Microtubule dysfunction precedes transport impairment and mitochondria damage in MPP+-induced neurodegeneration


Address correspondence and reprint requests to G. Cappelletti, Dipartimento di Biologia, Università degli Studi di Milano, Via Celoria 26, 20133 Milan, Italy. E-mail: graziella.cappelletti@unimi.it


J. Neurochem. (2010) 115, 247–258.


Dysfunction of the microtubule (MT) system is an emerging theme in the pathogenesis of Parkinson’s disease. This study was designed to investigate the putative role of MT dysfunction in dopaminergic neuron death induced by the neurotoxin 1-methyl-4-phenylpiridinium (MPP+). In nerve growth factor-differentiated PC12 cells, we have analyzed post-translational modifications of tubulin known to be associated with differently dynamic MTs and show that MPP+ causes a selective loss of dynamic MTs and a concomitant enrichment of stable MTs. Through a direct live cell imaging approach, we show a significant reduction of MT dynamics following exposure to MPP+ and a reorientation of MTs. Furthermore, these alterations precede the impairment of intracellular transport as revealed by changes in mitochondria movements along neurites and their accumulation into varicosities. We have also analyzed activation of caspase 3 and mitochondrial injury, well-known alterations induced by MPP+, and found that they are noticeable only when MT dysfunction is already established. These data provide the first evidence that axonal transport impairment and mitochondrial damage might be a consequence of MT dysfunction in MPP+-induced neurodegeneration, lending support to the concept that alterations of MT organization and dynamics could play a pivotal role in neuronal death in Parkinson’s disease.

Abbreviations used:

end-binding protein 3-green fluorescent protein


fast axonal transport


fluorescence recovery after photobleaching


horseradish peroxidase






phosphate-buffered saline


Parkinson’s disease


protein kinase C


region of interest


Reactive Oxygen Species


yellow fluorescent protein

Microtubules (MTs) are highly dynamic polymers that control many aspects of neuronal function: they provide a scaffold to sustain axonal and dendritic architecture and supply the railway for axonal transport. MTs are built up from tubulin subunits and switch rapidly between phases of growth, pause, and shrinkage, a behavior known as dynamic instability (Desai and Mitchison 1997). In developing neurons, MTs are highly dynamic during process outgrowth, whereas in mature neurons their stability increases to ensure and maintain cell shape and viability. However, proper MT dynamics remains fundamental for the survival of mature neuron since there is only a limited range of acceptable MT dynamic behaviors in neurons, outside of which MTs cannot function normally and the cells cannot survive (Feinstein and Wilson 2005). The peculiar architecture of neurons makes them particularly dependent on intracellular transport process, and the main mechanism to deliver cellular components is long-range MT-based transport. Axonal swelling and spheroids have been described in a number of neurodegenerative diseases, and damage to axonal transport may underlie the pathogenic accumulation of proteins and organelles (De Vos et al. 2008). Thus, alteration of axonal transport could be the link between MT dysfunction and neurodegeneration.

In Parkinson’s disease (PD), the most common motor neurodegenerative disorder affecting 1% of people over 60s, a growing body of evidence suggests a link between MT dysfunction and dopaminergic cell loss in the substantia nigra pars compacta. Tubulin is present in Lewy Bodies (Galloway et al. 1992), the typical cytoplasmic inclusions found in the substantia nigra of patients with PD. Recent data support a pathogenic role for tubulin in dopaminergic neuronal cell death: (i) tubulin interacts with some of the mutated proteins in PD, including α-synuclein (Alim et al. 2002), parkin (Yang et al. 2005), and leucine-rich repeat kinase 2 (Gillardon 2009); (ii) microtubular cytoskeleton is affected by parkinsonian toxins, such as 1-methyl-4-phenylpiridinium (MPP+) (Cappelletti et al. 2005) and rotenone (Ren et al. 2005); (iii) finally, dopaminergic neurons are selectively vulnerable to anti-MT drugs (Feng 2006) and intrastriatal injection of colchicine, a well-known MT disruptor, has been shown to induce striatonigral degeneration in mice (Liang et al. 2008). During the last years, axonal transport has been extensively investigated in neurodegenerative processes, but just few papers highlight its involvement in PD. Direct analyses of movement of mutant forms of α-synuclein, either associate with PD or mimicking define serine phosphorylation states, exhibit reduced axonal transport in respect to wild-type synuclein, following transfection into cultured neurons (Saha et al. 2003). Moreover, in isolated squid axoplasm, MPP+ produces significant alterations in fast axonal transport (FAT), increasing cytoplasmic dynein-dependent retrograde FAT and decreasing kinesin-1-mediated anterograde FAT (Morfini et al. 2007). Finally, in a rat model of synucleinopathy, anterograde transport motor proteins are decreased in the striatum, whereas retrograde motor proteins are increased, combined with dramatic changes in cytoskeletal protein levels (Chung et al. 2009).

