Characterization of C3a and C5a Receptors in Rat Cerebellar Granule Neurons during Maturation

There is now clear evidence that the Complement anaphylatoxin C3a and C5a receptors (C3aR and C5aR) are expressed in glial cells, notably in astrocytes and microglia. In contrast, very few data are available concerning the possible expression of these receptors in neurons. Here, we show that transient expression of C3aR and C5aR occurs in cerebellar granule neurons in vivo with a maximal density in 12-day-old rat, suggesting a role of these receptors during development of the cerebellum. Expression of C3aR and C5aR mRNAs and proteins was also observed in vitro in cultured cerebellar granule cells. Quantification of the mRNAs by real-time reverse transcription-PCR showed a peak of expression at day 2 in vitro (DIV 2); the C3aR and C5aR proteins were detected by Western blot analysis at DIV 4 and by flow cytometry and immunocytochemistry in differentiating neurons with a maximum density at DIV 4–9. Apoptosis of granule cells plays a crucial role for the harmonious development of the cerebellar cortex. We found that, in cultured granule neurons in which apoptosis was induced by serum deprivation and low potassium concentration, a C5aR agonist promoted cell survival and inhibited caspase-3 activation and DNA fragmentation. The neuroprotective effect of the C5aR agonist was associated with a marked inhibition of caspase-9 activity and partial restoration of mitochondrial integrity. Our results provide the first evidence that C3aR and C5aR are both expressed in cerebellar granule cells during development and that C5a, but not C3a, is a potent inhibitor of apoptotic cell death in cultured granule neurons.

The Complement is an important component of the immune system that has the capacity to recognize and eliminate a large range of pathogens. Activation of the Complement cascade results in the generation of several fragments, especially the anaphylatoxins C3a and C5a, which share several biological activities including mast cell degranulation, vasodilation, smooth muscle contraction, and recruitment of immune cells to the site of inflammation (1). C3a and C5a exert their biological effects through specific binding to membrane receptors, named, respectively, C3aR 1 and C5aR (CD88). These two receptors belong to the seven-transmembrane receptor superfamily and are coupled to a G i protein (2,3). C3aR and C5aR are expressed in myeloid cells (4 -8) and, more surprisingly, in non-myeloid cells and tissues (7)(8)(9)(10). In the central nervous system it has been shown that C3aR and C5aR are constitutively expressed by glial cells in vitro (11)(12)(13)(14) and in vivo (15) and that the expression of C3aR and C5aR is increased in inflammatory conditions (15)(16)(17).
Low constitutive expression of C3aR and/or C5aR in neurons has been recently demonstrated by in situ hybridization and immunohistochemistry analysis in the cerebral cortex (18,19), spinal cord (20), hippocampus (19), and Purkinje cells in the cerebellum (19). A strong up-regulation of C5aR expression in neurons has been observed under various pathological conditions such as experimental excitotoxic neurodegeneration (21), bacterial meningitis (18), multiple sclerosis (15,16), experimental autoimmune encephalomyelitis (20), and traumatic axonal injury (22). Functional roles for C3a and C5a have been described in neuronal cell death. For instance, C3a exerts a neuroprotective effect against excitotoxicity-induced death of neurons that are cultured with astrocytes (23). Recently, it has been reported that C5a mediates apoptosis in neuroblastoma cells (24,25). In contrast, C5a protects differentiated human neuroblastoma cells from the neurotoxic effect of the amyloid A␤ peptide (26), and a possible neuroprotective role for C5a has been suggested in Alzheimer's disease (27). Moreover, C5-deficient mice are more sensitive to kainic acid excitotoxicity (28), and C5a has been found to protect neurons from apoptotic cell death in vitro and in vivo in a model of intracerebroventricular kainic injection in mice (21). These observations suggest that the neurotoxic or neuroprotective effects of anaphylatoxins depend on many factors such as the neuronal population considered, the state of neuronal differentiation, and the pathological context. Development of the cerebellum involves a delicate balance between proliferation, differentiation, and programmed cell death (29,30). In the developing brain programmed cell death contributes to the removal of neuronal precursors that fail to establish appropriate synaptic connections (31) and, thus, plays a crucial role in morphogenesis of the cerebellum. During the first weeks of postnatal life, immature neurons generated in the external granular layer (EGL) migrate along radial processes of glial cells through the molecular layer to reach their final destination within the internal granular layer (32,33). Only half of granule cells give rise to mature neurons, because massive cell loss occurs in the EGL and, later on, in the internal granular layer (33,34). Immature cerebellar granule neurons from early postnatal rats can be maintained alive in serum-containing medium by elevating extracellular potassium concentration (25 mM) (35), i.e. in conditions that mimic in vitro the innervation that neurons receive in vivo from mossy fibers (36). This state of depolarization improves survival of granular neurons and allows neuronal differentiation and maturation in vitro. Apoptosis of cultured cerebellar granule cells can be reliably induced by removing serum and lowering the extracellular potassium concentration (37)(38)(39). Cerebellar granule neurons is, thus, a very suitable model to study the functions of anaphylatoxin receptors (40). The aim of the present study was to investigate the expression of C3aR and C5aR during maturation of cerebellar granule neurons and the possible effect of their ligands C3a or C5a on apoptosis induced by serum deprivation and low potassium concentration.

