Clusterin/ApoJ expression is associated with ... - Semantic Scholar

1635

Journal of Cell Science 110, 1635-1645 (1997) Printed in Great Britain © The Company of Biologists Limited 1997 JCS9526

Clusterin/ApoJ expression is associated with neuronal apoptosis in the olfactory mucosa of the adult mouse Denis Michel1,*, Emmanuel Moyse2, Alain Trembleau1,†, François Jourdan2 and Gilbert Brun1 1Laboratoire de Biologie Moléculaire et Cellulaire, UMR49 CNRS-Ecole Normale Supérieure de Lyon, 46 Allée d’Italie, 69364 Lyon cedex 07, France 2Laboratoire de Physiologie Neurosensorielle, CNRS ERS 5643, Université Claude Bernard-Lyon I, F-69622, Villeurbanne Cedex, France

*Author for correspondence †Present address: Ecole Normale Supérieure, Développement et Evolution du Système Nerveux, URA CNRS 1414, 46 rue d’Ulm, 75230 Paris cedex 05, France

SUMMARY The molecular events orchestrating neuronal degeneration and regeneration remain poorly understood. Attempts at identifying genes specifically expressed during these processes, have constantly led to the (re)isolation of the clusterin/ApoJ gene, whose expression is highly reactive to injury in a wide variety of tissues. To get insight into the function of clusterin in neuron loss, we have assessed in detail the clusterin gene expression in an experimental model of neurodegeneration, using the peripheral olfactory system of adult mouse. The sensory neurons of olfactory nasal mucosa can be massively induced to degenerate in vivo, by surgical removal of their only synaptic target: the olfactory bulb. We have previously shown that this neuron loss results from a near-synchronized induction of apoptosis genetic programs. We present here evidence that clusterin gene expression is tightly correlated to the onset of neuronal apoptoses in lesioned olfactory mucosae. The simultaneous preparation of DNA and RNA from the same

tissue samples reveals that a strong clusterin mRNA accumulation coincides with the wave of nucleosome-sized DNA fragmentation. However, double detection of apoptotic nuclei by the TUNEL method and of clusterin messengers by in situ hybridization revealed that the clusterin gene expression is not induced in dying neurons, but in the glial sheath surrounding the axon bundles of degenerating olfactory neurons. Clusterin immunocytochemistry reveals that the clusterin protein accumulates not only in these producing cells, but also in the olfactory epithelium, suggesting the possibility of clusterin internalization by cells located at a distance from the synthesis loci. In view of this localization and of the activities of the clusterin protein reported so far, possible functions of clusterin in nervous plasticity are discussed.

INTRODUCTION

lished (Michel et al., 1992; May and Finch, 1992). Although several normal populations of neuronal and glial cells contain clusterin mRNA (Michel et al., 1992; Danik et al., 1993), higher clusterin mRNA levels are observed after deleterious experimental treatments, like surgical (Lampertetchells et al., 1991; Pasinetti et al., 1993), or excitotoxic (Michel et al., 1992; Danik et al., 1993) brain injuries, but also in pathological brains from Alzheimer’s diseased human patients (May et al., 1990), scrapie infected hamsters (Duguid et al., 1989), or at epileptic foci (Danik et al., 1991). Clusterin mRNA accumulation has also been observed in retinitis-pigmentosa (Jones et al., 1992), and shown to coincide with the time of photoreceptor cell deaths in mouse models of this disease (Wong et al., 1994). Purkinje cells whose apoptosis is induced by the lurcher gene, were shown to contain high levels of clusterin mRNAs prior to their death (Norman et al., 1995). Besides, the clusterin protein has been found associated with dystrophic neurites (McGeer et al., 1992), amyloid plaques (Choi-Miura et al., 1992) and the soluble form of beta amyloid protein (Matsubara et al., 1995) from Alzheimer’s diseased human

The clusterin mRNA has been isolated on the basis of its accumulation in a classical model of tissue involution through apoptotic cell deaths: the ventral prostate regression after androgen deprivation (Leger et al., 1987). Clusterin gene expression has then been associated with numerous other cases of programmed cell death, such as those occurring during normal regression of temporary embryonic structures (Buttyan et al., 1989), in the mammary gland involution early after lactation (Strange et al., 1992; Guenette et al., 1994a; Lund et al., 1996), in the remodeling uterus (Brown et al., 1995) or in kidney after experimental obstruction (Buttyan et al., 1989; Pearse et al., 1992). Such striking coincidences between clusterin expression and the onset of programmed cell deaths have logically raised the hypothesis that clusterin could play a direct role in apoptosis (Buttyan et al., 1989; Guenette et al., 1994b; Norman et al., 1995). In the nervous system, a close relationship between neurodegeneration and clusterin gene expression has been estab-

Key words: Clusterin/ApoJ, Neuronal apoptosis, Olfactory system, Glial cell, Surgical synaptic target ablation

1636 D. Michel and others brains. The ischaemic, but not the normal, human Purkinje cells are also intensely immunostained with anti-clusterin antibodies (Yasuhara et al., 1994). In spite of these numerous observations, the precise function of clusterin in damaged nervous system remains to be elucidated. In fact, answers to several basic questions would be determinant for understanding the biological involvement of clusterin in neurodegeneration: (i) is clusterin expression linked to degeneration itself or to subsequent processes of nervous tissue reorganization? (ii) is clusterin expressed by degenerating or surviving cells in the lesioned nervous system? and (iii) where is the clusterin protein located relative to the clusterin-producing cells? We have addressed simultaneously these different points with a favourable in vivo model: the peripheral olfactory system of adult mammals. The olfactory turbinates are lined up by an almost pure population of easily accessible sensory neurons, which belong to a very simple network since they all connect to a single synaptic target: the mitral cells of the olfactory bulb. Their degeneration can be massively induced by surgical removal of the olfactory bulb (Monti-Graziadei and Graziadei, 1979). We have recently shown that this neurodegeneration results from the near synchronized induction of apoptotic genetic programs in vivo (Michel et al., 1994). The olfactory neuroepithelium is also unique in its ability to completely regenerate after lesion, even in the adult, due to the presence of neurogenetic precursor cells, named globose basal cells, residing in the depths of the epithelium (MontiGraziadei and Graziadei, 1992; Caggiano et al., 1994). In addition, because of the absence of anatomical connections between the two symmetrical sides of the peripheral olfactory system, it is possible to cause a unilateral lesion by eliminating a single olfactory bulb out of the two, and thus to obtain both control and targetless olfactory neuroepithelia in the same transverse tissue section of olfactory turbinates. Finally, the great number, the synchrony and the anatomical compartmentation of apoptotic neurons at the histological level render this model highly appropriate for the study of relationships between clusterin induction and apoptosis in vivo.