This study was designed to investigate the putative role of MT dysfunction in MPP+-induced axonal transport defects and dopaminergic neuronal death. We used MPTP, a neurotoxin that kills dopaminergic neurons and induces PD-like symptoms, and is widely used as a tool for studies on sporadic PD (Dauer and Przedborski 2003). We have previously shown that MPP+, the toxic metabolite of MPTP, influences the state of tubulin polymerization in PC12 cells (Cappelletti et al. 1999), and acts directly on purified MTs in vitro by affecting their assembly (Cappelletti et al. 2001) and dynamics (Cappelletti et al. 2005). Here, we have investigated MT dynamics in PC12 cells following MPP+ treatment by studying post-translational modifications of tubulin that are correlated with MT stability and by live cell imaging. To correlate MT, axonal transport, and mitochondrial activity, we have evaluated mitochondrial movement along neurites, we tested the possible protective effect of caspase 3 and protein kinase C (PKC) inhibitors on axonal transport impairment, and we have also investigated the functional state of mitochondria.

Materials and methods

Cell culture and transfection

PC12 cells were maintained in RPMI 1640 (HyClone, Logan, UT, USA) containing 10% horse serum and 5% fetal bovine serum (HyClone) supplemented with 2 mM l-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin, at 37°C in a humidified atmosphere, 5% CO2. Cells were plated at 1.5 × 104/cm2 onto poly-l-lysine (100 μg/mL in double distilled water)-coated cover glass for immunofluorescence or on coated petri dishes for protein biochemistry, and exposed to 50 ng/mL human β-nerve growth factor (PeproTeck, London, UK) in low serum medium (RPMI 1640 supplemented with 1% horse serum, 2 mM l-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin). For transfection experiments, PC12 cells were plated at 1.5 × 104/cm2 and transiently transfected, 24 h later, using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) (1 : 3 DNA to Lipofectamine ratio, 12.5 μg of DNA per 60 mm dish) with the enhanced yellow fluorescent protein-alpha tubulin (EYFP-Tub) plasmid (Clonetech, Mountain View, CA, USA), the end-binding protein 3-green fluorescent protein (EB3-GFP) construct (supplied by Dr Galjart, Medical Genetic Center, Department of Cell Biology, Erasmus University, Rotterdam, The Netherlands), or with the N-terminal mitochondrial signal sequence of cas2 fused to DsRed plasmid (kindly gifted by Dr Rugarli, Department of Neuroscience and Medical Biotechnologies, University of Milano-Bicocca, Milan, Italy). Three hours post-transfection, cells were plated onto poly-l-lysine-coated glass bottom dishes (MatTek, Ashland, MA) at a density of 1 × 104/cm2 in low serum medium with the addition of 50 ng/mL nerve growth factor.

For MPP+ treatment, cells were incubated for 6 or 24 h with different concentrations of freshly dissolved MPP+ (RBI, Natick, MA, USA). For PKCδ or Caspase 3 inhibition, cells were incubated for 24 h with 500 nm of Gö 6976 (Calbiochem, Darmstadt, Germany) or with 10 μM of Caspase 3 Inhibitor II (Calbiochem), respectively.

For neurite length assessment, living cells were viewed by phase-contrast microscopy with 10 random images captured per plate using Axiovert microscope equipped with AxiocamHR (Zeiss, Oberkochen, Germany) at 20× magnification. The length of the longest neurite of each differentiated cell was determined using digital image processing software (Interactive measurement module, Axiovision; Zeiss). All cells in each image were analyzed, with over 200 cells assessed per tissue culture plate. The experiments were repeated at least three times.

