|Home | About | Journals | Submit | Contact Us | Français|
The proinflammatory cytokines TNFα and Il-1β orchestrate the progression of CNS inflammation, which substantially contributes to neurodegeneration in many CNS pathologies. TNFα and Il-1β stimulate actin filament reorganization in non-neuronal cells often accompanied by the formation of reactive oxygen species (ROS). Actin filament dynamics is vital for cellular plasticity, mitochondrial function, and gene expression despite being highly susceptible to oxidative damage. We demonstrated that, in neuronal cells, TNFα and Il-1β stimulate a transient, redox-dependent reorganization of the actin cytoskeleton into lamellipodia under the regulation of Rac1 and a neuronal NADPH oxidase as the source of ROS. Persistent presence of intracellular ROS provoked oxidative damage (carbonylation) to actin coinciding with loss of lamellipodia and arrest of cellular plasticity. Inhibition of NADPH oxidase activity or Rac1 abolished the adverse effects of cytokines. These findings suggest that oxidative damage to the neuronal actin cytoskeleton could represent a key step in CNS neurodegeneration.
Inflammation accompanied by oxidative stress is widespread in acute, chronic, and psychiatric CNS pathologies and plays a key role in the progression of neurodegenerative processes (Mrak and Griffin, 2005; Lucas et al., 2006). Persistent expression of the two proinflammatory cytokines tumor necrosis factor α (TNFα) and interleukin-1β (Il-1β) compromises plasticity and survival of neuronal cells (Rothwell and Luheshi, 2000; Allan and Rothwell, 2001). Disruption of proper actin filament dynamics and oxidative damage to actin in neuronal cells is prevalent in both acute CNS injury and chronic CNS pathologies and might represent a pivotal step in neurodegeneration (Dalle-Donne et al., 2001; Beal, 2002). Many cellular processes aside from motility are intimately linked to actin filament dynamics including gene expression, mitochondrial homeostasis, and vesicle traffic (Lin et al., 1997; Gourlay and Ayscough, 2005; Percipalle and Visa, 2006). Munch et al. recently demonstrated that proinflammatory mediators released from microglia attenuated neurite outgrowth and caused neurite retraction (Munch et al., 2003). Exposure of hippocampal neurons to TNFα in vitro attenuated neurite outgrowth and reduced branching of neurites through a RhoA-dependent mechanism (Neumann et al., 2002). TNFα and Il-1β caused significant rearrangement of actin filament architecture in non-neuronal cells mediated by small Rho GTPases (Wojciak-Stothard et al., 1998; Peppelenbosch et al., 1999; Puls et al., 1999; Hanna et al., 2001). Rho GTPases are vital in converting a plethora of extrinsic stimuli into a coordinated reorganization of actin filaments in non-neuronal cells as well as neurons (Nobes and Hall, 1995; Hall and Nobes, 2000b; Luo, 2002). Recent observations in non-neuronal cells demonstrated that Rac1 and ROS are implicated in actin filament reorganization and cell motility (Moldovan et al., 1999; Karnoub et al., 2001; Nimnual et al., 2003). Actin filament dynamics is highly sensitive to oxidative modification and distinct ROS differentially affect assembly, disassembly, and stability of actin filament (Krieger-Brauer and Kather, 1995; Sulciner et al., 1996; Milzani et al., 1997; Dalle-Donne et al., 2001; Lassing et al., 2007). Notably, proinflammatory cytokines, growth factors, and hormones induce ROS formation in many non-neuronal cell types often under the regulation of Rac1 (Sundaresan et al. 1996). Indeed, Rac1-dependent ROS intermediates are key to many physiological processes including cellular proliferation, gene expression, ion channel modulation, and actin reorganization (Droge, 2002). Only recently members of the NADPH oxidoreductase family have emerged as primary sources of physiological redox signaling and pathological oxidative stress. The classic NADPH oxidase of phagocytes represents the prototypic member of this family defined by the large membrane subunit gp91phox (new nomenclature NOX2) (Lambeth, 2002). NOX isoforms have been identified in various non-phagocytic cell types including hematopoietic stem cells, endothelial cells, epithelial cells, myocardial cells, muscle cells, hepatocytes, and neurons (Bedard and Krause, 2007). NOX activities are multi-subunit protein complexes composed of two membrane-bound subunits (gp91phox and p22phox), and at least three cytosolic subunits (p67phox, p47phox, and p40phox) or their homologues. Most NOX isoforms (NOX1-4) require Rac1 or Rac2 as a key regulator for functional assembly of cytosolic subunits. In this study, we demonstrated that TNFα and Il-1β stimulate a transient and redox-dependent reorganization of the actin cytoskeleton into lamellipodia in neuronal cells under the regulation of Rac1. Persistent increases in ROS, likely superoxide generated by a NOX2 activity, are necessary for actin filament reorganization into lamellipodia yet also results in carbonylation of actin, an irreversible and damaging oxidative modification, accompanied by a loss of lamellipodia formation. These findings suggest that inflammation-mediated disruption of actin filament reorganization in neuronal cells through oxidative damage could represent a key step in CNS neurodegeneration.
Fibroblasts, endothelial cells, and smooth muscle cells all respond to the proinflammatory cytokines TNFα and IL-1β with a significant reorganization of the actin cytoskeleton into lamellipodia and filopodia (Wojciak-Stothard et al., 1998; Peppelenbosch et al., 1999; Hanna et al., 2001). Expanding on these findings, we examined the effects of TNFα and IL-1β on the actin cytoskeleton of SH-SY5Y human neuroblastoma cells since neuronal plasticity in the intact, developing, and regenerating CNS is almost universally linked to dynamics changes of the actin cytoskeleton (Luo, 2002).