MATERIALS AND METHODS Animals and tissue preparation Adult, three-month-old mice of the C57Bl/6J strain were used in the present study. Unilateral or bilateral bulbectomies were performed under anesthesia with equithesine (0.3 g/0.3 ml per 100 g body weight) as follows. A 1 mm wide hole was first dug with a dental drill in the dorsal face of the skull 2 mm rostro-laterally to the bregma; a curved glass pipet connected to a vacuum pump was then introduced through the hole, the underlying olfactory bulb was totally removed by aspiration and the resulting space filled up with gelfoam. Bulbectomized mice were housed and allowed to recover from anesthesia in individual cages with food and water ad libitum until sacrifice. For kinetic assessment of apoptosis and of gene expression, mice were killed at various intervals, ranging from one hour to 8 days, following bilateral bulbectomy; the bulk of olfactory turbinates was immediately dissected out of each mouse head and snap-frozen at −80°C. For in situ labeling of either clusterin mRNA or apoptotic nuclei, mice were killed by decapitation, and the bulk of the olfactory

turbinates was rapidly dissected out of the heads and immersed under vacuum in histological embedding medium (OCT, Miles, USA). OCTembedded turbinates were frozen by immersion in isopentane at −50°C, and stored at −80°C until sectioning. Serial 20 µm thick, coronal sections were cut at −20°C on a cryostat (Reichert) and touchmounted on slides that had been previously coated with 0.5% polyL-lysine (Sigma) after heat-sterilization. Slide-mounted sections were kept frozen at −20°C until use. From each animal, 15-20 sets of 4 consecutive, adjacent sections were collected for in situ hybridization with sense and antisense clusterin riboprobes and TUNEL labeling. For clusterin immunocytochemistry, animals taken 48 or 75 hours after bulbectomy were deeply anesthetized and perfused transcardially with 4% paraformaldehyde in 0.1 M phosphate buffered saline (PBS), pH 7.4. The bulk of the olfactory turbinates was dissected out of the head and rinsed in PBS, then immersed in a 30% sucrose solution in PBS. Following vacuum-embedding in OCT, the nasal bulk was frozen into isopentane at −50°C and stored at −80°C until sectioning. Serial 20 µm thick coronal sections were cut in a cryostat and collected on gelatin-coated slides. Five animals of the same strain were sham operated and submitted to the same protocol in order to assess the contingent expression of clusterin in the olfactory mucosa of intact animals. Reverse transcription-mediated PCR (RT-PCR) RNAs were extracted as previously described (Michel et al., 1994), from the same olfactory turbinates which were used to quantify DNA ladderization intensity, in order to compare the time-course of the genomic DNA fragmentation, with that of clusterin mRNA accumulation. Reverse transcriptions were done starting with 0.3 µg RNA from each sample in the presence of 50 µM random hexamer primers as described (Michel et al., 1994). PCRs were performed using 1/20 of total cDNA, 1 mM primers, 1 mM dNTP and 0.1 mCi of [α32P]dCTP. Forward and reverse primers were defined in different exons. They were, for clusterin: 5′-TCTCCAGCAGGGAGTCGATGCG-3′ and 5′-TGATGGCCCTCTGGGAGGAGTG-3′; for β-actin: 5′-TTGCTGATCCACATCTGCTG-3′ and 5′-GACAGGATGCAGAAGGAGAT-3′. Amplified fragments of clusterin and β-actin cDNAs are 177 bp and 146 bp long, respectively. To determine the clusterin/β-actin ratio, we amplified simultaneously the clusterin and β-actin cDNAs, by adding the four corresponding primers in the PCR tubes. 1/5 of the clusterin/β-actin co-amplifications were electrophoresed through a 4% polyacrylamide (38:2; acrylamide:bisacrylamide) gel. After UV-visualization, the gel was dried and autoradiographed for 2 hours with Amersham MP films to quantify the cDNA amplifications by 32P-incorporation. In situ hybridization Frozen cryostat sections of olfactory turbinates were brought to room temperature, fixed for 30 minutes at room temperature by immersion in 4% paraformaldehyde-containing PBS and rinsed three times (5 minutes each) in PBS. The sections were then treated with proteinase K, acetylated according to the method of Simmons et al. (1989). Sense and antisense clusterin riboprobes have been labeled with 33P during in vitro transcription of the coding region of the rat clusterin cDNA (generously supplied by Dr Michael Griswold, Washington State University). The transcription reaction contained 1 µg of linearized DNA template, 100 µCi [33P]UTP (3,000 Ci/mmol, Amersham), ATP, CTP, GTP (10 µM each), 2 µl transcription buffer (Boehringer Mannheim), 20 units of SP6 or T7 RNA polymerases (Boehringer Mannheim) and 20 units of RNase inhibitor (Boehringer Mannheim), in a final volume of 20 µl. The labeled probe was digested with DNase I, ethanol precipitated, and resuspended in Tris-EDTA (10 mM-1 mM) buffer. Following dehydration, tissue sections were incubated overnight at 52°C in hybridization buffer containing radioactive probe (107 cpm/ml), 50% formamide, 10 mM Tris-HCl, pH 8.0, 0.3 M NaCl, 1 mM EDTA, 1× Denhardt’s solution (0.02% BSA, 0.02% polyvinylpyrrolidone, 0.02% tri-sodium citrate), and 0.5 mg/ml