Western blot analysis

Whole cell extracts, Triton X-100 soluble and insoluble fraction of PC12 cells were made as previously reported (Cappelletti et al. 2003). Protein samples (15 or 40 μg per lane) were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis, blotted onto Polyvinylidene Difluoride membranes (Immobilon™-P; Millipore, Billerica, MA, USA), and probed with the following antibodies: α-tubulin mouse IgG (clone B-5-1-2; Sigma-Aldrich, St Louis, MO, USA), 1 : 1000; Glu tubulin rabbit IgG (kindly provided by Dr Lafanechere, Grenoble, France), 1 : 100 000; Tyr tubulin mouse IgG (clone TUB-1A2; Sigma-Aldrich), 1 : 1000; Ac tubulin mouse IgG (clone 6-11B-1; Sigma-Aldrich), 1 : 1000; actin mouse IgM (N350; Amersham, Little Chalfont, UK), 1 : 4000; Caspase 3 rabbit IgG (Stressgen Bioreagents Corporation, Ann Arbor, MI, USA), 1 : 1000. Immunoblots were developed for 1 h at 20°C with horseradish peroxidase (HRP) donkey anti-mouse IgG (Pierce, Rockford, IL, USA), 1 : 20 000; HRP goat anti-mouse IgM (Sigma-Aldrich), 1 : 2000; or HRP goat anti-rabbit IgG (Pierce), 1 : 40 000. Quantification was performed by scanning immunoblots with JX-330 color image scanner (Sharp Electronics, Europe) into ImageMaster VDS Software (GE Healthcare Bio-Sciences Corp., Piscataway, NJ, USA). All quantitative measurements used concurrent immunoblotting to verify linearity of film response to protein concentration and were normalized on α-tubulin.

Immunofluorescence and labeling

Cells were fixed with methanol or 4% paraformaldehyde and incubated with the following antibodies or probe: α-tubulin mouse IgG (clone B-5-1-2; Sigma-Aldrich), 1 : 500; Glu tubulin rabbit IgG (kindly provided by Dr Lafanechere), 1 : 500; Tyr tubulin mouse IgG (clone TUB-1A2; Sigma-Aldrich), 1 : 100; 5 μM Tetramethyl Rhodamine Isothiocyanate-phalloidin (Sigma-Aldrich). As secondary antibodies, we used Alexa FluorTM 488 goat anti-mouse, Alexa FluorTM 594 donkey anti-mouse, and Alexa FluorTM 488 goat anti-rabbit (Invitrogen), diluited 1 : 1000. Some slides were extracted before the fixation to remove unassembled tubulin (Slaughter and Black 2003). For staining of active mitochondria, cells were incubated for 40 min with 300 nM MitoTracker Red CMX-Ros (Invitrogen) and fixed for 10 min with 2% paraformaldehyde in phosphate-buffered saline (PBS); next total mitochondria were labeled using antibody against porin rabbit IgG (Voltage-Dependent Anion Channel; Abcam, Cambride, UK), 1 : 1000. The coverslips were mounted in Mowiol® (Calbiochem, San Diego, CA, USA)–DABCO (Sigma-Aldrich) and examined with a confocal laser scan microscope imaging system (TCS SP2 AOBS; Leica Microsystems, Heidelberg, Germany) equipped with an Ar, He–Ne, and UV lasers or with the Axiovert 200 M microscope (Zeiss).

Measurement of mitochondrial membrane potential and quantification of Reactive Oxygen Species (ROS)

For the comparison of the mitochondrial membrane potential (ΔΨ), cells were incubated for 40 min with 300 nM MitoTracker Red CMX-Ros. Oxidative stress and superoxide production were detected using ROS/Reactive Nitrogen Species kit (Enzo Life Sciences Ag., Lausen, Switzerland). Briefly, cells were loaded for 2 h with the detection mix containing specific probes for oxidative stress and superoxide, and washed two times with the 1× wash buffer. MitoTracker, oxidative stress probe and superoxide probe fluorescence was registered with the Axiovert 200 M microscope (Zeiss), and fluorescence intensity was analyzed offline using digital image processing software (Interactive measurement module, Axiovision; Zeiss). At least 15 randomly distributed fields were analyzed for each experimental condition.