EXPERIMENTAL PROCEDURES
Animals-Wistar rats (Charles River Laboratories, Arbresle, France) were kept in a temperature-controlled room (21 Ϯ 1°C) under an established photoperiod (lights on 07.00 -19.00 h) with free access to food and tap water. Animal manipulations were performed according to the recommendations of the French Ethics Committee and under the supervision of authorized investigators.
Reagents-Multiple-associated peptide (MAP)-C3a and MAP-C5a are eight peptidic monomers corresponding to the last 13 amino acids in the C-terminal region of the anaphylatoxins attached to a polylysine comb (a gift of Dr. Ischenko). These peptides were previously used and were shown to exhibit the same activity as recombinant or natural anaphylatoxins (41)(42)(43). C5aR antagonist (44) was synthesized by Dr J. Leprince (INSERM U413). Anti-mouse C3aR and anti-rat C5aR were obtained by immunization of rabbits with recombinant fusion proteins consisting of glutathione S-transferase fused to the region 161-333 of the large extracellular loop of mouse C3aR and to the 29-amino acid peptide corresponding to the 8 -36 sequence of the N-terminal region of rat C5aR (according to EMBL library index accession numbers NM_009779 and Y09613, respectively).
Cell Culture-Granule cell suspensions were prepared from cerebelli of 8-day-old Wistar rats, as described previously (45). Dispersed cells were seeded in multiwell plates or dishes (Falcon, BD Biosciences) coated with 10 M poly-L-lysine (Sigma) at a density of 2.5 ϫ 10 5 cells per cm 2 . Granule neurons were plated in a medium consisting of 75% Dulbecco's modified Eagle's medium (Invitrogen) and 25% Ham's F-12 medium (Sigma) containing 2 mM glutamine, 1 mM sodium pyruvate, 1% antibiotic-antimycotic solution (BioWhittaker, Verviers, Belgium) supplemented with 25 mM KCl and 10% fetal calf serum (SϩK25). Proliferation of non-neuronal cells was blocked by the addition of 10 M cytosine ␤-D-arabinofuranoside (Sigma) on day 1 in vitro (DIV 1). Cultures prepared by this method were enriched in granule neurons by more than 95%; the cell population was negative for glial fibrillary acidic protein (Sigma), OX 42 (ATCC, Manassas, VA), and calbindin D28K (Sigma) staining (data not shown). Cells were grown at 37°C in a humidified incubator with an atmosphere of 5% CO 2 .
RNA Isolation and Real-time Quantitative Reverse Transcription (RT)-PCR-Total RNAs were extracted from cultured granule cells by the guanidinium isothiocyanate method followed by ultracentrifugation onto a CsCl cushion as previously described (46). Total RNAs (2 g) were incubated at 70°C for 10 min, and RT was carried out in a thermocycler (Hybaid, Cera-Labo, Ecquevilly, France) at 37°C for 1 h with 5 mM dithiothreitol, 1 mM dNTPs, 500 pmol of random hexamer primers pdN6 (Amersham Biosciences), 120 units of RNasin (Promega, Charbonnières, France), and 400 units of Moloney murine leukemia virus (M-MLV) RT (Promega) in the reaction buffer (50 mM Tris-HCl, 75 mM KCl, 3 mM MgCl 2 , and 5 mM dithiothreitol). The absence of genomic contaminant was checked routinely by RT-PCR in negative control samples in which either the RNA samples were replaced with sterile water or the Moloney murine leukemia virus RT was omitted.
PCR was carried out by using the LightCycler FastStart DNA Master SYBR Green I kit (Roche Applied Science), which is specially adapted for the LightCycler instrument (Roche Applied Science), and the PCR reaction was performed in a final volume of 20 l in the LightCycler glass capillaries according to the manufacturer's recommendations. The primers used in our study were: RC3a forward (GAC CTA CAC TCA GGG C), RC3a reverse (ATG ACG GAC GGG ATA AG), RC5a forward (ATG CCT GCA GAT GGC GTT TA), and RC5a reverse (CAG AAA CCA AAT GGC GTT GAC). The housekeeping gene primers were glyceraldehyde-3-phosphate dehydrogenase forward (TGC CAT CAA CGA CCC CTT CA) and glyceraldehyde-3-phosphate dehydrogenase reverse (TGA CCT TGC CCA CAG CCT TG). Primers were chosen according to their cDNA sequences reported in the EMBL data library index accession numbers M33197 for glyceraldehyde-3-phosphate dehydrogenase, NM_032060 for C3aR, and Y09613 for C5aR. Southern Blot Analysis-RT-PCR products (5 l) were loaded onto a 1% agarose gel and separated by electrophoresis. The gel was treated for 10 min in 0.15 M HCl and neutralized for 30 min in 0.4 M NaOH. Southern blotting was performed by capillary transfer for 16 h in 0.4 M NaOH onto Nylon Plus membranes (Amersham Biosciences). The blot was neutralized for 10 min in 2ϫ SSPE (0.3 M NaCl, 17 mM Na 2 HPO 4 , and 50 mM EDTA). Rat C5aR and mouse C3aR cDNA probes were labeled by using Rediprime Kit (Amersham Biosciences) with [ 32 P]dCTP (Amersham Biosciences) at a specific activity of 2 ϫ 10 9 cpm/g. Membranes for C5aR were pre-hybridized in homologous conditions at 42°C for 4 h in a solution containing 50% formamide, 5ϫ SSPE, 1% SDS, 5ϫ Denhardt's, 5% dextran sulfate, and 100 g/ml herring sperm DNA, whereas membranes for C3aR were pre-hybridized in heterologous conditions in a solution containing 20% formamide, 4ϫ SSPE, 0.05 M Tris-HCl, pH 7.5, 1 M NaCl, 0.1% SDS, 10ϫ Denhardt's, and 100 g/ml herring sperm DNA. Hybridization was performed at 42°C for 16 h in the same solutions supplemented with 10 8 cpm of labeled probe. The membranes were washed briefly 3 times at room temperature in 2 ϫ SSPE, 0.1% SDS, for 1 h at 68°C for C5aR and 60°C for C3aR in 2ϫ SSPE, 0.1% SDS, and for 1 h at 68°C for C5aR and 60°C for C3aR in 1ϫ SSPE, 1% SDS and exposed for 4 h at room temperature onto X-Omat film (Eastman Kodak Co.).