Serum-starved SH-SY5Y cells displayed extensive lamellipodia formation and membrane ruffling accompanied by cell spreading upon acute exposure to TNFα or Il-1β (200 ng/ml each, 10 μl/500 μl medium) for 15 min, whereas exposure to an equal volume of PBS had not effect and SH-SY5Y cells retained a condensed, atrophic morphology (Fig. 1A–C). As a quantitative measure, we defined actively responding cells as cells exhibiting two or more distinct regions with lamellipodia and/or membrane ruffling. According to this criterion, we measured 80±4% (*p<0.0001, n=180) in response to 200 ng/ml TNFα and 83±4% (*p<0.0001, n=180) active cells in response to 200 ng/ml Il-1β, respectively, in stark contrast to 15±4% (n=180) active cells under control conditions. Both proinflammatory cytokines elicited a dose-dependent increase in the percentage of SH-SY5Y cells responding with lamellipodia formation and membrane ruffling (Table I). Remarkably, the reorganization of the neuronal actin cytokeleton ceased after prolonged exposure to either proinflammatory cytokine and SH-SY5Y cells reverted largely to their condensed, atrophic morphology typical for control conditions (Fig. 1D–G). After a 30 min presence of TNFα or Il-1β, the percentage of active SH-SY5Y cells was indistinguishable from control conditions (Table I). Taken together, these findings revealed that the proinflammatory cytokines TNFα and Il-1β elicited a dramatic yet transient reorganization of the neuronal actin cytokeleton into lamellipodia and membrane ruffles.
Reactive oxygen species (ROS) have emerged as key signaling intermediates in many vital cellular processes including mitogenesis, gene expression, stress responses, ion channel function as well as adhesion and motility (Kourie, 1998; Finkel and Sullivan, 2001; Droge, 2002; Lambeth, 2004). Several reports establish a link between cytokines, ROS formation and actin filament reorganization in non-neuronal cells (Sulciner et al., 1996; Sundaresan et al., 1996; Bonizzi et al., 1999; Deshpande et al., 2000; Moldovan et al., 2000). Therefore, we explored whether redox intermediates are implicated in the transient reorganization of the actin cytoskeleton in SH-SY5Y cells in response to proinflammatory cytokines.
Both disruption of ROS formation with Diphenylene iodonium (DPI, an NADPH oxidoreductase inhibitor) or ROS scavenging with Manganese (III) tetrakis (4-benzoic acid) porphyrin chloride (MnTBAP, a catalytic antioxidant and superoxide dismutase mimetic) abolished transient lamellipodia formation and membrane ruffling in response to TNFα or Il-1β (Fig. 2). As shown in Figure 2B, pretreatment of SH-SY5Y cells with 10 μM DPI for 30 min diminished the percentage of cells with lamellipodia and membrane ruffling upon acute addition of TNFα or Il-1β (200 ng/ml, 15 min) to control levels (13±4%, n=180) and 17±3%, n=180, respectively) compared to no DPI-pretreatment (65±5%, *p<0.001, n=180 and 57±4%, *p<0.001, n=180, respectively). DPI-treated SH-SY5Y cells exhibited the condensed morphology characteristic of control cells regardless of a presence of absence of cytokines (Fig. 2A). Basal levels of lamellipodia formation and membrane ruffling were not affected by DPI pretreatment. Similarly, MnTBAP-pretreatment of serum-starved SH-SY5Y cells (40 μM, 1 h) diminished lamellipodia formation and membrane ruffling upon acute addition of TNFα or Il-1β (15 min, 200 ng/ml) (Fig. 2A). As shown in Figure 2C, the extent of cytoskeletal reorganization in the presence of MnTBAP was substantially less (TNFα: 42±7%, **p<0.01, n = 180 and Il-1β: 40±7%, **p<0.01, n=180, respectively) compared to an absence of MnTBAP-pretreatment (TNFα: 85±7%, *p<0.01, n=180; and Il-1β: 80±7%, *p<0.01, n=180). Incubation of serum-starved SH-SY5Y cells with MnTBAP alone resulted in 17±7% active cells (n=180) identical to control conditions (28±7% active cells, n=180). Notably, SH-SY5Y cells in the presence of MnTBAP and to a lesser extend with DPI displayed considerable pseudopodia-like structures, which were not accounted for in our criterion. Taken together, a disruption of ROS formation (DPI) or scavenging of ROS (MnTBAP) abolished the transient reorganization of the neuronal actin cytoskeleton into lamellipodia in response to TNFα and Il-1β implying a role for ROS as signaling intermediates.
Member of the Rho family of small GTPases serve a pivotal function in the translation of extrinsic stimuli into a coordinated reorganization of the actin cytoskeleton with Rac1 activity tied to lamellipodia formation (Hall and Nobes, 2000a; Luo, 2000). Rac1 also serves as a key regulator of NADPH oxidase activities in non-phagocytic cell types stimulated by cytokines, hormones, and mitogenes (Freeman et al., 1996; Sulciner et al., 1996; Sundaresan et al., 1996). To assess whether Rac1 is implicated in the redox-dependent formation of lamellipodia in SH-SY5Y cells in response to cytokines, we transiently expressed dominant negative Rac1 (N17Rac1) or constitutively active Rac1 (V12Rac1) in serum-starved SH-SY5Y cells both Rac1 mutants carrying an N-terminal FLAG tag (> 70% transfection efficiency). N17Rac1-expressing SH-SY5Y cells exhibited a condensed, atrophic morphology with virtually no lamellipodia or membrane ruffling irrespective of a presence of TNFα or Il-1β (Fig. 3A). N17Rac1 was predominantly localized to the cell periphery. We focused our quantitative on FLAG-immunoreactive SH-SY5Y cells As shown in Figure 3B, N17Rac1 expression suppressed lamellipodia formation and membrane ruffling in SH-SY5Y cells upon acute exposure (200 ng/ml, 15 min) to either proinflammatory cytokine (TNFα: 40±7%, n=180 and Il-1β: 19±7%, n=180) compared to non-transfected SH-SY5Y cells (TNFα: 90±7% active cells, *p<0.01, n=180 and Il-1β: 70±7% active cells, *p<0.01, n=180, respectively). In the absence of cytokines, N17Rac1-transfected SH-SY5Y cultures contained 27±7% (n=180) active cells identical to 23±7% (n=180) active cells under control condition (non-transfected, PBS). FLAG immunoreactivity was readily detectable by immunocytochemistry as well as Western blotting (1 d postinfection) and colocalized with anti-Rac1 immunoreactivity in contrast to non-transfected cells (Fig. 3A and C).