Clusterin and neuronal apoptosis in vivo 1637 tRNA. Following hybridization, the sections were washed four times (5 minutes each) in 2× SSC (1× SSC = 0.15 M NaCl, 0.015 M trisodium citrate), and RNase A treated (30 minutes at 37°C) in a buffer containing 10 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 1 mM EDTA and 0.02 mg/ml RNase A (Sigma). Then, the sections were washed at room temperature as follows: 2× SSC, 2× 5 minutes; 1× SSC, 10 minutes; 0.5× SSC, 10 minutes, then 0.2× SSC for 30 minutes at 60°C. Finally, the sections were rinsed in 0.1× SSC at room temperature, dehydrated in graded ethanols and air-dried. Section-hybridized radioprobe molecules were detected by dry and wet autoradiography as follows. All sections were first autoradiographed simply by apposition onto Amersham MP films in light-proof cassettes for 2 to 4 days; films were developed in a developer apparatus (P2000, 3M). Sections were then autoradiographed by dipping into liquid nuclear emulsion (Amersham), exposed for 10-15 days in light-proof, sealed boxes, developed in Kodak D-19 for 3 minutes at 17°C, fixed in Kodak Unifix for 10 minutes at 4°C and rinsed under tap water and distilled water; sections were then Nissl-counterstained with 0.5% acidified Cresyl Violet, dehydrated-defatted in graded ethanols and xylene, and coverslipped with Depex. In situ labeling of nuclear DNA fragmentation Visualization of DNA laddering in situ was performed by the TUNEL method (Gavrieli et al., 1992). Frozen, slide-mounted tissue sections were brought to room temperature, fixed for 30 minutes at room temperature by immersion in 4% paraformaldehyde-containing 0.1 M PBS at 4°C and rinsed 2× 15 minutes in PBS. Slides were then incubated for 15 minutes at room temperature with proteinase K at 20 µg/ml PBS and rinsed 2× 5 minutes with PBS. Slides were further treated with 2% H2O2 in PBS for 5 minutes at room temperature in order to block endogenous peroxidases, and rinsed in PBS. After 5 minutes rinse in 30 mM Tris-HCl, pH 7.5, containing 140 mM sodium cacodylate and 1 mM cobalt chloride, slides were incubated with biotinylated dUTP (Boehringer, 6 nM/ml) and terminal transferase (Boehringer, 300 units/ml) in TdT buffer for 1 hour at 37°C. Slides were then rinsed for 15 minutes at room temperature in 300 mM sodium chloride-30 mM sodium citrate solution, further rinsed in PBS and incubated for 10 minutes with 2% bovine serum albumin (Eurobio) in PBS. After rinsing in PBS, slides were incubated for 30 minutes with a horseradish peroxidase-coupled ABC system (Vector) prepared in PBS as described in the manufacturer’s instructions, rinsed 5 minutes in PBS and 5 minutes in 50 mM Tris-HCl, pH 7.5. Slides were then reacted with 0.05% diaminobenzidine (Sigma) in 50 mM Tris-HCl, pH 7.5, containing 0.6% H2O2 and 0.03% nickel chloride. After 5-7 minutes reaction at room temperature, slides were rinsed in cold buffer, dehydrateddefatted through graded ethanols and xylene and coverslipped with Depex. Clusterin immunocytochemistry The mounted sections were incubated for 30 minutes in PBS containing Triton X-100 (0.3%) and normal serum (donkey or rabbit, depending on the secondary antibody used) and immersed for 20 hours at 4°C in the same medium containing clusterin antibody (polyclonal, sheep anti-rat clusterin, Quidel, San Diego, CA) at a dilution of 1/1,000 or 1/5,000. Following incubation in the primary antibody, the sections were rinsed several times in PBS and processed for visualization of the reactive sites. The first staining procedure used a FITC-conjugated secondary antibody (anti-sheep IgG, FITCconjugate; Sigma). A second set of slides was stained using the avidin-biotin-peroxidase technique (Vectastain kit, Vector Labs, USA). The tissue-bound peroxidase was visualized with diaminobenzidine (DAB, 25 mg/100 ml). Finally, a few sections adjacent to immunocytochemically stained ones were counterstained with Neutral Red or Hoechst. Control experiments were carried out by

omitting the primary antibody in the incubation medium of several slides. Clusterin/GFAP double immunofluorescence In order to assess the contingent glial nature of clusterin immunoreactive cells, some mucosae were submitted to a double immunofluorescence detection of clusterin and GFAP, using FITC and TRITC as fluorescent probes conjugated with respective secondary antibodies. The double immunofluorescence procedure was performed either sequentially on the same sections or separately on 20 µm thick adjacent sections. The procedure for clusterin immunofluorescence has been described previously. For GFAP immunocytochemistry, sections were pre-incubated in PBS containing Triton X-100 and normal goat serum (5%) for 30 minutes and then immersed for 20 hours at 4°C in the same medium containing the GFAP antibody (Dako) at a concentration of 1/500 or 1/1,000. Then, the sections were rinsed several times and immersed in PBS containing a TRITC-conjugated secondary antibody (anti-rabbit IgG, TRITC-conjugated, Sigma) at a concentration of 1/500. Finally, the sections were rinsed in PBS, mounted in Fluoprep (BioMérieux, France) and observed with a Zeiss fluorescence microscope. Control experiments were carried out by omitting the primary antibody in the incubation medium of several slides.

RESULTS Localization of apoptotic nuclei in olfactory mucosa involuting after bulbectomy The severe involution of the olfactory mucosa after synaptic target ablation is evidenced on a Nissl-stained transverse section of olfactory turbinates from a mouse killed 40 hours after unilateral bulbectomy (Fig. 1B). This neuroepithelium degeneration was previously shown to be preceded by a peak of genomic DNA fragmentation, typical of apoptotic programmed cell deaths (Michel et al., 1994). The cellular site of this DNA fragmentation was assessed here in situ, using the TUNEL method (Fig. 2). Comparison of each TUNELprocessed sample with the adjacent Nissl-stained section clearly shows that olfactory mucosa, on the side ipsilateral to the bulbectomy, is strongly enriched in apoptotic nuclei (Fig. 2A,B). By contrast, contralateral turbinates are almost devoid of TUNEL staining, except for a few scattered reactive nuclei (Fig. 2C,D). The TUNEL-stained nuclei of target-deprived mucosa were strictly confined to the neuroepithelium and more precisely to its inner part containing the nuclei of mature sensory neurons (Fig. 2A,B). No TUNEL reactivity is found in the superficial sub-layer of the olfactory epithelium containing the cell bodies of sustentacular cells, indicating that this non-neuronal cell population does not participate in the massive cell loss induced by bulbectomy. No more TUNEL reactivity is found in the lamina propria that underlies the basal lamina of the neuroepithelium and consists of olfactory axon bundles, Bowman glandular acini, blood vessels and connective tissue. Time-course of clusterin mRNA accumulation after olfactory bulbectomy Occurrence of clusterin mRNA was then assessed in olfactory turbinates of mice previously bulbectomized. Since the size of the tissue samples was not sufficient to allow northern blot experiments, we have used RNA minipreparations as

