Thy-1 is a developmentally regulated surface glycoprotein expressed on a number of tissues, including nerve where it is a major surface component of mature neurons. During neural development in the rat and mouse, expression of Thy-1 protein does not necessarily follow appearance of its mRNA, but additionally requires completion of the initial phase of axonal growth. Where there is a substantial lag phase between initial elongation and final axonal outgrowth into a terminal field (e.g. pontine projection to the cerebellum), Thy-1 protein appears at the cell body and dendrites of the neurons, but is excluded from their axons until the terminal phase of axonal growth is completed. In the more complex case of the vestibular ganglion neurons, whose axons project primarily to the vestibular nuclei in the brainstem before birth, and then 1–2 weeks later into the cerebellum, Thy-1 enters the proximal axonal regions where growth is completed, but not the distal growing ends. Thus complex controls govern the initial expression and distribution of Thy-1 so as to exclude it from growing regions of axons.

Neurons are exceptional in the extent to which different areas of their surface are specialised to perform different tasks. Beyond the major division into a transmitting (axonal) and receiving (dendritic) compartment lie complex subdivisions: many axons project to several targets, and most dendrites receive inputs from multiple areas that form synapses differing in their location, morphology, chemistry and physiological effect. In development, each of these specialised areas matures at its own rate, determined not just by events intrinsic to the neuron but also by the other cells in its immediate environment (e.g. Ghosh et al. 1990). At the molecular level, therefore, one would expect the maturation of different parts of a neuron to proceed at different rates appropriate to the particular microenvironments contacted.

Such local modulation of the developmental appearance of the neuronal surface glycoprotein, Thy-1, appears to occur in brain. In the olfactory bulb, Thy-1 mRNA is expressed in all mitral cells at the same stage of differentiation, when their soma have migrated to their final position and begun to grow dendrites. Expression of Thy-1 protein, however, does not simply follow that of its mRNA, but requires some additional signal, which we suggested was related to the cessation of axonal growth (Xue et al. 1990).

We examine this possibility here by studying Thy-1 expression during development of four different axonal systems (Fig. 1). Two are projections to the granule cell layer of the cerebellum, chosen because this is one of the few areas of brain where the vast majority of local cells (the granule interneurons) express very low levels of Thy-1 (Morris et al. 1985a; Bohn and Rouse, 1986). The acquisition of Thy-1 by the terminals of these axons, as they sprout into the granule layer and mature over the period P5 –28 (Altman, 1972; Hamori and Somogyi, 1983; Ito, 1984; Arsemo Nunes and Sotelo, 1985), can therefore be clearly followed. These fibres, collectively called mossy fibres because of their distinctively large terminals, come from diverse parts of the central and peripheral nervous systems. The first group that we have studied are those arising in the pons, axons whose only target is the cerebellar granule layer. These grow in two phases: initial elongation to cerebellar cortex in the perinatal period followed a week later by secondary sprouting into their terminal field. The second group are axons of the vestibular ganglion neurons. These are sensory neurons of the peripheral nervous system and, unlike the CNS neurons studied here, lack dendrites (Ballantyne and Engstrom, 1969). Early in neural development (E12 –13 in rat), their axons grow into the brain where they bifurcate, one branch going to the two descending vestibular nuclei, the other to the superior vestibular nucleus and thence into parts of the cerebellar cortex (Fig. 1). Their major targets are the neurons of the three vestibular nuclei in the brain stem, which are contacted from E15 by terminal sprouts of the axons; numerous large terminals are formed on them by birth (Morris et al. 1988). The initial elongation of the axons also brings them, at E13 –15, into the cerebellar cortex. Despite being the first axons to arrive here, they must wait for over 2 weeks for their terminal field, the granule cells, to mature. It was therefore of interest to see how this minor, and very late developing, arm of a complex axonal network would influence the appearance of Thy-1 on these cells.

Fig. 1.

The four axonal systems studied. In the sagittal (upper) plane is shown the pontine projection to lobule VII of cerebellar cortex (these also project to most of the other vermal lobules and the cerebellar hemispheres; for the course and timing of growth of these axons, see Ito, 1984; Flumerfelt and Hrycyshyn, 1985; Payne and Bower, 1988); the projection of the vestibular ganglion (VG) axons to lobule IX (location of the superior vestibular nucleus (SVN) and of the two descending vestibular nuclei (DVN) are indicated; the fourth vestibular nucleus (the lateral) is not a major target for these axons; see: Brodal and Hoivik, 1964; Gacek, 1969; Morris et al. 1988); and the projection of hippocampal granule cells to the CA3 pyramids (see: Bliss et al. 1974, Stirling and Bliss, 1978; Amaral and Dent, 1981; Gaarskjaer, 1985). In the coronal plane is shown the projection of hippocampal pyramidal neurons, ipsilaterally to other pyramidal neurons and the lateral septal nucleus (LSN), and contralaterally via the fimbria and ventral hippocampal fissure (VHC) to the other hippocampus (sec Fricke and Cowan, 1977; Zimmer and Haug, 1978; Swanson, 1978; Cowan et al. 1981; Frotscher et al 1988; Buchhalter et al. 1990).

Fig. 1.

The four axonal systems studied. In the sagittal (upper) plane is shown the pontine projection to lobule VII of cerebellar cortex (these also project to most of the other vermal lobules and the cerebellar hemispheres; for the course and timing of growth of these axons, see Ito, 1984; Flumerfelt and Hrycyshyn, 1985; Payne and Bower, 1988); the projection of the vestibular ganglion (VG) axons to lobule IX (location of the superior vestibular nucleus (SVN) and of the two descending vestibular nuclei (DVN) are indicated; the fourth vestibular nucleus (the lateral) is not a major target for these axons; see: Brodal and Hoivik, 1964; Gacek, 1969; Morris et al. 1988); and the projection of hippocampal granule cells to the CA3 pyramids (see: Bliss et al. 1974, Stirling and Bliss, 1978; Amaral and Dent, 1981; Gaarskjaer, 1985). In the coronal plane is shown the projection of hippocampal pyramidal neurons, ipsilaterally to other pyramidal neurons and the lateral septal nucleus (LSN), and contralaterally via the fimbria and ventral hippocampal fissure (VHC) to the other hippocampus (sec Fricke and Cowan, 1977; Zimmer and Haug, 1978; Swanson, 1978; Cowan et al. 1981; Frotscher et al 1988; Buchhalter et al. 1990).

Analysis of Thy-1 acquisition during axogenesis elsewhere in CNS is frustrated by the fact that Thy-1 is so ubiquitous a neuronal membrane component that its appearance on any one cell and its processes is obscured by the multitude of other similarly immunoreactive elements in its environment. Some analysis is possible in the hippocampus, however, where both the main projection neurons (pyramidal cells) and interneurons (granule cells) are clustered in distinct layers. The pyramidal cells show a complex pattern of projection, set up in late embryonic life (Fig. 1); their axons are dominant components of the large fibre tract called the fimbria, and its extension in the midfine, the ventral hippocampal commissure. The granule cell axons project, in the second postnatal week and later, to a single target in the ipsilateral hippocampus, the region immediately above the cell bodies of the large CA3 pyramids, but do not progress further around the pyramidal layer to contact the smaller CAI pyramids (Fig. 1).

To identify the growth status of axons in our material, we have used immunostaining with antibodies to the growth-associated protein (GAP) 43 (Goslin et al. 1988) and to the microtubule-associated protein (MAP) lx (Calvert and Anderton, 1985). The latter antibody (GIO) detects an epitope on MAPlx which is present only on growing axons and is rapidly down-regulated on cessation of growth (Calvert et al. 1987 ; Woodhams et al. 1989; Garner et al. 1989; Sato-Yoshitake et al. 1989; Reiderer et al. 1990).

A feature of Thy-1 expression in non-neural tissues is the extent to which significant interspecies differences occur, even between mouse and rat (Morris, 1985). In this study of neuronal expression, we have therefore examined both rodent species, in order to distinguish between characteristics of a particular species and properties that might relate more generally to neural differentiation.

The basic methodology for in situ hybridisation and immunohistochemistry, and much of the material, was identical to that already described (Xue et al. 1990). Pregnancies were dated from the appearance of a vaginal plug (embryonic day (E) 0), mice gave birth at E19, rats at E21. Day of birth was taken as postnatal day (P) 0. Thy-1 mRNA was detected with a probe to the third exon of mouse Thy-1, which is 90% homologous to the rat and gives equivalent signal in both species (Xue et al. 1990) Thy-1 was detected with monoclonal IgG antibodies against the Thy-1.1 (rat) and 1.2 (mouse) allelic determinants, specified by alternative Arg/Gin at residue 89 (Williams and Gagnon, 1982), and F(ab ′)2 fragments of polyclonal rabbit antibodies to rat Thy-1, which recognise epitopes different to the allelic one (Moms et al. 1980). In general, we show only the immunohistochemical results obtained with the monoclonal antibodies, except where these differed from those obtained with the polyclonal antibodies.