Since depletion of Rac1 activity (N17Rac1) as well as ROS scavenging or disruption of ROS formation abolished cytokine-mediated reorganization of the neuronal actin cytoskeleton, we tested whether Rac1 is implicated in the generation of ROS intermediates. As shown in Figure 4, expression of V12Rac increased lamellipodia formation and membrane ruffling accompanied by cell spreading in accordance with previous findings in many cell types. Yet, this function of Rac1 was exclusively dependent on redox intermediates. Whereas V12Rac-transfect SH-SY5Y cultures contained 93±7% active cells (*p<0.01, n=180), non-transfected SH-SY5Y cultures exhibited only 17±7% active cells (n=180). Inhibiting ROS formation in V12Rac-expressing SH-SY5Y cells with DPI (10 μM, 30 min) completely abolished lamellipodia formation and membrane ruffling (27±7% active cells, n=180). DPI-treated SH-SY5Y cells expressing V12Rac1 displayed an atrophic morphology typical for non-transfected SH-SY5Y cells yet in stark contrast to V12Rac1-expressing SH-SY5Y cells in the absence of DPI. Moreover, scavenging ROS intermediates with MnTBAP (40 μM, 1 h) in V12Rac1-expressing SH-SY5Y cells also negated lamellipodia formation and membrane ruffling (37±7% active cells, n=180) with cell morphologies characteristic for non-transfected cells. Taken together, these finding strongly suggest that Rac1 mediates the generation of ROS thus constituting a redox signaling pathway involved in the reorganization of the neuronal cytoskeleton in response to cytokines.
TNFα and Il-1β among other physiological stimuli, provoke the generation of ROS in several non-neuronal cell types others than phagocytes including fibroblast, endothelial cells, and hepatocytes (Sundaresan et al., 1996; Goossens et al., 1999; Chandel et al., 2001; Meier, 2001). We analyzed ROS formation in serum-starved SH-SY5Y human neuroblastoma cells utilizing the fluorescence indicators 2′,7′-dihydrodichlorofluorescein (H2DCF) and dihydroethidum (DHE), which both increase in fluorescence upon oxidation. Whereas H2DCF is oxidized to dichlorofluorescein (DCF) by H2O2, DHE is oxidized to ethidium (Eth) exclusively by superoxide. We found that TNFα exposure (100 ng/ml, 30 min) stimulated a sustained ROS formation in SH-SY5Y cells sensitive to both ROS scavenging and inhibition of NOX activity (Fig. 5A and B). Quantitative analysis demonstrated that TNFα caused a significant increase in relative DCF-fluorescence intensity (1.34±0.17, *p<0.05, n=5) in SH-SY5Y cells 7.5 minutes after addition that persisted for up to 60 min compared to control (0.95±0.08, n=3) indicative of ROS formation (Fig. 5A). TNFα-induced ROS formation was negated by a presence of 5 mM N-acetyl-L-cysteine (NAC; 0.86±0.04, n=3), or 5000 U/ml exogenous catalase (CAT; 1.14±0.10, n=3) but by exogenous superoxide dismutase (200 U/ml SOD; 1.20±0.08, n=3) (Fig. 5A). As expected, a presence of 10 μM DPI (0.97±0.06, n=4) blocked TNFα-mediated ROS formation (Fig. 5B). To further implicate a NOX activity in the cytokine-mediated formation of ROS, we targeted the lipid messenger arachidonic acid (AA), a potent stimulator of NOX activity. AA is generate by cytosolic phospholipase A2 (cPLA2), which is inhibited by arachidonyl trifluoromethylketone (ATK). Pretreatment of SH-SY5Y cells with 10 μM ATK completely prevented ROS formation (0.92±0.03, n=3). To verify the production of superoxide, the primary type of ROS generated by NOX activities, we included ethidium in our study, a superoxide-specific fluorescence indicator. TNFα (100 ng/ml, 15 min) also increased relative Eth-fluorescence intensity (1.53±0.09, *p<0.01, n=3) in SH-SY5Y cells indicative of superoxide production compared to control (1.00±0.1, n=5), which was abolished by a presence of 5 mM NAC (1.06±0.09, n=3), 10 μM DPI (1.05±0.08, n=3), or 10 μM ATK (0.90±0.06, n=3), respectively (Fig. 5C). Finally in SH-SY5Y cells loaded with both oxidation sensitive fluorescence indicators, TNFα primarily caused an increase in relative Eth-fluorescence intensity (1.47±0.07, *p<0.01, n=3) but not in relative DCF-fluorescence intensity (1.06±0.21, n=3) compared to their respective controls (Eth: 1.00±0.13, n=6 and DCF: 1.00±0.09, n=6) (Fig. 5D). As expected, TNFα-induced increases in relative Eth-fluorescence intensity (double loaded cells) were negated with 5 mM NAC (Eth: 1.00±0.08, n=3 and DCF: 0.63±0.09, n=3), 10 μM DPI (Eth: 0.92±0.03, n=3 and DCF: 0.95±0.27, n=3), or 10 μM ATK (Eth: 1.03±0.20, n=3 and DCF: 0.95±0.07, n=3). Taken together, these findings demonstrated that TNFα stimulated a sustained formation of ROS in SH-SY5Y human neuroblastoma cells. The sensitivity of ROS formation to DPI (NADPH oxidoreductase inhibitor) and ATK (cPLA2 inhibitor) together with the preferential oxidation of DHE even in the presence of H2DCF strongly implied a superoxide-generating mechanism; presumably a neuronal NOX activity
