Ultrastructural analysis of climbing fiber–Purkinje cell synaptogenesis in the rat cerebellum

PII: S 0 3 0 6 - 4 5 2 2 ( 0 1 ) 0 0 4 3 3 - X

Neuroscience Vol. 108, No. 4, pp. 655^671, 2001 ß 2001 IBRO. Published by Elsevier Science Ltd All rights reserved. Printed in Great Britain 0306-4522 / 01 $20.00+0.00

www.neuroscience-ibro.com

ULTRASTRUCTURAL ANALYSIS OF CLIMBING FIBER^PURKINJE CELL SYNAPTOGENESIS IN THE RAT CEREBELLUM S. MORARA,a J. J. L. VAN DER WANT,b H. DE WEERD,b L. PROVINIa;c and A. ROSINAa * a b

c

Istituto di Neuroscienze e Bioimmagini, CNR, Via Privata Bianco 9, 20131 Milan, Italy

Laboratory for Cell Biology and Electron Microscopy, Graduate School of Behaviour, Cognition and Neurosciences, University of Groningen, Oostersingel 69-2, 9713 EZ Groningen, The Netherlands

Istituto di Fisiologia Generale e Chimica Biologica, Via Trentacoste 2, Universita© degli Studi di Milano, 20134 Milan, Italy

AbstractöPrevious reports have described the transient expression of the neuropeptides calcitonin gene-related peptide and neuropeptide Y in selected subsets of rat olivocerebellar compartments during embryonic and postnatal development. Using these neuropeptides as endogenous markers for olivocerebellar ¢bers, the aim of this electron microscopic analysis was to reveal the synaptogenetic processes occurring between climbing ¢bers and their target Purkinje cells, from embryonic day 19 to postnatal day 16, the period during which Purkinje cells undergo intense emission and retraction of dendrites, and climbing ¢bers translocate their synapses along Purkinje cell membrane surfaces. The present ¢ndings provide the ¢rst direct evidence that climbing ¢ber synaptogenesis starts on embryonic day 19 and that these ¢rst synapses mainly involve the Purkinje cell embryonic dendrite rather than the Purkinje cell soma. At the same age, the presence of unlabeled synapses resembling calcitonin gene-related peptide-labeled synapses in the Purkinje cell plate makes it possible to conclude that climbing ¢bers form a major synaptic investment on embryonic Purkinje cells, a ¢nding that strongly supports the hypothesis of an early di¡erentiating role of climbing ¢bers on cerebellar development. Furthermore, during the period of intense dendritic remodeling of Purkinje cells, `myelin ¢gures' were often detected in Purkinje cell dendrites suggesting that they may at least in part represent real ultrastructural markers of membrane turnover that identi¢es the sites where Purkinje cell dendritic rearrangement is taking place. Finally the ¢nding that the climbing ¢ber terminals apposed to degenerating dendrites did not generally show signs of degeneration leads us to suggests that climbing ¢ber translocation from a perisomatic to a dendritic location may be driven by the Purkinje cell dendritic remodeling. ß 2001 IBRO. Published by Elsevier Science Ltd. All rights reserved. Key words: CGRP, NPY, Purkinje cell, olivocerebellar system, immunocytochemistry, electron microscopy.

from an initial mean ratio of three to four axons per Purkinje cell to a single axon per Purkinje cell in the adult (Crepel et al., 1976) and (b) the concomitant di¡erentiation of the Purkinje cell, which includes growing and orientation of its dendrite. Whereas olivocerebellar ¢bers already show a rough topographical arrangement when they ¢rst enter the cerebellar anlage (Chedotal and Sotelo, 1992, 1993), the dendritic rearrangements that contribute to de¢ning the adult connectivity of the climbing ¢ber^Purkinje cell synapse are thought to take place after the onset of synaptic activity. While there is evidence that neuropeptides may act as retrograde trophic factors regulating the component processes of neural ontogeny (as proposed by the neurotrophic theory that assumes the uptake of trophic factors by axons reaching the target and the subsequent retrograde transport to the cell body), it has recently been proposed that neuropeptides may also act as anterograde trophic factors (released by a¡erent nerve terminals onto their target cells). In this context, a series of neuropeptides (insulin-like growth factor-I, enkephalin, endorphin, substance P, somatostatin, cholecystokinin, motilin and calcitonin gene-related peptide (CGRP)) were recently found to be transiently expressed in the cerebellar cortex. In particular, our attention focused on the neuropeptide

The rodent cerebellum provides a useful model to analyze the developmental processes that transiently occur during the formation of neuronal connections, in particular the synapse formation and selection through which the speci¢city of a¡erent connections onto their targets is acquired. The ontogeny of the olivocerebellar system, terminating as climbing ¢bers and directly synapsing onto Purkinje cells, has been the subject of several anatomical and physiological studies on this speci¢c developmental process. Ever since the classical description by Ramo¨n y Cajal (1911), it has been well known that adult synaptic connectivity between olivary axons and Purkinje cells is achieved through a series of anatomically and physiologically de¢ned developmental steps, which involve (a) the reduction of olivocerebellar collaterals

*Corresponding author. Tel. : +39-2-26144220; fax: +39-228500036. E-mail address: [email protected] (A. Rosina). Abbreviations : CGRP, calcitonin gene-related peptide ; DAB, 3,3Pdiaminobenzidine; ED, embryonic day; GSSP, gold-substituted silver peroxidase; IR, immunoreactivity; LDCV, large dense core vesicle ; NPY, neuropeptide Y; PD, postnatal day; SCV, small clear vesicle; TBS, Tris-bu¡ered saline. 655

NSC 5274 29-11-01

656

S. Morara et al.

CGRP, which was shown to be involved in mechanisms of synapse stabilization at the neuromuscular junction (Fontaine et al., 1987) and to act as a di¡erentiating factor in the olfactory bulb (Denis-Donini, 1989). Transient expression of CGRP immunoreactivity (IR) in cerebellar cortical ¢bers during development was ¢rst described by Kubota et al. (1987). The peptide has subsequently been found to be transiently and selectively expressed in subsets of rat olivocerebellar compartments from late embryonic days up to the end of the second postnatal week (Morara et al., 1989, 1992; Chedotal and Sotelo, 1992). Moreover, in situ hybridization studies have shown that CGRP mRNAs are speci¢cally and transiently expressed in the same inferior olive subnuclei and that CGRP expression is regulated in inferior olivary neurons at the transcriptional level (Morara et al., 1995; Terrado et al., 1997, 1999). As no other precerebellar nuclei showed any CGRP expression during the same time window, it is possible to conclude that the CGRP expression only occurs in the olivocerebellar system during the intense phase of a¡erent ¢ber and target cell remodeling. The speci¢c expression of CGRP in subsets of climbing ¢bers of the rat cerebellar cortex thus o¡ered an invaluable tool to analyze at the ultrastructural level climbing ¢ber^Purkinje cell synaptogenesis. Moreover, to follow the entire synaptogenetic process, the expression of another peptide, neuropeptide Y (NPY), expressed up to adult stages in olivocerebellar compartments that partially overlap CGRP expression (Ueyama et al., 1994; Morara et al., 1997), was also used. The aim of this study was to analyze the ultrastructural appearance of the olivocerebellar ¢bers and terminals labeled by CGRP- or NPY-IR in the cerebellar cortex during development, and to de¢ne the features of the relationship between the identi¢ed terminals and their target cells. Part of this work has been presented in preliminary form (Morara et al., 1993).