Tissue for immunohistochemistry was fixed in cold 5 % acetic acid in 96 % ethanol, embedded in polyester wax, and 5 μm sections dewaxed in graded alcohols before use. Material was analysed in both the sagittal and coronal planes. To identify parvalbumin, sections were dewaxed in graded alcohols which from 70 % downwards contained 1 % paraformaldehyde; they were left a further 20 mm in 4 % paraformaldehyde in Ca2+,Mg2+-free PBS, then washed 3 times in PBS before use. This did not give labelling as intense as is obtained when the primary fixative is an aldehyde (Morris et al 1988), but was adequate to define the vestibular neurons and their axons, and allowed serial sections to be labelled with the range of other antibodies used. GAP43 was detected with mouse monoclonal 91E12 (Goslin et al. 1988) used at a dilution of 1:4000 of ascitic fluid; rabbit anti-parvalbumm antibodies (Kagi et al. 1987) were used at 1T000, rabbit antiN-CAM (Doherty et al. 1988) were used at 1:1000.

To identify antigenic determinants immobilised in the section via a phosphatidylinositol linkage, sagittal sections of mouse vestibular ganglia were rapidly frozen in crushed dry ice and sectioned on the cryostat. They were incubated for 15 min at 37°C in moist chambers with the phosphatidylinositol-specific phospholipase C (PI-PLC) from Bacillus thurmgiensis (Pemnsular Labs) at the concentration shown, washed m 4 changes of PBS over 15 mm, and stained with antibodies as before.

Thy-1 was solubilised from rat brain membranes treated sequentially with 2 % Triton X-100 and 2.5 % sodium deoxycholate (Morris et al. 1980), and purified using an immunoaffinity column to which the OX7 monoclonal was coupled. Protein concentration was determined using the Biorad dye binding kit. To inhibit antibody bmding, rabbit anti-Thy-1 antibodies were preincubated overnight with various concentrations of pure Thy-1, then used to stain sections as before

For immunoblot analysis, material was homogenised in 6 volumes of Ca2+, Mg2+-free PBS and the nuclear fraction removed by a brief pulse of centrifugation (5 s, 6500 revs min-1) in a microfuge. The membrane fraction was pelleted (13 500 revs min−1, 15 min at 4°C; no Thy-1 immunoreaction was detectable in the supernatant by dot blot analysis), dehpidated with 80 % acetone on solid CO2 and centrifuged at room temperature (13 500 revs min−1, 5 mm). The pellet was taken up in non-reducing Laemmh electrophoresis sample buffer, electophoresed on 15 % polyacrylamide gels (Laemmli, 1970) and transferred (5.0 mA cm−2 for 1h) to Immobilon P membrane (Milhpore) for western blotting. Lanes containing molecular weight markers were stained with amido black. The remainder of the filter was blocked by immersion for 20 min at room temperature in 10 % defatted dried milk in PBS containing 0.1 % NaN3 and incubated with antibodies diluted in 1 % milk in the same buffer. Secondary antibodies used were our own affinity-purified F(ab′)2 horse anti-rabbit, rabbit anti-rat and rabbit anti-mouse IgGs, bilabelled (20mCimg-1) by a modified chloramine T method (Morris and Raisman, 1983).

Axonal projection from the pons to cerebellar lobule VII

At birth, fibres from many pontine neurons can be seen with G10 immunohistochemistry to have ascended the mouse brainstem laterally and entered the cerebellum, where they disperse to multiple locations including coursing medially in the vermis to enter, among others, dorsal lobule VII (Fig. 2B). As later (ventrally located; Adams et al. 1980) pontine neurons matured (PO-5) their axons also entered this tract. Although the pontine axons are in general the largest group of mossy fibres terminating in the cerebellar granule layer, in lobule VII in particular they constitute the vast majority of such axons (Batini et al 1978; Kawamura and Hashikawa, 1981; Ito, 1984) and so it is here that we have focused our analysis.

Fig. 2.

Mouse hindbrain in sagittal (A; P5 animal) and coronal (B; PO animal) section after immunoperoxidase staining with GIO antibody. In sagittal section the pons (P) can be seen at the base of the brainstem, and lobules IX and VII in the cerebellum (Cb) are marked. The areas shown inB of three near adjacent sections illustrate the course of the pontine axons, crossing the midhne to ascend the brainstem laterally near the pial surface in a tract (brachium pontis, BP) which enters the cerebellum via the middle peduncle (mcp). Scale bars are 1mm (A) and 200μm (B).

Fig. 2.

Mouse hindbrain in sagittal (A; P5 animal) and coronal (B; PO animal) section after immunoperoxidase staining with GIO antibody. In sagittal section the pons (P) can be seen at the base of the brainstem, and lobules IX and VII in the cerebellum (Cb) are marked. The areas shown inB of three near adjacent sections illustrate the course of the pontine axons, crossing the midhne to ascend the brainstem laterally near the pial surface in a tract (brachium pontis, BP) which enters the cerebellum via the middle peduncle (mcp). Scale bars are 1mm (A) and 200μm (B).

Pontine neurons displayed a positive in situ hybridisation signal for Thy-1 mRNA from P0 (dorsal) to P5 (ventral) in the mouse (Fig. 3A-H). Thy-1 protein appeared on the surface membrane of their somata and dendrites from P5-8 (Fig. 3I-L), and by P12 was already of an intensity comparable to the adult.

Fig. 3.

Developmental appearance of Thy-1 mRNA (A-H) and protein (I-L) in the mouse pons. Thy-1 mRNA, detected by autoradiography after in situ hybridisation with 35S-labelled riboprobe, is shown at low power in dark-field photographs at PO (B; A is control incubated with sense probe), P5 (C), P12 (D) and P21 (E); the base of the pons is at the bottom of each photograph. Grains over individual cells are shown in bnght-field photographs for PO (F, these cells located dorsally in pontine nucleus), P5 (G) and P12 (H). Thy-1 protein is shown by immunoperoxidase labelling at P5 (I; high power DIC optics in J show labelling of surface membrane of soma and dendrite of a cell m 1), P8 (K) and P12 (L) Scale bars are 100μm in A (B-E same magnification), I,K and L, 10/on in F (G,H same), and 5μm in J.

Fig. 3.

Developmental appearance of Thy-1 mRNA (A-H) and protein (I-L) in the mouse pons. Thy-1 mRNA, detected by autoradiography after in situ hybridisation with 35S-labelled riboprobe, is shown at low power in dark-field photographs at PO (B; A is control incubated with sense probe), P5 (C), P12 (D) and P21 (E); the base of the pons is at the bottom of each photograph. Grains over individual cells are shown in bnght-field photographs for PO (F, these cells located dorsally in pontine nucleus), P5 (G) and P12 (H). Thy-1 protein is shown by immunoperoxidase labelling at P5 (I; high power DIC optics in J show labelling of surface membrane of soma and dendrite of a cell m 1), P8 (K) and P12 (L) Scale bars are 100μm in A (B-E same magnification), I,K and L, 10/on in F (G,H same), and 5μm in J.

The pontine axons (and other mossy fibres) form distinctively large synaptic terminals, called rosettes, readily seen in Thy-1 immunohistochemistry of adult cerebellum as intensely labelled structures interspersed within the granule cell layer (Fig. 4J-L). During postnatal development of lobule VII, the first, very weak, Thy-1 immunoreaction appeared at P5 on a few Purkinje cells (Fig. 4A, arrow). At this age, there was still no real internal granule layer, although medially coursing axons in this area were numerous and heavily stained with G10 and GAP43 antibodies (Fig. 4B,C). By P12, there was a well-developed internal granule layer, in which the characteristically large spaces of the mossy fibre terminals between granule cells were devoid of Thy-1 immunolabelling (Fig. 4D, arrowheads). These still reacted with G10 antibody (Fig. 4E), although this staining was already decreasing in intensity and by P14 had disappeared entirely (Fig. 4H; labelling of the parallel fibres also decreased markedly over these two days, remaining only in the most superficial (immature) axons at P14 in vermis) At P15, very faint Thy-1 immunolabelling became apparent in these rosettes (Fig. 4G); this increased in intensity by P21, and was nearly at the adult level at P28 (Fig. 4J-L). GAP43 staining of the rosettes remained strong at P14 (Fig. 41) and even at P28 (not shown).

Fig. 4.

Maturation of mossy fibre terminals in the granule layer of the cerebellum, as seen with Thy-1 (A,D,G,J-M), GIO (B,E,H) and GAP43 (C,F,I) immunoperoxidase labelling of mouse lobule IX (M) and VII (others) at P5 (A-C), P12 (D-F), P15 (G), P14 (H,I), P21 (J), P28 (K) and P56 (L,M). Open arrow in A points to 2 Purkinje cells just becoming Thy-1 positive; arrowheads in other Thy-1-stained sections point to examples of synaptic rosettes, the (presumptive) myelinated fibre layer is denoted with an asterisk. Scale bar is 50μm.

Fig. 4.