1638 D. Michel and others templates for reverse transcription-mediated PCR and tested the relative variations of clusterin and β-actin mRNAs. βActin and clusterin cDNAs from each tissue sample were coamplified in the same PCR tube, thus allowing us to obtain ideally normalized clusterin/β-actin ratios. This co-PCR procedure applied to olfactory turbinate extracts from mice killed at various post-bulbectomy intervals shows that the relative variations of the two mRNAs are clearly disconnected (Fig. 3). While the β-actin mRNA remains roughly constant before and after bulbectomy, the clusterin mRNA, barely detectable in control turbinates (Fig. 3, first lane), is strongly overaccumulated after bulbectomy, reaching transiently a level similar to that of the β-actin mRNA. However, clusterin mRNA induction is only transient and slowly returns to control values 120 hours post-operation and does not increase

Fig. 1. (A) Light microscopic structure of olfactory mucosa in control adult mouse, as observed on a semi-thin (1 µm-thick) section of prefixed olfactory organ. A basal lamina (bl) separates olfactory epithelium (above) from underlying lamina propria that harbors axon bundles of olfactory nerve (arrows) and Bowman glands (bg). Note the pseudostratified aspect of olfactory neuroepithelium, due to the laminar distribution of the respective nuclei of horizontal (hbc) and globose (gbc) basal cells, olfactory neurons (on) and supporting cells (sc). The upper sub-layer of olfactory epithelium is made up mainly of olfactory neuron dendrites at the tip of which dendrite knobs (dk) protrude into the nasal cavity (N). Bar, 25 µm. (B) Involution of the olfactory mucosa following unilateral olfactory. A section of the olfactory mucosa from a unilaterally bulbectomized mouse was Nissl-stained. This section includes the nasal septum (S) and olfactory turbinates of the two sides of the organ. Olfactory turbinates and septum are covered by olfactory neuroepithelium (OE) lying onto lamina propria (LP), as figures in A. Note the dramatic thickness reduction of olfactory neuroepithelium on the side that has been deprived of its synaptic target, ipsilateral to olfactory bulbectomy (left). N, nasal cavity. Bar, 100 µm.

again at longer time periods (data not shown). The time-course of quantitatively assessed clusterin mRNA accumulation is shown in Fig. 4. It can be compared with that of genomic DNA ladderization observed in the same conditions (Michel et al., 1994), prepared from the same tissue samples, dissected out from the same animals. Clusterin mRNA content increases significantly as soon as 8 hours after bulbectomy and then parallels the pattern of DNA fragmentation up to 32 hours. Clusterin mRNA decrease is delayed by about 40 hours from that of DNA fragmentation. In situ localization of clusterin mRNA in olfactory turbinates of unilaterally bulbectomized mice We further assessed the cellular site of clusterin gene expression by in situ hybridization of clusterin mRNAs on

Clusterin and neuronal apoptosis in vivo 1639 sections of olfactory turbinates from mice killed 75 hours after unilateral olfactory bulbectomy. Clusterin riboprobe hybridization was higher in the bulbectomized side of olfactory turbinates than in the control side, as shown by film autoradiography (Fig. 5A) and by microscopic autoradiograms (Fig. 5B,C). Clusterin gene expression is thus induced by olfactory bulbectomy within the mucosa of olfactory turbinates ipsilateral to the lesion, like neuronal apoptoses. Clusterin gene induction thus seems not only temporally, but also spatially related to olfactory neurodegeneration. In order to assess whether bulbectomy-induced apoptotic nuclei and clusterin mRNA accumulation within olfactory mucosa concern the same cells, we have detected these two parameters in adjacent tissue sections of olfactory mucosa from mice killed 40 hours after ipsilateral bulbectomy. In situ hybridization with the clusterin antisense riboprobe resulted in dense and heterogeneous labeling within the lamina propria, rows of silver grains conspicuously overlaying the glial sheath of olfactory axon bundles (Fig. 6B,C). No specific hybridization signal was found either above the basal lamina in olfactory epithelium, or in axon bundles per

se (Fig. 6B,C). Conversely, apoptotic nuclei were only observed in the neuroepithelium (Fig. 6A). This experiment demonstrates that the localizations of apoptotic nuclei and of clusterin messengers are mutually exclusive, and thus that clusterin mRNAs do not accumulate in dying neurons, but in surviving, non-neuronal cells located in the underlying lamina propria. Clusterin immunocytochemistry The mere detection of clusterin mRNA is not adequate to determine the localization of the related protein, since clusterin can either remain inside the producing cells (Reddy et al., 1996), or be targeted towards the secretion and even to the receptor-dependent internalization pathways (Kounnas et al., 1995). Hence, we wanted to visualize directly the clusterin protein. Fig. 7 shows the distribution pattern of clusterin immunoreactivity 75 hours after bulbectomy, as revealed by immunofluorescence. Specific labeling has been found widely distributed in the olfactory mucosa, but two main sites of clusterin accumulation could be identified, and obviously confirmed with the most sensitive avidin-biotin-

Fig. 2. Selective presence of apoptotic nuclei at the level of the involuting neuroepithelium. The DNA end labeling procedure (TUNEL method) allows us to specifically reveal nuclei of cells undergoing apoptosis, which appear in black. The distribution of apoptotic nuclei is clearly asymmetrical, with a strong enrichment in the lesioned side (A) as compared to the control side (C) from the same olfactory mucosa. The abundance of these apoptotic nuclei in the lesioned side is responsible for the striking reduction of the neuroepithelium thickness (compare hematoxylin-stained sections, B and D). Apoptotic nuclei are specifically located in the inner part of the neuroepithelium, at the level of the neuron body layer. By contrast, the lamina propria, located underneath the basal lamina of the neuroepithelium, is devoid of any apoptotic nucleus, indicating that apoptoses only concern olfactory neurons. N, nasal cavity; OE, olfactory epithelium. Bar, 50 µm.

1640 D. Michel and others concentrated in the unique cell type found in this area, i.e. glial cells which ensheath olfactory axons (Fig. 8B). Immunoreactive glial cell bodies were located either at the periphery or inside the olfactory nerve bundles whereas the weaker and more diffuse labeling observed inside the nerves (Figs 7A, 8B) was most likely due to clusterin accumulation in glial cell processes. In contrast to clusterin immunoreactivity in the epithelium, this deep clusterin accumulation in the lamina propia appeared more frequently in animals killed earliest (48 hours) after bulbectomy. Control sections incubated without primary antibody displayed only a faint diffuse autofluorescence in the superficial lamina propria which could not be mistaken for specific labeling (data not shown). While there is no doubt about the glial nature of immunoreactive cells located around or inside the olfactory nerve, the identity of large immunoreactive sites located in the deep olfactory epithelium cannot be simply inferred from their location. In order to assess their potential glial nature, we have studied the comparative distribution of clusterin and GFAP in the olfactory mucosa using double immunofluorescence. As shown in Fig. 9, GFAP immunoreactivity was confined to the lamina propia, and was obviously concentrated in glial cell bodies and processes ensheathing olfactory nerve bundles, in accordance with previous studies (Barber and Dahl, 1987). Following the clusterin immunofluorescence procedure, clusterin immunoreactivity appears to be of relatively low intensity in the lamina propria, probably due to the low concentration of clusterin stored in glial cell bodies and processes. While clusterin and GFAP immunoreactivities are regionally co-localized in the lamina propia, they are obviously mismatched in the epithelium. The most intense clusterin accumulation took place in large cell bodies located in the deep olfactory epithelium, just above the basal lamina, which proved completely devoid of GFAP immunoreactivity, demonstrating that cells accumulating clusterin in the epithelium were not glial cells.