To quantify the fraction of functional on total mitochondria, we made a ratio between the area covered by active mitochondria loaded with Mitotracker Red, and area covered by total mitochondria stained with anti porin rabbit IgG (Voltage-Dependent Anion Channel; Abcam). At least 15 randomly distributed fields were acquired for each experimental condition with the Axiovert 200 M microscope (Zeiss), and the areas were measured offline using digital image processing software (Interactive measurement module, Axiovision; Zeiss).

Fluorescence recovery after photobleaching (FRAP) analysis

Culture medium was replaced with registration buffer (25 mM Hepes and 4.5 g/L glucose in PBS, pH 7.4) and cells were kept at 37°C using a closed chamber system equipped with the objective heater (Bioptechs, Butler, PA, USA). FRAP analyses were performed using a confocal laser scanning microscope (TCS SP2 AOBS; Leica, 63×objective). Enhanced yellow fluorescent protein was excited at 514 nm and emission fluorescence was collected every 1.4 s before (10 s) and after (5 min) the bleaching using a 63×/1.4 numerical aperture oil-immersion objective. To increase the effectiveness of the enhanced yellow fluorescent protein photobleaching, three laser lines (488, 494, and 514 nm) at their maximum power were used during the bleaching. Background was subtracted and the data were corrected for the spontaneous decrease of fluorescence in an unbleached region of the same cell, and then normalized. According to Bulinski et al. (2001), normalized fluorescence was plotted against recovery time and fitted to an exponential recovery curve: Ft = Fe − (Fe − Fb)(ekt), where Ft is the fluorescence at time t, Fe is the fluorescence at the end of the experiment, Fb is the fluorescence after the bleaching. t½ was calculated as ln(2)/k and the percentage of recovery as ((Fe − Fb)/(Fp − Fb)) × 100, where Fp is the fluorescence before bleaching.

Live cell imaging

Cultures were transferred to a live cell imaging workstation composed of an inverted microscope (Axiovert 200; Carl Zeiss), a heated (37°C) chamber (Okolab, Naple, Italy), and a Plan neofluar 63×/1.25 numerical aperture oil-immersion objective (Carl Zeiss). Culture medium was replaced with a pre-warmed registration buffer and images were collected every 5 or 15 s with a cooled camera (Axiocam HRM Rev. 2; Carl Zeiss) for MT dynamics and every 10 s for mitochondrial transport. Distance traveled by EB3-GFP comets or by Mito-DsRed dashes that could be followed for at least three consecutive frames was measured in different neuronal areas, with over 150 spots from at least 10 different cells per condition analyzed, using Axiovision software (rel. 4.5; Carl Zeiss).

Mitochondrial complex I activity assay

For NADH dehydrogenase activity assay, intact cells were incubated for 1 h with 0.45 mg/mL of soluble thiazolyl blue tetrazolium bromide (Sigma-Aldrich), as NADH dehydrogenase substrate. After incubation, optical density was measured at 570 and 630 nm using a spectrophotometer and mitochondrial complex I activity was calculated as A570–A630, as previously reported (Mosmann 1983).

Complex I activity was further evaluated on purified mitochondria accordingly to Lannuzel et al. (2003). Briefly, control and MPP+-treated PC12 cells were scraped into 10 mM Tris–HCl pH 7.2, containing 225 mM mannitol, 75 mM saccharose, and 0.1 mM EDTA, sonicated and centrifuged at 600 g at 4°C for 20 min, to obtain the post-nuclear supernatant. Complex I activity was assessed following spectrophotometrically (340 nm) NADH oxidation, at 37°C, for 3 min. Assay medium contains 4 μg of proteins dissolved in 100 μL 25 mM phosphate buffer, pH 7.5, 2.5 mg/mL bovine albumin serum, 100 μM decylubiquinone, and 200 μM NADH. MPP+ (10 mM) was used to determine the fraction of NADH oxidized independently of complex I (blank values).

