Astrocyte-Derived Nitric Oxide Causes Both Reversible and Irreversible Damage to the Neuronal Mitochondrial Respiratory Chain

Authors

  • Victoria C Stewart,

  • Martyn A Sharpe,

  • John B Clark,

  • Simon J R Heales


  • Abbreviations used : FBS, fetal bovine serum ; IFN, interferon ; iNOS, nitric oxide synthase of the inducible type ; LDH, lactate dehydrogenase ; LPS, lipopolysaccharide ; MEM, minimal essential medium ; MS, multiple sclerosis ; l-NAME, N-nitro-l-arginine methyl ester ; NO, nitric oxide ; NOS, nitric oxide synthase ; ONOO-, peroxynitrite ; PTFE, polytetrafluoroethane ; TBE, trypan blue exclusion.

Address correspondence and reprint requests to Dr. V. C. Stewart at Department of Neurochemistry, Institute of Neurology, University College London, Queen Square, London WC1N 3BG, U.K. E-mail : V. Stewart@ion.ucl.ac.uk

Abstract

Cytokine-stimulated astrocytes produce nitric oxide (NO), which, along with its metabolite peroxynitrite (ONOO-), can inhibit components of the mitochondrial respiratory chain. We used astrocytes as a source of NO/ONOO- and monitored the effects on neurons in coculture. We previously demonstrated that astrocytic NO/ONOO- causes significant damage to the activities of complexes II/III and IV of neighbouring neurons after a 24-h coculture. Under these conditions, no neuronal death was observed. Using polytetrafluoroethane filters, which are permeable to gases such as NO but impermeable to NO derivatives, we have now demonstrated that astrocyte-derived NO is responsible for the damage observed in our coculture system. Expanding on these observations, we have now shown that 24 h after removal of NO-producing astrocytes, neurons exhibit complete recovery of complex II/III and IV activities. Furthermore, extending the period of exposure of neurons to NO-producing astrocytes does not cause further damage to the neuronal mitochondrial respiratory chain. However, whereas the activity of complex II/III recovers with time, the damage to complex IV caused by a 48-h coculture with NO-producing astrocytes is irreversible. Therefore, it appears that neurons can recover from short-term damage to mitochondrial complex II/III and IV, whereas exposure to astrocytic-derived NO for longer periods causes permanent damage to neuronal complex IV.

There is increasing evidence implicating excessive generation of nitric oxide (NO) and its toxic metabolite peroxynitrite (ONOO-) in the pathogenesis of several neurodegenerative disorders, including multiple sclerosis (MS) (Heales et al., 1999). Our observation that the CSF concentration of nitrate and nitrite (stable degradation products of NO/ONOO-) is elevated by 70% in MS patients (Johnson et al., 1995) supports this hypothesis. Furthermore, increased activity of NO synthase (NOS) of the inducible type (iNOS) and iNOS mRNA has been demonstrated in astrocytes associated with demyelinating lesions in postmortem MS brain (Bo et al., 1994 ; Bagasra et al., 1995). Nitrotyrosine residues have also been detected in MS brain (Bagasra et al., 1995), indicating the presence of ONOO-. Therfore, formation of NO/ONOO- may directly contribute to the oligodendrocyte/neuron damage observed in MS.

NO/ONOO- can inhibit components of the electron transport chain (Bolaños et al., 1994), and damage to the mitochondrial electron transport system has been implicated in the pathogenesis of several neurological disorders, including MS (Beal et al., 1993 ; Heales et al., 1999). Within the brain, there may be a differential susceptibility of various cell types to NO/ONOO-. Although astrocytes in culture show apparent resistance to these effects, neurons and oligodendrocytes appear particularly sensitive to NO/ONOO- in vitro (Mitrovic et al., 1994 ; Bolaños et al., 1995). Studies using NO donors have revealed an impairment of energy metabolism in cultured neurons (Bolaños et al., 1996) and oligodendrocytes (Mitrovic et al., 1994) and also reversible axonal conduction block (a characteristic of MS) in isolated nerve preparations (Redford et al., 1997). Furthermore, inhibition of the mitochondrial respiratory chain is associated with impairment, both in vitro and in vivo (Heales et al., 1995 ; Bates et al., 1996), of N-acetylaspartate metabolism, which is dependent on neuronal mitochondrial function (Moffet et al., 1991). Furthermore, NMR spectroscopy has revealed a potentially reversible reduction in N-acetylaspartate content in MS plaques (Davie et al., 1994), supporting the hypothesis of neuronal mitochondrial damage in MS.