Maturation of mossy fibre terminals in the granule layer of the cerebellum, as seen with Thy-1 (A,D,G,J-M), GIO (B,E,H) and GAP43 (C,F,I) immunoperoxidase labelling of mouse lobule IX (M) and VII (others) at P5 (A-C), P12 (D-F), P15 (G), P14 (H,I), P21 (J), P28 (K) and P56 (L,M). Open arrow in A points to 2 Purkinje cells just becoming Thy-1 positive; arrowheads in other Thy-1-stained sections point to examples of synaptic rosettes, the (presumptive) myelinated fibre layer is denoted with an asterisk. Scale bar is 50μm.

If these terminal sprouts of the pontine axons become Thy-1 immunoreactive a full week after their cell bodies do, what of their main shafts which grew to the cerebellar cortex in the first postnatal week? Their distal extent lies in the (at this stage presumptive) myelinated fibre layer of lobule VII, where they constitute 70% of the axons (Palkovits et al. 1972; Ito, 1984). The other components are the climbing fibres (at this stage Thy-1 negative, Morris et al. 1985b) and Purkinje cell axons. The latter became Thy-1 positive at the same time as the cell bodies and could be seen as the occasional Thy-l-labelled fibre in the myelinated fibre layer (eg Fig. 4D). The great majority of axons of the myelinated fibre layer became Thy-1 immunoreactive at the same time as the mossy fibre rosettes (Fig. 4J-L), suggesting that Thy-1 appears simultaneously on both the terminal region of the axon and its more proximal extent in the cerebellar cortex.

The pontine axons cannot be unambiguously identified more proximally in the peduncle, but nearer their cell bodies in the brainstem they cluster in laterally orientated fascicles and form a distinctive tract under the pial surface (brachium pontis; Fig. 2B). Their molecular maturation can be assessed by inspecting the superficial transverse fibres in the pons (Fig. 5). The deeper-lying fascicles started to become Thy-1 immunoreactive at P12, the more superficial by P21. GIO immunolabelling was already declining at P12, although even at P21 (but not P28) individual axons within the fascicles remained strongly immunoreactive (Fig. 5L). GAP43 antibodies labelled most of the pontine neuropil (including dendrites and axon fascicles) over this period (Fig. 5C,F,I).

Fig. 5.

Molecular maturation of the fascicles of pontine axons in the mouse assessed with immunoperoxidase labelling for Thy-1 (first column), the GIO epitope (second column) and GAP43 (third column exluding L) Parasagittal sections from animals at P5 (A-C), P12 (D-F), P15 (G-I) and P21 (J-L; hatched area in K is shown at higher power in L where arrowheads indicate GIO immunoreactive axons; arrowheads m A-K point to the pial surface). Examples of fascicles of pontine axons visible after GIO staining shown with arrows, as are fascicles after Thy-1 staining at P12-21. Scale bars are 20μm (A; B-K same) and 5 (L) μm

Fig. 5.

Molecular maturation of the fascicles of pontine axons in the mouse assessed with immunoperoxidase labelling for Thy-1 (first column), the GIO epitope (second column) and GAP43 (third column exluding L) Parasagittal sections from animals at P5 (A-C), P12 (D-F), P15 (G-I) and P21 (J-L; hatched area in K is shown at higher power in L where arrowheads indicate GIO immunoreactive axons; arrowheads m A-K point to the pial surface). Examples of fascicles of pontine axons visible after GIO staining shown with arrows, as are fascicles after Thy-1 staining at P12-21. Scale bars are 20μm (A; B-K same) and 5 (L) μm

Qualitatively the same pattern of Thy-1 acquisition occurred in the rat, but at a faster rate. Thy-1 mRNA appeared on the pontine neurons at E20-P2, and protein at the soma and dendrites at PO-2. Immunoreactivity of the axons, both near the cell bodies and at their terminals in the cerebellum, appeared from P12-21. The loss of GIO immunoreactivity from these axons also occurred earlier in this species, from P8-12.

Projection of vestibular ganglion axons to cerebellar lobule IX

This description of Thy-1 acquisition by the pontine axons in lobule VII applies to the mossy fibres in the other vermal lobules, except that it proceeded faster by 2–3 days in the most caudal and rostral lobules, where the granule cells themselves mature earliest (Altman, 1972). The majority of the mossy fibres of the most caudal lobules, IX and X, arise in the vestibular ganglion (Ito, 1984). The cell bodies and axons of this small ganglion can be identified by parvalbumin immunolabelhng during development (Fig. 6A; Morris et al. 1988).

Fig. 6.

Acquisition of Thy-1 by vestibular ganglion neurons and their axons. (A,B,B′) E20 rat showing vestibular ganglion (arrowed, brain stem is immediately below the area photographed) and its axons projecting peripherally to the ampulla (right of photograph), immunolabelled with antibodies to parvalbumin (A), polyclonal antibodies to Thy-1 (B) and monoclonal 0X7 antibodies to Thy-1.1 (B′). Even at low power ganglion cells and their axons can be seen strongly labelled m B, but not in B′. (C,D) Mouse vestibular ganglion at P6 (C) and P56 (D) labelled with polyclonal anti-Thy-1 antibodies; arrows indicate longitudinal profiles of axons (E,F) Mouse ganglion at P15, immunolabelled with monoclonal anti-Thy-1.2 (E) and polyclonal anti-Thy-1 (F) antibodies, axons (now myelinated) cut in cross-section. (G,H) Mouse P21 immunolabelled with the monoclonal anti-Thy-1.2 antibodies, showing m G the longitudinal profile of a labelled axon passing a poorly labelled vestibular neuronal cell body, and in H patches of immunolabelhng clustered at the axonal poles of the cells (arrows). (I,J) Rat ganglion at P2 (I) and P5 (J) immunolabelled with monoclonal anti-Thy-1 1 antibodies Arrow m I points to a neuron with weak, diffuse cytoplasmic labelling, and unstained axon and somatic surface membrane, typical of cells labelled with the monoclonal antibody at E18-20; other cells at this age are starting to show labelling of their axon and somatic surface membrane (K) Rat ganglion at P56, polyclonal anti-Thy-1 labelling. (L-N) Centrally directed rat vestibular axons at P2, immunolabelled with monoclonal anti-Thy-1.1 (L; arrow points to the single immunolabelled axon in these fascicles), polyclonal anti-Thy-1 (M) and anti-parvalbumm (N) antibodies. Scale bars are 250 μm in A (B same), and 10 μm m C (D,F,I-N same) and E (G,H same).

Fig. 6.

Acquisition of Thy-1 by vestibular ganglion neurons and their axons. (A,B,B′) E20 rat showing vestibular ganglion (arrowed, brain stem is immediately below the area photographed) and its axons projecting peripherally to the ampulla (right of photograph), immunolabelled with antibodies to parvalbumin (A), polyclonal antibodies to Thy-1 (B) and monoclonal 0X7 antibodies to Thy-1.1 (B′). Even at low power ganglion cells and their axons can be seen strongly labelled m B, but not in B′. (C,D) Mouse vestibular ganglion at P6 (C) and P56 (D) labelled with polyclonal anti-Thy-1 antibodies; arrows indicate longitudinal profiles of axons (E,F) Mouse ganglion at P15, immunolabelled with monoclonal anti-Thy-1.2 (E) and polyclonal anti-Thy-1 (F) antibodies, axons (now myelinated) cut in cross-section. (G,H) Mouse P21 immunolabelled with the monoclonal anti-Thy-1.2 antibodies, showing m G the longitudinal profile of a labelled axon passing a poorly labelled vestibular neuronal cell body, and in H patches of immunolabelhng clustered at the axonal poles of the cells (arrows). (I,J) Rat ganglion at P2 (I) and P5 (J) immunolabelled with monoclonal anti-Thy-1 1 antibodies Arrow m I points to a neuron with weak, diffuse cytoplasmic labelling, and unstained axon and somatic surface membrane, typical of cells labelled with the monoclonal antibody at E18-20; other cells at this age are starting to show labelling of their axon and somatic surface membrane (K) Rat ganglion at P56, polyclonal anti-Thy-1 labelling. (L-N) Centrally directed rat vestibular axons at P2, immunolabelled with monoclonal anti-Thy-1.1 (L; arrow points to the single immunolabelled axon in these fascicles), polyclonal anti-Thy-1 (M) and anti-parvalbumm (N) antibodies. Scale bars are 250 μm in A (B same), and 10 μm m C (D,F,I-N same) and E (G,H same).

Vestibular ganglion neurons in the mouse showed detectable Thy-1 mRNA at E13 (Fig. 7D) and, by E15 (Fig. 7C), all cells gave a strongly positive signal which had not increased noticeably by birth (Fig. 7A,B). GIO immunoreactivity of vestibular axons declined postnatally, and had essentially disappeared at P6 (not shown) when Thy-1 immunolabelhng was first seen on about 50% of these neurons. The actual staining pattern differed according to whether the polyclonal anti-Thy-1, or monoclonal anti-Thy-1.2, antibodies were used. With the polyclonal, some cell bodies and adjacent axons were positive at P6 (Fig. 6C; adult staining shown in Fig. 6D), six days before their terminals in the cerebellum By P8, 50–80 % of the cells were immunolabelled reasonably strongly, as were their centrally and peripherally directed axons. The proportion of labelled cells and axons increased thereafter: by P15 greater than 90% of the cells were Thy-1 positive, by P21 only very occasional cells were unlabelled; from P28 unlabelled cells or axons were not seen. When staining first appeared, it was predominantly on the surface membrane of soma and axon, and in a granular ring at the periphery of the somatic cytoplasm; cytoplasmic staining then spread more generally around the cell body and increased notably in intensity (Fig. 6C,D). From P28, all ganglionic neurons labelled with similar intensity.