The classic NADPH oxidase in phagocytes is a multi-subunit complex composed of two membrane-bound subunits (gp91phox or NOX2, and p22phox), and at least three cytosolic subunits (p67phox, p47phox, and p40phox). Only recently, NOX2 homologues were identified in many non-phagocyte cell types including SH-SY5Y cells (Nikolova et al., 2005; Bedard and Krause, 2007). Using human subunit-specific antibodies, we found immunoreactivity against all subunits of the NOX2 complex in whole cell lysates of human SH-SY5Y neuroblastoma cells with apparent molecular weights of 65 kDa for NOX2, 20 kDa for p22phox, 65 kDa for p67phox, 45 kDa for p47phox, and 40 kDa for p40phox (Fig 6A). Double bands were consistently detected for both p67phox and p47phox. In phagocytes, phorbol 12-myristate 10-acetate (PMA) and arachidonic acid (AA) potently stimulate NOX2 activity reflected by a translocation of cytosolic subunits such as p67phox to the plasma membrane. In analogy, treatment of SH-SY5Y cells with 200 nM PMA and 50 μM AA caused a time-dependent translocation of the cytosolic subunit p67phox to the plasma membrane (Fig. 6B). More importantly, TNFα (100 ng/ml, 15 min) also induced a significant translocation of p67phox to the plasma membrane (1.48 ± 0.09, *p<0.05, n=3) in SH-SY5Y cells compared to control (1.01 ± 0.02, n=3), which was completely negated by the cPLA2 inhibitor ATK (10 μM; 1.13 ± 0.03, n=3) (Fig. 6C). Similarly, phosphorylation of the cytosolic subunit p40phox (1.65 ± 0.21, *p<0.05, n=3) significantly increased compared to control (1.02 ± 0.08, n=3) upon TNFα stimulation (100 ng/ml, 15 min) (Fig. 6D). Taken together, these findings provided evidence that human SH-SY5Y neuroblastoma cells and E7 cortical neurons both express a functional NOX2 complex.
Actin both in its monomeric and filamentous state is highly susceptible to oxidative damage, which ultimately compromises proper dynamics of the actin cytoskeleton (Dalle-Donne et al., 2001; Beal, 2002). Oxidative damage of actin adversely affects many pivotal cellular processes including cellular plasticity, gene expression, mitochondrial viability, vesicle trafficking, and ion channel function (Kheradmand et al., 1998; Kourie, 1998; Moldovan et al., 2000; Lassing et al., 2007). In fact, actin oxidation is prevalent in neurodegeneration associated with many chronic CNS disorders and acute CNS trauma (Aksenov et al., 2001; Leski et al., 2001; Beal, 2002). We examined whether TNFα might cause oxidative damage to the neuronal actin cytoskeleton using carbonylation as a marker of oxidative stress (Dalle-Donne et al., 2003b). Exposure of purified rabbit actin to peroxide in vitro caused substantial carbonylation as revealed by Dinitrophenyl (DNP) hydrazone modification and detection of DNP-immunoreactivity by westernblotting (Fig. 7A). Interestingly, we detected considerable actin carbonylation in SH-SY5Y cells acutely exposed to TNFα or Il-1β (100 ng/ml, 1 h) analyzing whole cell lysates by 2D gel electrophoresis and westernblotting compared to our negative control (buffer) or positive control (100 μM H2O2) (Fig. 7B). It is noteworthy that a residual actin oxidation was detectable in cell extracts from control cultures. To quantify the extent of carbonylation, total cellular actin was immunoprecipitated from SH-SY5Y cell extracts, subjected to DNPH modification, and then analyzed by SDS gel electrophoresis followed by westernblotting and detection of anti-DNP immunoreactivity (chemiluminescence). Exposure of serum-starved SH-SY5Y cells to TNFα (100 ng/ml, 60 min) resulted in a dramatic increase of relative actin carbonylation (1.24 ± 0.02, *p<0.05, n=3) compared to control (0.99 ± 0.02, n=3), which was completely negated to control levels after pretreatment of SH-SY5Y cells with 10 μM DPI (1.09 ± 0.01, n=3) or 10 μM ATK (0.98 ± 0.01, n=3) (Fig. 7C). Neither pharmacological treatment altered actin carbonylation in the absence of TNFα and residual carbonylation was indistinguishable from control conditions (data not shown). These findings provide evidence that prolonged TNFα exposure inflicts irreversible oxidative damage (carbonylation) to the actin cytoskeleton by stimulating a NOX2-like activity in neuronal cells.
The proinflammatory cytokines TNFα and Il-1β exert pleiotropic effects in the developing, adult, and injured CNS yet persistent, high levels of expression are neurotoxic due to oxidative stress and accumulation of toxic lipid second messengers (Mrak and Griffin, 2005; Lucas et al., 2006). Our findings revealed that TNFα and Il-1β provoked three principal responses in neuronal cells: (1) a transient, redox-dependent actin filament reorganization into lamellipodia and membrane ruffling under the regulation of Rac1, (2) a sustained ROS production through a neuronal NOX2 activity, and (3) irreversible, oxidative damage (carbonylation) to the neuronal actin cytoskeleton accompanied by loss of lamellipodia and membrane ruffling. Indeed, actin carbonylation is prevalent after acute CNS injuries, in many chronic CNS disorders, as well as in normal CNS aging (Beal, 2002; Floyd and Hensley, 2002; Dalle-Donne et al., 2003a). Conceivably, neurotoxicity of inflammation could arise in part from actin oxidation since disruption of actin filament dynamics would disable many vital cellular processes including plasticity, mitochondrial function, vesicle traffic, gene expression, and ion channel function (Lin et al., 1997; Kourie, 1998; Luo, 2002; Gourlay and Ayscough, 2005; Percipalle and Visa, 2006).