Following cytokine exposure, astrocytes show increased NO production and a corresponding increase in NOS activity (Bolaños et al., 1994). Although the NO/ONOO- formed does not affect astrocyte survival, it may diffuse out and cause mitochondrial damage and possibly cell death to neighbouring NO/ONOO- -sensitive cells, such as neurons and/or oligodendrocytes.

In view of the ability of astrocytes to generate NO and ONOO- without astrocyte cell death occurring, it is possible that, in certain neurological disorders, astrocyte-derived NO/ONOO- may cause mitochondrial damage to susceptible target cells, such as neurons. We have previously shown that a 24-h coculture of neurons with NO/ONOO- -generating astrocytes results in significant neuronal mitochondrial damage (Stewart et al., 1998). Despite this mitochondrial damage, neuronal cell death did not occur (Stewart et al., 1998). In contrast, direct exposure of neurons in monoculture to ONOO- or a NO donor results in significant mitochondrial damage and is accompanied by neuronal cell death (Bolaños et al., 1995, 1996). The intracellular GSH concentration may be an important factor in determining cellular susceptibility to NO/ONOO-. The reason for the apparent decrease in neuronal sensitivity, under coculture conditions, may arise as a result of up-regulation of neuronal GSH synthesis due to the trafficking of GSH precursors from astrocytes to neurons (Bolaños et al., 1996 ; Dringen et al., 1999). As a coculture system may reflect the situation in vivo better than using monocultures of neurons alone, we believe that results obtained using monocultures should be treated with caution.

The aim of the present study was to expand on our previous observations and investigate whether neurons have the capacity to recover the activities of their mitochondrial complexes on removal of the iNOS-expressing astrocytes. In addition, the cells were cocultured for a longer period to see if prolonged exposure of neurons to astrocytic-derived NO/ONOO- caused further neuronal mitochondrial damage, possibly leading to neuronal cell death. We also used polytetrafluoroethane (PTFE) membranes, which are permeable to NO but impermeable to NO derivatives, to differentiate between effects due directly to astrocytic NO and those due to NO derivatives.

MATERIALS AND METHODS

Materials

Earle's balanced salts solution, interferon (IFN-γ) (mouse, recombinant), lipopolysaccharide (LPS), and N-nitro-l-arginine methyl ester (l-NAME) were obtained from Sigma Chemical Co. (Poole, U.K.). Minimal essential medium (MEM), fetal bovine serum (FBS), and tissue culture plastics were purchased from Life Technologies (Renfrewshire, U.K.). All other substrates were purchased from Sigma or Boehringer Mannheim.

Preparation of ONOO-

ONOO- was prepared from amyl nitrite and H2O2 using the method of Uppu and Pryor (1996). In brief, 26 ml of amyl nitrite (washed three times with water) was added to 100 ml of 2 M H2O2, 2 M NaOH, and 2 mM diethylenetriaminepentaacetic acid and stirred vigorously for 3 h at 4°C. The lower aqueous layer was removed from the organic phase using a separating funnel and washed three times with an equal volume of ice-cold hexane. To remove residual H2O2, 10 g of manganese dioxide was gradually added to the ONOO- solution while it was being stirred, on ice, for å 1 h. The ONOO- solution was filtered and then stored at Î 70°C. The concentration of ONOO- (90 mM) was determined spectrophotometrically using the extinction coefficient of 1,670 M-1 cm-1 at 302 nm. Nitrite contamination of the ONOO- solution was determined to be approximately one-third of the ONOO- concentration, as found by Uppu and Pryor (1996).