EXPERIMENTAL PROCEDURES

In this study, we used 37 albino Wistar embryo and pup rats at di¡erent developmental ages, embryonic day (ED) 19 and 21, and postnatal day (PD) 0, 1, 3, 5, 7, 9, 13 and 16 respectively. In the case of the embryos, the mother was deeply anesthetized with urethane (1.3 g/kg, i.p.) and her abdomen surgically opened by a small incision. Each embryo was removed, anesthetized on ice, transcardially perfused with a brief (45^60 s) rinse of Tyrode's solution (containing 136.9 mM NaCl, 2.7 mM KCl, 1.5 mM MgCl2 , 11.9 mM NaHCO3 , 5.6 mM glucose, 0.4 mM NaH2 PO4 ), followed by Zamboni's ¢xative (4% freshly depolymerized paraformaldehyde, 0.2% picric acid, 0.1% glutaraldehyde) for 15^20 min at 4³C. In a few cases, the glutaraldehyde concentration was increased to up to 4%. The pups were deeply anesthetized with Nembutal (5 mg/100 g body weight) injected i.p., and then perfused transcardially like the embryos. Light and confocal microscopy For light and confocal microscopy, the brains were kept in the same ¢xative for 2^3 h, and then soaked overnight in 20% sucrose at 4³C. Frozen 10-Wm sections of the cerebellum were obtained in the frontal or parasagittal plane, collected on gelatin-coated slides and air-dried. The immuno£uorescence experi-

ments were performed as previously described (Morara et al., 1997, 2000). In brief, the sections were incubated overnight at 4³C with anti-CGRP antibodies (generous gift of Catia Sternini, CURE/UCLA, Los Angeles, CA, USA or from Peninsula; 1/500^2000) in Tris-bu¡ered saline (TBS; Tris 100 mM, NaCl 0.9%, pH 7.4) containing 0.2% Triton X-100 and 10% normal serum. Sections were subsequently incubated with biotin-conjugated goat anti-rabbit immunoglobulins (Vector; 1/200) in the same bu¡er for 1 h and then with £uorescein-conjugated avidin DCS (Vector; 1/100) in the same bu¡er for 1 h. In order to detect calbindin D immunoreactivity (a marker of Purkinje cells), after initial incubation with anti-calbindin D monoclonal antibodies (Sigma; 1/5000^15 000), the sections were incubated with Texas Red-conjugated sheep anti-mouse immunoglobulins (Amersham Pharmacia; 1/10) for 1 h at room temperature. Between each incubation step, the sections were thoroughly rinsed in TBS containing 0.2% Triton X-100. The slides were ¢nally coverslipped using Vectashield mounting medium (Vector). A confocal argon ion laser scanning microscope (Sarastro 2000, Molecular Dynamics), equipped with a Zeiss Axioskop epi£uorescence microscope, was used to analyze the immuno£uorescence-processed sections using a 63U/1.4 oil immersion Plan-Apochromat Zeiss objective. The £uorescein-labeled structures were analyzed using a 488-nm excitation wavelength and a 510ELFP longpass barrier ¢lter; the Texas Red-labeled structures were analyzed using a 514-nm excitation wavelength and a 535ELFP longpass barrier ¢lter. The aperture size of the pinhole was set at 50 Wm, the laser power at 16.0^24.0 mW, the PM tube voltage at 700^900 V, and the scanning mode format at 512U512 (pixel size 0.2 Wm). Electron microscopy For electron microscopy, the brains were removed from the skull and post¢xed overnight at 4³C in the same ¢xative. Frontal and parasagittal Vibratome sections were cut at 60 Wm and collected in phosphate-bu¡ered saline, pH 7.4. The sections were incubated at room temperature for 16^24 h in rabbit anti-CGRP (1/2000^5000) in TBS containing 0.05% Triton X-100, subsequently incubated in goat anti-rabbit globulin serum (Nordic) 1/200 in TBS with 0.05% Triton X-100 for 1.5^2 h and then in rabbit peroxidase^anti-peroxidase 1/400 in TBS with 0.05% Triton X-100 for 1.5^2 h. The incubation bu¡ers were supplemented with 1% normal goat serum, 0.08% bovine serum albumin and 0.1% cold water ¢sh gelatin (Amersham Pharmacia). Between incubations, the sections were thoroughly rinsed three times for 15 min in TBS. Finally, the sections were incubated in 0.03% 3,3P-diaminobenzidine (DAB)/0.01% hydrogen peroxide for 10^15 min. The DAB reaction product was subsequently intensi¢ed using the gold-substituted silver peroxidase technique (GSSP; van den Pol and Gorcs, 1986). The following protocol was applied: the DAB-reacted sections were rinsed in sodium cacodylate bu¡er 0.1 M, pH 7.2^7.4, and post¢xed in 1% osmium tetroxide in 0.1 M sodium cacodylate bu¡er with 1% sodium ferricyanide for 20 min. The sections were then dehydrated and £at-embedded in epoxy resin. Light micrographs of the CGRP-labeled areas were taken from the embryonic material in order to follow the course of climbing ¢bers in the cerebellar cortex at these stages. From the £at-embedded sections, CGRP-IR-containing areas on the midline of the cerebellar cortex (i.e. in the parasagittal compartments A, B innervated by neurons located in the caudal part of the medial and dorsal accessory olives, respectively; Voogd and Bigare¨, 1980) were selected at the light microscope and mounted on Epon blocks. At embryonic stages, the selected areas were restricted to the rostral two-thirds of the median part of the cerebellar anlage; at postnatal stages, they were mainly selected from the vermis of the anterior lobe, or lobules VI^VIII, but the £occuli were also sampled. Ultrathin sections were cut, mounted on Formvar-coated grids, and ¢nally contrasted with uranyl acetate and lead citrate. The rationale for the GSSP procedure was extensively described in the original paper (van den Pol and Gorcs, 1986),

NSC 5274 29-11-01

Climbing ¢ber synaptogenesis

but it is worth commenting on a critical step. After the DAB reaction and before silver intensi¢cation, the sections are incubated in 10% thioglycolate in order to reduce tissue argyrophilia (i.e. non-speci¢c silver deposition in the tissue). The time of this incubation is critical as short incubation times are insu¤cient for the purpose and long incubation times can inhibit the catalytic capacity of DAB to reduce metallic silver on itself, thus leading to insu¤cient intensi¢cation of the lightest DAB staining and false negative results. The incubation time therefore needs to be adjusted to the speci¢c experimental conditions: we found that incubation times of 2.5^4 h at room temperature gave optimal results and therefore systematically used these for our experiments. The presence of endogenous peroxidase reactivity, which may give false positive labeling, was controlled by performing DAB/GSSP enhancement on a parallel series of immunolabel-free sections. All animal experiments were carried out in accordance with the European Communities Council Directive (86/609/EEC, 24 November 1986).

RESULTS

In this study, the embryonic and postnatal synaptogenesis of climbing ¢bers to Purkinje cells was analyzed using CGRP-IR or NPY-IR as markers. From ED 19 to PD 9, this was mainly done using antibodies against CGRP, a peptide shown to be expressed in speci¢c olivocerebellar compartments from ED 16 to PD 20, by both immunohistochemical (Morara et al., 1989, 1992; Chedotal and Sotelo, 1992) and in situ hybridization studies (Morara et al., 1995; Terrado et al., 1997, 1999). NPY-IR is expressed from ED 19 up to adult stages in olivocerebellar compartments (Ueyama et al., 1994; Morara et al., 1997) that partially overlap the CGRP-positive compartments, as shown by the distribution of CGRP- or NPY-IR in the inferior olive nuclei (Morara et al., 1989, 1997), where they may coexist in individual neurons (unpublished observations). NPY-IR gave results similar to CGRP-IR and was mainly used from the second postnatal week on, when CGRP-IR had decreased to almost undetectable levels. Both light and electron microscopy analyses were mainly made on areas selected from the rostral two-thirds of the median part of the cerebellar anlage at embryonic stages, and from the cerebellar cortex of the midline of the anterior lobe vermis at postnatal stages. Light and confocal microscopy CGRP-IR olivary axons were shown to enter the cerebellar anlage as early as ED 17 by Chedotal and Sotelo (1992). Our analysis, mainly restricted to the midline of the rostral part of the rat cerebellar anlage where the CGRP-IR olivocerebellar compartment is present, showed a dense plexus of CGRP-IR ¢bers as early as ED 19 (Fig. 1A). At this stage, the multilayered and disoriented calbindin-immunoreactive Purkinje cells showed an irregular shape and had long dendrites (Fig. 1B), most likely the embryonic dendrites described by Armengol and Sotelo (1991). CGRP-IR was clearly detectable in individual climbing ¢ber axons and presumptive terminal boutons, and the ¢bers could be followed throughout their course (Fig. 1A). They coursed

