Complement C5a Receptor-Mediated Signaling May Be Involved in Neurodegeneration in Alzheimer’s Disease¹

One of the factors considered to play a role in the death of the brain neurons in patients with Alzheimer’s disease (AD)⁴ is local inflammation (1, 2). In particular, the classical C pathway has been shown to be directly activated in AD by both fibrillar β amyloid deposits and neurofibrillary tangles (3, 4). In addition, all components of the classical C pathway have been identified in neurons of the AD brain (5, 6). The alternative C pathway has also been reported to be activated (7, 8). The end product of the terminal pathway, membrane attack complex (MAC), was also found in neuronal membranes of the AD brain (9).

Since MAC formation is initiated by C5b, which is generated when C5a is cleaved from C5 by C5 convertase of the alternative or classical C pathway, the presence of MAC in AD brain suggests that the C5a anaphylatoxin peptide, a 74-aa fragment of C5, has also been released (10). The expression of receptor for C5a (C5aR) has already been detected in pyramidal and granular cells of the hippocampus and pyramidal neurons of the cortex, astrocytes, oligodendrocytes, and microglia in humans and rodents (11, 12, 13, 14, 15, 16, 17). Human neuroblastoma cells were also shown to bear C5aR (17, 18, 19).

In the present study, we investigated the effects of abnormal activation of C5aR in C5aR-bearing human neuroblastoma cells as well as in living cells of primary cortical cultures and hippocampal and cortical slices of the rat brain. For this purpose, antisense homology box (AHB) fragment peptides of C5a and C5aR were used. The sequences were based on the molecular recognition theory: The peptides which are encoded on the opposite strands of the DNA in a given reading frame can have an affinity to bind each other and this binding would base on the hydropathic complementary of the peptides (20). Since these kinds of sense-antisense amino acid sequences might represent both putative intra- and intermolecular interaction sites, earlier we searched for existence of complementary regions using all possible sense-antisense combinations of amino acids in some proteins. Approximately 8- to 15-aa-long regions were found in them, which we termed AHB (21). When this search was applied to C5a protein, several AHBs were found representing complementary regions to each other (22). Numerous intramolecular AHBs were also found in the C5aR. Some of these AHBs even showed complementary properties between given regions of the C5a and C5aR, suggesting possible intermolecular interaction sites. The AHBs of the C5a/C5aR presenting the most numerous possible interactions were chosen and the sequences were then synthesized in multiple antigenic peptide (MAP) form. Our earlier results demonstrated that several of these AHB peptides interacted with the C5aR (22). When particular AHB peptide fragments of C5a and C5aR (named as PL37-MAP and PR226-MAP, respectively) were applied to cells, abnormal activation of C5aR was shown triggering an associated apoptotic pathway and causing the death of C5aR-expressing cells (18, 23). Although the function of C5aR in neurons has not been fully elucidated yet, this result suggests that C5aR-expressing neurons are susceptible to apoptosis and that therefore C5aR may play a role in neurodegeneration.

In addition to investigating the effects of abnormal activation of C5aR in neurons, in the present study we examined the presence of C5aR-expressing neurons in the hippocampus and cortex of healthy human brain. Its expression was also examined in the corresponding brain regions of AD and vascular dementia patients.

Peptides representing AHB or non-AHB regions of C5a and C5aR (21, 22) were synthesized with an AMS 422 Multiple Peptide Synthesiser (ABiMED, Langenfeld, Germany) using standard solid-phase synthesis techniques and 9-fluorenylmethoxycarbonyl (F-moc) amino acids (Watanabe Chemical Industry, Hiroshima, Japan). Amidated peptides (25-μm scale) were prepared using F-moc-PAL-PEG-PS resin (PerSeptive Biosystems, Warrington, U.K.). MAPs (10-μm scale) were prepared using 8-branched F-moc-MAP resin (Applied Biosystems, Foster City, CA). Purification was conducted by reversed-phase HPLC (Waters, Milford, MA) using linear gradient conditions of Milli-Q water/acetonitrile in 0.1% trifluoroacetic acid. Monomeric peptide molecular mass were confirmed using MALDI-TOF mass spectrometry on a Kompact Maldi II (Kratos Analytical, Shimadzu, Japan). MAP sequences were confirmed using N-terminal sequencing on a Procise cLC Protein Sequencing System (Applied Biosystems).