Fig. 3. Competitive amplification of β-actin and clusterin mRNAs by co-PCR. RNAs were prepared from a control mouse (first lane) or from mice previously bulbetomized for the indicated time periods, and reverse-transcribed using random hexamers as primers. The single-stranded cDNAs thus obtained were used as templates for PCR amplification specific of clusterin and β-actin. Both amplifications were carried out simultaneously in the same PCR tubes and starting from the same reverse-transcription mixtures. Amplification product amounts were quantified by ethidium intercalation (top panel) or radioactivity incorporation (bottom panel) yielded identical comparative results.

peroxidase method (Fig. 8). As shown in Figs 7 and 8, prominent immunoreactive sites are present in the deep part of the olfactory epithelium, just above the basal lamina (Figs 7A, 8A). A weaker and more diffuse labeling has also been occasionally observed more superficially in the epithelium. However, the most superficial layer containing supporting cell bodies was always immunoreactive. All these epithelial sites of clusterin accumulation have been found widely distributed in most areas of the olfactory epithelium, at both post-operative delays. However, their frequency and intensity was apparently increasing with time, being higher at the longest post-operative delay (72 hours). In addition to this unexpected epithelial labeling, clusterin immunofluorescence was also observed in the deep olfactory mucosa (lamina propia), in close association with olfactory nerve bundles (Fig. 7). Clusterin accumulation in this area appeared as either a diffuse labeling within nerve bundles or dense immunofluorescent patches located around or inside the nerve. Identification of reactive sites at the cellular level using the more sensitive ABC staining showed that labeling was

DISCUSSION Our work allows us to answer several questions about the involvement of clusterin in neuron degeneration, thanks to some remarkable features of the peripheral olfactory system in adult mammals. The in situ tissue localization of apoptotic cells, of clusterin mRNA and of clusterin protein, was clearly

c-fos/actine

MAX

Fig. 4. Comparative time-courses of clusterin mRNA accumulation and of genomic DNA degradation. Clusterin/β-actin PCR signal ratios were blotted together with the intensity of apoptotic DNA fragmentation as a function of time after bulbectomy. The intensity of DNA ladderization was determined by densitometric quantification of the two-nucleosome-sized fragments (Michel et al., 1994). For each time point, mRNA and laddering quantifications are done starting from the same animals.

MIN

DNA ladders DNA ladders clusterine/actine clusterin/ β -actin

0

8

16

24

32

40

48

56

64

72

80

Hours after bulbectomy

88

96

104

112

120

Clusterin and neuronal apoptosis in vivo 1641 established, at the light microscopic level, by the histological compartmentation of olfactory turbinates into segregated populations of sensory neuron cell bodies (85% of all neuroepithelial cells) responding massively and quasi-synchronously to synaptic target ablation, of bundled olfactory axons and of nerve sheathing glia. Although the TUNEL method labels all nuclei containing fragmented DNA, including those of necrotic cells, in the present case, the apoptotic character of TUNELstained nuclei in the olfactory epithelium of bulbectomized mice is validated by our previous visualization of nucleosomesized DNA fragmentation, unambigously related to the apoptotic process (Michel et al., 1994). The present in situ studies have allowed us to discriminate in vivo between dying and clusterin-expressing cells. They have demonstrated that the accumulation of clusterin mRNA is closely dependent on the presence of apoptotic cells, but does not occur at the level of these cells. As revealed by in situ hybridization, clusterin signals are completely absent from the olfactory epithelium which contains apoptotic cells. This clear disconnection demonstrates that clusterin can no longer be considered as a gene involved in the apoptosis genetic program, in the olfactory neuron. This conclusion is in line with results obtained outside the nervous system, such as the existence of apparently healthy clusterin-expressing cells (Michel et al., 1992; Aronow et al., 1993) or, conversely, with cases of apoptosis without clusterin expression, like in leucocytes (Pearse et al., 1992; French et al., 1992), or in cells irradiated in vitro with ultra violet (French et al., 1994). In turn, these findings are in apparent contradiction to recent data demonstrating that in the lurcher mutant mouse strain, Purkinje cells

strongly accumulate clusterin mRNA prior to their death by apoptosis (Norman et al., 1995). This observation, however, does not imply that clusterin gene expression plays an active role in the loss of Purkinje cells. We have shown that clusterin transcripts are exclusively located in the lamina propria of the olfactory mucosa, with the most intense hybridization signal in olfactory nerve bundles. This localization of hybridization signals is highly consistent with the assumption that clusterin synthesis occurs in glial cells of the olfactory nerve, since glial ensheathing cells are the only cell body type present in this area. This statement is in accordance with previous ones that support the hypothesis that clusterin gene induction in the lesioned brain occurs mainly, if not exclusively, in glial cells (Danik et al., 1993). A partial mismatch between clusterin mRNA and protein has been observed since immunoreactive sites to anticlusterin antibody are present not only in the lamina propia but also in the olfactory epithelium which proved completely devoid of any hybridization signal for the clusterin mRNA. The possibility that clusterin immunoreactive glial cells have migrated and penetrated the epithelium might be considered since GFAP-containing glial cells of the olfactory nerve have been shown to migrate towards the basal membrane lining the olfactory epithelium during the few days following olfactory nerve transection (Barber and Dahl, 1987). However, we can rule out this hypothesis since cell bodies accumulating clusterin in the olfactory epithelium are clearly GFAP immunonegative (Fig 9). The alternative, and most plausible interpretation of our data retains the possibility of

Fig. 5. In situ hybridization of clusterin mRNA in olfactory mucosa 75 hours after unilateral bulbectomy. Asymmetrical accumulation of clusterin mRNA, more important in the ipsilateral side of the lesion, is visualized at the level of the global autoradiogram obtained by apposition onto an X-ray film (A) or of the emulsion microautoradiograms (B,C). More grains are visible in the targetless side (B) than in the control one (C). B and C derive from the same unilateraly lesioned olfactory mucosa, corresponding to that shown in A. Bar, 1 mm.