657

dorsally at the midline from the presumptive white matter underlying the Purkinje cell plate to reach the border of the external granular layer after intermingling with the Purkinje cells (Fig. 1A, B). This stage of climbing ¢ber innervation onto Purkinje cells corresponds to the so-called creeper stage (Chedotal and Sotelo, 1993). During the following days, CGRP labeling in climbing ¢bers became progressively restricted to presumptive terminal boutons (PD 6, Fig. 1C), and the Purkinje cells underwent an intense process of dendrite remodeling that consisted in the retraction of the embryonic dendrite and the emission of several short `disoriented' perisomatic dendrites (PD 6, Fig. 1D). In conjunction with this process, the CGRP-labeled climbing ¢ber terminal boutons concentrated around the Purkinje cell somata to design the `nest phase' of climbing ¢ber innervation ¢rst described by Ramo¨n y Cajal in 1911 (Fig. 1C) which, in CGRP-labeled climbing ¢bers, could be detected in the ¢rst two postnatal weeks (Morara et al., 1992). The creeper and nest stages are the two main phases of CGRP-labeled climbing ¢ber^Purkinje cell innervation observed at the light microscope. After the second postnatal week, CGRP-IR decreases to almost undetectable levels in the olivocerebellar system (Morara et al., 1989; Chedotal and Sotelo, 1992), whereas NPY-IR reaches its peak expression during the second postnatal week but is still detectable thereafter (Morara et al., 1997). At the end of the second postnatal week both CGRP- and NPY-labeled climbing ¢ber terminals were seen to progressively cluster on the apical part of the Purkinje cell somata (data not shown). Subsequently NPY-labeled terminals could be observed to translocate to the de¢nitive Purkinje cell dendrite (data not shown). These two phases are respectively called the `capuchon' and `climbing' phases of climbing ¢ber innervation onto Purkinje cells (Ramo¨n y Cajal, 1911). Electron microscopy In order to visualize the immunochemical reaction product of CGRP- or NPY-IR, we applied the GSSP technique (van den Pol and Gorcs, 1986), in which the DAB reaction product is ¢rst intensi¢ed by silver and then substituted by gold. The GSSP reaction product is organized in dense black precipitates with de¢ned contours, which have the advantage of giving an increased visibility, a better de¢nition of their localization and a clearer identi¢cation of the ultrastructure over the di¡use DAB precipitate. At the midline of the cerebellar anlage, CGRP-IR already labeled growth cones (Fig. 2B^D) and synaptic boutons (Figs. 2D and 3A, B) by ED 19. Between ED 19 and PD 0, labeled unmyelinated ¢bers were thin and slender, with a central dense core of microtubules and granular material, and relatively few rami¢cations, and gave rise to numerous varicose pro¢les (Fig. 2A). They most frequently ended in labeled growth cones that were ¢lled to various degrees by tubular elements (Fig. 2B^D), large, round or ellipsoid, electron-lucent vesicles, generally referred to as `growth cone vesicles' (Del Cerro and Snider, 1968; see Fig. 2B^D), and they could form syn-

NSC 5274 29-11-01

658

S. Morara et al.

Fig. 1. Di¡erential climbing ¢ber innervation of Purkinje cells on ED19 and PD 6, as revealed by CGRP- and calbindin-IR. In A and B, frontal sections through the rostral part of the rat cerebellar anlage on ED 19 show a dense plexus of CGRPIR climbing ¢bers (A) at the level of the calbindin-immunoreactive Purkinje cell plate (B). In A, individual ¢bers appear either as fully labeled by CGRP-IR throughout their course, or as rows of labeled puncta. They design the outline of putative Purkinje cells, and can be followed (in B) to reach the dorsal part of the Purkinje cell multilayer (PcL), at the border with the external granular layer (EGL). In B, Purkinje cells show elongated, irregular shapes and long embryonic dendrites : one of them is indicated by the arrow. CGRP-IR climbing ¢ber terminals (C) and calbindin-IR Purkinje cell bodies (D) are shown in C and D respectively, on adjacent frontal sections through the anterior lobe of rat cerebellar anlage on PD 6. CGRP labeling is restricted to terminal-like structures (C) that seem to surround Purkinje cell bodies (D). At this age Purkinje cells that are almost aligned in a monolayer have only short and randomly oriented dendrites (D). Scale bars = 20 Wm.

aptic contacts characterized by a dense cluster of labeled small clear vesicles (SCVs; Fig. 2D). Labeled ¢bers also gave rise to labeled boutons forming synaptic contacts of the asymmetrical type (Fig. 3A, B), showing thickenings of the synaptic membranes and clustering of spherical SCVs in close vicinity to the presynaptic membrane specialization (Fig. 3A, B). The labeled synaptic boutons frequently contained large vesicles similar to growth cone vesicles (Fig. 3A), and a few dense core vesicles were occasionally found, but CGRP labeling was mainly apposed to the SCVs throughout the terminal.

As a whole, the immunoreactive synapses showed the typical features of climbing ¢ber synapses at an immature stage, as the thickenings of the synaptic membrane specializations were clear but thinner than at later stages. Furthermore, the vesicles were not restricted to the cytoplasm adjacent to the synaptic membrane specialization, but could be dispersed throughout the terminal, their number and clustering being generally less than in the synapses at later stages. At the same age, unlabeled synapses (Fig. 3C) were very rarely detected in the midline of the cerebellar

NSC 5274 29-11-01

Climbing ¢ber synaptogenesis

659

Fig. 2. CGRP-immunoreactive climbing ¢bers and growth cones from ED 19 to PD 0. In A, a survey of the Purkinje cell plate shows several CGRP-labeled varicose pro¢les apposed to Purkinje cell (Pc) somata at PD 0. In B^D, three labeled growth cones on ED 19. Within the growth cones note the presence of CGRP labeling, several types of vesicular and tubular structures as well as growth cone vesicles (small arrows). In D, a synapse (arrowhead) is also present, showing pre- and postsynaptic thickenings and a dense cluster of labeled synaptic vesicles. Note the presence of dilated extracellular spaces, frequently observed at these early stages of development. Scale bars = 1.0 Wm in A, 0.5 Wm in B^D.

anlage; the unlabeled boutons made synaptic contacts of the asymmetrical type and showed a dense cluster of round clear vesicles, thus resembling climbing ¢ber synapses. The postsynaptic structures contacted by the labeled terminals at ED 19 were rarely cell somata (Fig. 3A) and more frequently small-caliber dendrites (Fig. 3B). The number of CGRP-IR unmyelinated ¢bers and synaptic terminals considerably increased by PD 0: as shown in Fig. 2A, the number of labeled pro¢les in a single microscopic ¢eld could be very high, like the CGRP-IR granular content in individual labeled ¢bers (Fig. 2A) and synaptic boutons (Fig. 3D). Moreover, the CGRP-labeled synaptic pro¢les generally showed more evident clustering of the small clear vesicles and thicker synaptic membrane specializations (Figs. 2D and 3D), with features of more mature climbing ¢ber synapses. The synapses were still made on cellular pro¢les identi¢able as small-caliber dendrites (Fig. 3D), but were now more frequently made on cell somata than at E19. The cellular pro¢les to which CGRP-labeled ¢bers and boutons were apposed at these early developmental stages had darkly stained nuclei, smooth and spherical in outline, with occasional deep indentations, surrounded

by a cytoplasm that contained few organelles but a rich accumulation of free ribosomes and cisternae of endoplasmic reticulum (Fig. 2A). In contrast, the cytoplasm of the apical part of the somata was densely ¢lled with endoplasmic reticulum and mitochondria (Fig. 2A, top left). On the basis of these characteristics, the observed postsynaptic pro¢les were considered Purkinje cells that were still at immature stages, since the hypolemmal cisternae, characteristically present in mature Purkinje cells, could not yet be detected. The extracellular spaces often appeared to be dilated during these early stages of development (see Figs. 2 and 3). Moreover, from PD 3 on, in the same ¢elds where labeled boutons were found, signs of interaction between neuronal and glial elements (Fig. 4A) and degenerating pro¢les (Figs. 3D, 4B and 5A^C) were observed. Glial pro¢les are at this stage identi¢able as thin membrane sheets, which are closely apposed to the Purkinje cell and barely exhibit further cytoplasmic details (see Figs. 4A, 5B, C and 6C). In Fig. 4A an example of functional interaction between a glial pro¢le and a labeled climbing ¢ber is shown, in which the glial pro¢le is undergoing an endocytotic process including part of the membrane of the labeled terminal. Such endocytotic processes were also found between Purkinje cell dendrites and labeled

NSC 5274 29-11-01

660

S. Morara et al.

Fig. 3. CGRP-labeled and -unlabeled climbing ¢ber synapses from ED 19 to PD 0. Surveys of the Purkinje cell plate on ED 19 (A^C) or PD 0 (D). In A, next to a Purkinje cell (Pc) soma on the right, a CGRP-positive terminal forms a synapse (arrowhead) on the soma of a second Purkinje cell on the left. The CGRP-labeled terminal contains small spherical vesicles and several larger electron-lucent growth cone vesicles (small arrows). In B, a CGRP-positive terminal forms a synapse (arrowhead) with a thin postsynaptic structure, probably a distal Purkinje cell dendrite. In C, an unlabeled interrupted synapse `en passant' (arrowheads) on ED 19 shows asymmetrical membrane thickenings, a dense cluster of numerous synaptic vesicles and a single large dense core vesicle (white-headed arrow), thus resembling a climbing ¢ber synapse. In D, a heavily labeled, CGRP-positive terminal forms an asymmetrical synapse (arrowhead) with a Purkinje cell dendrite (Pcd). Two regressing pro¢les consisting of electron-dense material delimited by a membrane are indicated by arrows. Note that on both ED 19 and PD 0, CGRP-IR is localized over the cluster of small, clear vesicles that ¢ll the terminal and that on PD 0 an increased thickening of the synaptic membranes is present. Scale bars = 1.0 Wm in A, 0.5 Wm in B^D.