Labeling of the peptides was as follows: PR refers to regions of C5aR; PL refers to regions of C5a; the subscript s refers to a scrambled amino acid sequence; MAP refers to peptides in MAP form; otherwise, peptides are in monomeric form; and the number after PR or PL refers to the first N-terminal amino acid of the C5aR or C5a peptide.

Naming and sequences of the peptides were as follows: PR226 and PR226-MAP, LRTWSRRATRSTKTLKVV (C5aR fragment, AHB region); PL37 and PL37-MAP, RAARISLGPRCIKAFTE (C5a fragment, AHB region); PL37-MAP_S, AREICIRLSAARTGPKF (control scrambled peptide); PR1-MAP, MNSFNYTTPDYGHYDDKD (control peptide, non-AHB region of C5aR); and PL6-MAP, IEEIAAKYKHSVVKKCC (control peptide, non-AHB region of C5a).

A New Zealand White rabbit was immunized with PR226-MAP. The serum Ab titer was checked by indirect ELISA and the IgG fraction was purified by protein A affinity chromatography. The specificity of the anti-C5aR polyclonal Ab (pAb) was tested by immunoblot analysis using platelet-rich plasma. In addition, an immunoabsorption test was conducted with the PR226-MAP peptide.

Anti-human C5aR mAb was prepared as previously described (19). Briefly, using spleen cells from a mouse immunized with the PR226-MAP peptide, hybridomas were made and the clones secreting mAb to C5aR were selected by testing for binding of the Ab to the peptide. The mAb was purified by salting out with ammonium sulfate followed by gel filtration. The Ab isotype was identified as IgM. Reactivity of the mAb to C5aR was confirmed by flow cytometry (FACScan, BD Biosciences, Franklin Lakes, NJ) on dibutyryl cAMP-stimulated U-937 cells expressing C5aR.

Brain samples (24 patients, males and females; 82.1 ± 11.4 years) were obtained from the Brain Bank of the Choju Medical Institute of Fukushimura Hospital with the agreement of the Fukushimura Hospital Ethical Committee. These consisted of 12 neurologically normal controls (83.5 ± 8.2 years), 10 cases of AD (78.7 ± 8.4 years), and 2 cases of vascular dementia (80.0 ± 4.2 years). All of the neuropathological cases were clinically diagnosed premortem and diagnosis was confirmed at autopsy by a neuropathologist. All brains were obtained within 1–2 h after death.

The hippocampus and cortex were dissected and fixed in 4% formaldehyde and 30-μm sections were then cut from frozen samples. The sections were treated with 10% thioglycolic acid (for mAb labeling), 0.2% Triton X-100, and 3% H₂O₂ for 15 min. Following blocking with FCS, the sections were incubated for 2 days at 4°C with C5aR Abs (1/100 diluted mAb and 1/10,000 diluted pAb). Biotinylated goat anti-mouse IgM and goat anti-rabbit IgG Abs (Vector Laboratories, Burlingame, CA) were used as secondary Abs (1/200) and incubated for 2 h at room temperature (RT). The sections were then incubated for 2 h at RT in peroxidase-conjugated streptavidin (Jackson ImmunoResearch Laboratories, West Grove, PA; diluted 1/4000). The cells were visualized with nickel-enhanced 3,3′-diaminobenzidine (DAB-Ni). Silver intensification of mAb-labeled slices by the modified method of Gallyas (24) or counterstaining with neutral red was used with some of the slices.

Anti-mouse Thy 1.2 (Sigma-Aldrich, St. Louis, MO) and anti-β-galactosidase (Sigma-Aldrich) mAbs (IgM) were used for the C5aR mAb control. Brain sections were incubated with the control Abs and treated as above.