Cell culture

Primary cortical astrocyte cultures were prepared from neonatal Wistar rats (1-2 days old) as described by Bolaños et al. (1995). Cells were cultured for 7 days in d-valine-based MEM, supplemented with 10% (vol/vol) FBS and 2 mMl-glutamine, followed by 7 days in l-valine-based MEM. Primary forebrain neuronal cultures were prepared from fetal Wistar rats (day 17 of gestation) as described by Bolaños et al. (1995). Cell suspensions were plated at a density of 2.5 Í 105 cells/cm2 in six-well culture dishes coated with poly-l-ornithine. Neuronal cells were cultured in d-valine-based MEM supplemented with 10% (vol/vol) FBS, 2 mMl-glutamine, and 25 mM KCl. Three days after plating, the neuronal medium was replaced with fresh medium containing 10 μM cytosine arabinofuranoside. Neuronal cultures were used at day 6 in vitro. Immunocytochemistry using neurofilament antibody confirmed the cell population to be 85% neuronal (Tabernero et al., 1993).

Neuronal-astrocyte coculture

The coculture system used was described by Stewart et al. (1998). In brief, astrocytes at day 13 in vitro were collected by trypsinisation, resuspended, and plated at a density of 1 Í 105 cells/cm2 onto polyester cell culture inserts (membrane pore size, 0.4 μm ; Costar, High Wycombe, U.K.) in a total volume of 4 ml of fresh astrocytic medium. The next day, the inserts were incubated for 24 h in the presence or absence of IFN-γ (100 U/ml) plus LPS (1 μg/ml). In addition, some wells also contained the NOS inhibitor l-NAME (1 mM). After the incubation period, the inserts were thoroughly washed and transferred to wells containing neurons (neuronal culture day 6) in fresh neuronal medium. Some wells also contained l-NAME (1 mM). The neuronal-astrocytic cocultures were incubated for 24 h, after which the inserts containing astrocytes were removed.

In addition, some experiments involved a 48-h coculture during which neurons and astrocytes were cocultured for 24 h and then the astrocytes were removed and replaced with fresh control/iNOS-expressing astrocyte inserts, which were cocultured with the neurons for a further 24 h. This was done to ensure that the neurons were being exposed to maximal astrocyte-derived NO, although in parallel experiments where some cultures were incubated with the same insert comparable results were obtained. At the end of the 24- or 48-h coculture period, some wells were terminated immediately, whereas others were given fresh medium and allowed to recover for a further period of up to 72 h.

Additional coculture experiments were performed using inserts containing a PTFE membrane (Rank Brothers, Bottisham, Cambridge, U.K.) below the polyester membrane on which the astrocytes were seeded. PTFE is permeable only to gases such as NO (Dobson and Taylor, 1986), and so NO derivatives, such as ONOO-, NO2-, and NO3-, are unable to penetrate PTFE membranes over the time course of our experiment. Therefore, these coculture experiments would enable us to assign effects arising directly from astrocyte-derived NO.

Neuronal viability

As a measure of neuronal viability, lactate dehydrogenase (LDH ; EC 1.1.1.27) activity was determined in medium as described by Vassault (1983). LDH released was expressed as a percentage of total cell LDH as described previously (Bolaños et al., 1995). LDH released was used as an index of cell death (Koh and Choi, 1987). In addition, neuronal cell viability was assessed by trypan blue exclusion (TBE) (Koh and Choi, 1988) at the end of each coincubation.

Enzyme activity determination

On termination of the experiments, neurons were trypsinised and pelleted. The activities of the mitochondrial complexes—NADH-CoQ1 reductase (complex I ; EC 1.6.99.3), succinate-cytochrome c reductase (complex II/III ; EC 1.8.3.1), cytochrome c oxidase (complex IV ; EC 1.9.3.1), and citrate synthase (EC 4.1.3.7)—were measured in cell homogenates as previously described (Bolaños et al., 1994). Enzyme activities were expressed as nanomoles per minute per milligram of protein except that of cytochrome c oxidase, which was expressed as the first-order reaction constant, k per minute per milligram.

Determination of nitrate plus nitrite content

As a measure of NO generation, the astrocytic cell culture medium was assayed for nitrate plus nitrite. Nitrate was reduced by incubation with nitrate reductase and then quantified as nitrite by the Griess reaction (Green et al., 1982).

Protein content determination

Protein concentration was determined by the method of Lowry et al. (1951).

Statistical analysis

Results are expressed as mean ê SEM values for the number of independent cell culture preparations indicated. Multiple comparisons were made by one-way ANOVA followed by the least significant difference multiple range test. In all cases, pÔ 0.05 was considered significant.

