An ex vivo murine model to study poliovirus-induced apoptosis in nerve cells

Paralytic poliomyelitis results from destruction of motor neurons owing to poliovirus (PV) replication. Using a mouse model, we have previously shown that PV kills neurons of the central nervous system (CNS) as a result of apoptosis (Girard et al., Journal of Virology 73, 60666072, 1999). We report the development of mixed mouse primary nerve cell cultures from the cerebral cortex of neonatal mice transgenic for the human PV receptor. These cultures contained all three main cell types of the CNS, i.e. neurons, astrocytes and oligodendrocytes. All three cell types were susceptible to PV infection and virus replication in the cultures led to DNA fragmentation characteristic of apoptosis. PV-induced apoptosis was inhibited by the caspase inhibitor benzyloxycarbonyl-Val-Ala-Asp(O-Me) fluoromethyl ketone (Z-VAD.FMK), indicating that this process involved caspases. Thus, these mixed mouse primary nerve cell cultures are a new in vitro model for studying the molecular mechanisms of PV-induced apoptosis in nerve cells.

Poliovirus (PV), an enterovirus belonging to the family Picornaviridae, is the aetiological agent of paralytic poliomyelitis in humans. PV strains are classified into three serotypes (PV-1, PV-2 and PV-3). The viral particle has a capsid composed of four proteins, VP1VP4, enclosing a single-stranded RNA of positive polarity. PV first infects the oropharynx and the gut, is released into the blood and then reaches the central nervous system (CNS). The destruction of motorneurons, a consequence of PV replication (Couderc et al., 1989 ), results in paralysis (Bodian & Howe, 1955 ). Experimentally, poliomyelitis can be transmitted to monkeys and in some cases to mice by inoculating PV directly into the CNS. Monkeys and TgCD155 mice [transgenic (Tg) mice expressing the human PV receptor, CD155] are susceptible to all three PV serotypes (Ren et al., 1990 ; Koike et al., 1991 ). In non-TgCD155 mice, only a small number of PV strains induce poliomyelitis. By using both TgCD155 and non-TgCD155 mouse models, we have recently shown that CNS injury, notably the neuronal cell death that occurs during paralytic poliomyelitis, is associated with an apoptotic process (Girard et al., 1999 ). Moreover, this apoptosis correlates with viral load and coincides with the onset of paralysis. However, the molecular mechanism by which PV infection triggers apoptosis in nerve cells remains unknown.

Apoptosis is an active process of cell death, which occurs in response to a variety of stimuli, including viral infections. It is characterized by a number of distinct morphological features and biochemical processes, such as cell shrinkage, plasma membrane blebbing, chromatin condensation and intranucleosomal cleavage (Roulston et al., 1999 ). A family of proteases, called caspases (cysteine proteases with aspartate specificity), are universal effectors of apoptotic cell death (Cryns & Yuan, 1998 ; Earnshaw et al., 1999 ). In vitro, PV can either induce or inhibit apoptosis in HeLa cells, an epithelial cell line, according to the conditions of the viral infection (Tolskaya et al., 1995 ; Agol et al., 1998 , 2000 ). Moreover, PV-induced apoptosis has been observed in the CaCo-2 enterocyte-like cell line and in the U937 promonocyte cell line (Ammendolia et al., 1999 ; Lopez-Guerrero et al., 2000 ). However, no in vitro model has been available to investigate PV-induced apoptosis in nerve cells. The primary cultures of human foetal brain cells that have been described (Pavio et al., 1996 ) are not a convenient model for investigating PV-induced apoptosis in nerve cells because of the difficulty of obtaining human foetal tissues. Here we describe the development of a model of mixed nerve cell cultures prepared from the cerebral cortex of neonatal TgCD155 mice.