1642 D. Michel and others a fast secretion of clusterin from glial cells of the olfactory nerve, followed by internalization of the protein by target cells located mainly, if not exclusively, in the olfactory neuroepithelium. This hypothesis is highly consistent with previous demonstrations that clusterin synthesized by glial cells is secreted (Pasinetti et al., 1994) and that extracellular clusterin could be internalized by target cells (Kounnas et al., 1995). Further investigations, using immunocytochemical studies at the cellular and ultrastructural levels, are now needed for an accurate identification of cells accumulating clusterin in the olfactory mucosa, and particularly in the olfactory neuroepithelium. At all events, we must emphasize the fact that the highest concentration of clusterin was observed in the area where degenerating and regenerating processes (olfactory neuron degeneration and stem cell proliferation) occur concomitantly,

Fig. 6. Co-detection of clusterin mRNA and apoptotic nuclei. Adjacent sections of an olfactory mucosa from a mouse bulbectomized for 40 hours were submitted to the three following treatments: (A) TUNEL detection of apoptotic nuclei. (B) In situ hybridization for clusterin. (C) Hematoxylin staining. Comparison of the three pictures, corresponding to the same tissular region and observed at identical magnification, shows that clusterin mRNA and apoptotic nuclei are detected in the lamina propria and in the neuroepithelium, respectively. This mutually exclusive distribution demonstrates that clusterin is not synthesized in the dying neurones, but by lamina propria cells, informed about the physiological state of olfactory neurones through an unknown mechanism. b, olfactory nerve bundles; N, nasal cavity; OE and arrowheads, olfactory epithelium. Bar, 250 µm.

Fig. 7. Immunofluorescence detection of clusterin in the olfactory mucosa of a bulbectomized mouse, 75 hours after surgery (A). The adjacent section has been stained with Neutral Red (B) in order to display the main anatomical subdivisions. In the olfactory epithelium (oe), dense patches of immunofluorescence are visible in the deepest zone, just above the basal lamina. Smaller immunoreactive sites are also scattered in the intermediate zone of the epithelium. By contrast, the upper layer containing the supporting cell bodies appears as completely immunonegative. In the lamina propria (lp), clusterin immunoreactivity concentrates in the olfactory nerve bundles (on, arrows), as either a diffuse fluorescent signal or small dots of cellular appearance. In the Nissl stained section (B), olfactory nerve bundles correspond to the pale circular areas located in the deep mucosa, while the stained areas are Bowman acini (bg). Bar, 50 µm.

Clusterin and neuronal apoptosis in vivo 1643

Fig. 8. Immunocytochemical detection of clusterin in the olfactory mucosa 75 hours following olfactory bulb ablation, using the avidin/biotin-peroxidase method. In some areas of the olfactory epithelium (oe), large immunoreactive patches are observed in the basal area, just above the basal lamina (A, arrows). In the lamina propria (lp), immunoreactive cell bodies are present in olfactory nerve bundles (B, large arrows), either at the periphery, or inside the nerve. A fine network of immunoreactive processes can also be distinguished inside the olfactory nerve bundle (small arrows). Bars: 50 µm (A) and 25 µm (B).

in the lower part of the olfactory epithelium (Monti-Graziadei and Graziadei, 1979, 1992). Among the different purported activities of the clusterin protein, it is tempting to retain its possible involvement in a local lipid homeostasis since clusterin, also named apolipoprotein J, has been isolated as a component of a minor class of circulating lipoproteins (de Silva et al., 1990). In this respect, it can be involved in the lipid recycling between degenerating and regenerating structures, as already proposed for apolipoprotein E during reactive synaptogenesis (Poirier et al., 1991; Goodrum, 1991). Alternatively, since clusterin has a potent complement-inhibiting activity (Kirszbaum et al., 1989; Jenne and Tschopp, 1989), it has been proposed to be upregulated as a defense mechanism against immunological agressions in wounded brains (McGeer et al., 1992; Zhan et al., 1994). Indeed, the complement pathway has been implicated in brains of Alzheimer patients (McGeer et al., 1989). This hypothesis is further supported by observations that expression of complement components and clusterin are co-induced after experimental lesioning in the nervous system (Pasinetti et al., 1992; Liu et al., 1995; Rozovsky et al., 1994), probably by glial cells (Liu et al., 1995; Gasque et al., 1995). Finally, taking into account that clusterin does not complex only with complement molecules but also with other extracellular components as the secreted amyloid (Matsubara et al., 1995), one may propose clusterin as a ‘molecular cleaner’

Fig. 9. Regional distribution of clusterin (A) and GFAP (B) immunoreactive sites in two adjacent sections of an olfactory turbinate, 75 hours after bulbectomy. The pattern of clusterin accumulation is similar to the one shown in Fig. 7, with the most prominent reactives sites located in the basal part of the olfactory epithelium (oe), and a diffuse or patchy fluorescence at the level of olfactory nerve bundles. GFAP immunoreactivity (B) is exclusively present in the lamina propia (lp). Olfactory nerve bundles contain numerous immunoreactive glial cell bodies and processes. It is noteworthy that clusterin immunoreactive sites located in the neuroepithelium are obviously not immunoreactive and cannot be interpreted as migrating glial cells.

ensuring the solubilization of a variety of compounds potentially cytotoxic. Indeed, clusterin is thought to maintain amyloid (Matsubara et al., 1996; Boggs et al., 1996) as well as complement membrane attack complexes (Jenne and Tschopp, 1992) in soluble forms. Previous hypotheses assuming that clusterin is expressed by apoptotic cells, have interpreted the decrease of the clusterin mRNA level at the end of the degeneration period, as a mere consequence of the loss of clusterin-producing cells (Guenette et al., 1994a). The demonstration, in the olfactory system, that the apoptotic and the clusterin-expressing cells are clearly separated, means that the clusterin gene is in fact subjected to a stringent regulation in surviving cells, and probably depends on cell-cell interactions. The way in which the ensheathing cells are informed about the physiological

1644 D. Michel and others state of the olfactory neurons remains to be elucidated, but our comparative time-course experiments already show that this information is needed early on, just before the beginning of DNA fragmentation and when the neuron bodies are still intact. In conclusion, the present observations rule out the direct involvement of clusterin expression in the apoptosis genetic program, but at the same time strengthen the existence of a link between clusterin and apoptosis. This work was supported by the Association pour la Recherche sur le Cancer (ARC) and by the Region Rhône-Alpes.