Flow Cytometry Analysis-Neurons were harvested from cultures by incubation in phosphate-buffered saline (PBS) containing 10 mM EDTA. Cells were washed and resuspended in PBS containing 1% bovine serum albumin (BSA) and incubated with 10 g/ml non-immune mouse IgG for 15 min. After washing, cells were incubated with 2 g/ml anti-C3aR or anti-C5aR IgG or anti-␥-aminobutyric acid, type A receptor ␣6 subunit (Chemicon, Euromedex, Souffelweyersheim, France) diluted 1:100 in PBS containing 1% BSA at 4°C for 30 min. After washing, cells were incubated at 4°C for 30 min with fluorescein isothiocyanate-conjugated goat anti-rabbit IgG (Jackson Immuno Research, Interchim, Montluçon, France) diluted at 1:200 and measured in a FL-1 channel (530 Ϯ 15 nm bandpass filter). For two-color flow cytometric analysis, cells were incubated in PBS containing 1% BSA and 0.1% saponin for 30 min with biotinylated anti-C3aR or biotinylated anti-C5aR (2 g/ml) and monoclonal anti-neurofilament 68 clone NR4 (Sigma) diluted 1:100. After washing, cells were incubated for 30 min with streptavidin cyanine 5 (measured in FL-4 channel, 661 Ϯ 8 nm bandpass filter) diluted 1:500 and with fluorescein isothiocyanateconjugated goat anti-mouse IgG (FL-1) diluted 1:200. Cells were washed before analysis on a FACScalibur flow cytometer (BD Biosciences) operating with the Cell Quest TM software. Dead cells and debris were excluded from the analysis by gating living neurons from size/structure density plots. Data were displayed on a logarithmic scale with increasing fluorescence intensity. Each histogram plot was recorded for at least 10,000 gated events.
Immunocytochemistry-Granule cells were fixed with 1% formaldehyde in PBS at room temperature for 20 min. Fixed cells were incubated 1 h at room temperature with 2 g/ml anti-C3aR or anti-C5aR or with anti-neurofilament 68 diluted 1:100 in PBS containing 1% BSA. After several rinses in PBS, cells were incubated at room temperature for 2 h with peroxidase-conjugated Affinipure goat anti-rabbit or anti-mouse IgG (Jackson Immuno Research) diluted 1:1000 in PBS containing 1% bovine serum albumin. Cells were washed twice with PBS. The immunoreactivity was revealed by adding 0.5 mg/ml diaminobenzidine (Sigma) and 0.015% H 2 O 2 in PBS. The reaction was stopped by removing the diaminobenzidine solution and by rinsing in a large volume of water. The immunolabeling was analyzed using a computer-assisted microscope AxioVert (Zeiss, Le Pecq, France).
Immunohistochemistry-Rats were deeply anesthetized with chloral hydrate (400 mg/kg intraperitoneal) and transcardially perfused and fixed with 4% paraformaldehyde in PBS. The cerebellum was removed, post-fixed for 12 h in 4% paraformaldehyde, and placed in 15 and 30% sucrose until they sank. Tissue sections (10-m thick) were cut with a cryostat (Leica, Rueil-Malmaison, France). For immunohistochemical detection, endogenous peroxidase was blocked by incubating sections in 3% H 2 O 2 in methanol at room temperature for 10 min. Then fixed tissues were incubated at 4°C overnight with 2 g/ml anti-C3aR or anti-C5aR or anti-neurofilament 68 in PBS containing 0.3% Triton X-100 and 1% BSA. After several rinses in PBS, the tissues were incubated at room temperature for 2 h with peroxidase-conjugated Affinipure goat anti-rabbit or anti-mouse IgG diluted 1:1000. After several rinses in PBS, the sections were incubated at room temperature for 2 h with the peroxidase-anti peroxidase complex (Sigma) diluted 1:200 in PBS. The sections were washed twice with PBS and rinsed with 0.1 M Tris-HCl buffer. The immunoreactivity was revealed by using a 3,3Ј-diaminobenzidine substrate kit containing 0.02% H 2 O 2 (Sigma) and nickel ammonium sulfate. The reaction was stopped by removing the diaminobenzidine solution and by rinsing with a large volume of water. The specificity of immunolabeling was checked by omitting the primary antibody and by replacing the antiserum with pre-immune serum. Microphotographs were acquired on a computer-assisted image analyzer (Biocom 2000, Les Ulis, France).