SH-SY5Y neuroblastoma cells grown on collagen displayed, after serum starvation, an atrophic, condensed morphology with virtually no lamellipodia, membrane ruffling, or pseudopodia. Upon acute exposure to cytokines, SH-SY5Y cells displayed a dramatic increase in cellular plasticity reflected by extensive lamellipodia formation and membrane ruffling followed by substantial cell spreading. However after 30 min, SH-SY5Y cells reverted back to their original morphology with mostly peripheral actin stress fibers. Interestingly, SH-SY5Y cells were no longer responsive to cytokine exposure even after a prolonged recovery phase (data not shown) suggesting a long lasting, irreversible effect on the actin cytoskeleton. Reorganization of actin filaments was demonstrated in several non-neuronal cell lines exposed to TNFα or Il-1β and consisted predominantly of lamellipodia formation followed by stress fiber formation (Wojciak-Stothard et al., 1998; Peppelenbosch et al., 1999). In regard to neuronal cells, TNFα reduces neurite branching, inhibits neurite outgrowth, and causes even neurite retraction in cultures of hippocampal neurons, reflecting a disruption of proper actin cytoskeleton reorganization (Neumann et al., 2002; Munch et al., 2003). It is well established that Rho GTPases (Cdc42, Rac1, and RhoA) are implicated in converting extrinsic stimuli into a coordinated reorganization of actin filament often in a hierarchical fashion (Nobes and Hall, 1995; Hall and Nobes, 2000b; Luo, 2000). Not surprisingly, depletion of Rac1 activity in SH-SY5Y cells by expressing N17Rac1 abolished lamellipodia formation and membrane ruffling in response to cytokines and SH-SY5Y cells retained their typical atrophic morphology in accordance with previous findings in non-neuronal cells. N17Rac1-expressing SH-SY5Y cells exhibited a more condensed morphology compared to control cells (qualitative observation). In fact, Rho GTPases do mediate cytoskeletal rearrangement induced by TNFα or Il-1β in various cellular systems (Peppelenbosch et al., 1999; Puls et al., 1999). In several non-neuronal cells, TNFα and Il-1β but also growth factors were shown to induce a Rac1-dependent ROS formation (Sulciner et al., 1996; Sundaresan et al., 1996; Bonizzi et al., 1999; Deshpande et al., 2000). And Rac1 and RhoA seem to have opposing roles in ROS generation in endothelial cells (Wojciak-Stothard et al., 2005). A connection was revealed among Rac1 activation, ROS formation, and TNF receptor associated factor (TRAF) recruitment (Chandel et al., 2001; Li et al., 2006). Based on reports by Moldovan et al. (1999), we demonstrated that (i) redox intermediates were required for cytokine-stimulated reorganization of actin filaments in SH-SY5Y cells, and (ii) Rac1 regulated the formation of redox intermediates. Sensory neurons responded to a NGF-withdrawal with a Rac1-dependent ROS formation (Suzukawa et al., 2000). We quantified cytokine-dependent ROS formation using the peroxide-sensitive fluorescence indicator H2DCF in conjunction with the superoxide-sensitive fluorescence indicator DHE (Tarpey and Fridovich, 2001). Acute exposure of SH-SY5Y cells to TNFα caused oxidation of both indicators as revealed by increases in DCF or Eth-fluorescence. The fact that DHE was preferentially oxidized in response to cytokines even in the presence of H2DCF suggested a superoxide-generating mechanism (Rothe and Valet, 1990). Several findings further support this conclusion. The 1e− transport inhibitor DPI, generally used as an NOX inhibitor, and the cPLA2 inhibitor ATK both abolished TNFα-mediated ROS formation. Addition of exogenous SOD did not quench maximum DCF fluorescence since dismutation of superoxide to peroxide is in favor of DCF oxidation. However, addition of exogenous catalase inhibited increases in DCF fluorescence suggesting that superoxide dismutation is naturally occurring. In addition, we demonstrated the presence of a functional NOX activity in SH-SY5Y cells based on immunological and biochemical evidence. Using human-specific antibodies, we found immunoreactivity against all subunits of NOX2 in human SH-SY5Y neuroblastoma cells in agreement with previous reports (Nikolova et al., 2005). We consistently detected p67phox immunoreactivity as double bands in SH-SY5Y cells whereas the same antibody revealed a single band when testing RAW 264 cell extracts (data not shown). The NOX2 activity in our assay system was both stimulus-dependent and functional as revealed both on the level of enzyme activity and protein subunit assembly. First, NOX disruption using DPI and ATK negated TNFα-stimulated ROS formation. Second, cytokine-stimulated ROS production was strictly Rac1-dependent but, in the absence of cytokines, enhanced by expressing constitutively active Rac1. Importantly, Rac1 is a key regulator of NOX isoforms and can substitute for Rac2 in phagocytes (Hordijk, 2006). Third, SH-SY5Y cells exposed to PMA and AA, two well-documented activators of NOX enzymes, induced the translocation of the cytosolic subunit p67phox to plasma membranes (Groemping and Rittinger, 2005). Fourth, a presence of ATK abolished plasma membrane translocation of p67phox induced by TNFα. Fifth, TNFα stimulated the phosphorylation of p40phox. These findings provide strong evidence for the presence of a neuronal NADPH oxidase as a pivotal source of oxidative stress in neurodegeneration (Lambeth, 2007). Several recent studies describe the expression of members of the multi subunit enzyme NADPH oxidase (NOX) in non-neuronal as well as neuronal