Fig. 7.

Appearance of Thy-1 mRNA in vestibular ganglion and hippocampus. (A-D) Mouse vestibular ganglion, showing tn situ hybridization signa! in darkfield at PO (A, light area at base of photo is bone), and at higher power in bright field at PO (B), E15 (C) and E13 (D) (E-J) Dark-field photos of mouse (E,G,I) and rat (F,H,J) hippocampi at PO (E,F; pyramidal cells in layer marked Py), P5 (G,H) and P21 (I,J; DG, dentate granule cells; CA3, pyramidal field CA3). (K,L) Rat dentate granule cells, bnght-field, at P5 (K) and P12 (L), showing the progression across the layer of mRNA-expressing cells Scale bars are 50μm (A, E (F same), and G (H-J same) and 10μm for the bnght-field photographs.

Fig. 7.

Appearance of Thy-1 mRNA in vestibular ganglion and hippocampus. (A-D) Mouse vestibular ganglion, showing tn situ hybridization signa! in darkfield at PO (A, light area at base of photo is bone), and at higher power in bright field at PO (B), E15 (C) and E13 (D) (E-J) Dark-field photos of mouse (E,G,I) and rat (F,H,J) hippocampi at PO (E,F; pyramidal cells in layer marked Py), P5 (G,H) and P21 (I,J; DG, dentate granule cells; CA3, pyramidal field CA3). (K,L) Rat dentate granule cells, bnght-field, at P5 (K) and P12 (L), showing the progression across the layer of mRNA-expressing cells Scale bars are 50μm (A, E (F same), and G (H-J same) and 10μm for the bnght-field photographs.

The monoclonal anti-Thy-1.2 antibody gave a different picture. Part of this difference was simply quantitative - the polyclonal antibodies gave stronger staining. In addition, there was an important qualitative difference. At the earliest ages (P6-8), the monoclonal antibody gave very weak, diffuse labelling of somatic cytoplasm. Thereafter (including in the adult), it labelled axonal Thy-1 and a minor component of the cytoplasmic Thy-1, but failed to label the bulk of the cytoplasmic Thy-1 or that of the surface of the cell body (Fig. 6E-H). Where immunolabelling of the vestibular cell bodies was apparent, it was usually polarised towards the axon initial segment (Fig. 6H).

In the rat, ganglionic neurons first displayed Thy-1 immunolabelling with the polyclonal antibody at E18, and more strongly so at E20 (Fig. 6B). Labelling was of the axonal and somatic plasma membranes, with some additional cytoplasmic labelling. The axons still stained with the GIO antibody at this time, but this immunoreactivity decfined sharply after birth. Thy-1 immunolabelhng, however, spread to include virtually all ganglionic neurons and their axons at P2, and by P5 the acquisition of Thy-1 was complete. Cytoplasmic labelling increased in intensity relative to the surface membrane at P2-5, and remained a dominant feature of somatic staining in the adult (Fig. 6I-K). Immunolabelling with the monoclonal antibody gave a more complex picture only for the first few days of its appearance. At E18-20, weak cytoplasmic labelling was seen, but neither the somatic surface nor the axons were labelled (Fig. 6B and B′). At P2, some cells and axons showed labelling of their surface membrane (Fig. 61,L-N), and by P5 (Fig. 6J) all cells showed immunolabelfing which was indistinguishable from that of the polyclonal antibodies - strong surface and cytoplasmic labelling of somata, and of the axonal surface. The rosettes in the cerebellar granular layer of lobules IX and X were weakly Thy-1 labelled (with either antibody) at P12. The intensity of labelling increased in the third week of life, and by P21 was comparable to that of proximal regions of the axons.

The discrepancy in staining patterns between the polyclonal and monoclonal antibodies (also seen with the sensory neurons of the nearby trigeminal ganglion) requires the specificity of their reaction with the sections to be demonstrated. This was done in three ways. Increasing dilutions of pure Thy-1 (Fig. 8A, lane 1) were preincubated with the polyclonal antibodies before they were applied to sections of mouse vestibular ganglion. Complete inhibition was observed down to very low levels of Thy-1 (15 ng ml−1, or 1.2 nwt), occurring equally on the cell bodies and axons (Fig. 8B,C). The second test was based on the fact that Thy-1 is anchored to the membrane by a phosphatidylinositol tail (Homans et al. 1988) and can be selectively removed by phosphatidylinositol-specific phospholipase C (PI-PLC; Low and Kincade, 1985). Treatment of unfixed sections of mouse vestibular ganglion with PI-PLC removed immunoreactivity for both the monoclonal (not shown) and polyclonal anti-Thy-1 antibodies (Fig. 8D, upper panels) without affecting staining (in this myelinated peripheral nerve, of the Schwann cell basement membrane (Martini and Schachner, 1986)) of another glycoprotein, N-CAM (Fig. 8D, lower panels). Finally, immunoblot analysis revealed no extra bands detected by the polyclonal antibody on this material (Fig. 8A, lanes 2-9).

Fig. 8.

Specificity of immunological reaction with vestibular ganglion. (A) SDS-PAGE analysis of pure Thy-1 (lane 1, Comassie blue staining of 12 μg of protein) and of immunoblots of adult rat pons (lanes 2,4) and vestibular ganglion (lane 3) labelled with polyclonal anti-Thy-1 (lanes 3,4) and mouse monoclonal O×7 anti-Thy-1.1 (lane 2) antibodies; lanes 5-9 are immunoblots of mouse vestibular ganglion (P8, lane 6; P56, lane 5) and pons (P8, lanes 7 and 8; P56, lane 9) labelled with rat monoclonal 30H12 anti-Thy-1 2 (lane 7) and polyclonal anti-Thy-1 antibodies (lanes 5,6,8,9) For the immunoblottmg, 25 μg of protein was loaded per well, longer exposure times of the autoradiographic film have been used to bring up the weak bands. (B,C) P12 mouse vestibular ganglion cells (column B) and axons (column C) immunolabelled with polyclonal anti-Thy-1 antibodies, preincubated with 0, 7.5, 15 and 30ngml-1 of Thy-1 Very weak brown labelling was evident at 7.5 ng ml−1 of inhibiting Thy-1, with just a trace remaining at 15ngml-1. At 30ngml-1 the sections were devoid of immunoperoxidase reaction. Nuclei were not counterstained on any of the sections used in this figure. Scale bar is 20μm, applies throughout. (D) P21 mouse vestibular ganglion, unfixed cryostat sections, preincubated for 15 min with 0 or 150mlJ of PI-PLC as indicated, immunolabelled with polyclonal anti-Thy-1 antibodies (two upper photos of cell bodies) or anti-N-CAM (two lower photos of cross-section of axons)

Fig. 8.

Specificity of immunological reaction with vestibular ganglion. (A) SDS-PAGE analysis of pure Thy-1 (lane 1, Comassie blue staining of 12 μg of protein) and of immunoblots of adult rat pons (lanes 2,4) and vestibular ganglion (lane 3) labelled with polyclonal anti-Thy-1 (lanes 3,4) and mouse monoclonal O×7 anti-Thy-1.1 (lane 2) antibodies; lanes 5-9 are immunoblots of mouse vestibular ganglion (P8, lane 6; P56, lane 5) and pons (P8, lanes 7 and 8; P56, lane 9) labelled with rat monoclonal 30H12 anti-Thy-1 2 (lane 7) and polyclonal anti-Thy-1 antibodies (lanes 5,6,8,9) For the immunoblottmg, 25 μg of protein was loaded per well, longer exposure times of the autoradiographic film have been used to bring up the weak bands. (B,C) P12 mouse vestibular ganglion cells (column B) and axons (column C) immunolabelled with polyclonal anti-Thy-1 antibodies, preincubated with 0, 7.5, 15 and 30ngml-1 of Thy-1 Very weak brown labelling was evident at 7.5 ng ml−1 of inhibiting Thy-1, with just a trace remaining at 15ngml-1. At 30ngml-1 the sections were devoid of immunoperoxidase reaction. Nuclei were not counterstained on any of the sections used in this figure. Scale bar is 20μm, applies throughout. (D) P21 mouse vestibular ganglion, unfixed cryostat sections, preincubated for 15 min with 0 or 150mlJ of PI-PLC as indicated, immunolabelled with polyclonal anti-Thy-1 antibodies (two upper photos of cell bodies) or anti-N-CAM (two lower photos of cross-section of axons)

Thy-1 expression during hippocampal axogenesis

Pyramidal neurons first showed detectable signal for Thy-1 mRNA after they had migrated to their definitive position and started to form their layer, at E16-17 in the mouse and E17 in the rat. The mRNA signal remained similar in intensity in the two species (Fig. 7E-J), and in the different hippocampal fields, during postnatal development.