Immunoprecipitation and Western Blot-Total cellular proteins were extracted from DIV 4 neurons in lysis buffer containing 0.15 M NaCl, 0.05 M EDTA, 4% CHAPS, and 0.2% of the protease inhibitor mixture (0.5 M EDTA, 2.5 M benzamidine, 2.18 mM pepstatin A, 11.7 M leupeptin) for C3aR detection and 1% Triton X-100, 25 mM Tris-HCl, 5 mM EDTA, 250 mM NaCl, 10% glycerol and 0.2% of the protease inhibitor mixture for C5aR, Bax, and Bcl-2 detections. Spleen or cerebellum of 12-day-old rat were lyophilized, and the dry powders were resuspended in the corresponding lysis buffer (10 ml per g of lyophilized tissue) and incubated overnight at 4°C. These extracts were centrifuged (15,000 ϫ g, 4°C, 15 min), and the supernatant was immunoprecipitated before Western blotting to enrich extracts in C3aR or C5aR proteins. Freshly made cell lysates were incubated with anti-C3aR or anti-C5aR at 4°C for 30 min. The lysates were then incubated with protein A-Sepharose at 4°C for 1 h under gentle rotation. After centrifugation at 10,000 g for 5 min, the supernatant was discarded. Immunoprecipitates were washed 5 times with PBS. The resulting pellet was solubilized in Laemmli sample buffer, reduced with the addition of mercaptoethanol (47), and boiled for 5 min. The immunoprecipitate proteins were separated by 10% SDS-PAGE on 12% gels and electrotransferred to polyvinylidene difluoride membranes (Immobilon-P Millipore, Saint-Quentin en Yvelines, France). Nonspecific sites were blocked with 5% nonfat milk at room temperature for 30 min, and the membrane was incubated overnight at 4°C with biotinylated anti-C3aR, biotinylated anti-C5aR (2 g/ml), anti-Bax (1:200, sc-493, Santa Cruz Biotechnology, CA), or anti-Bcl-2 (1:200, sc-492, Santa Cruz Biotechnology) with rocking. After washing, bands were visualized by incubation with streptavidin-horseradish peroxidase or conjugated Affinipure goat anti-rabbit IgG (1:1000) for 2 h followed by chemiluminescence detection (ECL detection kit, Amersham Biosciences).
Measurement of Calcium Flux-Cells were loaded with 4 M fluo-4AM (Molecular Probes, Interchim) in HBK buffer (120 mM NaCl, 5 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 6 mM glucose, 10 mM HEPES, pH 7.4) for 30 min at 37°C. After washing twice with HBK buffer, fluorescence intensity was measured with a FL600 microplate reader (Bio-Tek Instruments, Winooski, VT) at an excitation wavelength of 485 nm and an emission wavelength of 530 nm every 15 s for 6 min.
Apoptosis Induction-Cerebellar granule neurons were cultured for 4 days in SϩK25. The cells were washed twice and cultured in serum-free medium containing low (5 mM) potassium concentration (S-K5). Under these conditions cerebellar granule neurons degenerate and die by apoptotic cell death within 8 -24 h (37-39).
Cell Viability-Cells were incubated with 1.3 M calcein AM (Molecular Probes) in PBS for 15 min. After washing twice with PBS, fluorescence intensity was measured with a FL600 microplate reader (Bio-Tek Instruments) at an excitation wavelength of 485 nm and an emission wavelength of 530 nm. Pilot experiments have shown that the fluorescence intensity is proportional to the number of viable cells (in the range 5 ϫ 10 4 to 1 ϫ 10 6 cells/ml).
DNA Fragmentation-Internucleosomal DNA cleavage was assessed by conventional gel electrophoresis after extraction of nuclear DNA by using the Wizard Plus Minipreps DNA purification system (Promega). Granule neurons were incubated for 10 min at room temperature in a lysis buffer consisting of 50 mM Tris-HCl, 10 mM EDTA, 1% Triton X-100, and 50 g/ml RNase A. After centrifugation at 14,000 ϫ g for 15 min, the cleared lysates were mixed with the Wizard resin and transferred into a vacuum manifold column. Three washes with the columnwash solution were performed, and DNA fragments were eluted with 50 l of water at 60°C. DNA ladders were visualized on a 1.5% agarose gel.