RESULTS

Astrocytic NO generation

Following cytokine exposure, astrocytes show increased NOS activity and a corresponding increase in NO production (Bolaños et al., 1994 ; Brown et al., 1995). In the present study, exposing cultures of astrocytes to IFN-γ (100 U/ml) plus LPS (1 μg/ml) for 24 h resulted in production of 111.7 ê 11.2 nmol of nitrate Ï nitrite (stable degradation products of NO/ONOO-)/mg of protein (n ë 3), compared with 14.3 ê 0.7 nmol/mg (n ë 3) for control unstimulated astrocytes. Following removal of the cytokine-containing medium, washing of the cell surface, and replacement with fresh medium, the astrocytes were incubated for a further 24 h, and then nitrate/nitrite was assayed. Astrocytes previously treated with IFN-γ plus LPS produced 98.8 ê 12.3 nmol/mg (n ë 3), whereas 17.1 ê 1.0 nmol/mg total nitrite (n ë 3) was detected in untreated cells. Furthermore, after removal of the medium again and replacement with fresh medium for another 24 h, i.e., now 72 h post-cytokine stimulation, 86.0 ê 12.2 nmol/mg total nitrite (n ë 3) was detected in the wells previously treated with cytokines compared with 17.2 ê 1.5 nmol/mg (n ë 3) in untreated wells. Therefore, we were able to use cytokine-stimulated astrocytes as a source of NO/ONOO- in our coculture system.

In experiments using PTFE filters, 24 h after exposure of astrocytes to IFN-γ plus LPS, 90.9 ê 4.5 nmol/mg total nitrite was detected on the “neuronal” side of the PTFE filter. This level of nitrite generation is comparable with that detected previously in our cytokine-stimulated astrocyte cultures indicating that the astrocyte-derived NO is penetrating the PTFE membrane and degrading on the “neuronal” side of the membrane because the PTFE membrane is not permeable to nitrite or nitrate.

To confirm that the PTFE membrane was impermeable to nitrite/nitrate, 50 μM sodium nitrite or 50 μM sodium nitrate was added to the upper chamber of the PTFE filter, and 24 h later, the medium on either side of the PTFE membrane was assayed for nitrite/nitrate. In both cases, 50 μM nitrite was detected in the upper chamber, whereas there was no detectable nitrite on the other side of the PTFE membrane. These results confirm that nitrate/nitrite does not penetrate the PTFE membrane over the time course of our experiment.

Similarly, 90 mM ONOO- was unable to penetrate the PTFE membrane. ONOO- (90 mM) was added to the upper chamber of both “normal” and PTFE-coated inserts. Within 30 min, there were no viable neurons in the wells containing “normal” filters, whereas in the wells containing PTFE filters, 24 h later the neurons were still 100% viable as assessed by trypan blue exclusion. These results indicate ONOO- does not penetrate the PTFE membrane over the time course of our experiment.

Neuronal mitochondrial complex activities

We have previously shown that coculture of neurons with unstimulated astrocytes does not affect the activity of the neuronal mitochondrial respiratory chain complexes (Stewart et al., 1998). However, exposure to iNOS-expressing astrocytes for 24 h leads to a degree of neuronal respiratory chain damage. Expanding on these studies, we have now made the following observations :

Complex I. Coculture of neurons with iNOS-expressing astrocytes for 24 or 48 h did not affect the activity of neuronal mitochondrial complex I (Fig. 1).

Figure 1.

Effect of coculture with astrocytes on activity of neuronal mitochondrial complex I. Neurons were cocultured for either 24 or 48 h with control unstimulated astrocytes (Control astros) or activated NO-producing astrocytes (Activated astros). Data are mean ê SEM (bars) values (n ë 7), expressed as nmol/min/mg.

FIG. 1.

Complex II/III. Coculture of neurons with iNOS-expressing astrocytes for 24 h caused a significant 50% loss of neuronal complex II/III activity (6.4 ê 0.4 vs. 12.5 ê 0.4 nmol/min/mg ; pÔ 0.05 ; Fig. 2). In cultures where l-NAME was included the complex II/III activity was comparable with that in control neurons (11.8 ê 0.7 nmol/min/mg, n ë 4), confirming that NOS activity was responsible for the loss of neuronal complex II/III activity.