The early immunohistochemical staining pattern of the hippocampal area was dominated by the presence of the entorhinal afferents, which in the mouse were immunolabelled for Thy-1 from PO (Fig. 9A), and fibres from the subiculum which were Thy-1 positive from P5. The pyramidal neurons had already grown an extensive apical dendrite at PO (Fig. 9A), and in addition basal dendrites by P5-8, when the vast majority of these, cells were entirely Thy-1 negative in the mouse. The first Thy-1-positive cells were seen at P5 in lateral hippocampus (Fig. 9B; we have used silvergold enhancement of the peroxidase product to demonstrate this) on the somatic and dendritic surface membranes of cells located on the basal (alvear) side of the layer. Some superficially located cells remained negative at P12 (Fig. 9C), although all appeared labelled at P14. No difference was apparent in the rate of Thy-1 acquisition by fields CAI or CA3.

Fig. 9.

Development of Thy-1 immunoreactivity on hippocampal pyramid and granule cells, and their axons (A) Mouse hippocampus at PO, Thy-1 immunolabelling enhanced with silver-gold development labels (a dense black) only the entorhinal axons terminating on the distal ends of the pyramidal dendrites at the top of the photo, the course of some dendrites can be seen by virtue of the Nissl counterstain (e.g. arrowhead). (B) Mouse CAI pyramidal cells at P5, Thy-1 immunolabelhng with silver-gold enhancement, two cells at base of layer (arrowhead) have labelled apical dendrites, most pyramidal cells are unlabelled at this age Immunolabelling above and below the pyramidal layer is due predominantly to afferent fibres, with contributions from branching dendrites of the few Thy-1 positive cells (C) Mouse Thy-1 immunolabelhng (without enhancement) of CAI pyramids at P12, not all cells in layer are yet Thy-1 positive. (D) Rat CA3 pyramids at P8, most have strongly Thy-l-labelled somata and dendrites (no enhancement) (E,F) Adjacent coronal sections (phase-contrast photography) of mouse at P8 showing the area of the ventral hippocampal commissure (vhc, corpus callosum (cc) and longitudinal fibres of the dorsal fornix (df) are also shown) labelled for Thy-1 (enhanced immunohistochemistry, only occasional axons cut in cross-section in the fornix are positive) and GIO (F). (G,H) Thy-1 labelling of dentate granule cells at P12 in the mouse (G; arrowhead shows immunolabelled dendrites of a Thy-1-positive cell on the superficial face of the layer, most cells are unlabelled) and rat (H; most cells labelled). (I-L) Mouse P8 coronal sections showing CA3 (I,J) and CAI (K,L) pyramidal fields labelled for Thy-1 (I,K; these sections have been silver-gold enhanced) and GIO (J,L). The granule cell axons grow above the somata of CA3 (stratum lucidum, demarced by arrowheads) but not CAI pyramids; so, stratum oriens; sm, stratum moleculare; sr, stratum radiatum, containing the entorhinal input Scale bars are 10μm (A-D, G,H) and 50μm (E,F, I-L).

Fig. 9.

Development of Thy-1 immunoreactivity on hippocampal pyramid and granule cells, and their axons (A) Mouse hippocampus at PO, Thy-1 immunolabelling enhanced with silver-gold development labels (a dense black) only the entorhinal axons terminating on the distal ends of the pyramidal dendrites at the top of the photo, the course of some dendrites can be seen by virtue of the Nissl counterstain (e.g. arrowhead). (B) Mouse CAI pyramidal cells at P5, Thy-1 immunolabelhng with silver-gold enhancement, two cells at base of layer (arrowhead) have labelled apical dendrites, most pyramidal cells are unlabelled at this age Immunolabelling above and below the pyramidal layer is due predominantly to afferent fibres, with contributions from branching dendrites of the few Thy-1 positive cells (C) Mouse Thy-1 immunolabelhng (without enhancement) of CAI pyramids at P12, not all cells in layer are yet Thy-1 positive. (D) Rat CA3 pyramids at P8, most have strongly Thy-l-labelled somata and dendrites (no enhancement) (E,F) Adjacent coronal sections (phase-contrast photography) of mouse at P8 showing the area of the ventral hippocampal commissure (vhc, corpus callosum (cc) and longitudinal fibres of the dorsal fornix (df) are also shown) labelled for Thy-1 (enhanced immunohistochemistry, only occasional axons cut in cross-section in the fornix are positive) and GIO (F). (G,H) Thy-1 labelling of dentate granule cells at P12 in the mouse (G; arrowhead shows immunolabelled dendrites of a Thy-1-positive cell on the superficial face of the layer, most cells are unlabelled) and rat (H; most cells labelled). (I-L) Mouse P8 coronal sections showing CA3 (I,J) and CAI (K,L) pyramidal fields labelled for Thy-1 (I,K; these sections have been silver-gold enhanced) and GIO (J,L). The granule cell axons grow above the somata of CA3 (stratum lucidum, demarced by arrowheads) but not CAI pyramids; so, stratum oriens; sm, stratum moleculare; sr, stratum radiatum, containing the entorhinal input Scale bars are 10μm (A-D, G,H) and 50μm (E,F, I-L).

Pyramidal axons cannot be seen clearly in Thy-1 immunohistochemistry at their proximal or distal ends, due to the abundance of other Thy-1-positive elements. However, they are major components of the axonal tract, the fimbria, where the first few Thy-l-immunoreactive fibres appeared at P5-8, although the majority did not acquire Thy-1 until the second and third postnatal week. (Fibres from the septum and early maturing subiculum also run in this tract, and the subicular fibres almost certainly constitute the earliest labelled axons seen here). The axons of the ventral hippocampal fissure were entirely Thy-1 negative for the first postnatal week (Fig. 9E) when they strongly immunolabelled with the GIO (Fig. 9F) and GAP43 (not shown) antibodies. The intensity of GIO immunolabelling of the commissure decfined towards the end of the second week when, at P12-15, Thy-1 immunolabelling first became apparent. The area of immunoreaction had spread to the whole tract, and increased in intensity to the adult level, by P21.

Hippocampal granule cells first showed detectable Thy-1 mRNA at P5 in both species, on the superficially located cells of the suprapyramidal blade. This spread in a gradient of maturation, both around the layer to the infrapyramidal blade, and across the layer to the deeper (hilar) face (Fig. 7K,L), being complete at P21. Within the layer, the least mature cells (small nucleus staining dark blue with the Nissl counterstain (Gaarskjaer, 1985)) were unlabelled for Thy-1 mRNA. As with pyramidal cells, substantial dendritic growth could be seen by the Nissl counterstain to have occurred before Thy-1 immunolabelling (of both soma and dendrites) appeared. This labelling followed a similar gradient, starting in the most mature regions at P5 and more generally by P12 (Fig. 9G) and being complete by P28. The granule axons run in a distinctive layer above the cell bodies of the CA3, but not CAI, pyramids. At P8 in the mouse these axons strongly labelled with GIO antibody, and conversely failed to be labelled with Thy-1 antibodies (Fig. 9I-L; the reciprocal staining by Thy-1 and GIO antibodies is particularly evident in the hippocampal layers). GIO labelling of granule axons declined in the third postnatal week, and the axons could be seen in coronal sections to be Thy-1 immunolabelled by P21.

As in hindbrain, in hippocampus also the expression of rat Thy-1, and loss of GIO immunoreactivity, followed a similar pattern, but with a faster tempo (eg Fig. 9D,H), than in the mouse. Thus pyramidal neurons became Thy-1 positive at PO-2, their axons in the fimbria and ventral hippocampal fissure did so at P8-21. The granule cells were immunolabelled over the period P5-21, their axons above the CA3 pyramids acquired immunoreactivity over P5-21.

The pattern of Thy-1 expression in the four sets of neurons examined in this work can be summarised in five simple statements (see also Fig. 10) which apply equally to the rat and mouse.

Fig. 10.

Summary diagram showing onset of expression of Thy-1 mRNA (filled in nucleus denotes expression) and protein (heavy line denotes expression of Thy-1 at that region of the cell, dashed heavy lines indicates the postulated presence of Thy-1 on the course of the vestibular axons in the brainstem) for the pontine and vestibular ganglion axons. The dates shown indicate the relevant period in the mouse, VN, the 3 vestibular nuclei in the brainstem; Cb, cerebellum.

Fig. 10.

Summary diagram showing onset of expression of Thy-1 mRNA (filled in nucleus denotes expression) and protein (heavy line denotes expression of Thy-1 at that region of the cell, dashed heavy lines indicates the postulated presence of Thy-1 on the course of the vestibular axons in the brainstem) for the pontine and vestibular ganglion axons. The dates shown indicate the relevant period in the mouse, VN, the 3 vestibular nuclei in the brainstem; Cb, cerebellum.