Measurement of Caspase Activity-For measurement of caspase-3 activity, cultured cells were washed twice with PBS at 37°C, resuspended in Dulbecco's modified Eagle's medium (100 l), and treated with the fluorometric caspase-3 assay system (Promega). In brief, caspase substrate and homogeneous caspase buffer (100 l) were mixed and added to the cells (100 l) in 96-well plates. The fluorescence intensity was measured with a Bio-Tek FL600 microplate reader (Bio-Tek Instruments) at an excitation wavelength of 485 nm and an emission wavelength of 530 nm every 15 min for 3 h. For measurement of caspase-9 activity cultured cells were washed twice with PBS at 37°C and resuspended in PBS at 4°C. Cells were harvested by centrifugation (350 ϫ g, 4°C, 9 min) and treated with the fluorometric caspase-9 assay systems (R&D Systems, Minneapolis, MN). In brief, the cell pellet was resuspended in 95 l of hypotonic cell lysis buffer and incubated with 95 l of reaction buffer and caspase-9 substrate in 96-well plates. Fluorescence intensity was measured at an excitation wavelength of 360 nm and an emission wavelength of 530 nm every 15 min for 3 h.
Measurement of Mitochondrial Activity-Cultured cells were incubated in the presence of the JC-1 probe (7.5 g/ml; Molecular Probes) at 37°C for 15 min and then washed with PBS. In viable granule neurons, the intact membrane potential allows the lipophilic dye JC-1 to enter into the mitochondria, where it accumulates and aggregates, producing an intense orange signal. In apoptotic cells, the mitochondrial membrane potential collapses so that the monomeric JC-1 probe remains cytosolic and stains the cell in green. Fluorescence intensity was measured with a Bio-Tek FL600 microplate reader and expressed as a ratio of the emission at 590 nm (orange) over 530 nm (green).
Statistical Analysis-All values are expressed in means Ϯ S.E. A one-way analysis of variance and a Tukey-Kramer multiple comparisons test were used to determine statistical differences between mean values.

Occurrence of the C3aR and C5aR
Proteins in the Rat Cerebellum during Development-The presence of the C3aR and C5aR proteins in cerebellum sections from 4, 8, 12, and 30 day-old rats was studied by immunohistochemistry using specific anti-C3aR and anti-C5aR antibodies. The expression of the two receptors followed similar temporal patterns (Fig. 1). A low density of C3aR and C5aR was detected in the cerebellum of 4 and 8 day-old rats. Intense immunoreactivity for both C3aR and C5aR was observed in the cerebellar cortex at 12 days. The staining was mainly localized in the EGL, in which granule cells constitute the major population, and in the molecular layer. In 30-day-old rats, C5aR immunoreactivity vanished, whereas a faint C3aR staining was observed in the molecular layer, which is composed of axons of granule neurons also called parallel fibers. Specificity controls showed the absence of staining when the primary antisera were replaced by PBS or pre-immune serum (Fig. 1). The C3aR-and C5aR-positive cells were clearly identified as neurons by staining of adjacent sections with the anti-neurofilament 68 antibody (data not shown).

Expression of C3aR and C5aR mRNA during Differentiation of Rat Cerebellar Granule Neurons in Vitro-Because
C3aR and C5aR were expressed in neurons of EGL, further experiments were performed on cultured cerebellar granule cells. RT-PCR analysis of mRNAs from granule neurons cultured in SϩK25 revealed that the expression of C3aR and C5aR was maximum in DIV 2 neurons with a 2-fold increase in C3aR mRNA level and a 5-fold increase in C5aR mRNA level (Fig. 2 A and B). At this stage (DIV 2) neurons outgrew and emitted neuritic extensions. Thereafter, the expression of C3aR and C5aR mRNAs gradually decreased (Fig. 2, A and  B). Specificity of the amplicons was assessed by Southern blotting with hybridization of RT-PCR products with the mouse C3aR cDNA probe (Fig. 2C) and the rat C5aR cDNA probe (Fig. 2D); amplicons for C3aR and C5aR mRNAs were detected in rat granule cells as well as in rat lung used as a positive control. No amplification product was seen in the negative control (ϪMoloney murine leukemia virus (ϪM-MLV)) in which the RT step had been omitted, ruling out the amplification of genomic DNA. These data clearly show a transient up-regulation of gene expression for C5aR and, to a lesser extent, C3aR at DIV 2, when neurons are beginning to differentiate.
Expression of the C3aR and C5aR Proteins during Differentiation of Rat Cerebellar Granule Neurons in Vitro-The occurrence of C3aR and C5aR at the plasma membrane level was studied by flow cytometry on cultured granule cells at different stages of development (Fig. 3, A-C). In immature cells (DIV 0) taken 3 h after seeding, C3aR and C5aR were only expressed in 35 and 22% of the cell population (Fig. 3A). The proportion of C3aR-and C5aR-positive neurons increased rapidly during the differentiation process; more than 80% of the cells were positive for both receptors on DIV 2, and nearly all granule neurons expressed C3aR and C5aR at DIV 4 and DIV 9. It was also noticed that the homogeneity of the positive population increased with time (Fig. 3A). Replacement of the C3aR and C5aR antisera with pre-immune serum did not produce any specific staining in DIV 4 cells (Fig. 3B). Preincubation of the C3aR and C5aR antisera with their respective antigens markedly reduced cell staining (Fig. 3B). In contrast, incubation of the antisera with the glutathione S-transferase protein did not affect the flow cytometric pattern (data not shown). Two-color flow cytometry analysis showed that DIV 4 neurons were positive both for C3aR or C5aR and the neuronal marker antineurofilament 68 (Fig. 3C, upper right quadrant). In addition, DIV 9 neurons were labeled by antibodies against the ␣6 subunit of the ␥-aminobutyric acid, type A receptor, a marker of mature cerebellar granule neurons (data not shown). Immunohistochemical labeling of cultured granule neurons showed weak diffuse staining in DIV 1 cells and strong membrane staining in DIV 4 and DIV 9 cells (Fig. 3D). No staining was observed when the primary antibody was omitted or replaced by pre-immune serum.