Figure 2.

Effect of coculture with astrocytes on activity of neuronal mitochondrial complex II/III. Neurons were cocultured for either 24 or 48 h with control unstimulated astrocytes (Control astros) or activated NO-producing astrocytes (Activated astros). Data are mean ê SEM (bars) values (n ë 7), expressed as nmol/min/mg. *p Ô 0.05 compared with neurons cocultured with control astrocytes.

FIG. 2.

Furthermore, the NO-mediated damage appeared to be reversible as on removal of the astrocytes, the complex II/III activity had recovered to 85% of control values by 24 h, and by 72 h there was 100% recovery of activity compared with incubations with unstimulated (control) astrocytes (Fig. 3).

Figure 3.

Neuronal complex II/III activity over a 72-h period following a 24-h coculture with control astrocytes (Control Astros) or activated NO-producing astrocytes (Activated Astros). After the 24-h coculture period, the astrocytes were removed, and the neurons were given fresh medium and monitored for a further 72 h. Coculture with activated astrocytes results in a significant loss of neuronal complex II/III activity. However, on removal of the astrocytes, the neurons show a time-dependent recovery of complex II/III activity. Complex II/III activity is expressed as nmol/min/mg. Data are mean ê SEM (bars) values (n ë 4).

FIG. 3.

Increasing the period of coculture with iNOS-expressing astrocytes from 24 to 48 h did not cause any further loss in complex II/III activity (Fig. 2). As before, on removal of astrocytes from the coculture system, there was a complete recovery of neuronal complex II/III activity (data not shown).

In experiments involving PTFE inserts, the damage to neuronal complex II/III on coculture with inducible iNOS-expressing astrocytes was comparable with that seen in cocultures involving “normal” inserts (Table 1). These results indicate that the damage to neuronal complex II/III activity was caused by a molecule that was able to penetrate the PTFE membrane, e.g., NO. Furthermore, inclusion of the NOS inhibitor l-NAME prevented the observed loss of complex II/III activity (10.7 ê 0.2 nmol/min/mg, n ë 4). Based on these findings, we hypothesise that astrocyte-derived NO, rather than ONOO-, is responsible for the observed neuronal complex II/III damage.

Table 1. Effect of PTFE membrane on neuronal mitochondrial complexes II/III and IV following 24- or 48-h coculture with control or iNOS-expressing (IFN-γÏLPS-treated) astrocytes
Interval, treatmentComplex II/III activity (nmol/min/mg) Complex IV activity (k/min/mg)
  1. In all cases, coculture of neurons with IFN-γÏ LPS-treated astrocytes resulted in a significant loss of complex II/III and IV activity. Furthermore, there was no difference between neuronal complex activities in the presence or absence of PTFE. Complex II/III activity is expressed as nmol/min/mg of protein. Complex IV activity is expressed as the first-order rate reaction constant k/min/mg of protein. Data are mean ê SEM values (n ë 4).

24 h  
Control12.5 ê 0.41.13 ê 0.03
IFN-γ/LPS Î PTFE7.6 ê 1.00.79 ê 0.02
IFN-γ/LPS Ï PTFE7.6 ê 1.00.78 ê 0.04
48 h  
Control10.8 ê 0.31.07 ê 0.03
IFN-γ/LPS Î PTFE7.7 ê 0.90.69 ê 0.02
IFN-γ/LPS Ï PTFE7.3 ê 0.30.73 ê 0.05

TABLE 1.

Complex IV. Coculture of iNOS-expressing astrocytes with neurons for 24 h caused a significant 40% loss of neuronal complex IV activity (0.69 ê 0.04 vs. 1.13 ê 0.03 k/min/mg ; pÔ 0.05 ; Fig. 4). This was prevented on inclusion of l-NAME during the cytokine stimulation of astrocytes (1.07 ê 0.04 k/min/mg, n ë 4). Furthermore, on removal of astrocytes from the coculture system, the neuronal complex IV activity recovered to 80% of control values after 24 h (Fig. 5). Extending the recovery period to 72 h did not allow for any further recovery of complex IV activity (Fig. 5).