Neuron-rich primary cultures of cortical neurons were prepared from rat embryonic cortex (E18) as described earlier (25, 26). Briefly, papain-digested and dissociated cortical cells were filtered and then washed with DMEM containing a high concentration of glucose (4500 mg/L). The cells were seeded in DMEM containing a high concentration of glucose supplemented with 20% FCS and kept in an incubator at 37°C in an atmosphere of 93% air/7% CO₂. Experiments were conducted on the 10th day of culture.

TGW human neuroblastoma cells were cultured in RPMI 1640 medium supplemented with 10% FCS. The cells were kept in an incubator at 37°C in an atmosphere of 95% air/5% CO₂.

The cultured cells were treated with 10 μM PR226-MAP in HBSS at 37°C for 8 h. In the control culture, the medium was changed to PR226-MAP-free HBSS or control peptide (10 μM PR1-MAP in HBSS). A Mebstain apoptosis kit (Medical & Biological Laboratories, Nagoya, Japan) was used for nick-end labeling and the samples were visualized with streptavidin-peroxidase conjugate and DAB-Ni.

Cortical slices of the rat brain were prepared as described earlier with slight modifications (27). Briefly, the brains of 10- to 13-day-old postnatal Wistar rats were removed under deep pentobarbital anesthesia and placed in an ice-cold hypertonic calcium-free solution (containing 126 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 12 mM MgSO₄, 26 mM NaHCO₃, 170 mM sucrose, and 30 mM glucose), through which 95% O₂/5% CO₂ was bubbled. Frontal slices with a thickness of 400 μm were cut with a vibratome. Cortical pieces were removed and allowed to recover in hypertonic calcium-free solution for 30 min and then in artificial cerebrospinal fluid (ACSF containing 126 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 2 mM MgSO₄, 2 mM CaCl₂, 26 mM NaHCO₃ and 20 mM glucose) for another 60 min. under a 95% O₂/5% CO₂ atmosphere at RT.

The cortical slices were treated with ACSF containing 10 μM PL37-MAP for 8 h at RT under the 95% O₂/5% CO₂ atmosphere. Control slices were incubated in ACSF. We then fixed all slices in 4% formaldehyde and embedded them in paraffin. Sections of 5-μm thickness were cut and labeled with diluted (30×) anti-neurofilament (NF) mAb at RT for 60 min (clone 2F11; DAKO, Carpinteria, CA) using an OptiMax Plus Consolidated System (BioGenex, San Ramon, CA). An LSAB2 kit/HRP (DAKO) was used for secondary Ab with DAB chromogen. After NF staining, the apoptotic cells were labeled with an Apoptag peroxidase kit (Intergen, Purchase, NY) and cobalt chloride-enhanced DAB. Other sections were stained with H&E.

The protocols for the animal experiments were approved by the Institutional Animal Care and Use Committee of the Nagoya City University Medical School.

Cultured cells were loaded with the calcium-sensitive fluorescent dye fura 2-acetoxymethyl ester (fura 2-AM, 1 μM; Molecular Probes, Eugene, OR) in HBSS containing 0.1% DMSO and 0.02% Pluronic F-127 (Molecular Probes) in 1.5 h at 28°C. After washing with HBSS, the experiments were conducted at 30°C. The peptides were pipetted directly onto the cells in HBSS after a 5-min baseline recording. In the case of pretreatment, the cells were pretreated with the monomeric peptides or C5a (Sigma-Aldrich) after a baseline recording, then the MAP peptides were introduced into the bath fluid after 5 min.

Preparation of hippocampal slices of rat brain was similar to that described for the nick-end labeling. The slices were then loaded with 10 μM fura 2-AM in ACSF containing 0.0125% Pluronic F-127 and 0.1% DMSO for 1.5 h. The peptides were then applied directly to the slices in the ACSF after a 5-min baseline recording.