cells (Lambeth, 2002; Lambeth, 2004; Bedard and Krause, 2007). The NOX2 isoform was found primary cortical neurons, hippocampal neurons, and sensory neurons (Noh and Koh, 2000; Tammariello et al., 2000; Tejada-Simon et al., 2005). We demonstrated that actin represents one target of oxidative damage resulting from a cytokine-stimulated ROS formation in neuronal cells. Both cytokines increased carbonyl residues in total actin of SH-SY5Y neuroblastoma cells, which was completely abolished by inhibiting NOX activity (DPI or ATK). Interestingly, we consistently detected residual oxidation in actin under control conditions even if performed in a strongly reducing environment. This finding might suggest that oxidative modification of actin could represent a physiological regulatory mechanism. It is noteworthy that ROS formation, actin carbonylation, and irreversible arrest of actin filament reorganization strongly coincided. SH-SY5Y cells displayed a redox-dependent formation of lamellipodia, highly motile subcellular structures, within 15 min upon cytokine addition. Irreversible arrest of actin filament reorganization occurred only after an exposure time of 30 min or longer. TNFα elicited ROS formation within 10 min after addition, which persisted for up to 60 min. Translocation of p69phox to plasma membranes, indicating NOX activation and thus ROS production, was found after 10 min following stimulation with PMA or TNFα. And lastly, actin carbonylation was readily detectable after incubation of SH-SY5Y cells with cytokines for over 30 min. Taken together, early increase in ROS produced by NOX2 in response to cytokines paralleled lamellipodia formation (dynamic actin filament structures) whereas sustained ROS generation (> 30 min) was accompanied by actin oxidation and arrest of motility. Presumably, limited exposure to cytokines could produce beneficial effects on neuronal motility and survival implying initially a physiological role for ROS intermediates whereas a prolonged presence of cytokines disrupts cellular plasticity via actin oxidation suggesting a pathological role for persistent high levels of ROS intermediates. Consequently, neurotoxicity of TNFα and Il-1β could be attributed, at least in part, to the oxidation of actin since (1) both cytokines elicit an irreversible reorganization of actin filaments, (2) both cytokines stimulate ROS formation, (3) actin exhibits a strong vulnerability to oxidative damage both in vivo and in vitro, (4) actin filament dynamics is essential role for many processes vital to cellular homeostasis, and (5) a presence of oxidized actin in CNS trauma and many chronic CNS disorders. A persistent inflammatory reaction is highly prevalent in most CNS pathologies whether acute, chronic, psychiatric, or even in normal aging with activated microglia and astrocytes releasing numerous neurotoxic mediators including cytokines and ROS (Mrak and Griffin, 2005; Lucas et al., 2006). Our findings provided evidence that the proinflammatory cytokines TNFα and Il-1β exert their neurotoxicity at least partially through a NOX-dependent, irreversible oxidative modification of actin.
Recombinant human tumor necrosis factor alpha (TNFα), interleukin-1β (Il-1β), a polyclonal rabbit anti-human p67phox antibody, a monoclonal mouse anti-human p40phox antibody, and a monoclonal mouse anti-FLAG antibody were purchased from Millipore (Temecula, CA). All other antibodies were obtained from Santa Cruz (Santa Cruz, CA). DMEM and Penicillin/Streptomycin were obtained from Mediatech (Herndon, VA). Glutamax, Hank’s Balanced Salt Solution (HBSS), trypsin/EDTA solution, and a polyclonal rabbit anti-DNP (dinitrophenyl) primary antibody were from Invitrogen (Carlsbad, CA). Fetal bovine serum (FBS) was received from Atlanta Biologicals (Atlanta, GA). Streptavidin-agarose beads, protease inhibitor cocktail (PIC), Pico Super Signal chemoluminescent kit, and a BCA protein assay kit were obtained from Pierce (Rockland, IL). Diphenylene iodonium chloride (DPI) and Manganese(III)tetrakis (4-benzoic acid) porphyrin chloride (MnTBAP) were obtained from EMD (Gibbstown, NJ) and stock solutions (200x) of DPI or MnTBAP were prepared in DMSO or 10 mM NaOH, respectively, sterile filtered, and stored at −20°C protected from light. Rhodamine phalloidin was from Cytoskeleton Inc. (Denver, CO). Rattail collagen was from Roche Diagnostics (Indianapolis, IN). All other reagents were from Sigma (St. Louis, MO).
SH-SY5Y human neuroblastoma cells were grown in 100 mm tissue culture dishes (Falcon) in medium composed of DMEM, 10% Fetal Bovine Serum (FBS), 1% Glutamax, 100 U/ml Penicillin and 100 U/ml Streptomycin (humidified atmosphere, 5% CO2, 37°C). For amplification, 80% confluent SH-SY5Y cultures were washed with HBSS incubated with 0.5 mg/ml trypsin/0.2mg/ml EDTA (5 min 37°C). Cells were rinsed off, collected by centrifugation (200×gmax), and plated at a dilution of 1:3 (100 mm dishes), 105 cells/well (96 well plates), or 2×105 per glass cover slip (confocal microscopy). Glass cover slips (22×22 mm2, #1) were mounted over a 1.5 cm diameter hole drilled into the bottom of 35 mm culture dishes. Glass cover slips were coated with 5 μg/cm2 collagen. As our acute exposure paradigm, SH-SY5Y cells were serum-starved overnight in DMEM, 1% Glutamax, 100 U/ml Penicillin and 100 U/ml Streptomycin (humidified atmosphere, 5% CO2, 37°C). Cytokines, pharmacological agents, and controls (PBS) were added to serum-free culture medium as volumes accounting for 2% or less of the total culture medium.