ATP determination

ATP bioluminescent somatic cell assay kit (Sigma-Aldrich) was used. The cells were washed once with PBS, 100 μL of somatic cell ATP-releasing reagent were added and immediately collected. The solution was diluted 1 : 50 in the same buffer and mixed 1 : 2 with water or ATP standard solution; 50 μL mix was added to 50 μL working solution. After 10 s, the ATP content was determined using a luminometer and integrated over 20 s. The remaining cells were scraped into PBS, sonicated, the protein concentration was measured, and used to normalize the ATP measurement.

Statistical analysis

The statistical significance of MPP+ effects was assessed by one-way anova with Dunnett 2-sided post hoc testing. χ2 test was used to analyze qualitative variables. Student’s t-test for unpaired data was used in FRAP experiments. All experiments were repeated at least three times and data were expressed as means ± SE.


MPP+ impairs mitochondrial transport without inducing caspase 3 activation or damaging mitochondria

Knowing that axonal transport is impaired in isolated squid axoplasm exposed to MPP+ (Morfini et al. 2007), we first evaluated the effects evoked by the neurotoxin on mitochondrial transport in PC12 cells transfected with the N-terminal mitochondrial signal sequence of cas2 fused to DsRed, that allowed us to follow mitochondria movement in living cells (Fig. 1a, Movies S1 and S2). In control cells, mitochondria were uniformly distributed along neurites (Movie S1, Fig. 1b); we analyzed their movement and found that around 45% of mitochondria moved toward growth cone with a mean velocity of 7.92 ± 0.18 μm/min, whereas 22% of mitochondria displayed a retrograde movement and a mean velocity of 8.04 ± 0.3 μm/min. After 6 h of treatment with 45 μM MPP+, we did not observe any change in the distribution of mitochondria, in the percentage of mitochondria exhibiting anterograde or retrograde movements, or any significant variation of the mean velocities (Fig. 1). On the contrary, after 24 h of incubation with the neurotoxin a great part of the mitochondria was accumulated into axonal swellings and varicosities (Movie S2, Fig. 1b). Nevertheless, we were able to measure the movement of the mitochondria that were not accumulated into swellings and found that MPP+ induces the increase in the number of mitochondria moving toward the cell body (Fig. 1c), and significantly alters the mean velocities (Fig. 1d), reducing the anterograde one (6.36 ± 0.36 μm/min) and increasing the retrograde velocity (10.68 ± 0.36 μm/min). These data show that axonal transport is impaired in our experimental model, and are consistent with the results reported by Morfini et al. (2007), having shown that the microinjection of MPP+ increases retrograde axonal transport in isolated squid axoplasm.

Figure 1.

 1-Methyl-4-phenylpiridinium (MPP+) impairs mitochondrial transport. PC12 cells were transfected with N-terminal mitochondrial signal sequence of cas2 fused to DsRed (pMito-DsRed) and differentiated 3 days with nerve growth factor (NGF). (a) Frames from typical time-lapse series of control cells and of cells exposed 24 h to MPP+. Gray arrows indicate the starting positions and white arrows show the actual position of mitochondria moving toward the growth cone, whereas gray arrowheads indicate the starting positions and white arrowheads show the actual position of mitochondria moving toward the cell body. In the treated cell is evident fluorescence accumulation into the varicosity (asterisk). Elapsed time is shown. Scale bar: 10 μm. (b) White bar indicates the quantity of fluorescence that is uniformly distributed along neurite, whereas black bar indicates the percentage of fluorescence accumulated into axonal swellings. Graphs representing the percentage of immobile and anterograde or retrograde moving mitochondria (c), and their respective velocities (d) in control cells or in cells treated 6 or 24 h with 45 μM MPP+. **p < 0.005, ***p < 0.0005 versus control.

To evaluate the effect of increasing concentrations of MPP+ (from 15 to 75 μM) on the axonal transport, we used PC12 cells loaded with Mitotracker Red and analyzed the frequency of cells displaying mitochondria conventionally distributed (Fig. 2a), or accumulated into axonal swellings (Fig. 2b, arrows). The results show that mitochondria distribution does not change in cells treated with 15 μM MPP+ with respect to controls, whereas mitochondria accumulation increases significantly at higher MPP+ concentrations (Fig. 2c), as expected in a condition where the axonal transport is already impaired.