The experiments were conducted with an ARGUS-HiSCA calcium imaging system (Hamamatsu Photonics, Hamamatsu, Japan) using the ratio of the fluorescent signals obtained at 30°C at excitation wavelengths of 340 and 380 nm.

Statistical analyses (Student’s unpaired one-tailed t test) of the calcium imaging measurements were conducted on n = 8 or more cells at each recording using a Prism software package (GraphPad Software, San Diego, CA). Data are presented as means ± SEM. Integration of the recorded curves was conducted after baseline correction.

Administration of 31.2 nM PR226-MAP triggered marked calcium increase in fura 2-AM-loaded TGW cells (Fig. 1,a). Elevation started within 1.5 min after introducing the peptide into the bath fluid. The averaged ratio value peak reached 1.19 ± 0.021 and the integrated area of the graphs representing the net free calcium content during measurement was 9.39 ± 0.26 min. Similarly, the calcium elevation was recorded when 250 nM PL37-MAP was applied to the TGW cells (Fig. 1,b). The intracellular calcium content increased almost immediately and its peak amplitude was 0.57 ± 0.01 and the integrated area was 2.82 ± 0.13 min. Although the control peptide PR1-MAP (naming and sequences of the peptides can be found in Materials and Methods) was applied at a high concentration (1 μM), it did not evoke a calcium response in the TGW cells (Fig. 1,c). The PL6-MAP peptide (5 μM) and the scrambled peptide PL37-MAP_S (250 nM) that were used as control peptides of PL37-MAP also failed to trigger an elevation of the intracellular calcium content (Fig. 1, d and e), showing that the response was specific for the sequences of PR226-MAP and PL37-MAP.

The TGW cells were pretreated with 5 μM PL37 (monomeric) peptide, then 5 min later 31.2 nM PR226-MAP was added to the cells (Fig. 1,f). The response was delayed by ∼5 min and the amplitude (1.09 ± 0.11) and integrated area (0.63 ± 0.05 min) decreased significantly compared with results from the recording made without pretreatment (p < 0.01). A similar significant delay was observed when the cells were pretreated with 5 μM PR226 peptide 5 min before PR226-MAP administration (31.2 nM; Fig. 1,g). The amplitude (0.58 ± 0.04) and integrated area (1.73 ± 0.30 min) also decreased (p < 0.01), compared with no pretreatment. When the cells were pretreated with 5 μM PL37 or PR226 and 5 min later 250 nM PL37-MAP was applied to the TGW cells, a significant decrease (p < 0.01) in the calcium content was also recorded (integrated areas: 1.24 ± 0.23 min and 0.68 ± 0.07 min; Fig. 1, h and i) and the amplitude of the curves diminished as well (0.41 ± 0.01, p < 0.01; Fig. 1 i).

PR226-MAP and PL37-MAP peptides were applied to TGW cells in calcium-free bath fluid. However, neither of these peptides even at a high concentration (5 μM) were able to trigger any change in the calcium content under such conditions, demonstrating that extracellular calcium stores were involved in these processes (data not shown).

The TGW cells were pretreated with 1–50 nM C5a and PL37-MAP (1.25 μM; Fig. 2, a–d) or PR226-MAP (250 nM; Fig. 2, e and f) was then administered. This pretreatment diminished the fluorescence signal, decreasing both the amplitude (Fig. 2 g) and the integrated area (data not shown) significantly (p < 0.05) and dose dependently, showing that effects of the peptides were mediated by C5aR.

The effect of the two peptides was also apparent on cells of a neuron-rich primary culture of rat cortex. Both PL37-MAP (62.5 nM; Fig. 2,h) and PR226-MAP (31.2 nM; Fig. 2 i) triggered an almost immediate calcium response in ∼10% of the cells.