REFERENCES Aronow, B. J., Diane Lund, S., Brown, T. L., Harmony, J. A. K. and Witte, D. P. (1993). Apolipoprotein J expression at fluid-tissue interfaces: possible role in barrier cytoprotection. Proc. Nat. Acad. Sci. USA 90, 725-729. Barber, P. C. and Dahl, D. (1987). Glial fibrillary acidic protein (GFAP)-like immunoreactivity in normal and transected rat olfactory nerve. Exp. Brain Res. 65, 681-685. Boggs, L. N., Fuson, K. S., Baez, M., Churgay, L., McClure, D., Becker, G. and May, P. C. (1996). Clusterin (Apo J) protects against in vitro amyloidbeta (1-40) neurotoxicity. J. Neurochem. 67, 1324-1327. Brown, T. L., Moulton, B. C., Baker, V. V., Mira, J. and Harmony, J. A. K. (1995). Expression of apolipoprotein J in the uterus is associated with tissue remodeling. Biol. Reprod. 52, 1038-1049. Buttyan, R., Olsson, C. A., Pintar, J., Chang, C., Bandyk, M., Ng, P.-Y. and Sawczuk, I. S. (1989). Induction of the TRPM-2 gene in cells undergoing programmed death. Mol. Cell Biol. 9, 3473-3481. Caggiano, M., Kauer, J. S. and Hunter, D. D. (1994). Globose basal cells are neuronal progenitors in the olfactory epithelium: a lineage analysis using a replication-incompetent retrovirus. Neuron 13, 339-352. Choi-Miura, N. H., Khara, Y., Fukuchi, K., Takeda, M., Nakano, Y., Tobe, T. and Tomita, M. (1992). SP-40, 40 is a constituent of Alzheimer’ amyloid. Acta Neuropathol. 83, 260-264. Danik, M., Chabot, J.-G., Mercier, C., Benabib, A.-L., Chauvin, C., Quirion, R. and Suh, M. (1991). Human gliomas and epileptic foci express high levels of a mRNA related to rat testicular sulfated glycoprotein 2, a purported marker of cell death. Proc. Nat. Acad. Sci. USA 88, 8577-8581. Danik, M., Chabot, J.-G., Hassan-Gonzalez, D., Suh, M. and Quirion, R. (1993). Localization of sulfated glycoprotein-2/clusterin mRNA in the rat brain by in situ hybridization. J. Comp. Neurol. 334, 209-227. de Silva, H., Stuart, W. D., Duvic, C. R., Wetterau, J. R., Ray, M. J., Ferguson, D. G., Albers, H. W., Smith, W. R. and Harmony, J. A. K. (1990). A 70-kDa apolipoprotein designated ApoJ is a marker for subclasses of human plasma high density lipoproteins. J. Biol. Chem. 265, 1324013247. Duguid, J. R., Bohmont, C. W., Liu, C. W. and Tourtelotte, W. W. (1989). Changes in brain gene expression shared by scrapie and Alzheimer disease. Proc. Nat. Acad. Sci. USA 86, 7260-7264. French, L. E., Sappino, A.-P., Tschopp, J. and Schifferli, J. A. (1992). Distinct sites of production and deposition of the putative cell death marker clusterin in the human thymus. J. Clin. Invest. 90, 1919-1925. French, L. E., Wohlwend, A., Sappino, A.-P., Tschopp, J. and Schifferli, J. A. (1994). Human clusterin gene expression is confined to surviving cells during in vitro programmed cell death. J. Clin. Invest. 93, 877-884. Gasque, P., Fontaine, M. and Morgan, B. P. (1995). Complement expression in human brain: Biosynthesis of terminal pathway components and regulators in human glial cells and cell lines. J. Immunol. 154, 4726-4733. Gavrieli, Y., Sherman, Y. and Ben-Sasson, S. A. (1992). Identification of programmed cell death in situ via specific labeling of nuclear DNA fragmentation. J. Cell Biol. 119, 493-501. Goodrum, J. F. (1991). Cholesterol from degenerating nerve myelin becomes associated with lipoproteins containing apolipoprotein E. J. Neurochem. 56, 2082-2086. Guenette, R. S., Corbeil, H. B., Léger, J., Wong, K., Mézl, V., Mooibroek, M. and Tenniswod, M. (1994a). Induction of gene expression during involution of the lactating mammary gland of the rat. J. Mol. Endocrinol. 12, 47-60.