Mixed mouse primary nerve cell cultures were prepared from the cerebral cortex of neonatal TgCD155 mice, as described by Kiss et al. (1994) . Briefly, the cerebral hemispheres were gently dissociated by passage through a Pasteur pipette to obtain a single-cell suspension in Hanks balanced salts solution (Gibco). Cells were resuspended in DMEM containing 4·5 g/l glucose (Sigma) and 10% foetal calf serum, and then filtered through 80 µm pore size nylon mesh to remove tissue clumps. Viable cells were plated at approximately 1·5x10⁵ cells/cm² and incubated at 37 °C in a humidified atmosphere of 5% CO₂. Culture medium was changed every 3 days until cell confluence, and then replaced with serum-free medium [DMEM containing 4·5 g/l glucose, 5 µg/ml insulin, 20 µg/ml transferrin, 20 nM progesterone, 100 µM putrescine, 30 mM sodium selenite, 1% penicillinstreptomycin (Gibco)]. Cell cultures were used 811 days after plating.

These cultures consisted of a bilayer made up of mixed neuronal and glial cells. To identify further the neural cell types in the primary cultures, we performed triple immunofluorescence labelling using specific cell markers for neuronal and glial cell lineages, as previously described (McKinnon et al., 1990 ; Ben-Hur et al., 1998 ). Neuronal cells were detected with TUJ1, a mouse monoclonal antibody (IgG2a) against neuron-specific class III β-tubulin (1/500 dilution; BAbCO). Oligodendrocytes and astrocytes were identified with a mouse monoclonal antibody (IgM, 1/5 dilution; Boehringer Mannheim) recognizing the cell surface sulfatide O4 and a rabbit antibody to glial fibrillary acidic protein (GFAP) (1/200 dilution; DAKO), respectively. Primary antibodies were stained with the appropriate fluorescent secondary antibody. Nuclei were stained with DAPI. Control experiments included omitting the primary antibodies from the staining procedure. No non-specific labelling was observed in controls.

The confluent underlying layer comprised flat type 1 astrocytes with an epitheloid shape; these cells strongly expressed GFAP (Fig. 1A). They constituted about 50% of the total cell population. The surface of this monolayer was predominantly populated by neuronal cells (about 30%) characterized by expression of β-tubulin antigen. The majority of the neurons were grouped in clusters of small, round cells with short processes, but some more mature neurons with long neurites were dispersed among these clusters. Other glial cells (about 20%), including oligodendrocyte progenitors, were also observed dispersed within the upper monolayer. These cells expressed O4 antigen on their surface. Rare non-neural cells such as macrophages, fibroblasts and endothelial cells were also occasionally found in the upper monolayer (Kiss et al., 1994 ).