Hippocampal slices of rat brain were also examined. Recordings demonstrated that 500 nM PR226-MAP initiated elevation of the calcium concentration in numerous neurons (Fig. 3,a) with an average amplitude of 1.30 ± 0.05 and an integrated area of 4.13 ± 0.50 min. A photograph showed that the neurons responding to PR226-MAP were pyramidal cells (Fig. 4,b). Another photograph (Fig. 4,a) was taken of the same slice before peptide administration. Similarly, many cells in the slices responded to 500 nM PL37-MAP (Fig. 3,b) and exhibited an elevated calcium level (average amplitude, 1.29 ± 0.05; integrated area, 4.00 ± 0.67 min). These neurons were also located in the pyramidal cell layer (data not shown). Experiments with control peptides (500 nM PR1-MAP and 500 nM PL6-MAP) failed to trigger calcium signals in cells of the brain slices (Fig. 3, c and d). In addition, the scrambled PL37-MAP_S evoked a significantly delayed and attenuated response (p < 0.01) in some cells only (amplitude, 0.98 ± 0.07; integrated area, 1.52 ± 0.50 min; Fig. 3 e), whereas the majority of the cells remained silent.

A neuron-rich primary culture of the rat cortex was treated with 10 μM PR226-MAP for 8 h and the cells were then stained using an apoptosis kit. In the presence of the applied peptide, the cells retracted their processes, the cell bodies became rounded and black spots appeared where apoptosis had occurred in the nucleus (Fig. 4 c). In 8 h, ∼8–10% of the cells exhibited extensive DNA damage. Many of these cells displayed swollen nuclei.

Control cultures treated either with the control peptide PR1-MAP (10 μM; Fig. 4,d) or with HBSS (Fig. 4 e) preserved the original cellular network morphology of well-established cell-cell connections and long processes. In addition, the incidence of spontaneous apoptosis in these cultures was rare (<0.1%).

The PL37-MAP-treated and H&E-stained cortical rat brain slices presented numerous pyramidal cells with a shrunken, atrophied cell body, and dark staining of their nucleus showing aggregation of chromatin (Fig. 4,g). In control slices stained with H&E, the pyramidal neurons preserved the morphology of their soma and retained their pale nucleus (Fig. 4,f). Nick-end labeling and NF staining of the peptide-treated slices demonstrated abundant apoptotic cells (Fig. 4,i), including pyramidal cells (inset), whereas cells of the control slices did not exhibit apoptosis (Fig. 4 h).

Immunoblot analysis verified the specificity of the anti-C5aR pAb. The homogenate and supernatant fractions of platelet-rich plasma derived from a normal control exhibited a single band (40 kDa) with the pAb (data not shown).

In normal aging brains, the third and fifth layers of the neocortex were selectively stained with the anti-human C5aR pAb (Fig. 5 a).

In the hippocampus, large pyramidal neurons were strongly stained, especially in the CA2–4 regions (Fig. 5, b and c). Granular cells were also immunoreactive. Glial cells and vessel walls were not stained at all. In the CA1 area, staining of pyramidal neurons decreased slightly, revealing a peculiar type of neuronal process staining resembling a tadpole tail (Fig. 5 d). In the subiculum and parahippocampal gyrus, such neuronal process staining, most likely corresponding to apical dendrites, was common with the staining of perikarya of pyramidal neurons. Such thread-like staining was also seen in the neocortical region.

An immunoabsorption assay demonstrated that intense staining of hippocampal pyramidal neurons was abrogated by preincubation with the synthesized peptide (data not shown).

In AD brain, neuronal staining was not seen in any of the areas examined. Instead, dotted or patch-like staining was seen around the amyloid core of typical senile plaques (Fig. 5 e) which were thought to be due to C5a/C5aR fragments in dystrophic neurites. Diffuse plaques exhibited the same staining.

In vascular dementia, staining of the pyramidal cell layer in the hippocampus diminished only slightly (data not shown).

