populations are obtained with different labelling techniques. This is particularly important in cases where motoneutons cannot be identified in Nissl-stained material, e.g., like in development, or are not confined to a single nucleus or may represent only a part of the nucleus. Therefore, counting of motoneurons labelled from their target avoids two potential sources of error of Nissl-stained material, false positive or negative identification of cells as motoneurons within a nucleus and false identification of boundaries of a nucleus. The most serious challenge for counting labelled neurons is clearly the problem of incomplete labelling (McLachlan and Jiinig, 1983). This problem is obviously more pronounced after injection of H R P into a muscle than after applying H R P to the cut nerve (Murphy et al., 1986). If cutting of a nerve and application of a tracer was an unreliable procedure, we would expect a large difference in double-labelled as compared to single-labelled cells because of an accumulation of technical problems. Contrary to this expectation, we found that the level of reliability of the double application technique is such that no significant differences are obtained compared to any single labelling procedure employed. This result is in contrast with the inter-individual differences of about 50% obtained with any technique. Our data show that regeneration can almost completely restore the original pattern of innervation, that is about 88% of motoneurons reach their target after a period of 42 days or longer. Data on the trochlear system of mammals suggest a successful regeneration of only about 50% of cells after peripheral nerve section (Murphy et

al., 1990), of only 63% in regenerating salamander limb motoneurons (Wigston and Kennedy, 1987) and of only 5% of supraspinal neurons in lizards (Duffy et al., 1990). However, using double labelling with fluorescent tracers Madison et al. (1990) showed greater than 90% success of regeneration in the femoral nerve of mammals. Using the approach of sequential double labelling we can demonstrate that about 70% of the labelled motoneurons have already regenerated at 5 - 8 days after surgery. This result is in agreement with claims of regeneration of the trochlear nerve fibers to the muscle after 5 days in same aged Xenopus (Fangboner, 1979). Using sequential double labelling we currently assess the success of regeneration and degeneration over time quantitatively while identifying at the same time the position of successfully regenerated motoneurons and the termination of regenerated fibers in the target. From these data we conclude three things: (a) Single labelling of motoneurons through their cut axon is a reliable procedure to estimate numbers of neurons projecting to a given target, provided the sample size is large enough to compensate for inter-individual differences. Sequential double labelling results in numbers which are less than 10% lower than after single labelling, i.e., double labelling is almost as successful as single labelling thereby arguing against large scale technical problems of the application technique. (b) Differences in motoneuron numbers of different animals (as much as 50%) seem to reflect real differences rather than technical problems related to the application of the dye. A comparable inter-individual variability of 50% or more was reported for spinal motoneurons (Sperry, 1987) and seems to be a general feature of frogs rather than of the technique used to obtain the numbers. (c) Sequential double labelling is a good technique to estimate the degree of regeneration obtained after a given time. This is particularly important after an embryonic cut of a nerve when, at least in the trochlear system of Xenopus, both cell death and continuous proliferation (Fritzsch and Sonntag, 1987; Sonntag and Fritzsch, 1989; submitted) may change the composition of the regenerating population over time. Using different sur-

16

vival times we are currently using double labelling to estimate the degree of cell death caused by axotomy at different stages. In summary, we propose the double-labelling technique as a useful test to ascertain the reliability of a labelling technique in systems with accessible axons such as motoneurons. Comparing double with single-labelled cells allows to estimate the degree of error introduced by the application procedure employed. Double labelling with fluorescent dextran amines is also a simple way to estimate success of regeneration over time. Double labelling with dextran amines was recently employed successfully to reveal the distribution of regenerating nerve cells in a more complex system, the reticulo-spinal projection of birds (Hasan et al., 1990) and we are currently testing whether it can be used to assess quantitatively the success of regeneration in the reticulo-spinal system of frogs.

Acknowledgements This work was supported by the German Science Foundation (SFB 223). We thank the reviewers of Journal of Neuroscience Methods for helpful criticism on an earlier version and Dr. T. Neary for linguistic advice.