(73K): Fig. 1. Detection of cell markers, viral antigens and apoptotic nuclei in mixed mouse primary nerve cell cultures. (A) Identification of neural cell types. Neuronal cells, astrocytes and oligodendrocytes were detected by immunofluorescence labelling in the same culture with TUJ1 antibodies and rhodamine conjugate (red), anti-GFAP and biotin/streptavidinAMCA (blue), and anti-O4 and FITC conjugate (green), respectively. Cells were observed by epifluorescence under a Leica microscope with a double filter for FITC and rhodamine, and under UV for AMCA. Bar, 20 µm. (B) Identification of neural cell types infected with PV. Cell cultures were mock-infected (a, b, c) or infected with the PV-1/Mahoney strain at an m.o.i. of 10 p.f.u./cell (d, e, f) for 16 h. Immunofluorescence staining was then performed. Neuronal cells (a, d), astrocytes (b, e) and oligodendrocytes (c, f) were identified separately with the specific primary antibodies described above, then stained with secondary antibody tagged with rhodamine (red). Viral antigens were detected with anti-PV-1 rabbit serum (a, c, d, f) or with anti-PV-1 mouse monoclonal antibody C3 (b, e), then with the appropriate FITC-labelled secondary antibody (green). Infected neurons (d) and oligodendrocytes (f) showed viral antigen staining in cell bodies (black arrowheads) and small dots in processes (white arrowheads) but no detectable modification of their morphology in comparison with control cells (a, c). In contrast, infected astrocytes (e) showing viral antigen staining (black arrowheads) presented a decrease in the intensity of GFAP staining and a lost of the fibrillary aspect of this staining in comparison with astrocytes negative for viral antigens (arrows in e) and with mock-infected astrocytes (b). Bar, 20 µm. (C) Identification of PV-infected and apoptotic cells. Cell cultures were infected with PV-1/Mahoney strain for 16 h, then nuclei were stained by DAPI incubation (blue in a, b, c) or TUNEL reactions performed with biotin-16-dUTP and streptavidin CY3-conjugated (blue in d, e, f), as described (Girard et al., 1999 ). Neuronal cells (a, d), astrocytes (b, e) and oligodendrocytes (c, f) were detected with the primary antibodies described above, then stained with the appropriate secondary antibody tagged with FITC (green). Viral antigens were detected with anti-PV-1 rabbit serum (a, d, c, f) or with antibody C3 (b, e), then with the appropriate secondary antibody conjugated either to rhodamine (red in a, b, c) or to CY5 (red in d, e, f). Cells were observed by epifluorescence as described above (a, b, c; bar, 20 µm) or under a confocal microscope (d, e, f; bar, 5 µm). The majority of infected cells showed a nucleus condensation (black arrowheads in a, b, c) visualized by intense DAPI staining.

To test the susceptibility of mixed mouse nerve cell cultures to PV infection, we followed the growth kinetics of the PV-1/Mahoney strain infecting these cells (Fig. 2). The yield of infectious particles was determined at the times indicated up to 24 h after infection by determining TCID₅₀. The pattern of growth was similar to that observed in fully permissive human cells such as HEp-2 cells (Couderc et al., 1996 ). Although the virus yield was slightly lower than that obtained in HEp-2 cell cultures (about 1 log₁₀ lower) (data not shown), these mouse primary neural cell cultures were productively infected with PV.

(15K): Fig. 2. Single growth curve of PV in mixed mouse primary nerve cell cultures. Cells were infected at an m.o.i. of 10 p.f.u./cell with PV-1/Mahoney. At the indicated times, cells and supernatants were harvested and total yield of infective particles was determined by the TCID50 on HEp-2 cells. Each point represents the mean and the standard error of the mean (bars) of two separate experiments.

We identified infected cells in the mixed primary nerve cell cultures by double-immunofluorescence labelling with antibodies specific for PV capsid proteins and for the markers of neural cells described above. The anti-viral antibodies used were anti-PV-1 rabbit serum (1/800 dilution) or the mouse monoclonal antibody C3 (1/200 dilution of mouse ascitic fluid) (Blondel et al., 1983 ) directed against the PV capsid. Control experiments included mock-infected cell cultures as negative controls for viral antigen detection and omission of the cell-specific primary antibodies from the staining procedure as negative controls for cellular antigen detection. No non-specific labelling was observed in controls.

In infected cell cultures, viral antigen staining was observed in the cell cytoplasm, the site of virus replication, in all three cell types (Fig. 1B). While most astrocytes and oligodendrocytes (more than 80%) were infected at 8 h and 16 h post-infection, only 1520% of neurons were positive for viral antigens. Viral antigens were also detected in the processes of infected oligodendrocytes and neurons, as illustrated in Fig. 1(B) for the 16 h time point. At later time points (22 h and 28 h post-infection), the percentages of infected cells were similar to those at earlier time points. However, most astrocytes exhibited an altered morphology with a decrease in both the intensity and the fibrillary aspect of the GFAP staining. Moreover, neurons and the rare surviving oligodendrocytes showed degeneration of cell processes.