Immunohistochemical analysis of the aged-control human brain sections using the anti-human C5aR mAb also demonstrated expression of C5aR in the hippocampus. C5aR-IR was restricted to layers of the pyramidal and granular cells (Fig. 6,a, magnification, ×4). Higher magnification (×10) showed that the immunoreactive cells could be clearly identified as neurons in the fascia dentata and in the layer of pyramidal cells (Fig. 6,b) and granular cells due to their characteristic location and morphology. C5aR-immunoreactive pyramidal cells were demonstrated in the entorhinal cortex as well (Fig. 6,c). The third and fifth layers of pyramidal cells of the temporal cortex were also found to be immunoreactive (Fig. 6 d). C5aR-IR could be observed in the soma, although almost none was observed in the processes.

The use of control Abs did not reveal any staining in the entorhinal cortex (anti-mouse Thy 1.2: Fig. 6 e), in the hippocampus, or in the temporal cortex (data not shown).

In contrast to the healthy brain, the staining intensity decreased dramatically in the pyramidal cell layers of the AD hippocampus using anti-C5aR mAb (Fig. 6,f). In addition, pyramidal cells of the entorhinal cortex (Fig. 6,g) and the temporal cortex (Fig. 6 h) also exhibited strongly reduced immunostaining. IR was also diminished in granular cells of the hippocampus (data not shown).

Toluidine blue staining demonstrated that numerous pyramidal neurons of the hippocampus (Fig. 6,i) and the entorhinal cortex (Fig. 6 j) were still present in sections from AD brain, showing that the diminished immunostaining was at least partially the result of reduced labeling of C5aR and not due to the loss of C5a-immunoreactive neurons. Nevertheless, many of the nerve cells still showed the abnormal morphology of shrunken and atrophied cell bodies, reflecting consequences of the degenerative processes.

Although C5a itself has not yet been demonstrated to induce cell death directly in neurons, a C5a peptide fragment synthesized in MAP form (PL37-MAP) triggered apoptosis in neuroblastoma cells expressing C5aR via a C5aR-associated pathway (23). The C5aR-derived fragment peptide (PR226-MAP) also caused programmed cell death in TGW cells utilizing this pathway (18). Our present experiments also showed that these peptides triggered a strong increase in the intracellular calcium content in TGW cells, in cells of the primary neuronal culture of rat brain, and in living pyramidal cells of rat brain slices. The observations therefore suggested that this activation disturbed the calcium metabolism of the cells. The effects of these MAPs were specific for particular amino acid sequences and mediated via C5aR because monomeric peptides as well as native C5a interfered with the calcium influx induced by the MAPs. The failure to trigger an increase in the calcium concentration by control peptides or scrambled amino acid sequence peptides in MAP form also supports this specificity. Furthermore, the lethal consequence of abnormal activation of C5aR was also shown in cells of the primary neuronal culture and in cortical cells of rat brain slices since administration of the C5a/C5aR fragment peptides to these cells evoked apoptosis. The H&E staining of these brain slices showed numerous affected pyramidal cells with aggregated chromatin, hence we supposed that pyramidal neurons were involved in the apoptotic process. The NF and nick-end labeling further confirmed the presence of apoptotic pyramidal cells. Nevertheless, several NF-negative cells in the peptide-treated slices were also positive for nick-end labeling. We presume that those cells were probably glial cells, which also express C5aR constitutively (12, 13, 16). The results therefore further suggest that abnormal activation of C5aR can evoke apoptosis and that neurons bearing this receptor are particularly vulnerable to this type of death. Recent data showing IR to C5aR in human hippocampal and cortical pyramidal cells and in hippocampal granular cells (17) supports the immunohistochemical results presented here and suggests that these cells are also susceptible to apoptosis and consequent neurodegeneration in diseases in which complement activation occurs. The finding that C5aR immunostaining decreases or sometimes even disappears from AD neurons, but not in the neuronal cells of vascular dementia (where no C activation has been reported as yet), supports this idea.

One of the characteristic features of AD is that only specific types of neurons are affected by the neuropathological changes, whereas other, sometimes adjoining, cell types remain undamaged or exhibit only secondary injury (28, 29). These affected neurons may be particularly vulnerable to cell death because they possess receptors through which the degenerative processes can be triggered. Results indicating that only specific groups of neurons exhibit IR for C5aR in examined brain regions and that C5aR is associated with an apoptotic pathway support this explanation.