Figure 4.

Effect of coculture with astrocytes on activity of neuronal mitochondrial complex IV. Neurons were cocultured for either 24 or 48 h with control unstimulated astrocytes (Control astros) or activated NO-producing astrocytes (Activated astros). Data are mean ê SEM (bars) values (n ë 7), expressed as k/min/mg. *pô 0.05 compared with neurons cocultured with control astrocytes.

Figure 5.

Neuronal complex IV activity over a 72-h period following a 24-h coculture with control astrocytes (Con Astros) or activated NO-producing astrocytes (Activated Astros). After the 24-h coculture period, the astrocytes were removed, and the neurons were given fresh medium and monitored for a further 72 h. Coculture with activated astrocytes results in a significant loss of neuronal complex IV activity. On removal of the astrocytes, the neurons show only a partial recovery of complex IV activity. Complex IV activity is expressed as k/min/mg. Data are mean ê SEM (bars) values (n ë 4).

FIG. 4.

FIG. 5.

Increasing the period of coculture with iNOS-expressing astrocytes, from 24 to 48 h, did not cause further damage to neuronal complex IV activity (Fig. 4). Again, inclusion of l-NAME prevented the loss of complex IV activity (1.06 ê 0.04 vs. 0.61 ê 0.02 k/min/mg ; Pô 0.05). However, on removal of the astrocytes from the coculture system, the activity of neuronal complex IV did not recover (24-h recovery period, 0.72 ê 0.01 k/min/mg ; 48-h recovery period, 0.65 ê 0.06 k/min/mg).

In experiments involving PTFE inserts, the damage to neuronal complex IV on coculture with iNOS-expressing astrocytes was comparable with that seen in cocultures involving “normal” inserts (Table 1), indicating that the damage to neuronal complex IV activity was caused by a substance that was able to penetrate the PTFE membrane, e.g., NO.

Citrate synthase. In all cases, the specific activity of citrate synthase, an indicator of the mitochondrial enrichment of the cell homogenates used in the above investigations, was unaffected (data not shown).

Neuronal viability

We have previously shown that coculture of neurons with cytokine-stimulated astrocytes for 24 h does not affect neuronal viability (Stewart et al., 1998). In the present study, a 24-h coculture of neurons with cytokinestimulated astrocytes did not affect neuronal viability, confirming our previous results. However, increasing the period of coculture from 24 to 48 h did cause a significant increase in neuronal death, as judged by TBE (78.0 ê 3.3% viability compared with controls, n ë 5 ; Pô 0.05) and LDH leakage (53.8 ê 12.0% increase over control levels, n ë 5 ; Pô 0.05). Inclusion of l-NAME protected against this loss of neuronal viability (TBE, 96.0 ê 2.5% viability compared with controls, n ë 4 ; LDH, 5.7 ê 2.9% increase over control levels, n ë 4).

Although LDH leakage was assayed in medium that had been incubated with both neurons and astrocytes, we have previously shown that cytokine stimulation of astrocytes does not affect astrocytic LDH leakage (Bolaños et al., 1994 ; Stewart et al., 1997). Therefore, we are confident that the neurons are the source of the increase in LDH seen in our coculture system. It should be noted that LDH leakage was detected in all neuron cultures under normal conditions, and therefore our LDH leakage results are expressed as a percentage over and above the control value of 25.7 ê 1.2%. Although the value may appear relatively high, it does not indicate that ò 25% of the untreated neuronal cultures are dead. Furthermore, it has been suggested that LDH release is perhaps better considered to be an indicator of a slowly evolving degenerative process rather than an indicator of cell death, i.e., a process that may normally occur in neuronal cultures with increasing time in vitro (Keilhoff and Wolf, 1993). Consequently, we do not rely on LDH release alone as a measure of cell viability and always use TBE to confirm our viability data.

DISCUSSION

In vivo, neurons are in close proximity to astrocytes, which under certain conditions are postulated to provide a stabilising milieu (Magistretti et al., 1997). The coculture system used here mimics this and allows two distinct populations of cells to exist in close proximity but with no direct contact, i.e., enabling the diffusion of soluble factors between the two cell types.