(1) Thy-1 mRNA expression is determined by somatic differentiation

The early developmental history of these neurons is quite diverse: the hippocampal pyramids migrate from primary germinal neuroepithelium of the lateral ventricle (Bayer, 1980); most of the hippocampal granule cells are generated in situ from a secondary neuroepithelium without migration (Cowan et al. 1981); the pontine neurons migrate from a secondary, precerebellar neuroepithelium (Altman and Bayer, 1987); and the vestibular ganglion neurons are not even derived from neural tube, but are placodal in origin (D’Amico-Martel and Noden, 1983; Altman and Bayer, 1982). Yet these, like the mitral cells of the olfactory bulb (Xue et al. 1990), all express Thy-1 mRNA at the same stage of development, when they have finished migration and begun dendritic growth (Fig. 10). This is a very basic developmental stage for neurons, and can be seen morphologically as an enlargement of both soma and nucleus, with the latter becoming round and muchlighter stained with thionin. Part of this general differentiation of the neuronal soma involves the expression of Thy-1 mRNA.

(2) mRNA expression does not by itself lead to appearance of Thy-1 protein

For rat pontine neurons, or dentate granule cells in either species, appearance of Thy-1 mRNA and protein were closely linked, protein following its message within 24–48 h. However, for mouse vestibular ganglion neurons there was a delay of 2 weeks, even though the level of Thy-1 mRNA signal for most of this period was substantially higher than on the mouse pontine neurons when they were expressing Thy-1 protein (c.f. Fig. 7B,C with Fig. 3G,H). Rat and mouse hippocampal pyramids showed similar levels of mRNA signal, followed in 5 days in the rat, and 10 days in the mouse, by appearance of protein. Taken together with similar observations in the olfactory system (Xue et al. 1990), it is clear that some signal in addition to expression of its mRNA is required for expression of Thy-1 protein.

(3) Initial expression of Thy-1 protein is restricted to the dendritic compartment and follows cessation of primary axon elongation

The soma and dendrites of the pontine neurons became Thy-1 immunoreactive a full week before their axons immunolabelled (Fig. 10). A similar delay was evident in the case of the hippocampal pyramids and granule cells.

Although Thy-1 on its first appearance was restricted to the dendrite, there is no correlation between its time of appearance and dendritic growth. On some neurons (e.g. rat Purkinje cells, Morris et al. 1985a) Thy-1 is present at all stages on all developing dendrites; for olfactory mitral cells, Thy-1 is present on part of the growing dendritic tree (Xue et al. 1990); for hippocampal pyramidal cells, dendritic growth occurs for over a week, during which they receive their mam afferent inputs (Zimmer and Haug, 1978), before Thy-1 is expressed.

However, appearance of Thy-1 protein closely follows completion of the initial phase of axonal growth (except for vestibular neurons, see below). In the rat, there is very little delay between the period of primary axon elongation and the appearance of Thy-1 protein. Rat pontine axons, for instance, reach the cerebellar cortex by birth (Payne and Bower, 1988); their cell bodies and dendrites become Thy-1 positive from P0 to P2. In the mouse, Thy-1 protein was detected on the cells studied 3–6 days later than in the rat (taking birth as the reference point) However, GIO immunoreactivity also disappeared from mouse axons a few days later than in the rat. Whether mouse axons actually grow more slowly (or start later), or the neurons respond molecularly more slowly to the completion of growth, is unclear. (The 2 day longer gestation of the rat could obviously contribute to, but not entirely account for, this difference. Most studies of the development of these fibre pathways have been done in the rat and not the mouse, so direct comparison of the timing of axonal growth in the two species is not possible).

A temporal relationship between completion of the initial phase of axon elongation, and appearance of Thy-1 protein, does not imply a causal relationship between the two, although it is worth noting an established mechanism exists whereby this could occur, when an axon reaches its terminal zone, further molecular maturation of the neuron is induced by factors such as NGF produced by the terminal field (Barde, 1989). Part of this maturation could be acquisition of competence to accumulate Thy-1 protein in dendrites.

(4) Thy-1 is only allowed into axons after they have completed both their initial elongation, and subsequent terminal sprouting

These two phases of axonal growth are quite distinct, often in timing but also in the signals they respond to and substrates used (see eg Morris et al. 1988; O’Leary and Terashima, 1988; Ghosh et al. 1990). The appearance of Thy-1 on the pontine axons in the third and fourth postnatal weeks follows the formation and maturation of the mossy fibre terminals, which starts in the second week (Altman, 1972; Hamori and Somogyi, 1983; Arsenio Nunes and Sotelo, 1985). Indeed, although we have concentrated upon two of the mossy fibre projections to cerebellum, the point is perhaps best made by considering this group as a whole. They are a diverse group of fibres, arriving in the cerebellum from E12 (vestibular; Morris et al. 1988) to nearly two weeks later (part of the spinocerebellar group; Arsenio Nunes and Sotelo, 1985). Yet all acquire Thy-1 in their terminal sprouts (and the nearby distal region in the myelinated fibre layer) at the same time, dictated by the state of maturation of their targets, the late-developing granule cells, rather than by that of their cells of origin. The late acquisition of Thy-1 by the cerebellar climbing fibres, after they have finished terminal growth (Morris et al. 1985ft), demonstrates the same effect. Although we cannot be as precise in identifying the acquisition of Thy-1 by the hippocampal axons, it is clear that this occurs some days after their cell bodies/dendrites display Thy-1 immunoreactivity, and after the axons have lost GIO immunoreactivity. It therefore seems probable that cessation of terminal growth also occurs with these before Thy-1 appears on their axons.

(5) Where axons grow at different times to multiple targets, Thy-1 is allowed into regions of completed growth, but remains excluded from growing regions

The growth of the vestibular axons into the granule layer of the cerebellum is a relatively small and late burst of terminal sprouting by axons that have principally terminated on the three vestibular nuclei of the brainstem two weeks earlier (Morris et al. 1988). In fact, their initial phase of axonal growth was substantially complete by E14-15 (Morris et al. 1988). The delay (in the rat, of 4–6 days; in the mouse, of more than 10 days) before Thy-1 protein became detectable on these cells was presumably because they lack dendrites, and therefore have no appropriate compartment in which to place Thy-1 until the axon becomes permissive. Nevertheless, the proximal regions of their axons became Thy-1 positive 6–12 days before their terminals in the cerebellum (Fig. 10). These axons follow a diffuse course across the brainstem (Morris et al 1988), and the boundary between the Thy-1-positive and -negative areas cannot be determined. We suspect that within the brainstem they are Thy-1 positive, and only the growing cerebellar region is negative (Fig. 10). Such partitioning of Thy-1 along a shorter process, the growing mitral cell dendrite, has been demonstrated and occurs at defined cellular boundaries (Xue et al. 1990).

Specificity of the anti-Thy-1 antibodies

There was a discrepancy between the staining patterns of the monoclonal and polyclonal antibodies on the vestibular ganglion cells, especially in the mouse. We suggest the polyclonal antibodies reveal the true extent of Thy-1 on these cells, since their reaction is inhibited by pure Thy-1, removed by PI-PLC, and they stain no additional bands in western blots. The failure of the monoclonals to detect some somatic Thy-1 might be relatively trivial in origin. For instance, our current work shows that Thy-1 can be nicked by endogenous tissue proteases (two disulphide bonds hold the molecule together) and these forms are detected by the polyclonal, but not monoclonal, antibodies (B Pliego Rivero, in preparation). If the sensory ganglia have an unusually high concentration of such proteases in their soma, then postmortem nicking of the Thy-1 would produce the observed staining pattern. Alternatively, it may demonstrate the presence of another molecule, intimately associated with Thy-1 at early stages of its biosynthesis, which masks the allele-determining residue 89 (Williams and Gagnon, 1982).

Implications for Thy-1 biosynthesis

In general, newly synthesised cell surface glycoproteins are carried internally to the axon terminal by rapid transport where they are incorporated into the plasma membrane (Grafstein and Forman, 1980; Forman, 1984). This model is compatible with the observed compartmentahsation of certain cell surface glycoproteins on growing axons, as described for instance for TAG-1 and LI on developing commissural axons of rat spinal cord (Dodd et al. 1988), if the proximally located protein (in this case, TAG-1) is transiently expressed at an earlier stage than the distally located protein. However, it clearly cannot explam the situation seen here with Thy-1, where the glycoprotein on its first appearance is specifically excluded from the growing (distal) region of the axon. Given the lengths of the axons studied here, fast axonal transport would deposit Thy-1 at the growing tip within hours of its synthesis at the cell body. Our data require a model that can explain the initial vectorial phase of Thy-1 synthesis, and its developmental switch to include the axon; and a mechanism for allowing Thy-1 into proximal, but not the growing distal, axonal regions.

A key role in the initial vectorial synthesis of Thy-1 almost certainly follows its unusual mode of insertion into the plasma membrane, via a glycophospholipid moiety (Homans et al. 1988). The latter serves in epithelial cells to direct incorporation of such lipid-anchored proteins to the apical (rather than basolateral) surface (Lisanti et al. 1989). Hippocampal pyramidal neurons in culture show the same vectorial synthesis of viral glycoproteins as do epithelial cells, those proteins that are inserted into the apical epithelial surface being directed to the axonal surface in neurons (Dotti and Simons, 1990). These workers find Thy-1 appears only on the axons (and not dendrites) of rat hippocampal pyramids after two weeks in culture (equivalent to P11), so this endogenous glycophospholipid-linked membrane protein shows the same vectorial insertion as the viral glycoproteins (Dotti et al. 1991).