As the majority of neurons did not stain positive for viral antigen, we verified that PV receptor was expressed on their surface. CD155 was detected in cultures by immunofluorescence with the mouse monoclonal antibody IgG1 404.19 (1/100 dilution, kindly supplied by M. Lopez) (Lopez et al., 1997 ), followed by biotinylated antibody (1/100 dilution, Southern Biotechnology Associates, Inc.) and with streptavidinCY2 conjugate (1/200 dilution, Jackson ImmunoResearch). Neurons were identified with TUJ1 and the appropriate fluorescent secondary antibody in the same culture. CD155 was detected at the surface of all cells expressing neuron-specific β-tubulin (data not shown), indicating that the low level of neuron infection was not due to the lack of PV receptors but rather to a restriction in a post-binding event of the virus cycle. The role of the stage of maturation of the neuronal population in the culture on the susceptibility to PV needs to be studied further.

In the mixed mouse primary nerve cell cultures, the three main cell types of the CNS, neurons, astrocytes and oligodendrocytes, are thus all susceptible to PV infection. In contrast, in the CNS, the main cell target of PV is the motor neuron. However, no clear experiments using glial cell markers have yet been described and it would be interesting to perform double labelling for glial cell markers and viral antigens in the CNS of infected TgCD155 mice, even if the CD155 mRNA could not be detected by in situ hybridization in the glial cells of the TgCD155 mouse CNS (Koike et al., 1994 ).

To investigate whether PV infection of neural cells was associated with apoptosis, we analysed DNA fragmentation in infected cultures by testing for oligonucleosomal laddering, an indicator of apoptosis. At 16 h and 28 h post-infection, cells (6x10⁶) were washed and suspended in a buffer containing 50 mM TrisHCl, pH 7·5, and 20 mM EDTA and lysed by addition of NP40 to a final concentration of 1%. After pelleting intact chromatin in an Eppendorf centrifuge (1200 g, 5 min, 4 °C), SDS was added to the supernatant to a final concentration of 0·5%. Supernatants were treated with 0·1 mg/ml proteinase K for 2 h at 50 °C and DNA was precipitated and end-labelled with terminal transferase (25 U) and digoxigenin-11-dUTP (50 µM final concentration) (Boehringer Mannheim) according to the manufacturers instructions. The DNA was treated with RNase A (0·1 mg/ml, 37 °C, 30 min) and the samples were electrophoresed on a 1·8% agarose gel, transferred to a Hybond-N nylon membrane (Amersham Life Science) and visualized by the digoxigenin luminescent detection kit (Boehringer Mannheim) with alkaline phosphatase-conjugated antibody and Lumin-phos Plus as the chemiluminescent substrate for the alkaline phosphatase.

No DNA laddering was visualized in mock-infected mixed primary nerve cell cultures (Fig. 3, lane 2). In cultures infected for 16 h, only a slight DNA laddering was detected (data not shown), whereas distinct DNA laddering, manifest as mono- (180200 bp), di-, and trinucleosome fragments, was clearly observed in cultures infected for 28 h (Fig. 3, lane 3). Thus, PV infection caused apoptosis in mixed nerve cell cultures and DNA fragmentation appeared to be delayed, lagging behind the virus growth curve for PV-induced death in HEp-2 cells. In fact, late PV-induced apoptosis has also been previously observed in the CaCo-2 enterocyte-like cell line and in the U937 promonocyte cell line (Ammendolia et al., 1999 ; Lopez-Guerrero et al., 2000 ). The delay in PV-induced death in these cells might be based on the fact that CaCo-2 and U937 cells, as well as mixed nerve cells in our cultures, are more differentiated than HEp-2 cells.

(42K): Fig. 3. Detection of DNA ladders in PV-infected mixed mouse primary nerve cell cultures. DNA fragmentation was assayed in cell cultures infected with PV-1/Mahoney at an m.o.i. of 10 p.f.u./cell for 28 h. Lane 1, DNA molecular mass markers; lane 2, mock-infected culture; lane 3, culture infected with PV-1/Mahoney.