Guenette, R. S., Daehlin, L., Mooibroek, M., Wong, K. and Tenniswood, M. (1994b). Thanatogen expression during involution of the rat ventral prostate after castration. J. Androl. 15, 200-211. Jenne, D. E. and Tschopp, J. (1989). Molecular structure and functional characterization of a human complement cytolysis inhibitor found in blood and seminal plasma: identity to sulfated glycoprotein 2, a constituent of rat testis fluid. Proc. Nat. Acad. Sci. USA 86, 7123-7127. Jenne, D. E. and Tschopp, J. (1992). Clusterin: the intriguing guises of a widely expressed glycoprotein. Trends Biochem. Sci. 17, 154-159. Jones, S. E., Meerabux, J. M. A., Yeats, D. A. and Neal, M. J. (1992). Analysis of differentially expressed genes in retinitis-pigmentosa retinas – Altered expression of clusterin messenger RNA. FEBS Lett. 300, 279-282. Kirszbaum, L., Sharpe, J. A., Murphy, B., d’Apice, A. J. F., Classon, B., Hudson, P. and Walker, I. D. (1989). Molecular cloning and characterization of the novel, human complement-associated protein, SP-40, 40: a link between the complement and reproductive systems. EMBO J. 8, 711-718. Kounnas, M. Z., Loukinova, E. B., Stefansson, S., Harmony, J. A. K., Brewer, B. H., Strickland, D. K. and Argraves, W. S. (1995). Identification of glycoprotein 330 as an endocytic receptor for apolipoprotein J/clusterin. J. Biol. Chem. 270, 13070-13075. Lampertetchells, M., McNeil, T. H., Laping, N. J., Zarow, C., Finch, C. E. and May, P. C. (1991). Sulfated glycoprotein-2 is increased in rat hippocampus following entorhinal cortex lesioning. Brain Res. 563, 101106. Leger, J. G., Montpetit, M. L. and Tenniswood, M. R. (1987). Characterization and cloning of androgen-repressed mRNAs from rat ventral prostate. Biochem. Biophys. Res. Commun. 147, 196-203. Liu, L., Tornqvist, E., Mattsson, P., Eriksson, N. P., Persson, J. K. E., Morgan, B. P., Aldskogius, H. and Svensson, M. (1995). Complement and clusterin in the spinal cord dorsal horn and gracile nucleus following sciatic nerve injury in the adult rat. Neuroscience 68, 167-179. Lund, L. R., Romer, J., Thomasset, N., Solberg, H., Pyke, C., Bissel, M. J., Dano, K. and Werb, Z. (1996). Two distinct phases of apoptosis in mammary gland involution: proteinase-independent and -dependent pathways. Development 122, 181-193. Matsubara, E., Frangione, B. and Ghiso, J. (1995). Characterization of apolipoprotein J-Alzheimer’s A beta interaction. J. Biol. Chem. 270, 75637567. Matsubara, E., Soto, C., Governale, S., Frangione, B. and Ghiso, J. (1996). Apolipoprotein J and Alzheimer’s amyloid beta solubility. Biochem. J. 316, 671-679. May, P. C., Lampert-Etchells, M., Johnson, S. A., Poirier, J., Masters, J. N. and Finch, C. E. (1990). Dynamics of gene expression for hipocampal glycoprotein elevated in Alzheimer’s disease and in response to experimental lesions in rat. Neuron 5, 831-839. May, P. C. and Finch, C. E. (1992). Sulfated glycoprotein 2: new relationships of this multifunctional protein to neurodegeneration. Trends Neurosci. 15, 391-396. McGeer, P. L., Akiyama, H., Itagaki, S. and McGeer, E. G. (1989). Activation of the classical complement pathway in brain tissue of Alzheimer patients. Neurosci. Lett. 107, 341-346. McGeer, P. L., Kawamata, T. and Walker, D. G. (1992). Distribution of clusterin in Alzheimer brain tissue. Brain Res. 579, 337-341. Michel, D., Chabot, J.-G., Moyse, E., Danik, M. and Quirion, R. (1992). Possible functions of a new genetic marker in central nervous system: the sulfated glycoprotein-2 (SGP-2). Synapse 11, 105-111. Michel, D., Moyse, E., Brun, G. and Jourdan, F. (1994). Induction of apoptosis in rat olfactory neuroepithelium. Neuroreport 5, 1329-1332. Monti-Graziadei, G. A. and Graziadei, P. P. C. (1979). Neurogenesis and neuron regeneration in the olfactory system of mammls. II. Degeneration and reconstitution of the olfactory sensory neurons after axotomy. J. Neurocytol. 8, 197-213. Monti-Graziadei, A. G. and Graziadei, P. P. C. (1992). Sensory reinnervation after partial removal of the olfactory bulb. J. Comp. Neurol. 316, 32-44. Norman, D. J., Feng, L., Cheng, S. S., Gubbay, J., Chan, E. and Heintz, N. (1995). The lurcher gene induces apoptotic death in cerebellar Purkinje cells. Development 121, 1183-1193. Pasinetti, G. M., Johnson, S. A., Rozovsky, I., Lampert-Etchells, M., Morgan, D. G., Gordon, M. N., Morgan, T. E., Willoughby, D. and Finch, C. E. (1992). Complement C1qB and C4 mRNAs responses to lesioning in rat brain. Exp. Neurol. 118, 117-125. Pasinetti, G. M., Cheng, H. W., Moran, D. G., Lampertetchells, M.,

Clusterin and neuronal apoptosis in vivo 1645 Mcneill, T. H. and Finch, C. E. (1993). Astrocytic messenger RNA responses to striatal deafferentation in male rat. Neuroscience 53, 199-211. Pasinetti, G. M., Johnson, S. A., Oda, T., Rozovsky, I. and Finch, C. E. (1994). Clusterin (SGP-2): A multifunctional glycoprotein with regional expression in astrocytes and neurons of the adult rat brain. J. Comp. Neurol. 339, 387-400. Pearse, M. J., O’Bryan, M., Fisicaro, N., Rogers, L., Murphy, B. and d’Apice, A. J. (1992). Differential expression of clusterin in inducible models of apoptosis. Int. Immunol. 4, 1225-1231. Poirier, J., Hess, M., May, P. C. and Finch, C. E. (1991). Astrocytic apolipoprotein E mRNA and GFAP mRNA in hippocampus after entorhinal cortex lesioning. Mol. Brain Res. 11, 97-106. Reddy, K. B., Jin, G., Karode, M. C., Harmony, J. A. K. and Howe, P. H. (1996). Transforming growth factor β (TGFβ)-induced nuclear localization of apolipoprotein J/clusterin in epithelial cells. Biochemistry 35, 6157-6163. Rozovsky, I., Morgan, T. E., Willoughby, D. A., Dugichi-Djordjevich, M. M., Pasinetti, G. M., Johnson, S. A. and Finch, C. E. (1994). Selective expression of clusterin (SGP-2) and complement C1qB and C4 during responses to neurotoxins in vivo and in vitro. Neuroscience 62, 741-758. Simmons, D. M., Arriza, J. L. and Swanson, L. W. (1989). A complete

protocol for in situ hybridization of messenger RNAs in brain and other tissues with radio-labeled single-stranded probes. J. Histotechnol. 12, 169-181. Strange, R., Li, F., Saurer, S., Burkhardt, A. and Friis, R. R. (1992). Apoptotic cell death and tissue remodelling during mouse mammary gland involution. Development 115, 49-58. Wong, P., Borst, D. E., Farber, D., Danciger, J. S., Tenniswood, M., Chader, G. J. and Vanveen, T. (1994). Increased TRPM-2/clusterin mRNA levels during the time of retinal degeneration in mouse models of retinitis pigmentosa. Biochem. Cell Biol. 72, 439-446. Yasuhara, O., Aimi, Y., Yamda, T., Matsuo, A., McGeer, E. G. and McGeer, P. L. (1994). Clusterin as a marker for ischaemic Purkinje cells in human brain. Neurodegeneration 3, 325-329. Zhan, S.-S., Veerhuis, R., Janssen, I., Kamphorst, W. and Eikelenboom, P. (1994). Immunohistochemical distribution of the inhibitors of the terminal complement complex in Alzheimer’s disease. Neurodegeneration 3, 111194.

(Received 22 July 1996 – Accepted 15 May 1997)