References Aitken, A.R. (1987) The preparation and use of embryonic brain wholemounts, J. Neurosci. Methods, 19: 297-304. Basbaum, A.I. and Menetrey, D. (1987) Wheat germ agglutinin-apoHRP Gold: a new retrograde tracer for light- and electron-microscopic single- and double-label studies, J. Comp. Neurol., 261: 306-318. Baulac, M. and Meininger, V. (1983) Postnatal development and cell death in the sciatic motor nucleus of the mouse, Exp. Brain Res., 50: 107-116. Braekvelt, C.R., Beazley, LD., Dunlop, S.A. and Darby, J.E. (1986) Numbers of axons in the optic nerve and of retinal ganglion cells during development in the marsupial Setonix brachyurus, Dev. Brain Res., 25: 117-125. Cabrera, B., Portillo, F., P~isaro, R. and Delgado-Garcia, J.M. (1988) Location of motoneurons and internuclear neurons within the rat abducens nucleus by means of horseradish peroxidase and fluorescent double labelling, Neurosci. Lett., 87: 1-6.

Coggeshall, R.E., La Forte, R. and Klein, C.M. (1990) Calibration of methods for determining numbers of dorsal root ganglion cells, J. Neurosci. Methods, 35: 187-194. Dunlop, S.A. and Beazley, L.D. (1987) Axonal morphology in the developing mammalian optic nerve - towards an explanation for transiently high axon numbers, Neurosci. Lett. Suppl.. 27: 71. Dunlop, S.A., Ross, W.M. and Beazley, L.D. (1990) Retinal projections and axonal morphology in the developing visual system of the quokka, Setonix brachyurll~, Soc. Neurosci. Abst. 16: 335. Duffy, M.T., Simpson, S.B., Liebich, D.A. and Davis, B.M. (1990) Origin of spinal cord axons in the lizard regenerated tail: supernormal projections from local spinal neurons, J. Comp. Neurol., 293: 208-222. Ebbesson, S.O.E. and Tang, D. (1964) A method for estimating the number of cells in histological sections, J. Roy. Micr. Soc., 84: 449-464. Fangboner, R.F. (1979) Trochlear-oculomotor nerve interactions in Xenopus laeL,is tadpoles: a temporal study, J. Exp. Zool., 209: 355-366. Fritzsch, B., Nikundiwe, A.M. and Will, U. (1984) Projection patterns of lateral-line afferents in anurans. A comparative study using transganglionic transport of HRP, J. Comp. Neurol., 229: 451-469. Fritzsch, B. and Sonntag, R. (1987) The trochlear nerve of amphibians and its relation to proprioceptive fibres: a qualitative and quantitative HRP study, Anat. Embryol.. 177: 105-114. Fritzsch, B. and Sonntag, R. (1988) The trochtear motoneurons of lampreys (Lampetra fluviatilis): location, morphology and numbers as revealed with horseradish peroxidase, Cell Tiss. Res., 252: 223-229. Fritzsch, B. and Wilm, C. (1990) Dextran amines in neuronal tracing. Trends Neurosci., 13: 14. Fritzsch, B. Niemann, U. and Bleckmann, H. (1990) A discrete projection of the saeculus and lagena to a distinct brainstem nucleus in a catfish, Neurosci. Lett., 111: 7-11. Gimlich, R.L. and Braun, J. (1985) Improved fluorescent compounds for tracing cell lineage, Dev. Biol., 109: 509514. Glover, J.C., Petursdottir, G. and Jansen, J.K.S. (1986) Fluorescent dextran amines as neuronal tracers in the nervous system of the chicken embryo, J. Neurosci. Methods, 18: 243-254. Harris, A.J. and McCaig, C.D. (1984) Motoneuron death and motor unit size during embryonic development of the rat, J. Neurosci., 4: 13-24. Hasan, S.J., Keirstead, H.S. and Steeves, J.D. (1990) Regeneration of brainstem-spinal axons after spinal cord cut in embryonic chicken, Soc. Neurosci. Abst., 16: 1002. Holt, C.E. (1989) A single-cell analysis of early retinal ganglion cell differentiation in Xenopus: from soma to axon tip, J. Neurosci., 9: 3123-3145. Honig, M.G. and Hume, R.I. (1989) Dil and DiO: versatile fluorescent dyes for neuronal labelling and pathway tracing, Trends Neurosci., 12: 333-341.