Characterization of the C3aR and C5aR proteins was performed by Western blot analysis after immunoprecipitation. The C3aR antibodies revealed a single band with an apparent molecular mass of 75 kDa in DIV 4 granule cell extracts (Fig.  4A, lane 1) and 12-day-old rat cerebellar extracts (Fig. 4A, lane  2). Rat spleen extracts were used as positive control (Fig. 4A,  lane 3). The C5aR antibodies revealed a major band with an apparent molecular mass of 47 kDa in DIV 4 granule cell extracts (Fig. 4B, lane 1) and 12-day-old rat cerebellar extracts (Fig. 4B, lane 2). The apparent molecular masses of the immunoreactive proteins were in good agreement with the predicted masses of 53.6 kDa with 4 potential N-glycosylation sites for mouse C3aR and 41.3 kDa with 1 potential N-glycosylation site for rat C5aR.
Functionality of C3aR and C5aR was assessed by measurement of cytosolic calcium flux. C3aR and C5aR agonists MAP-C3a and MAP-C5a induced a sharp increase of intracellular calcium concentration in DIV 4 granule neurons (Fig. 5, A and  B). MAP-C3a and MAP-C5a induced calcium waves from 1 to 10 nM and from 10 to 100 nM, respectively, and the same experiment performed with EGTA in HBK buffer did not modify these calcium fluxes (data not shown). Taken together, these results show that C3aR and C5aR are transiently ex- pressed in granule neurons, suggesting a role of these anaphylatoxins during the development of the cerebellar cortex.
Protective Effects of MAP-C5a against Death of Cerebellar Granule Neurons-Apoptosis of cerebellar granule cells can be reliably induced by removing serum and lowering KCl concentration in the culture medium. Therefore, after a 4-day culture in SϩK25, granule neurons were incubated in S-K5 to induce apoptosis, and the possible neuroprotective effects of the C3aR and C5aR agonists MAP-C3a and MAP-C5a were investigated. MAP-C3a had no effect on granule cell apoptosis in this model (data not shown). In contrast, exposure of granule neurons, cultured for 9 h in S-K5, to graded concentrations of MAP-C5a induced a dose-related increase in the number of surviving cells, as determined by measurement of calcein AM-induced fluorescence (Fig. 6A). The maximum effect of MAP-C5a was observed at a concentration of 10 Ϫ8 M. Moreover, this protective effect of MAP-C5a was prevented by pretreatment of DIV 4 granule neurons with the C5aR antagonist (10 Ϫ7 M)-derived macrocycle AcPhe(L-ornithine-Pro-D-cyclohexylalanine-Trp-Arg) (44) (Fig. 6A). MAP-C5a and C5aR antagonist were not toxic for neurons at the concentrations used. Time-course experiments revealed that MAP-C5a significantly increased the number of surviving cells after 9 -24 h of incubation in S-K5 medium (Fig.  6B). The neuroprotective effect of MAP-C5a was visualized by phase contrast microscopy (Fig. 6C). DIV 4 neurons cultured for 12 h in SϩK25 medium looked healthy and exhibited long processes. Treatment with MAP-C5a (10 Ϫ8 M) had no effect on the density and morphological aspect of these neurons. In contrast, DIV 4 neurons cultured for 12 h in S-K5 medium showed massive cell loss, and the remaining neurons exhibited the morphological characteristics of apoptotic cell death, i.e. cell shrinkage, nuclear condensation, and regression of the neuritic network. Treatment of these neurons with 10 Ϫ8 M MAP-C5a increased the number of surviving cells and restored the morphological characteristics of these neurons.