Figure 2.

 1-Methyl-4-phenylpiridinium (MPP+) induces axonal swellings without caspase 3 activation. PC12 cells were differentiated for 5 days with nerve growth factor (NGF) and stained with MitoTraker Red. Mitochondria distribution was analyzed by superimposing images to high-contrast bright-field background obtained from PC12 cells incubated in the absence (a) or presence of 45 μM MPP+ (b). Scale bar: 10 μm. Percentage of neurites showing a conventional distribution (black bar) or accumulation of mitochondria (white bar) is evaluated after treatment with different MPP+ concentrations (c) or in the presence of Caspase 3 or PKCδ inhibitor (d). *p < 0.05, **p < 0.01 versus control.

It has been recently reported that inhibition of caspase 3 or PKCδ could prevent the alteration of axonal transport induced by MPP+ (Morfini et al. 2007). To ascertain the role of these two proteins in the regulation of axonal transport impairment in our model, we treated PC12 cells with 45 μM MPP+ in the presence of specific inhibitors of PKCδ or caspase 3 (Fig. 2d). We did not observe any significant effect of these inhibitors, suggesting that, at least in our experimental model, axonal transport block occurs before activation of PKC and caspase. Furthermore, there was no a significant conversion of procaspase 3 into active enzyme up to very high concentrations (Fig. S1), highlighting that transport impairment precedes caspase 3 activation.

It is well accepted that MPP+ acts on complex I of the mitochondrial respiratory chain (Nicklas et al. 1985), resulting in depletion of ATP and ROS production. Because both cargo alterations and ATP deficiency could influence axonal transport (De Vos et al. 2008), we tried to correlate the functional state of mitochondria and axonal transport in PC12 cells exposed to increasing concentrations of MPP+. We analyzed the fraction of active mitochondria (Fig. 3a), loaded with Mitotracker Red (red, Fig. S2a), on total mitochondria, stained with antibody against porin (green, Fig. S2a), and the membrane potential of active mitochondria (Fig. 3b). We also evaluated complex I activity in intact cells and in purified mitochondria (Fig 3c), and ATP synthesis (Fig. 3d). Finally, by using probes specific for oxidative stress (green, Fig. S2b) and superoxide (red, Fig. S2b), we quantified ROS production (Fig. 3e). Under our experimental paradigm, mitochondria did not show any sign of damage up to 45 μM MPP+, but alteration of mitochondrial membrane potential, complex I dysfunction, ATP depletion, and ROS production, became relevant at the highest concentration. These data suggest that mitochondrial activity unbalance does not precede and cause axonal transport impairment; rather it could be a consequence of impaired mitochondrial transport along neurites.

Figure 3.

 Dose-dependent effect of 1-methyl-4-phenylpiridinium (MPP+) on mitochondrial functionality. PC12 cells were differentiated for 5 days with nerve growth factor (NGF), and incubated with increasing MPP+ concentrations. (a) Histograms of the ratio between polarized (active) and total mitochondria. (b) Evaluation of ΔΨ. MitoTracker Red fluorescence is expressed as arbitrary unit. (c) Evaluation of complex I efficiency in purified mitochondria (black bars) and from intact cells (white bars). (d) Evaluation of ATP synthesis. (e) Quantification of oxidative stress (black bars) and superoxide production (white bars). *p < 0.05 versus control.

The fact that we excluded mitochondrial damage, reduction of ATP, and alteration in the signal cascade as possible causes of axonal transport impairment in our model system, opens the question: what can it lend to axonal transport alterations in PC12 cells exposed to MPP+?

MPP+ induces the overall loss of dynamic MTs and increase of stable MTs

Among the possible causes of the alteration of axonal transport, MTs can have a fundamental role, being the railways along which protein and organelle move. Novel and highly dynamic MTs are enriched in tyrosinated (Tyr) tubulin whereas long-lived and stable MTs are enriched in detyrosinated (Glu) and acetylated (Ac) tubulin (Werstermann and Weber 2003), and these modifications seem to be implicated in binding and velocity regulation of MT-associated motor proteins (Reed et al. 2006; Dunn et al. 2008).