NSC 5274 29-11-01

Climbing ¢ber synaptogenesis

661

Fig. 4. Glia^neuron interactions and regressing processes in the rat cerebellar cortex on PD 3. In A, interaction between a glial process and a labeled climbing ¢ber: the glial process (g) incorporates (arrow) part of the membrane of the apposed CGRP-labeled terminal. A Purkinje cell (Pc) body is present on the right side. In B, a CGRP-labeled terminal in synaptic contact (arrowhead) with a regressing pro¢le that is probably a Purkinje cell dendrite (Pcd). Note that the climbing ¢ber apposed to the regressing dendrite does not show any sign of degeneration. Scale bars = 1.0 Wm in A, 0.5 Wm in B.

olivocerebellar ¢bers and terminals and are generally interpreted as cellular recognition activities (PalaciosPru« et al., 1981). The degenerating pro¢les (Figs. 3D, 4B and 5A^C) consisted of either multiple concentric membrane whorls (generally referred to as `myelin ¢gures'), with dilated intermembranous spaces or highly electron-dense spots of membrane cohorts surrounded by separated membrane, or intermediate forms. These pro¢les were mainly con¢ned to the intracellular areas bordered by the somata and neuropil, which did not show any degenerative changes such as cytoplasmic vacuolization, mitochondrion swelling, membrane disruption and accumulation of secondary lysosomes (Figs. 3D, 4B and 5A^C). Myelin ¢gures were mainly found in pro¢les identi¢ed as Purkinje cell dendrites and, as in the example illustrated in Fig. 4B, the climbing ¢ber presynaptic terminal apposed to the degenerating dendrite did not generally show any sign of degeneration. These pro¢les, very seldom observed also in presynaptic

structures, were detected with progressively decreasing frequency at later developmental stages. From PD 3 to 7, the synaptic complex formed by the labeled boutons rapidly matured. Signs of maturation mainly concerned the site of apposition of the boutons (smooth surface versus spine of Purkinje cell somata or dendrites) and the glial investment of the synaptic complex: the two processes occurred in parallel on both somata and dendrites. On PD 3^5 labeled terminal boutons were now found embedded in a network of growing protrusions emanating from the soma of Purkinje cells (Fig. 5A), and synapses could be made either directly on the surface of Purkinje cell somata (Fig. 5B) or on their somatic protrusions (Fig. 5C). At this stage, the Purkinje cells had a very elaborate morphology, with complex perisomatic processes and tortuous indented surfaces to which multiple synaptic contacts were apposed (Fig. 5C). Labeled terminals with two attachment sites in the same sectional plane were commonly observed and, in such

NSC 5274 29-11-01

662

S. Morara et al.

Fig. 5.

NSC 5274 29-11-01

Climbing ¢ber synaptogenesis

cases, the two synapses could be made either on individual spiny protrusions or on the surface of the Purkinje cell soma and a somatic spiny protrusion (Fig. 5C). It is worth noting that when the labeled bouton made synaptic contact on a somatic protrusion the glial investment of the synaptic complex was more prominent (compare Fig. 5B and C). Both the somatic protrusions and the somatic surface of the Purkinje cells showed clusters of growth cone vesicles (Fig. 5B, C), indicating the existence of an intense budding activity. The developmental stage here described corresponds to the Cajal phase of the Purkinje cells with `disoriented dendrites', i.e. protrusions emanating in all directions, that allow for an increase in the number of climbing ¢ber^Purkinje cell contacts per individual Purkinje cell (see Fig. 5). The morphology and location of the labeled terminals changed from PD 5 to PD 7. A maturational process parallel to that observed for the somatic synapses at PD 5 occurred for the synapses on the dendritic location on PD 7^9 (Fig. 6). Both the dendritic shaft (Fig. 6A, B) and, less frequently, the dendritic spines (Fig. 6C) of Purkinje cell proximal dendrites were synaptically contacted by labeled climbing ¢ber boutons. In the synaptic boutons a CGRP-IR localization to the dense cluster of synaptic vesicles docked to the presynaptic membrane was more frequently observed than at earlier stages (see Fig. 6A, B). The cytoplasm of the Purkinje cell dendrites contained numerous tubular elements (Fig. 6A) and some multivesicular bodies. The glial investment of the synapses was generally greater when they were formed on dendritic spines than on dendritic shafts. Moreover, when the labeled synapses showed a prominent glial investment they were generally not associated with the presence of growth cone vesicles (Fig. 6C). Thus, at this stage the climbing ¢ber^Purkinje cell synaptic complexes displayed most of the features of adult synapses. Up to PD 9, the CGRP-IR and NPY-IR provided essentially similar ¢ndings. At this stage both CGRPIR and NPY-IR climbing ¢bers could be found on proximal dendrites (Fig. 6A). However, because of the greater abundance of NPY at this age, the NPY-IR climbing synapses showed a higher level of labeling (Fig. 6B). A further maturation of the climbing ¢ber^Purkinje cell synaptic complex occurs after PD 9. On PD 13 (Fig. 7), labeled ¢bers showing clear vesicles and microtubules characteristic of climbing ¢bers form varicosities ¢lled with clusters of SCVs. Climbing ¢ber synapses almost completely surround Purkinje cell dendritic spines

663

and show clusters of SCVs docked to the presynaptic membrane specialization (Fig. 7), thus featuring the classical asymmetrical climbing ¢ber^Purkinje cell synapse. Few regressing pro¢les were still found at this developmental age. At the beginning of the third postnatal week (PD 16, the latest stage analyzed), the morphology of NPY-IR climbing ¢ber^Purkinje cell synapse shows the typical features of the adult synaptic complex (Fig. 8). Also in terms of localization, most of the climbing ¢ber terminals, both labeled and unlabeled, further penetrated the molecular layer, as indicated by the concomitant presence of numerous parallel ¢ber axons and by the fact that climbing ¢ber synapses were exclusively apposed to small-caliber Purkinje cell dendrites (Fig. 8A, B). It has to be noted that there were more unlabeled than labeled climbing ¢bers in the same microscopic ¢eld, indicating the ongoing decrease of NPY expression in the olivocerebellar ¢bers.

DISCUSSION

In the present study, the neuropeptides CGRP and NPY were used as endogenous climbing ¢ber markers to analyze, at the ultrastructural level, the development of climbing ¢ber synaptogenesis onto their target, the Purkinje cells. It has previously been shown at the optical level that CGRP-IR is transiently and selectively expressed in speci¢c compartments of the rat olivocerebellar system, during development (Morara et al., 1989, 1995). So far, CGRP is the only endogenous marker that selectively labels rat olivocerebellar ¢bers. The presence of NPY-IR has been described in the adult cerebellum, in speci¢c olivocerebellar compartments and in a small number of mossy ¢bers (Ueyama et al., 1994), and its expression in the olivocerebellar system is up-regulated during the second postnatal week (Morara et al., 1997). Although the ultrastructural CGRP- and NPY-IR results are superimposable, CGRP is more abundantly expressed up to the second postnatal week ^ and was therefore mainly used for the analysis from ED 19 to PD 9 ^ and NPY is more abundantly expressed from the second postnatal week ^ and was therefore mainly used after PD 9. Technical considerations As already described above in the Results section, the

Fig. 5. Somatic localization of climbing ¢bers during the ¢rst postnatal week. In A, a CGRP-labeled terminal contacts both the soma (Pc) and a somatic protrusion (star) of a Purkinje cell, on PD 3. The protrusion contains vesicular and tubular elements and a granular matrix. In B, a CGRP-labeled terminal forms an asymmetrical synapse (arrowhead) directly on the soma of a Purkinje cell (Pc) on PD 5. The terminal is partially surrounded by glial processes (g). Lucent growth cone vesicles (small arrows) and dark, swollen, degenerating elements (arrows) are present in an otherwise normal, well preserved tissue. In C, three somatic spines (stars) protruding from a Purkinje cell (Pc) soma form synaptic contacts (arrowheads) with CGRPpositive terminals, on PD 5. A double synaptic contact (two arrowheads) with both the soma and a spine of the Purkinje cell is formed by one of the terminals. Within a spine neck aggregates of growth cone vesicles (small arrow) and several tubular elements are present. Growth cone vesicles are also present in a dendrite (small arrows). Note the glial processes (g) ensheathing the Purkinje cell somatic spines and the climbing ¢bers. Widened extracellular spaces containing electron-dense material, suggestive of regressing processes, are present (one of them is indicated by an arrow). Scale bars = 1.0 Wm in A and B, 0.5 Wm in C.