Massive DNA fragmentation was observed in DIV 4 granule cells incubated for 15 h in S-K5 medium but not in cells incubated in SϩK25 medium (Fig. 7A). The addition of 10 Ϫ8 M MAP-C5a to the S-K5 medium markedly reduced DNA laddering (Fig. 7A). As previously reported (48,49), a 6-h incubation of DIV 4 granule cells with S-K5 medium caused a robust activation of caspase-3 (Fig. 7B). Administration of graded concentrations of MAP-C5a (10 Ϫ10 to 10 Ϫ7 M) provoked a dose-dependent reduction of caspase-3 activity (Fig.  7B). It has been recently shown that the extrinsic pathway is not involved in (S-K5)-induced apoptosis in cerebellar granule cells (48,50). We have, thus, investigated the possible implication of the intrinsic apoptotic pathway by measuring the activity of caspase-9, one of the initiator caspases, and mitochondrial membrane potential. Incubation of DIV 4 cells with S-K5 medium for 6 h markedly increased caspase-9 activity (Fig. 7C). The addition of 10 Ϫ8 M MAP-C5a significantly reduced S-K5-evoked caspase-9 activation (Fig. 7C). MAP-C5a had no effect on mitochondrial integrity in DIV 4 granule neurons cultured in SϩK25 medium, as revealed by measurement of the 590 nm/530 nm ratio using the JC-1 probe (Fig. 7D). Incubation of DIV 4 cells in S-K5 medium reduced by 3-fold the proportion of active mitochondria. Exposure S-K5-cultured cells to MAP-C5a (10 Ϫ8 M) induced a modest but significant increase of mitochondrial membrane potential (Fig. 7D). Real-time PCR revealed that the expression of various members of the Bcl-2 family, including the anti-apoptotic members bcl-2 and bcl-xL, and the proapoptotic members bax, bak, and bad was not modified during a 1-8-h incubation with S-K5 in the absence or presence of MAP-C5a (data not shown). Similarly, Western blot analysis showed that incubation in S-K5 medium in the absence or presence of MAP-C5a had no effect on the levels of Bcl-2 and Bax proteins (Fig. 7E).

DISCUSSION
The cerebellar cortex, which undergoes profound morphogenetic transformations after birth, is a very well suited model to identify factors that control neuronal apoptosis (40). In particular, dissociated cerebellar granule cells from postnatal rats form a homogeneous population of nerve cells that can either survive and differentiate when cultured in serum-containing medium under depolarized conditions (SϩK25) or undergo apoptosis when cultured in serum-free medium and low potassium concentration (S-K5) (37,39). The present study has shown that the anaphylatoxin receptors C3aR and C5aR are expressed in vivo in the EGL of the cerebellum of young rats but are virtually absent in the adult cerebellum, suggesting that C3a and C5a may play a physiological role in the histogenesis of the cerebellar cortex. In vitro studies on cultured cells confirmed the expression of C3aR and C5aR by cerebellar granule neurons; (i) measurement of C3aR and C5aR mRNAs revealed maximum expression of both receptor genes at DIV 2, and (ii) flow cytometry and immunocytochemical experiments showed that the C3aR and C5aR proteins occurred in the majority of granule cells from DIV 4 to DIV 9. Up to now the presence of C3aR and C5aR has been documented in the adult rodent brain, where the two receptors are expressed by glial cells, astrocyte (11)(12)(13), and microglia (14,15). C3aR and C5aR are present both in vitro (11)(12)(13)(14) and in vivo (15) and are upregulated in these cells after inflammation (15)(16)(17). In situ hybridization and immunohistochemistry analysis also revealed a constitutive expression of mouse and rat C3aR in cortical and hippocampal neurons (17,19) as well as in cerebellar Purkinje cells (19). Controversial results have been published for C5aR, which was first hardly detected by in situ hybridization or immunohistochemistry in neurons from rodent adult brains (18,22) but was recently identified on mouse pyramidal neurons in hippocampus and neocortex as well as on Purkinje cells in cerebellum (26). C5aR expression seemed induced by inflammatory stimuli (18,22) and is found in cultured murine corticohippocampal neurons (21,28) and neuroblastoma cells (24 -26). Our study, which was the first to investigate the expression of anaphylatoxin receptors during brain development, provides evidence for a transient expres-sion of C5aR in the developing cerebellum, which then vanishes during neuronal differentiation, supporting the idea that C5aR could be involved in brain development and would then disappear in mature neurons. So far, an intriguing question that remains is the activation of C5aR during cerebellar development. It is well known that local Complement biosynthesis occurs in normal brain (51) and particularly C3 and C5 are produced by postnatal astrocytes and neurons (52,53). However, we did not know if Complement could be activated during postnatal cerebellar development by either mechanism of activation. Nevertheless, peptide fragments generated by proteolysis at dibasic sites have been described in the cerebrospinal fluid, suggesting an action of pro-hormone convertases (54). Thus, C4a anaphylatoxin was found in the cerebrospinal fluid, and we can suppose that C3a and C5a could be cleaved from C3 and C5 in brain during development by a similar pro-hormone convertase mechanism. Another possibility would be that C5aR could bind during neuron maturation to another ligand not yet identified. Finally, the receptor itself could have a basal activity in the absence of ligand as observed for interleukin-12 expression (55). Thus, C5aR might play a role in neurons during maturation to protect them against apoptotic death and improve their lifetime in order to establish connections.
C5aR expression is maximum on cerebellar granule neurons after 4 days of culture in serum medium supplemented with depolarizing concentrations of potassium. Under these conditions, immature granule cells differentiate into mature neurons within a week of culture. Removing serum and lowering potassium concentration from the culture medium has been reported to induce apoptosis of cerebellar granule cells (37)(38)(39). Under these conditions, in accordance with previous studies (37,49,50,56,57), our results showed an increase of DNA fragmentation and caspase-3 activity, which could be prevented in a dose-and time-dependent manner by a peptidic agonist of C5aR (MAP-C5a), which highlights the role of anaphylatoxins on tissue remodeling, as has been hypothesized by Mastello and Lambris (58). Concerning C3a, in which the receptor was also expressed by granule cells, no effect on cell survival could be observed in this model. Apoptosis is only one of the phenomena among others such as cell proliferation, migration, or differentiation that occurs during brain development, so the function of C3aR remains an open question that may be addressed at least in part by gene profiling. An inhibitory effect of C5aR on caspase-3 activity had been reported on differentiated murine corticohippocampal neurons exposed to exitotoxic concentrations of glutamate (21).