Results of our experiments demonstrated that although C5aR immunostaining decreased or even disappeared in neuronal cells of the AD brain, numerous pyramidal neurons were still present. This shows that the decreased immunostaining was caused either by the diminished immunoreactivity or by selective disappearance of neurons. These alternatives are in accordance with other reports, since the affected neurons, although severely damaged, can survive for a long time before they eventually die (28).

It was reported that C5aR is up-regulated in inflammation at the periphery (10). Other authors also found increased neuronal expression of the receptor in rodents under conditions of experimental inflammation (15, 30, 31, 32, 33). However, the time period over which they examined expression of C5aR was significantly shorter than the time course of AD. Furthermore, there is a sustained inflammation in the AD brain which persists over a period of years, which can hardly be modeled by the experimental conditions used in those reports. Nevertheless, O’Barr et al. (17) reported no significant change in C5aR-IR in the AD brain as compared with the aged normal brain. A possible explanation for the decreased immunostaining found in our study may be that our Abs and that used by O’Barr et al. (17) were generated against different epitopes of C5aR (ours, aa 226–243; O’Barr et al., aa 9–29). Our earlier results demonstrated that pretreatment of C5aR-bearing neuronal cells with C5a decreased C5aR staining in an Ab-dependent manner (19). The epitope against which our Abs were generated is involved in signal transduction (10). The observed diminished immunostaining may therefore be the result of interaction of the epitope resulting from activation of the receptor. However, this hypothesis requires further examination.

The idea that C5aR is involved in neurodegeneration is supported by recent results demonstrating that C5a/C5aR can contribute to apoptotic cell death of thymocytes and multiorgan failure during sepsis, although those reports did not establish whether this occurs by the direct or indirect pathway (34, 35, 36). However, recent findings showed a neuroprotective role for C5a/C5aR in a glutamate neurotoxicity model (37, 38, 39). This discrepancy can nevertheless be resolved considering the multiple signal transduction pathways of the C5aR (40). Depending on the presence of various agonists, different pathways could be activated (40). Therefore, it is reasonable to suppose that abnormal activation of C5aR by polyvalent fragments could induce apoptosis, whereas monomeric C5a activates the “normal” pathway suppressing apoptotic signals and exerting a neuroprotective influence.

Role of other mediators in inducing the apoptotic process and the calcium increase found in our experiments cannot be excluded, since effect of C5a/C5aR can be related for example to glutamate (38), TNF-α (15), and IL-6 (41). However, C5aR-associated apoptosis also exists in other cell types for example thymocytes of a sepsis model (34, 42) in which glutamate has not been described to play an apoptotic role yet. Supposing existence of similar pathways in the C5aR-associated cell death process in neurons and thymocytes, we may therefore conclude that it is mediated rather directly through C5aR than via the glutamate pathway. Although TNF mediated neuronal expression of C5aR in mice (15), it did not influence expression of C5aR in TGW cells (19), suggesting that TNF is probably not required for this apoptotic process. IL-6 is up-regulated parallel with C5aR in the sepsis model (41) and therefore having a harmful effect probably because of a positive feedback mechanism between IL-6 level and C5aR expression. In contrast, other studies described an antiapoptotic effect of IL-6 (43). Hence, the role of IL-6 is controversial even in thymocytes and its influence in the C5aR-mediated neuronal apoptosis requires further elucidation.

We thank Dr. Hitoo Nishino for providing equipment for preparing the brain slices; Dr. Endre Dobo for his useful advice; Mihaly Dezso and Csaba Bohata for the artwork; Monika Kara, Seizo Nagaya, and Hisashi Takino for their technical assistance; Balintne Farkas Judit for writing the software used for scrambling peptide sequences; and Catherine Campbell and Dr. William Campbell for English editing of this manuscript. We greatly appreciate the help of Dr. Eliada Lazoura in designing and synthesizing the control peptides.