Constitutively active Rac1 (V12Rac1) or dominant negative Rac1 (N17Rac1) containing a N-terminal FLAG tag were cloned into the pShuttleCMV expression plasmid (Qbiogene, Carlsbad, CA) under a CMV promoter. SH-SY5Y cells grown on collagen-coated glass cover slips were transfected using Lipofectamine PLUS (Invitrogen, Carlsbad, CA). Briefly, plasmid DNA was digested with PmeI, treated with Calf Intestinal Phosphatase (Promega, Madison, WI), and subjected to 0.8% agarose electrophoresis. The linearized plasmid DNA was recovered and purified on glass milk (QBiogene, Carlsbad, CA). Purified, linearized plasmid DNA (0.5 μg) was resuspended in 20 μl of Opti-MEM serum-free media, mixed with 5 μl of PLUS reagent, and incubated at RT for 15 min. The transfection solution was combined with Lipofectamine (5 μl) in Opti-MEM (20 μl), incubated for 15 min at RT, and then added to SH-SY5Y cultures in 200 μl Opti-MEM. Transfection efficiencies were routinely greater than 70%. Following a 24 h expression period, cultures were treated with proinflammatory cytokines in the presence or absence of pharmacological inhibitors.
Human SH-SY5Y neuroblastoma cells were grown on collagen-coated glass cover slips (0.13 mm thick German glass) in serum-free medium for 24 h. Prior to acute addition of TNFα or Il-1β, serum-starved SH-SY5Y cultures were incubated with pharmacological inhibitors or transfected Rac1 mutants prior to acute exposure to 200 ng/ml TNFα or Il-1β for 15 min or 30 min unless indicated otherwise. SH-SY5Y cultures were fixed (RT, 30 min) with 4% paraformaldehyde in 10 mM MES, pH 6.1, 138 mM KCl, 3 mM MgCl2, 2 mM EGTA, 0.5% Triton X-100, a cytoskeleton stabilizing buffer (Symons and Mitchison, 1991). After rinsing with 0.1% Triton X-100 in Tris-buffered saline (TX-TBS), cultures were incubated with rhodamine phalloidin in TX-TBS-BSA (1:10 dilution, 2% BSA, 20 min, RT). To reveal FLAG expression, SH-SY5Y cultures were incubated with a monoclonal mouse-anti-FLAG-antibody in TBS-TX-BSA (1:100, 1 h, RT) followed by incubation with a secondary FITC-conjugated goat-anti-mouse antibody (1:1000, 1 h, RT, light protected). Finally, cultures were rinsed with TX-TBS, and transferred into 60% glycerol/40% PBS. Images were acquired (40x, oil, Plan Fluor) with a Zeiss confocal microscope LSM 510 equipped with a He/Ne laser and an Argon laser using filter combinations for fluorescein and rhodamine fluorescence. Zeiss LSM Software was used for image acquisition and analysis. For each treatment condition, random fields of view were analyzed and 60 cells each from three independent sets of experiments (n=180) were scored for the presence of lamellipodia formation and/or membrane ruffling. As our criterion, SH-SY5Y cells exhibiting at least two distinct regions with extensive lamellipodia formation or membrane ruffling were considered active cells whereas pseupodia-like structures were not accounted for. Statistical significance between the means of control and treatments were calculated using Dunnett’s t-test. A Kruskal-Wallis test was employed to calculate significance between control distributions and treatment distributions.
Increases in reactive oxygen species (ROS) were detected with the oxidation-sensitive fluorescence indicators 2′,7′-dihydrodichlorofluorescein diacetate (H2DCFDA) or Dihydroethidium (DHE). H2DCFDA is retained in the cytosol after deacetylation to dihydrodichlorofluorescein and increases in fluorescence upon oxidation by H2O2 to dichlorofluorescein (DCF). In contrast, DHE is oxidized to ethidium (Eth) preferentially by superoxide. SH-SY5Y cells (105 cells/well, 96-well plates, Falcon) were incubated for 1 h either with 50 μM H2DCFDA, 50 μM DHE, or a combination of both (160 μM DHE and 10 μM H2DCFDA) in the presence or absence of pharmacological inhibitors or ROS scavenging enzymes. Following a medium exchange, SH-SY5Y cells were exposed to 100 ng/ml TNFα, washed with PBS, and lysed (2 M Tris-Cl pH 8.0, 2% SDS, 10 mM Na3VO4). Maximum Eth or DCF-fluorescence intensity was quantified (100 μl cell lysate) using a Beckman Coulter Multimode DTX 880 microplate reader (Eth: 485 nm excitation filter, 610 nm emission filter and DCF: 495 nm excitation filter, 525 emission filter). All Eth or DCF-fluorescence intensity data were normalized to the average maximum Eth or DCF-fluorescence under control conditions (relative fluorescence values).