One of the most relevant upstream inducers able to activate caspase-3 in this model was the intrinsic pathway since caspase-9 rather than caspase-8 seems to trigger granule cell death after serum and potassium withdrawal (48,50,59). Indeed, under apoptotic conditions, caspase-9 was activated, and this induction was significantly reduced in the presence of MAP-C5a. This effect of MAP-C5a on caspase-9 activity is in accordance with the inhibition of caspase-9 by C5a on neutro-phil-induced apoptosis (60). In these neutrophils, C5a has been reported to inhibit apoptosis at least partially through the phosphatidylinositol 3-kinase/Akt pathway. Because phosphatidylinositol 3-kinase activation is known to inhibit apoptosis of granule neurons (61,62), this pathway could mediate the neuroprotective effect of MAP-C5a. C5aR is a G i protein-coupled receptor of the rhodopsin family (2, 3) known to activate the Ras/Raf/mitogen-activated protein kinase cascade through the phospholipase C/protein kinase C pathway in human neutrophils (63). The activation of this mitogen-activated protein kinase pathway by C5a has also been reported in neurons. In particular, inhibition of MEK (mitogen-activated protein kinase kinase) phosphorylation by PD98059 prevented the inhibition of caspase-3 by C5a (64). The mitogen-activated protein kinase pathway has been reported to be involved in the control of granule cells apoptosis (65) and differentiation, so we can speculate that this cascade can be another alternative for the antiapoptotic effect of C5a. Mitogen-activated protein kinase and phosphatidylinositol 3-kinase pathways are currently being investigated.
Because caspase-9 is known to be activated by cytochrome c released from mitochondria, the integrity of this organelle was investigated. As expected, serum removal with low potassium concentration induced a collapse of mitochondria potential, but surprisingly, C5a only exhibited a weak protection of the mitochondria integrity. This observation suggests that MAP-C5a inhibits caspase-9 activity in a mitochondria-independent manner that would reinforce the possible involvement of the phosphatidylinositol 3-kinase/Akt pathway in the neuroprotective effect of MAP-C5a. Indeed, it has been shown that Akt can regulate apoptosis at a post-mitochondrial level (66). Akt can phosphorylate pro-caspase 9, which blocks its activation by cytochrome c release (67). Another possible regulation of capase-9 by Akt involves the phosphorylation of Bad, a pro-apo- and caspase-9 (C) activities in granule cells cultured in SϩK25 medium (gray bars) or exposed 6 h to S-K5 medium (black bars). The values are expressed as percentages of control cells cultured in SϩK25 medium alone (100%). D, effect of MAP-C5a on mitochondrial membrane potential in granule cells cultured in SϩK25 medium (gray bars) or exposed 15 h to S-K5 medium (black bars). The values are expressed as percentages of control cells cultured in SϩK25 medium alone (100%). Each value represents the mean (ϮS.E.) of three independent experiments performed in triplicate. ***, p Ͻ 0.001 versus control (SϩK25 alone); #, p Ͻ 0.05; ##, p Ͻ 0.01 versus S-K5 alone. E, Western blot showing the effect of MAP-C5a (10 Ϫ8 M) on the expression of Bcl-2 and Bax in DIV 4 neurons cultured in SϩK25 medium or exposed to S-K5 medium. ptotic member of the Bcl-2 family (68). Phosphorylation of Bad results in its dissociation from the anti-apoptotic Bcl-xl, which in turn prevents caspase-9 activation (69,70). This direct regulation by protein phosphorylation of caspase 9 could explain the strong inhibition of the protease activity by MAP-C5a without any significant improvement of MAP-C5a on mitochondrial integrity. Moreover, in our present study phosphorylation phenomenon might explain the unaffected constitutive expression of Bcl-2 family members either on mRNA or protein levels, detected in DIV 4 granular neurons after (S-K5)-induced apoptosis. This result reinforces our hypothesis of involvement of phosphatidylinositol 3-kinase/Akt pathway since Akt is known to rescue cells from apoptosis without altering the expression levels of endogenous Bcl-2, Bcl-x, or Bax (66). This result was in agreement with the observation of cytochrome c release without alteration of expression several members of the Bcl-2 family in granule neurons subjected to serum withdraw and potassium deprivation (71).
In conclusion the present study demonstrated that the anaphylatoxin receptors C3aR and C5aR are transiently expressed by granular neurons during maturation both in vivo and in vitro. A neuroprotective effect of C5a was established on granule cells cultured under conditions that promote apoptosis. Also both receptors are expressed by granule neurons, the function of C3aR remains an open question, but our findings suggest a novel neurodevelopmental role for C5a during cerebellum maturation.