17 Illert, M., Fritz, N., Aschoff, A, and Holliinder, H. (1982) Fluorescent compounds as retrograde tracers compared with horseradish peroxidase (HRP). II. A parametric study in the peripheral motor system of the cat, J, Neurosci. Methods, 6: 199-218, Johnson. J.K., Jinks, R.N. and Chamberlain, S.C. (1990) Transneuronal diffusion of DiI in the visual system of the horseshoe crab, Limulus, Soc. Neurosei. Abst., 16: 53. Leonard, R.B. and Willis, W.D. (1979) The organization of the electromotor nucleus and extraocular motor nuclei in the star gazer (Astorscopus y-graecum), J. Comp. Neurol., 183: 397-413. Madison, R.D., Archibald, S.J. and Meadows, S. (1990) Permanent labeling of motoneuron pools: accuracy of regeneration analyzed at the single neuron level, Soc. Neurosci. Abst., 16: 806. McLachlan, E.M. and J~inig, W. (1983) The cell bodies of origin of sympathetic and sensory axons in some skin and muscle nerves of the cat hindlimb, J. Comp. Neurol., 214: 115-13(I.

Murphy, E.H., Garone, M, Tashayyod, D. and Baker, R.B. (1986) lnnervation of extraocular muscles in the rabbit, J. Comp. Neurol., 254: 78-90. Murphy, E.H., Brown, J., Iannuzzelli, P.G. and Baker, R. (1990) Regeneration and soma size changes following axotomy of the trochlear nerve, J. Comp. Neurol., 292:524 536. Nieuwkoop, P.D. and Faber, J. (1967) Normal table of Xenopus laeris, North Holland, Amsterdam. Oppenheim, R.W. (1986)The absence of significant postnatal motoneuron death in the brachial and lumbar spinal cord of the rat, J. Comp. Neurol., 246: 281-286.

Oppenheim, R.W. (1989) The neurotrophic theory, and naturally occurring motoneuron death, Trends Neurosci., 12: 252-255. Sohal, G.S. and Weidman, T.A. (1978) Development of the trochlear nerve: loss of axons during normal ontogeny, Brain Res., 142: 455-465. Sonntag, R. and Fritzsch, B. (1987) The development of the trochlear nucleus of amphibians. A HRP study, Neurosci. Lett., 72: 143-148. Sonntag, R. and Fritzsch, B. (1989) Addition of trochlear motoneurons in the clawed frog, Xenopus laet'is, during early development and midlarval stages. In: H. Rahmann (Ed.), Fundamentals of Memory Formation. Neuronal Plasticity and Brain Function. Vol. 37, Progress in Zoology, Fischer, Stuttgart, pp. 109 11(1. Sonntag, R. and Fritzsch, B. (1990) Numerical development of trochlear motoneurons of Xenopus laetis, J. Neurobiol., submitted. Sperry, D.G. (1987) Relationship between natural variations in motoneuron number and body size in Xenopus laez'is: a test for size matching, J. Comp. Neurol., 264: 250-267. Steward, O. (1981) Horseradish peroxidase and fluorescent substances and their combination with other techniques. In: L. Heimer and M.J. Robards (Eds.), Neuroanatomical Tract-Tracing Methods, Plenum Press, New York, pp. 279-311). Williams, R.W. and Herrup, K. (1988) The control of neuron number, Annu. Rev. Neurosci., 11:423 453. Wigston, D.J. and Kennedy, P.R, (1987) Selective reinnervation of transplanted muscles by their original motoneurons in the axolotl, J. Neurosci., 7: 1857-1865.