We have previously demonstrated that coculture of neurons with cytokine-stimulated astrocytes for 24 h does not affect neuronal viability (Stewart et al., 1998). However, in the present study we have shown that increasing the period of coculture from 24 to 48 h does cause a significant increase in neuronal death. This is in agreement with the findings of Dawson et al. (1994), who used mixed neuronal-glial cultures and observed delayed NO-mediated neuronal cell death. Furthermore, we demonstrated that the NOS inhibitor l-NAME protected against this loss of neuronal viability, suggesting that NO, or one of its derivatives, was responsible for the neuronal death. Similarly, Chao et al. (1996) demonstrated that NO released from cytokine-stimulated human astrocytes is toxic to human neurons.

We have shown that coculture of neurons and activated NOS-expressing astrocytes for 24 h causes significant damage to complexes II/III and IV of the mitochondrial respiratory chain (Bolaños et al., 1996 ; Stewart et al., 1998). The present study confirmed these results. Furthermore, we have now demonstrated that on removal of the NOS-expressing astrocytes from the coculture system, there is complete recovery of complex II/III activity and partial recovery of complex IV activity with time. Increasing the coculture period from 24 to 48 h does not result in further damage to the neuronal mitochondrial complexes. However, whereas the activity of damaged complex II/III recovers completely with time, the damage to complex IV caused by a 48-h coculture with activated NOS-expressing astrocytes is irreversible.

The damage to both complexes II/II and IV was prevented by the NOS inhibitor l-NAME, indicating involvement of NOS. However, we have so far not differentiated between effects due to NO itself and those due to NO derivatives such as ONOO-. PTFE is permeable only to gases such as NO, and so NO derivatives, such as ONOO-, NO2-, or NO3-, are unable to penetrate PTFE membranes (Dobson and Taylor, 1986). Therefore, we repeated our coculture experiments using inserts containing a PTFE membrane. This membrane was shown to be both permeable to NO and impermeable to ONOO- and NO2-Ï NO3-. We demonstrated that the PTFE membrane did not prevent the damage to the neuronal mitochondrial respiratory chain induced by cytokine-stimulated astrocytes. Therefore, it appears that astrocytic NO, and not astrocytic ONOO-, is responsible for the damage to the neuronal mitochondrial respiratory chain observed in our coculture model. However, subsequent formation of ONOO- on the other side of the PTFE membrane may still be responsible for the observed neuronal mitochondrial damage.

Astrocytic NO may react with superoxide to generate ONOO-. Superoxide may be generated in our system via autooxidation of glucose in the medium (Wolff and Dean, 1987). Furthermore, the mitochondrial respiratory chain itself produces appreciable quantities of superoxide (Turrens, 1997). Therefore, the astrocytic-derived NO may react with superoxide to form ONOO-, and the ONOO- may be responsible for the observed neuronal mitochondrial respiratory chain damage.

NO and ONOO- may exert inhibitory effects on the respiratory chain in several ways. Assay of complex II/III requires endogenous ubiquinone (Rustin et al., 1994), and it is known that NO reacts directly with ubiquinol (Poderoso et al., 1999). Furthermore, ONOO- reacts directly with complex II (Brookes et al., 1998). NO/ONOO- can inhibit complex IV activity both reversibly and irreversibly (Brown et al., 1995). In addition, complex IV activity correlates directly with the concentration of cardiolipin (Paradies et al., 2000). Complex IV activity is particularly sensitive to lipid peroxidation, and as ONOO- may initiate lipid peroxidation (Beckman et al., 1990), this may explain the observed irreversible loss of complex IV in the NO/ONOO- - exposed neurons.

In conclusion, it appears that neurons can recover from short-term mitochondrial complex II/III and IV damage, whereas exposure of neurons to astrocytic-derived iNOS activity over a longer interval results in permanent damage to neuronal complex IV. These observations may have important implications for the situation in vivo, suggesting that short-term exposure of neurons to oxidative stress may lead only to short-term mitochondrial respiratory chain damage, rather than a permanent deficit. In contrast, prolonged exposure to oxidising species may result in mitochondrial respiratory chain damage, leading to neuronal cell death. This may have implications for acute infection versus chronic inflammatory disease states, such as MS, and hence therapeutic opportunities.

Ancillary