NSC 5274 29-11-01

664

S. Morara et al.

Fig. 6 (caption overleaf).

NSC 5274 29-11-01

Climbing ¢ber synaptogenesis

665

Fig. 7. CGRP labeling of climbing ¢ber terminals at the end of the second postnatal week. Survey of a CGRP-labeled ¢ber, with clear vesicles and microtubules characteristic for climbing ¢ber (cf), on PD 13. On the left, an unlabeled climbing ¢ber terminal forms an asymmetrical synapse (arrowhead) on a dendritic spine. Scale bar = 0.5 Wm.

GSSP reaction product of CGRP-or NPY-IR is organized in dense black precipitates with de¢ned contours, which have the advantage of an increased visibility and

a better de¢nition of their localization over the di¡use DAB precipitate. It is worth noting that, in the cases in which the precipitate was su¤ciently de¢ned, the GSSP^

Fig. 6. Dendritic localization of climbing ¢bers during the second postnatal week. In A, a CGRP-positive terminal forms an asymmetric synaptic contact (arrowhead) with the shaft of a dendrite close to a Purkinje cell (Pc) soma, on PD 7. The cytoplasm of the dendrite contains abundant vesicular and tubular structures (arrow). In B, an NPY-labeled terminal, containing small electron-lucent vesicles, forms synaptic contact (arrowhead) with the shaft of a small dendrite, on PD 9. Note the presence of an unlabeled climbing ¢ber (cf) terminal and of several parallel ¢bers. In C, a CGRP-labeled terminal forms a synapse (arrowhead) on a spine of a primary Purkinje dendrite (Pcd), on PD 9. Note the presence of glial processes (g) close to the spine. Scale bars = 1.0 Wm in A, 0.5 Wm in B and C.

NSC 5274 29-11-01

666

S. Morara et al.

Fig. 8. NPY-labeled climbing ¢ber terminals during the third postnatal week. In A and B, NPY-labeled climbing ¢bers, intermingled with unlabeled ones, show only low levels of intensity, on PD 16. Climbing ¢ber synapses (arrowheads) are localized in the molecular layer, as judged from the presence of numerous parallel ¢bers. Note, in A, a single postsynaptic pro¢le simultaneously contacted by an NPY-labeled climbing ¢ber and an unlabeled one. Scale bars = 0.5 Wm.

peptide-IR reaction product was always found to be apposed to round, SCVs in the climbing ¢ber terminals (see Figs. 4^8). While it is generally accepted that the neurotransmitter speci¢c for the pathway is localized to SCVs, whereas the sites of neuropeptide storage are large dense core vesicles (LDCVs), in our material with the GSSP technique, CGRP- or NPY-IR was consistently found associated with SCVs in the climbing ¢bers during development. It has to be noted that a localization of the peptides also to LDCVs cannot be discarded: indeed, the climbing ¢ber terminals contain only a few LDCVs (one or two per terminal; Palay and Chan-Palay, 1974; van der Want and Voogd, 1987), and so any such localization

might be missed in the absence of a serial reconstruction analysis (not performed here). In conclusion, our ¢nding that CGRP- or NPY-IR is associated to SCVs may indicate an exception to the vesicular peptide localization speci¢c for the climbing ¢ber pathway, or be a re£ection of a transient SCV localization during the developmental stages considered in this study. A few reports describing an association of neuropeptide immunoreactivity with SCVs, using the DAB technique, in other systems and adult animals (see Maley, 1990 for a review), and an immunogold study on axotomized neurons by Zhang et al. (1995) showing that NPY-IR can also be associated with constitutive secretory vesicles, suggest that neuropeptide localization may not be con¢ned to LDCVs. In

NSC 5274 29-11-01

Climbing ¢ber synaptogenesis

this context it seems that our experimental paradigm, in which GSSP^neuropeptide-IR is found apposed to SCVs in terminals that contain numerous SCVs but few LDCVs, provides a unique opportunity to test the hypothesis of an extra-LDCV localization of neuropeptides. Further investigations are currently in progress. Climbing ¢ber synaptogenesis The localization of CGRP-IR in olivocerebellar climbing ¢ber terminals is con¢rmed by the present ultrastructural observations. Indeed, it is generally accepted that the asymmetrical synapses with densely packed synaptic vesicles, apposed to perisomatic protrusions of Purkinje cells during postnatal development, are climbing ¢ber synapses (Ramo¨n y Cajal, 1911; Larramendi, 1969; Mugnaini, 1969; Altman, 1972). Our results are also in keeping with those of a previous immunocytochemical study using parvalbumin as climbing ¢ber marker (Chedotal and Sotelo, 1993), which also showed that these synapses stem from thin ¢bers containing parallel arrays of microtubules and dispersed vesicular pro¢les. In addition to con¢rming these previous ¢ndings on the postnatal stages of climbing ¢ber development, our results show that climbing ¢ber synapses already express a number of their features during the embryonic phases of synaptogenesis. However, these embryonic climbing ¢ber synapses are characterized by reduced electron-density of synaptic membrane specializations, reduced clustering and docking of synaptic vesicles, and thus show the features of immature climbing ¢ber synapses. One of the major ¢ndings of the present study is the experimental evidence that climbing ¢ber synapses onto Purkinje cells start to be seen on ED 19, thus shifting the onset of climbing ¢ber synaptogenesis to an embryonic stage. The CGRP-labeled climbing ¢ber synapses (Fig. 2A, B) were found at the level of the Purkinje cell layer, along the rostral part of the midline extent of the cerebellar anlage. Most of these rare, but consistently observed, CGRP-labeled synapses were apposed to Purkinje cell dendrites, although some were apposed to Purkinje cell somata. It is worth noting that unlabeled synapses were observed ^ albeit very rarely ^ in the sectors of cerebellar cortex analyzed, at this same stage of development (ED 19, Fig. 3C). In this context, during mouse cerebellar postnatal development mossy ¢bers have been described to give rise also to `climbing ¢berlike' branches that directly contact Purkinje cells (Mason and Gregory, 1984; Mason et al., 1990; see also Takeda and Maekawa, 1989). However, these branches were only observed after PD 0, their expression peaked during the second postnatal week, and their synapses had the appearance of `elementary contacts'. The unlabeled synapses observed in this study on ED 19 had, on the contrary, the same morphological features as the labeled ones, i.e. the features characteristic of climbing ¢bers (see Palay and Chan-Palay, 1974) and it is therefore very likely that they are synapses of unlabeled climbing ¢ber axons originating from unlabeled inferior olive neurons adjacent to the labeled ones. Thus, it is here sug-