This encoding of axonal insertion of Thy-1 presumably operates in vivo, and its activation could initiate access of Thy-1 to the axonal compartment. The earher direction of Thy-1 to dendrites would presumably occur if, at this stage, newly synthesised Thy-1 retained its amino acid transmembrane tail (which is normally cleaved and replaced within seconds by the lipid anchor; Conzelmann et al. 1987). Such alternate forms of lipid-linked molecules are produced by alternative splicing giving rise to different mRNA species, or by association early in biosynthesis with another protein subunit that protects against cleavage of the original transmembrane polypeptide tail (see Kurosaki and Ravetch, 1989). It would therefore be in keeping with known mechanisms of membrane protein biosynthesis if alternate membrane-anchored forms of Thy-1 resulted in dendritic and axonal insertion. We are currently examining this possibility, although alternative mechanisms, such as differential location of mRNA (as occurs with the dendritic cytoskeletal protein, MAP2 (Gamer et al. 1988), and certain myelin proteins (Campagnoni and Macklin, 1988; Shiota et al. 1989; Gillespie et al. 1990)) are not excluded by our present observations.

Whatever the mechanism that establishes the initial separation of Thy-1 between dendrite and axon, how is this compartmentation maintained over the long time that we have observed? The glycophospholipid tail confers on Thy-1 the higher mobility in the plane of the membrane typical of a lipid (Ishihara et al. 1987), and it would be expected to diffuse along the axon unless some other mechanism acted to retard its progress. Since Thy-1 does not span the membrane, this barrier cannot be directly cytoskeletal; it could be an interaction with some other molecule, perhaps present on glial cells in the environment of more mature axons, as we have suggested to underlie the partitioning seen in growing mitral cell dendrites (Xue et al 1990). If this is correct, then ghal maturation would determine the progression of Thy-1 protein along the axon.

Biological implication of this pattern of Thy-1 expression

Although the molecular mechanisms underlying this biosynthetic partitioning of Thy-1 are unknown, their end result is clearly to exclude Thy-1 from regions of axonal growth. This in vivo observation is particularly interesting in the light of recent experimental evidence indicating that Thy-1 inhibits process outgrowth by neural cell lines (Mahanthappa and Patterson, 1989) especially on astrocytes (Morris et al. 1990). If this is a model for axonal (rather than dendritic) growth, then Thy-1 would be inhibitory for axogenesis. Since adult neurons do not prohibit Thy-1 from their axons, it will be interesting to assess whether this molecule is present on, and contributes to the failure of growth by, the abortive sprouts produced by lesioned axons in the astrocytic environment of adult CNS.

It is a pleasure to thank Carlos Dotti and Kai Simons for communicating results prior to publication, Rosie Calvert for GIO antibody, Pate Skene and David Schreyer for antiGAP43 antibody 91E12, Claus Heizmann for the anti-parvalbumin antibodies, Pat Doherty for the anti-N-CAM antibodies, Joe Brock for the drawings, and helpful discussions with Geoff Raisman, Frank Grosveld and Eugenia Spanopoulou.