SH-SY5Y cells were treated with pharmacological inhibitors and then exposed either to a mixture of 200 nM PMA/50 μM arachidonic acid (AA) or 100 ng/ml TNFα. Plasma membrane fractions were enriched either by density centrifugation or biotinylation and streptavidin-affinity chromatography as described (Li and Shah, 2002). Regarding centrifugation, cell lysates were obtained by scraping SH-SY5Y cells into Buffer A (250 mM sucrose, 20 mM HEPES pH 7.4, 2 mM EDTA, 5 mM MgCl2, 1 mM dithiothreitol, 1 mM AEBSF, and 1% PIC), lysis by sonication, and collection of supernatants (1,200 × gmax, 5 min). Centrifugation (100,000 × gmax, 30 min, 4°C) of lysates separated membrane fractions (i.e. pellets) and cytosolic fractions (i.e. supernatants). Pellets were a resuspended in Buffer B (Buffer A containing 1% Triton X-100 and 0.01% saponin) and incubated on ice for 15 min. Following centrifugation (100,000 × gmax, 30 min, 4°C), membrane fractions were collected as supernatants. Regarding biotinylation, SH-SY5Y cultures (5×106 cells/well, 6-well plates) were washed with HBSS-CM (HBSS, 0.1 g/l CaCl2, 0.1 g/l MgCl2, pH 7.5), and incubated with 0.5 mg/ml Sulfo-NHS-Biotin (prepared in 20 mM HEPES, HBSS-CM, pH 8.0) for 40 min on ice. Excess Sulfo-NHS-Biotin was neutralized (on ice, 15 min) with 50 mM glycine (prepared in HBSS-CM), and cells were scraped into HBSS-CM. After centrifugation (200 × gmax, 2 min), cells were resuspended in 500 μL Buffer C (20 mM HEPES pH 7.5, HBSS, 1% Triton X-100, 0.2 mg/ml saponin, 1% PIC), sonicated, and cell lysates incubated with streptavidin-agarose beads (25 μl, 2 h, 4°C). Beads collected (2,500 × gmax, 2 min), and resuspended in 150 μL of Buffer C. After heating beads (5 min, boiling water bath), membrane proteins (2,500 × gmax, 2 min) were collected in the supernatant and total protein content was determined using a BCA protein assay.
Protein samples were prepared either using purified rabbit muscle actin, total protein of SH-SY5Y cell extracts, or actin immunoprecipitated from SH-SY5Y cells. SH-SY5Y cultures were exposed to 100 ng/ml TNFα (1 h), 100 ng/ml Il-1β (1 h), 100 μM H2O2 (30 min) or buffer (1 h) in serum-free culture medium following a preincubation with 10 μM DPI or 10 μM ATK (1 h), and then washed with Tris/sucrose (10 mM Tris-Cl pH 7.0, 250 mM sucrose). For 1D SDS gel electrophoresis, SH-SY5Y cultures were extracted with 2 M Tris-Cl pH 8.0, 2% SDS, 1 mM Na3VO4 (SDS) whereas for 2D gel electrophoresis, cultures were extracted with 8 M urea, 2% CHAPS, 80 μM DTT, 0.5 % Ampholytes pH 3–10, 5% glycerol, 0.001% bromophenol blue. Cell extracts were centrifuged (3,000 × gmax, 4°C, 20 min) and soluble protein subjected to carbonyl derivatization. For immunoprecipiation of actin, cell lysates (100 μl) were incubated with 1 μg normal mouse serum (45 min, 25°C, agitation), and normal IgG removed with protein A/G agarose beads (20 μl). Cleared lysates (100 μl) were mixed with 1 μg monoclonal mouse anti-human β-actin primary antibody (2 h, 4°C, agitation) followed by overnight incubation with protein A/G agarose beads (20 μl, 4°C, agitation). Protein A/G beads were collected by centrifugation (200 × gmax, 2 min), washed three times in lysis buffer, and bound proteins were released by boiling for 5 min. Carbonyl residues in protein samples were derivatized with 2,4-dinitrophenylhydrazine (DNPH) as described (Levine et al., 1994). Briefly, protein samples were mixed with an equal volume of 12% SDS, and further incubated with two sample volumes of 20 mM DNPH in 2M HCl (30 minutes, 25°C). Protein was recovered by chloroform/methanol precipitation and washed with ethylacetate/ethanol (1:1) followed by acetone (Wessel and Flugge, 1984). Protein samples were resuspended in 1D or 2D gel sample buffer and protein concentration was determined using a BCA assay.
For SDS-polyacrylamide gel electrophoresis, equal amounts of total protein (30–50 μg total cellular protein, or 5 μg actin) were separated (125 volts, 50 watts, 75 mA). For 2D gel electrophoresis, 100 μg total protein per sample was loaded onto rehydrated IPG strips (Invitrogen, Calsbad, CA) and isoelectric focusing performed at RT (Castegna et al., 2002). After equilibrating IPG strips with 1D sample buffer, strips were placed on NUPAGE 4–12% gels and separated (125 V, 30 mM) using a MES running buffer (50 mM MES, 50 mM Tris base, 0.1% SDS, 1 mM EDTA, pH 7.3). For western blotting, proteins were transferred onto PVDF membranes (2.5 hours, 50 volts, 50 watts, 250 mA) and membranes were blocked with TBST-BSA (50 mM Tris-Cl pH 7.4, 150 mM NaCl, 0.1% Tween-20, 5 mg/ml BSA) (Towbin et al., 1979). After overnight incubation with the respective primary antibody (1 μg/μl in TBST), membranes were washed (TBST) and incubated with the corresponding secondary antibody (0.2 μg/μl in TBST, 45 min). Immunoreactivity was detected by chemiluminescence and quantified using a GE Healthcare Typhoon Imager running ImageQuant software (United Kingdom) or by colorimetric detection.
Analysis of variance (ANOVA) was used to determine statistically significant differences between treatments (p<0.05). At least three independent experiments were performed for each condition. Post hoc comparisons of specific treatments were performed using Scheffe’s test to determine statistical significance based on the calculated ANOVA data. A Dunette’s test was used for multiple comparisons between treatments and a single control. A Kruskal-Wallis test was employed to calculate significance between control distributions and treatment distributions (confocal analysis). Error bars represent standard deviations of the mean.
We are grateful for the experimental support received from Dr. M. Wright and R. Parsley regarding the biochemical analysis of the membrane translocation of p67phox. We would like to extend our thanks to Dr. Schulte, Dr. Taylor, Dr. O’Brien and Dan Kirschner for their review of the manuscript and many constructive discussions of the research. This study was supported by National Institutes of Health U54 grant NS41069 (SNRP: NINDS, NIMH, NCRR, NCMHD), United States Department of Agriculture grant 2005-34495-16519 (TBK), and Christopher Reeve Paralysis Foundation grant KA1-0004-2 (TBK).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.