667

gested that the onset of climbing ¢ber synaptogenesis coincides with the onset of synaptogenesis in the cerebellar cortex. Other cerebellar a¡erents, such as the mossy ¢bers of vestibular and spinal origin as well as some monoaminergic and peptidergic ¢bers, enter the cerebellar anlage earlier than climbing ¢bers (Martin et al., 1983; Bishop et al., 1985; Morris et al., 1988; Ashwell and Zhang, 1992; Cummings et al., 1994) and were suggested to be the ¢rst synaptic investment in the embryonic cerebellum. However, on ED 19, in the Purkinje cell plate along the midline of the rostral cerebellar anlage, synapses whose morphology clearly di¡ered from the climbing ¢ber synapse morphology were not observed (Palay and Chan-Palay, 1974). The lack of evidence of other types of synapses in the Purkinje cell plate analyzed here could be due to: (1) their distribution in regions of the cerebellar anlage that do no overlap those of the CGRPpositive climbing ¢bers (see e.g. Morris et al., 1988) or (2) the possibility that they innervate cellular elements other than Purkinje cells (Morris et al., 1988). The onset of synapses on ED 19 is supported by the observations of West and Del Cerro (1976) that synaptogenesis in the rat cerebellum starts on ED 19 and by the comparable ¢ndings by Mason (1986) in the mouse cerebellum, where the onset of synaptogenesis dates from ED 17. Altogether these results support our conclusion that the synapses ¢rst established in the embryonic cerebellum are made by climbing ¢bers. The fact that climbing ¢ber synapses provide a major synaptic investment in the Purkinje cell plate as early as ED 19 suggests that they may in£uence the embryonic di¡erentiation of their targets. The results reported here, showing that CGRP is expressed in climbing ¢ber axons, growth cones and synapses and is localized to vesicular structures ^ i.e. in a compartment from which it can be released ^ indicate the possibility that climbing ¢bers might exert their proliferating/di¡erentiating role through the release of CGRP. In this context, it has been shown that CGRP receptors are expressed in the white matter oligodendrocytes and in Bergmann glial cells at the Purkinje cell layer, starting from late embryonic days (Morara et al., 2000 and unpublished observations). The CGRP receptor, like in general neuropeptide receptors, is a member of the G protein-coupled receptor superfamily, and may therefore produce long-lasting e¡ects (Iversen, 1995). The peptide has been shown to induce morphological di¡erentiation, second messenger modulation and immediate early gene expression in astrocytes (Lazar et al., 1991; Haas et al., 1991; Priller et al., 1995; Parsons and Seybold, 1997). Taken together, these data support the hypothesis that CGRP released by climbing ¢bers may promote the embryonic di¡erentiation of cerebellar astrocytes through the activation of CGRP receptors, and indirectly a¡ect the di¡erentiation of Purkinje cells through glia^neuron interactions (see Discussion in Morara et al., 2000). Along the same lines, NPY, like other peptides developmentally expressed in climbing ¢bers (Ha et al., 2000), represents other potential candidates for a di¡erentiating role by climbing ¢bers on the embryonic cerebellum. The

NSC 5274 29-11-01

668

S. Morara et al.

hypothesis that a¡erent systems may exert a developmental role on their target has previously been advanced (see Rubenstein, 2000, for a review). More speci¢cally, the hypothesis that climbing ¢bers may play a role in the morphogenesis of cerebellar Purkinje cells was previously considered; however, the earliest experimental lesions on the olivocerebellar system of cats and rats were made postnatally, i.e. later than the embryonic onset of climbing ¢ber synaptogenesis, and gave contradictory results (Kornguth and Scott, 1972; Kawaguchi et al., 1975; Bradley and Berry, 1976; Sotelo and Arsenio-Nunes, 1976). Purkinje cell dendrite regression The maturation of the cerebellar cortex is achieved as a result of a complex sequence of outgrowth and regression events. In particular, it has been shown that Purkinje cells undergo sequential processes of dendritic arbor remodeling: a long dendrite is ¢rst emitted and reabsorbed embryonically, then `disoriented' perisomatic processes are given o¡ and retracted during the ¢rst postnatal week, and ¢nally the mature oriented dendrite starts growing from the second postnatal week (Ramo¨n y Cajal, 1911; Larramendi and Victor, 1967; Ha¨mori, 1969; Mugnaini, 1969; Altman, 1982; Mason and Gregory, 1984; Mason et al., 1990). However, the events leading to the regression of the embryonic dendrite and the postnatal perisomatic processes have not had an ultrastructural correlate. The hypothesis has been made that `the dendritic involution consists of a progressive retraction of the long dendritic segments, with retrieval of cytoplasmic organelles' (Chedotal and Sotelo, 1993). Multiple concentric whorls of membranes, generally referred to as myelin ¢gures, were consistently found during the ¢rst postnatal week, in Purkinje cell pro¢les, and occasionally in axonal varicosities; they were not found in glial pro¢les. Such structures were also documented in other authors' ultrastructural studies on the developing cerebellum (see Meller and Glees, 1969, plate 7; Chedotal and Sotelo, 1993, ¢gure 10), but generally not commented upon. The interpretation of myelin ¢gures is controversial: they have been considered to represent either in vitro artifacts occurring during poor ¢xation procedures or in vivo markers of sites where intense lipid tra¤c or intense membrane turnover (Williams, 1980; White and Michaud, 1981) is occurring. In the latter cases myelin ¢gures have been associated either with the rearrangement phenomena that occur during development (Hildebrand et al., 1994), or with intense membrane turnover which accompanies degeneration processes occurring in pathological conditions (see e.g. Tanaka et al., 1988; Higashi et al., 1995). An alternative possibility, which reconciles the two interpretations reported above, stems from the results of the extensive studies by Blanchette-Mackie and Scow (1981, 1983) who showed that myelin ¢gures may form in vitro, even after excellent glutaraldehyde ¢xation, in those sites where high membrane turnover and intense lipid tra¤c are taking place in vivo, resulting in a strong accumulation of lipids. In our study, myelin ¢gures could

be detected in cerebellar specimens that showed an optimal ultrastructural preservation and in areas of the tissue that did not show any signs of classical degeneration. We are therefore inclined to suggest that they represent, at least in part, a real marker of Purkinje cell membrane turnover that identi¢es the sites where processes of membrane rearrangement are taking place, as in the Purkinje cell dendrite during early developmental stages. Climbing ¢ber translocation During the phases of intense Purkinje cell dendritic rearrangement that lead to the maturation of adult dendritic arborization, two major processes take place at the presynaptic site: the regression of climbing ¢ber multiple innervation (Crepel et al., 1976) and the concomitant change in position of climbing ¢ber synapses from a perisomatic to the adult peridendritic location (Ramo¨n y Cajal, 1911; Mugnaini, 1969; Larramendi, 1969; Altman, 1972; Mason et al., 1990; Schoen et al., 1991). Direct evidence for the existence of a transient stage of Purkinje cell multiple innervation by climbing ¢bers was elegantly provided by Crepel et al. (1976, 1981). After Bourrat and Sotelo (1984) and DelhayeBouchaud et al. (1985) showed that the phase of neuronal death in the olivary complex precedes the regression of redundant climbing ¢ber innervation, it was concluded that the regression of multiple innervation is due to the retraction of axon collaterals by the surviving olivary neurons, as later supported by tract-tracing studies (Rosina et al., 1989). The change in position of climbing ¢ber terminals from a perisomatic to a peridendritic location ¢rst described in the pioneering work of Ramo¨n y Cajal (1911) was later corroborated by several Golgi impregnation and ultrastructural studies (Larramendi, 1969; Mugnaini, 1969; Altman, 1972; Schoen et al., 1991). Recent observations have provided a more complex view of the development of the climbing ¢ber^ Purkinje cell synapse, by showing that: (I) climbing ¢bers enter the Purkinje cell plate as soon as they arrive in the cerebellar anlage, rather than waiting close to the target (Mason et al., 1990; Chedotal and Sotelo, 1992); (II) climbing ¢bers give rise to a dendritic innervation of Purkinje cells (creeper phase; Chedotal and Sotelo, 1993) that precedes the establishment of the perisomatic innervation (nest phase; Ramo¨n y Cajal, 1911); and (III) the start of the climbing phase of olivocerebellar ¢bers onto Purkinje cell dendrites is independent of the outgrowth of the ¢nal Purkinje cell dendritic tree (Mason et al., 1990; our unpublished observations). However, it is still debated whether the climbing ¢ber position change from the perisomatic to the peridendritic location that starts in the second postnatal week is due to the translocation of the perisomatic climbing ¢ber synapses (Larramendi, 1969; Chedotal and Sotelo, 1993), or to the regression of the old synapses and the formation of new ones (Chedotal and Sotelo, 1993). In this context, our study indicates that unlike the intense regressive phenomena commonly observed in the postsynaptic Purkinje cell pro¢les, regressive phenomena were rarely observed in the presynaptic climbing

NSC 5274 29-11-01

Climbing ¢ber synaptogenesis

¢ber pro¢les during the ¢rst two postnatal weeks. Although it has been observed that olivocerebellar ¢bers degenerate rapidly (Desclin and Colin, 1980) and thus some degenerating synaptic terminals may have been missed, our ¢ndings that the climbing ¢ber presynaptic terminals apposed to the degenerating dendrite did not generally show any sign of degeneration lead us to suggest that the change in the position of climbing ¢ber synapses is driven by the Purkinje cell, i.e. that at least some of the climbing ¢ber synapses on the Purkinje cell somata are moved along the Purkinje cell membrane toward their ¢nal dendritic location. This conclusion is supported by the observations that climbing ¢ber maturational processes take place in parallel at both Purkinje cell soma and dendrite levels. As early as PD 5, it is possible to observe `mature' somatic climbing ¢ber synapses that have a glial investment (Fig. 5C), which is thought to be a hallmark of the end of synaptogenesis (Altman, 1972), and lack growth cone vesicles (Fig. 5C) that are a hallmark of immature, growing ¢bers (Del Cerro and Snider, 1968). However, given the large number of climbing ¢ber synapses on the dendritic arbor of an individual Purkinje cell, the contribution of new synapses to the ¢nal size of the axonal terminal ¢eld of a climbing ¢ber should also be taken into account.