To identify infected cells dying by apoptosis, cell cultures were immunostained for both viral antigen and cell markers and simultaneously tested for apoptosis by either DAPI staining or terminal deoxynucleotidyltransferase-mediated dUTP nick end-labelling (TUNEL), as described previously (Girard et al., 1999 ). Eight hours post-infection, most nuclei showed no condensation or fragmentation after DAPI staining and were negative for TUNEL, as observed in mock-infected cultures (data not shown). At 16 h post-infection, PV-infected cells of the three cell types were seen to exhibit a morphology characteristic of apoptosis with nuclear condensation or fragmentation (Fig. 1C). The number of these cells increased until 22 h post-infection, reaching 90% of infected cells. Thus, PV replication led to apoptosis in the three cell types of the CNS: neurons, astrocytes and oligodendrocytes.

To determine whether PV-induced apoptosis was caspase-dependent, we studied the effect of the irreversible and cell-permeable pan-caspase inhibitor Z-VAD.FMK (benzyloxycarbonyl-Val-Ala-Asp-(OMe) fluoromethylketone) on apoptosis in PV-infected cell cultures. The inhibitor was used at a concentration of 100 µM, which has been shown to inhibit caspases completely in cultured mammalian cells (Slee et al., 1996 ) but which did not affect PV growth (data not shown). Apoptosis was determined by evaluating DNA fragmentation in samples 18 h post-infection by quantitative ELISA. Inhibition of caspases was not analysed after 18 h post-infection because Z-VAD.FMK had a toxic effect on these cell cultures after this time point. Infected cell cultures exhibited substantial oligonucleosome DNA fragmentation in the absence of caspase inhibitor, but in the presence of Z-VAD.FMK there was no detectable DNA fragmentation (Fig. 4). Thus, the PV-induced apoptosis was caspase-dependent.

(21K): Fig. 4. DNA fragmentation following PV-induced apoptosis and its inhibition by Z-VAD.FMK in mixed mouse primary nerve cell cultures. Cell cultures were incubated with or without 100 µM Z-VAD.FMK (Bachem) for 90 min and infected with PV-1/Mahoney at an m.o.i. of 10 p.f.u./cell for 18 h. DNA fragmentation was estimated by using the Cell Death Detection Elisaplus kit (Roche).

This primary cell culture from neonatal mouse cortex containing all three of the major CNS cell types is thus a relevant cellular model for studying the molecular mechanisms of PV-induced apoptosis in nerve cells. Studies of the molecular mechanisms involved in apoptosis of HeLa cells infected with PV have revealed that caspases and also serine proteases such as chymotrypsin are involved during the executive phase of apoptosis (Agol et al., 1998 , 2000 ). Moreover, the anti-apoptotic proteins Bcl-2 and Bcl-X_L (Castelli et al., 1998 ) and the inhibition of RNase L block PV-induced apoptosis in HeLa cells (Castelli et al., 1997 ). Thus, apoptosis triggered by PV in HeLa cells may require the activity of RNase L and involve both a Bcl-X_L sensitive pathway and caspases. In U937 cells, caspases are also involved in PV-induced apoptosis (Lopez-Guerrero et al., 2000 ). Presumably, certain PV proteins induce apoptosis: it has been shown recently that expression of PV protease 2A in human embryonic kidney epithelial 293 cells (Goldstaub et al., 2000 ) or PV protease 3C in HeLa cells (Barco et al., 2000 ) is sufficient to trigger apoptosis. Moreover, 3C-induced apoptosis seems to depend on the caspase pathway, whereas 2A-induced apoptosis may be caspase-independent. Here, we have shown that the apoptotic pathway induced by PV in mixed primary nerve cell cultures involves caspases. This cell model will be useful for further investigations of the biochemical pathways leading to PV-induced apoptosis in nerve cells as well as the PV proteins involved in this process.