Adams
,
C. E.
,
Parna Velas
,
J G
,
Mihailoff
,
G A
and
Woodward
,
D J
(
1980
)
The neurons and their postnatal development in the basilar pontine nuclei of the rat
Brain Res Bull
5
,
277
283
Altman
,
J
(
1972
)
Postnatal development of the cerebellar cortex in the rat. in. Maturation of components of the granular layer
J comp Neuro!
145
,
465
514
Altman
,
J
and
Bayer
,
S A
(
1982
)
Development of the cranial nerve ganglia and related nuclei in the rat
Adv Anal Embryol Cell Biol
74
,
1
90
.
Altman
,
J
and
Bayer
,
S A
(
1987
).
Development of the precerebellar nuclei in the rat IV The anterior precerebellar extramural migratory stream and the nucleus reticularis tegmenti pontis and the basal pontine grey
J comp Neurol
257
,
529
552
Amaral
,
D
and
Dent
,
J A
(
1981
)
Development of the mossy fibers of the dentate gyrus I A light and electron microscopic study of the mossy fibres and of their expansions
J comp Neurol
195
,
51
86
Arsenio Nunes
,
M L
and
Sotelo
,
C
(
1985
)
Development of the spinocerebellar system in the postnatal rat-
J comp Neurol
.
237
,
291
306
Ballantyne
,
J
and
Engstrom
,
H
(
1969
)
Morphology of the vestibular ganglion cells
J Laryngol Otol
83
,
19
42
Barde
,
Y -A
(
1989
)
Trophic factors and neuronal survival
Neuron
2
,
1525
1534
Batini
,
C
,
Buisseret-Delmas
,
C
,
Corvisier
,
J
,
Hardy
,
O
and
Jassik-Gerschenfeld
,
D
(
1978
)
Brain stem nuclei giving fibers to lobules VI and VII of the cerebellar vermis
Brain Res
153
,
241
261
Bayer
,
S A
(
1980
)
Development of the hippocampal region in the rat II Morphogenesis during embryonic and early postnatal life
.
J. comp Neurol
190
,
115
134
Buss
,
T V. P.
,
Chung
,
S H
and
Stirung
,
R V
(
1974
)
Structural and functional development of the mossy fibre system in the hippocampus of the post-natal rat
.
J Physiol, Lond
139
,
92P
94P
Bolin
,
L M
and
Rouse
,
R V
(
1986
)
Localisation of Thy-1 expression during postnatal development of the mouse cerebellar cortex
J Neurocytol
15
,
29
36
Brodal
,
A
and
Hoivik
,
B
(
1964
)
Site and mode of termination of primary vestibulo-cerebellar fibers in the cat An experimental study with silver impregnation methods
.
Arch ital Biol
102
,
1
21
Buchhalter
,
J R
,
Fieles
,
A
and
Dichter
,
M. A
(
1990
)
Hippocamal commissural connections in the neonatal rat
Dev Brain Res
56
,
211
216
Calvert
,
R
and
Anderton
,
B H.
(
1985
)
A microtubule-associated protein (MAPI), which is expressed at elevated levels during development of the rat cerebellum
EMBO J
4
,
1171
1176
Calvert
,
R. A
,
Woodhams
,
P L
and
Anderton
,
B. H.
(
1987
)
Localization of an epitope of a microtubule-associated protein lx in outgrowing axons of the developing rat central nervous system
Neuroscience
23
,
131
141
.
Campagnoni
,
A T
and
Mackun
,
W B.
(
1988
).
Cellular and molecular aspects of myelin protein gene expression
Mol Neurobiol
2
,
41
89
Conzelmann
,
A
,
Spiazzi
,
A
and
Bron
,
C
(
1987
)
Glycolipid anchors are attached to Thy-1 glycoprotein rapidly after translation
Biochem. J
246
,
605
610
Cowan
,
W M
,
Stanfield
,
B B
and
Amaral
,
D G
(
1981
)
Further observations on the development of the dentate gyrus
. In
Studies tn Developmental Neurobiology
(ed
W M
Cowan
), pp
395
—435 Oxford Oxford University Press
D’amico-martel
,
A.
and
Noden
,
D M
(
1983
)
Contributions of placodal and neural crest cells to avian cranial peripheral ganglia
Am J Anat
166
,
445
468
Dodd
,
J
,
Morton
,
S B
,
Karagogeos
,
D
,
Yamamoto
,
M
and
Jessel
,
T M
(
1988
)
Spanal regulation of axonal glycoprotein expression on subsets of embryonic spinal neurons
Neuron
1
,
105
116
.
Doherty
,
P
,
Mann
,
D A.
and
Walsh
,
F S
(
1988
)
Comparison of the effects of NGF, activators of protein kinase C, and a calcium ionophore on the expression of Thy-1 and N-CAM in PC12 cell cultures
J Cell Biol
107
,
333
340
.
Dorn
,
C G
,
Parton
,
R G
and
Simons
,
K
(
1991
)
Polarized sorting of glypiated proteins in hippocampal neurons
Nature
349
,
158
161
Dotti
,
C G
and
Simons
,
K
(
1990
)
Polarized sorting of viral glycoproteins to the axon and dendrites of hippocampal neurons in culture
Cell
62
,
63
72
Flumerfelt
,
B A
and
Hrycyshyn
,
A W
(
1985
)
Precerebellar nucleus and red nucleus
In
The Rat Nervous System
. Volume
2
Hindbrain and Spinal Cord (ed
G
Paxinos
), pp
221
250 Sydney Academic Press
Forman
,
D S.
(
1984
)
New approaches to the study of the mechanism of fast axonal transport
.
TINS
4
,
112
116
.
Fricke
,
R
and
Cowan
,
W. M
(
1977
)
An autoradiographic study of the development of the entorhinal and commissural afferents to the dentate gyrus of the rat
J comp Neur
173
,
231
250
Frotscher
,
M
,
Kugler
,
P
,
Misgeld
,
U
and
Zilles
,
K
(
1988
)
Neurotransmission in the hippocampus
Adv Anat Embryol Cell Biol
111
,
2
19
Gaarskjaer
,
F B.
(
1985
)
The development of the dentate area and the hippocampal mossy fiber projection of the rat
J. comp Neurol
241
,
154
170
Gacek
,
R R
(
1969
).
The course and central termination of first order neurons supplying vestibular end organs in the cat
.
Acta Oto-Laryngologica Suppl
253
,
1
66
Garner
,
C C
,
Matus
,
A.
,
Anderton
,
B
and
Calvert
,
R
(
1989
)
Microtubule-associated proteins MAP5 and MAPlx closely related components of the neuronal cytoskeleton with different cytoplasmic distributions in the developing brain
Mol Brain Res
5
,
85
92
Garner
,
C G
,
Tucker
,
R P
and
Matus
,
A
(
1988
).
Selective localisation of messenger RNA for cytoskeletal protein MAP2 in dendrites
Nature
336
,
674
677
Ghosh
,
A.
,
Antonini
,
A
,
Mcconnell
,
S K
and
Schatz
,
C J
(
1990
)
Requirement for subplate neurons in the formation of thalamocortical connections
Nature
347
,
179
181
Gillespie
,
C S.
,
Bernier
,
L
,
Brophy
,
P J
and
Colman
,
D R
(
1990
)
Biosynthesis of the myehn 2′,3′-cychc nucleotide 3′-phosphodiesterases
.
J Neurochem
54
,
656
661
Goslin
,
K
,
Schreyer
,
D. J
,
Skene
,
J H P
and
Banker
,
G
(
1988
)
Development of neuronal polarity GAP-43 distinguishes axonal from dendritic growth cones
Nature
336
,
672
674
Grafstein
,
B.
and
Forman
,
D. S
(
1980
)
Intracellular transport in neurons
Physiol Rev
60
,
1167
1283
Hamori
,
J
and
Somogyi
,
J
(
1983
)
Differentiation of cerebellar mossy fiber synapses in the rat a quantitative electron microscope study
J comp Neurol
220
,
365
377
Homans
,
S W
,
Ferguson
,
MAJ
,
Dwek
,
R A
,
Rademacher
,
T W.
,
Anand
,
R.
and
Williams
,
A F
(
1988
)
Complete structure of the glycosyl phosphatidylinositol membrane anchor of rat brain Thy-1 glycoprotein
Nature
333
,
269
272
Ishihara
,
A
,
Hou
,
Y
and
Jacobson
,
K
(
1987
)
The Thy-1 antigen exhibits rapid lateral diffusion in the plasma membrane of rodent lymphoid cells and fibroblasts
.
Proc natn Acad Sci U S.A
84
,
1290
1293
Ito
,
M
(
1984
)
The Cerebellum and Neural Control NY Raven Press
.
Kagi
,
U
,
Berchtold
,
M W.
and
Heizmann
,
C W
(
1987
)
Ca2+-bindmg parvalbumin in the rat testis. Characterization, localization and expression during development
J biol Chem
262
,
7314
7320
.
Kawamura
,
K
and
Hashikawa
,
T
(
1981
).
Projections from the pontine nuclei proper and reticular tegmental nucleus onto the cerebellar cortex in the cat An autoradiographic study
J comp Neurol
201
,
395
413
Kurosaki
,
T.
and
Ravetch
,
J. V
(
1989
)
A single ammo acid in the glycosyl phosphatidylinositol attachment domain determines the membrane topology of Fc RIH
Nature
342
,
805
807
Laemmu
,
U
(
1970
)
Cleavage of structural proteins during the assembly of bacteriophage T4
Nature
227
,
680
681
Lisanti
,
M P.
,
Caras
,
I W
,
Davitz
,
M. A
and
Rodriguez-Boulan
,
E
(
1989
)
A glycophosphohpid membrane anchor acts as an apical targeting signal in polarized epithelial cells
J Cell Biol
109
,
2145
2156
Low
,
M G.
and
Kincade
,
P W
(
1985
)
Phosphatidylinositol is the membrane-anchoring domain of the Thy-1 glycoprotein
Nature
318
,
62
64
Mahanthappa
,
N K
and
Patterson
,
P H
(
1989
)
Thy-1 dimerization, Thy-1 derived peptides and neunte outgrowth
Soc Neurosa Abst
15
,
651
.
Martini
,
R
and
Schachner
,
M
(
1986
)
Immunoelectron microscopic localization of neural cell adhesion molecules (LI, N-CAM, and MAG) and their shared carbohydrate epitope in developing sciatic nerve
J Cell Biol
103
,
2739
2748
Morris
,
R J
(
1985
)
Thy-1 in developing nervous tissue
Dev Neurosa
7
,
133
160
Morris
,
R J
,
Beech
,
J N
,
Barber
,
P C
and
Raisman
,
G
(
1985a
)
Early stages of Purkinje cell maturation demonstrated by Thy-1 immunohistochemistry on postnatal rat cerebellum
J Neurocytol
14
,
427
452
Morris
,
R J
,
Beech
,
J N
,
Barber
,
P C
and
Raisman
,
G
(
1985b
)
Late emergence of Thy-1 on climbing fibres demonstrates a gradient of maturation from the fissures to the fohal convexities in developing rat cerebellum
J Neurocytol
14
,
453
467
Morris
,
R J
,
Beech
,
J N
,
Gormley
,
A M
and
Barboni Clarke
,
E
(
1990
)
Selective inhibition of process outgrowth on astrocytes by neuronal Thy-1
.
Neuroscience Lett
38
,
S72
Morris
,
R J
,
Beech
,
J N
and
Heizmann
,
C W.
(
1988
)
Two distinct phases and-mechanisms of axonal growth shown by primary vestibular fibres in the brain, demonstrated by parvalbumin immunohistochemistry
Neurosa
27
,
571
596
Morris
,
R J
,
Mancini
,
P E
and
Pfeiffer
,
S E
(
1980
)
Thy-1 cell surface antigen on cloned nerve cell lines of the rat and mouse amount, location and origin of the antigen on the cells
Brain Res
182
,
119
135
Morris
,
R J
and
Raisman
,
G
(
1983
).
Estimation of Thy-1 in cryostat sections of nervous tissue
J Neurochem
40
,
637
644
O’leary
,
D D M
and
Terashima
,
T.
(
1988
).
Cortical axons branch to multiple subcortical targets by interstitial axon budding implications for target recognition and ‘waiting periods’
.
Neuron
1
,
901
910
Palkovits
,
M
,
Magyar
,
P
and
Szentagothai
,
J
(
1972
)
Quantitative histological analysis of the cerebellar cortex in the cat. IV. Mossy fibre-Purkinje cell numerical transfer
Brain Res
45
,
15
29
Payne
,
J. N.
and
Bower
,
A J
(
1988
)
Cerebellar afferents in early postnatal rats a retrograde fluorescence study
Dev Brain Res
39
,
313
318
Reiderer
,
B
,
Guadano-Ferraz
,
A.
and
Innocenti
,
G.
(
1990
)
Difference in the distribution of microtubule-associated proteins 5a and 5b during the development of cerebral cortex and corpus callosum in cats dependence on phosphorylation
Dev Brain Res
56
,
235
243
Sato-Yoshitake
,
R.
,
Shiomura
,
Y
,
Miyasaka
,
H
and
Hirokawa
,
N
(
1989
)
Microtubule-associated protein IB molecular structure, localization, and phosphorylation-dependent expression in developing neurons
Neuron
3
,
229
238
.
Shiota
,
C
,
Miura
,
M
and
Mikoshiba
,
K
(
1989
)
Developmental profile and differential localization of mRNAs of myehn proteins (MBP and PLP) in oligodendrocytes in the brain and in culture
Dev Brain Res
.
45
,
83
94
Stirling
,
R V
and
Buss
,
T. V P.
(
1978
)
Hippocampal mossy fibre development at the ultrastructural level
.
Prog Brain Res
48
,
191
198
Swanson
,
L. W.
(
1978
).
The anatomical organisation of septo-hippocampal projections
In
Functions of the Septo-Hippocampal System, Ciba Foundation Symposium
58
,
pp 25
43 Amsterdam Elsevier
Williams
,
A F
and
Gagnon
,
J.
(
1982
)
Is the Thy-l-hke glycoprotein of neuronal cell membranes like the primordial immunoglobulin domain9
Science
216
,
696
703
Woodhams
,
P L
,
Calvert
,
R
and
Dunnett
,
S B
(
1989
)
Monoclonal antibody GIO against microtubule-associated protein lx distinguishes between growing and regenerating axons
Neuroscience
28
,
49
59
.
Xue
,
G P
,
Calvert
,
R
and
Morris
,
R. J.
(
1990
)
Expression of the neuronal surface glycoprotein Thy-1 is under post-transcnptional control, and is spatially regulated, in the developing olfactory system
Development
109
,
851
864
.
Zimmer
,
J.
and
Haug
,
F -M
(
1978
)
Laminar differentiation of the hippocampus, fascia dentata and subiculum in developing rats, observed with the Timm sulphide silver method
J comp Neur
179
,
581
618