669 CONCLUSION

The present ¢ndings provide the ¢rst direct evidence that climbing ¢ber synaptogenesis starts at ED 19 and that the ¢rst synapses mainly involve the Purkinje cell embryonic dendrite rather than the Purkinje cell soma. At the same age, the presence of unlabeled synapses resembling CGRP-labeled synapses in the Purkinje cell plate allows us to conclude that climbing ¢bers form the ¢rst major synaptic investment on Purkinje cells, a ¢nding that strongly supports the hypothesis of their early di¡erentiating role on cerebellar development. Finally, it is here suggested that the so-called myelin ¢gures, representing at least in part ultrastructural markers of the sites in which intense membrane turnover and lipid transport are taking place, may be considered the ultrastructural correlate of the Purkinje dendrite regression occurring during development.

AcknowledgementsöThe authors are indebted to Ms. Laura Zambusi for her skillful technical assistance, and to Mr. Bert Hellinga and Mr. Peter van de Sijde for their excellent photographic assistance. This work was supported by CNR Biotechnology Targeted Project, MURST-CNR Biotechnology Strategic Project L 95/95 and Graduate School of Behavior, Cognition and Neurosciences, University of Groningen.

REFERENCES

Altman, J., 1972. Postnatal development of the cerebellar cortex in the rat. II. Phases in the maturation of Purkinje cells and of the molecular layer. J. Comp. Neurol. 145, 399^464. Altman, J., 1982. Morphological development of the rat cerebellum and some of its mechanisms. In: Palay, S.L., Chan-Palay, V. (Eds.), The Cerebellum ^ New Vistas. Springer-Verlag, New York, pp. 8^46. Armengol, J.A., Sotelo, C., 1991. Early dendritic development of Purkinje cells in the rat cerebellum. A light and electron microscopic study using axonal tracing in `in vitro' slices. Dev. Brain Res. 64, 95^114. Ashwell, K.W., Zhang, L.L., 1992. Ontogeny of a¡erents to the fetal rat cerebellum. Acta Anat. 145, 17^23. Bishop, G.A., Ho, R.H., King, J.S., 1985. An immunohistochemical study of serotonin development in the opossum cerebellum. Anat. Embryol. 171, 325^338. Blanchette-Mackie, E.J., Scow, R.O., 1981. Lipolysis and lamellar structures in white adipose tissue of young rats: lipid movement in membranes. J. Ultrastruct. Res. 77, 295^318. Blanchette-Mackie, E.J., Scow, R.O., 1983. Movement of lipolytic products to mitochondria in brown adipose tissue of young rats: an electron microscope study. J. Lipid Res. 24, 229^244. Bourrat, F., Sotelo, C., 1984. Postnatal development of the inferior olivary complex in the rat. III. A morphometric analysis of volumetric growth and neuronal cell number. Dev. Brain Res. 16, 241^251. Bradley, P., Berry, M., 1976. The e¡ects of reduced climbing and parallel ¢bre input on Purkinje cell dendritic growth. Brain Res. 109, 133^151. Chedotal, A., Sotelo, C., 1992. Early development of olivocerebellar projections in the fetal rat using CGRP immunocytochemistry. Eur. J. Neurosci. 4, 1159^1179. Chedotal, A., Sotelo, C., 1993. The `creeper stage' in cerebellar climbing ¢ber synaptogenesis precedes the `pericellular nest' ^ ultrastructural evidence with parvalbumin immunocytochemistry. Dev. Brain Res. 76, 207^220. Crepel, F., Delhaye-Bouchaud, N., Dupont, J.L., 1981. Fate of the multiple innervation of cerebellar Purkinje cells by climbing ¢bers in immature control, X-irradiated and hypothyroid rats. Dev. Brain Res. 1, 59^71. Crepel, F., Mariani, J., Delhaye-Bouchaud, N., 1976. Evidence for a multiple innervation of Purkinje cells by climbing ¢bers in the immature rat cerebellum. J. Neurobiol. 7, 567^578. Cummings, S.L., Scott Young, W., III, King, J.S., 1994. Early development of cerebellar a¡erent systems that contain corticotropin-releasing factor. J. Comp. Neurol. 350, 534^549. Del Cerro, M.P., Snider, R.S., 1968. Studies on the developing cerebellum. Ultrastructure of the growth cones. J. Comp. Neurol. 133, 341^362. Delhaye-Bouchaud, N., Geo¡roy, B., Mariani, J., 1985. Neuronal death and synapse elimination in the olivocerebellar system. I. Cell counts in the inferior olive of developing rats. J. Comp. Neurol. 232, 299^308. Denis-Donini, S., 1989. Expression of dopaminergic phenotype in the mouse olfactory bulb induced by the calcitonin-gene related peptide. Nature 339, 701^703. Desclin, J.C., Colin, F., 1980. The olivocerebellar system. II. Some ultrastructural correlates of inferior olive destruction in the rat. Brain Res. 187, 29^46. Fontaine, B., Klarsfeld, A., Changeux, J.-P., 1987. Calcitonin gene-related peptide and muscle activity regulate acetylcholine receptor K-subunit mRNA levels by distinct intracellular pathways. J. Cell Biol. 105, 1337^1342. Ha, B.K., Bishop, G.A., King, J.S., Burry, R.W., 2000. Corticotropin releasing factor induces proliferation of cerebellar astrocytes. J. Neurosci. Res. 62, 789^798.

NSC 5274 29-11-01

670

S. Morara et al.

Haas, C.A., Reddington, M., Kreutzberg, G.W., 1991. Calcitonin gene-related peptide stimulates the induction of c-fos gene expression in rat astrocyte cultures. Eur. J. Neurosci. 3, 708^712. Ha¨mori, J., 1969. Development of synaptic organization in the partially agranular and in the transneuronally atrophied cerebellar cortex. In: Llina©s, R. (Ed.), Neurobiology of Cerebellar Evolution and Development. AMA/ERF, Chicago, IL, pp. 845^858. Higashi, Y., Murayama, S., Pentchev, P.G., Suzuki, K., 1995. Peripheral nerve pathology in Niemann-Pick type C mouse. Acta Neuropathol. 90, 158^163. Hildebrand, C., Bowe, C.M., Nilsson Remahl, I., 1994. Myelination and myelin sheath remodelling in normal and pathological PNS nerve ¢bers. Prog. Neurobiol. 43, 85^141. Iversen, L.L., 1995. Neuropeptides : promise unful¢lled? Trends Neurosci. 18, 49^50. Kawaguchi, S., Yamamoto, T., Mizuno, N., Iwahori, N., 1975. The role of climbing ¢bers in the development of the Purkinje cell dendrites. Neurosci. Lett. 1, 301^304. Kornguth, S.E., Scott, G., 1972. The role of climbing ¢bers in the formation of Purkinje cell dendrites. J. Comp. Neurol. 146, 61^82. Kubota, Y., Inagaki, S., Shimada, S., Kito, S., Zaidi, M., Girgis, S.I., MacIntyre, I., Tohyama, M., 1987. Transient appearance of calcitonin generelated peptide-like immunoreactive ¢bers in the developing cerebellum of the rat. Brain Res. 415, 385^388. Larramendi, L.M.H., 1969. Analysis of synaptogenesis in the cerebellum of the mouse. In: Llina©s, R. (Ed.), Neurobiology of Cerebellar Evolution and Development. AMA/ERF, Chicago, IL, pp. 803^843. Larramendi, L.M.H., Victor, T., 1967. Synapses on the Purkinje cell spines in the mouse. An electron microscopic study. Brain Res. 5, 15^30. Lazar, P., Reddington, M., Streit, W., Raivich, G., Kreutzberg, G.W., 1991. The action of calcitonin gene-related peptide on astrocyte morphology and cyclic AMP accumulation in astrocyte cultures from neonatal rat brain. Neurosci. Lett. 130, 99^102. Maley, B.E., 1990. Ultrastructural identi¢cation of neuropeptides in the central nervous system. J. Electron Microsc. Tech. 15, 67^80. Martin, G.F., Culberson, J.L., Hazlett, J.C., 1983. Observations on the early development of ascending spinal pathways. Studies using the North American opossum. Anat. Embryol. 166, 191^207. Mason, C.A., 1986. Axon development in mouse cerebellum : embryonic axon forms and expression of synapsin I. Neuroscience 19, 1319^1333. Mason, C.A., Christakos, S., Catalano, S.M., 1990. Early climbing ¢ber interactions with Purkinje cells in the postnatal mouse cerebellum. J. Comp. Neurol. 297, 77^90. Mason, C.A., Gregory, E., 1984. Postnatal maturation of cerebellar mossy and climbing ¢bers: transient expression of dual features on single axons. J. Neurosci. 4, 1715^1735. Meller, K., Glees, P., 1969. The development of the mouse cerebellum. A Golgi and electron microscopical study. In: Llina©s, R. (Ed.), Neurobiology of Cerebellar Evolution and Development. AMA/ERF, Chicago, IL, pp. 783^801. Morara, S., Marcotti, W., Provini, L., Rosina, A., 1997. Neuropeptide Y (NPY) expression is up-regulated in the rat inferior olive during development. NeuroReport 8, 3743^3747. Morara, S., Provini, L., Rosina, A., 1989. CGRP expression in the rat olivocerebellar system during postnatal development. Brain Res. 504, 315^ 319. Morara, S., Rosina, A., Provini, L., 1992. CGRP as a marker of the climbing ¢bers during the development of the cerebellum in the rat. Ann. NY Acad. Sci. 657, 461^463. Morara, S., Rosina, A., Provini, L., Forloni, G., Caretti, A., Wimalawansa, S.J., 2000. Calcitonin gene-related peptide receptor expression in the neurons and glia of developing rat cerebellum : an autoradiographic and immunohistochemical analysis. Neuroscience 100, 381^391. Morara, S., Sternini, C., Provini, L., Rosina, A., 1995. Developmentally regulated expression of K- and L-calcitonin gene-related peptide mRNA and calcitonin gene-related peptide immunoreactivity in the rat inferior olive. J. Comp. Neurol. 354, 27^38. Morara, S., van der Want, J.J.L., de Wolf, A., Provini, L., Rosina, A., 1993. Ultrastructural study of CGRP immunoreactivity in the neonatal cerebellum. Soc. Neurosci. Abstr. 19, 1215. Morris, R.J., Beech, J.N., Heizmann, C.W., 1988. Two distinct phases and mechanisms of axonal growth shown by primary vestibular ¢bres in the brain, demonstrated by parvalbumin immunohistochemistry. Neuroscience 27, 571^596. Mugnaini, E., 1969. Ultrastructural studies on the cerebellar histogenesis. II. Maturation of nerve cell populations and establishment of synaptic connections in the cerebellar cortex of the chick. In: Llina©s, R. (Ed.), Neurobiology of Cerebellar Evolution and Development. AMA/ERF, Chicago, IL, pp. 749^782. Palacios-Pru«, E.L., Palacios, L., Mendoza, R.V., 1981. Synaptogenetic mechanisms during chick cerebellar cortex development. J. Submicrosc. Cytol. 13, 145^167. Palay, S.L., Chan-Palay, V., 1974. Cerebellar Cortex. Cytology and Organization. Springer-Verlag, Berlin. Parsons, A.M., Seybold, V.S., 1997. Calcitonin gene-related peptide induces the formation of second messengers in primary cultures of neonatal rat spinal cord. Synapse 26, 235^242. Priller, J., Haas, C.A., Reddington, M., Kreutzberg, G.W., 1995. Calcitonin gene-related peptide and ATP induce immediate early gene expression in cultured rat microglial cells. Glia 15, 447^457. Ramo¨n y Cajal, S., 1911. Histologie du Syste©me Nerveux de l'Homme et des Verte¨bre¨s, vol. II. Maloine, Paris. Rosina, A., Provini, L., Morara, S., 1989. Development of olivocerebellar somatotopic maps. In: Strata, P. (Ed.), The Olivocerebellar System in Motor Control. Exp. Brain Res. Ser. 17. Springer-Verlag, Berlin, pp. 150^154. Rubenstein, J.L., 2000. Intrinsic and extrinsic control of cortical development. Novartis Found. Symp. 228, 67^75. Schoen, S.W., Graeber, M.B., To¨th, L., Kreutzberg, G.W., 1991. Synaptic 5P-nucleotidase is transient and indicative of climbing ¢ber plasticity during the postnatal development of rat cerebellum. Dev. Brain Res. 61, 125^138. Sotelo, C., Arsenio-Nunes, M.L., 1976. Development of Purkinje cells in absence of climbing ¢bers. Brain Res. 111, 389^395. Takeda, T., Maekawa, K., 1989. Transient direct connection of vestibular mossy ¢bers to the vestibulocerebellar Purkinje cells in early postnatal development of kittens. Neuroscience 32, 99^111. Tanaka, J., Nakamura, H., Miyawaki, S., 1988. Cerebellar involvement in murine sphingomyelinosis : a new model of Niemann-Pick disease. J. Neuropathol. Exp. Neurol. 47, 291^300. Terrado, J., Gerrikagoitia, I., Dominguez, L., Raldua, D., Martinez-Milla¨n, L., Sarasa, M., 1999. Expression of the genes for K-type and L-type calcitonin gene-related peptide during rat embryogenesis. Neuroscience 92, 713^727. Terrado, J., Gerrikagoitia, I., Martinez-Milla¨n, L., Pascual, F., Climent, S., Muniesa, P., Sarasa, M., 1997. Expression of the genes for K-type and L-type calcitonin gene-related peptide during postnatal rat brain development. Neuroscience 80, 951^970. Ueyama, T., Houtani, T., Nakagawa, H., Baba, K., Ikeda, M., Yamashita, T., Sugimoto, T., 1994. A subpopulation of olivocerebellar projection neurons express neuropeptide Y. Brain Res. 634, 353^357. van den Pol, A.N., Gorcs, T., 1986. Synaptic relationships between neurons containing vasopressin, gastrin-releasing peptide, vasoactive intestinal polypeptide, and glutamate decarboxylase immunoreactivity in the suprachiasmatic nucleus : dual ultrastructural immunocytochemistry with gold-substituted silver peroxidase. J. Comp. Neurol. 252, 507^521.

NSC 5274 29-11-01

Climbing ¢ber synaptogenesis

671

van der Want, J.J., Voogd, J., 1987. Ultrastructural identi¢cation and localization of climbing ¢ber terminals in the fastigial nucleus of the cat. J. Comp. Neurol. 258, 81^90. Voogd, J., Bigare¨, F., 1980. Topographical distribution of olivary and cortico nuclear ¢bers in the cerebellum : a review. In: Courville, J., De Montigny, C., Lamarre, Y. (Eds.), The Inferior Olivary Nucleus: Anatomy and Physiology. Raven, New York, pp. 207^234. West, M.J., Del Cerro, M., 1976. Early formation of synapses in the molecular layer of the fetal rat cerebellum. J. Comp. Neurol. 165, 137^160. White, R.H., Michaud, N.A., 1981. Disruption of insect photoreceptor membrane by divalent ions : dissimilar sensitivity of light- and darkadapted mosquito rhabdomeres. Cell Tissue Res. 216, 403^411. Williams, D.S., 1980. Ca‡‡ -induced structural changes in photoreceptor microvilli of diptera. Cell Tissue Res. 206, 225^232. î man, K., Hokfelt, T., 1995. Secretory pathways of neuropeptides in rat lumbar dorsal root ganglion neurons and e¡ects of peripheral Zhang, X., A axotomy. J. Comp. Neurol. 352, 481^500. (Accepted 22 August 2001)

NSC 5274 29-11-01