Using antibodies specific for modified tubulin, we examined the effect of MPP+ on the level of Tyr, Glu, and Ac tubulin by western blotting and densitometric analyses (Fig. 4a, 4b). In whole cell extracts, MPP+ induces a transient increase in the total level of Tyr tubulin followed by a significant decrease, which ends with halving of Tyr tubulin level at the highest MPP+ concentration. In contrast, the total amount of both Glu and Ac tubulin increases progressively with MPP+ treatment, showing an enrichment of stable MTs in MPP+-treated cells.

Figure 4.

 1-Methyl-4-phenylpiridinium (MPP+) induces microtubule (MT) over-stabilization. Immunoblot (a) and densitometric analyses (b) of Tyr (white bar), Glu (black bar), Ac (light gray bar), total tubulin, and actin (dark gray bar), were performed in whole-cell extracts from differentiated PC12 cells incubated 24 h in the presence of increasing MPP+ concentrations. *p < 0.05. Free Tyr tubulin (D) and Tyr tubulin incorporated into MTs (MT) were investigated by immunoblot (c), analyzed by densitometry and showed as ratio (d). *p < 0.05 versus control.

It is well known that Tyr tubulin is the native form of α-tubulin and is found in the non-assembled pool of tubulin and in highly labile MTs (Bulinski and Gundersen 1991). We have analyzed cytoskeletal fractions obtained by extracting cells with Triton X-100, calculated the ratio of Tyr tubulin incorporated into the MTs versus free Tyr tubulin, and found a significant decrease of the ratio in MPP+ treated cells (Fig. 4c and d), that means Tyr tubulin accumulates in non-assembled pool following MPP+ treatment. This is consistent with our previous in vitro results (Cappelletti et al. 2005) showing that MPP+ depolymerizes dynamic MTs; however, the enrichment of stable MTs could be viewed as the extreme effort of the cell to stabilize a collapsing system.

Because MTs and actin filaments act in concert to maintain neuronal shape and viability, we decided to investigate also the level of actin in cells exposed to the neurotoxin (Fig. 4a and b). We did not observe any significant change in the level of total actin after the treatment with MPP+, even with the highest concentration.

Taken together, these results indicate that MPP+ induces a shift in Tyr/Glu tubulin equilibrium that causes the accumulation of Tyr tubulin in the non-assembled pool and that the neurotoxin causes the overall enrichment of stable MTs at doses lower than those are necessary to induce mitochondrial damage or axonal transport block.

MPP+ causes the reduction of MT dynamics

The above reported loss of Tyr MTs and the enrichment in Glu and Ac tubulin content strongly suggest that MPP+ affects MT dynamics. We therefore analyzed the dynamic behavior of the MT cytoskeleton in living PC12 cells expressing yellow fluorescent protein (YFP)-α-tubulin using FRAP.

FRAP experiments (Fig. S3were performed bleaching the region of interest (ROI), along neurites or at growth cones, and then measuring the fluorescence recovery. To verify that differences in the area of the growth cone do not influence the recovery, we bleached either the entire growth cone region or a fixed circular ROI in the middle of it. Using the equations reported in the ‘Materials and methods’, we have extrapolated two parameters: the half time of fluorescence recovery (t½) and the mobile fraction (Mf), that respectively represent how faster tubulin moves and how much tubulin moves. At the growth cone, as shown in Table 1, t½ of FRAP was significantly increased in MPP+-treated cells with respect to the controls, whereas the Mf of tubulin was not affected by the neurotoxin. These results clearly show that MPP+ delays the fluorescence recovery in growth cones. To discriminate between passive diffusion of tubulin subunits and dynamic exchange between free tubulin and MTs, we performed distinct FRAP analyses in the middle or at the edge region of the bleached zone. The results reported in Table 1 do not show a faster recovery at the edge of the circular ROI, that would have been consistent with lateral diffusion of YFP-tubulin from the unbleached regions to the beached ones, and suggest that the observed FRAP may be owing to the dynamic exchanges between free tubulin dimers and MTs underlying MT dynamics. We conclude that increasing of t½ in MPP+-treated cells is consistent with a significant reduction of MT dynamics at growth cones. Similarly in neurites, t½ of FRAP was significantly increased in MPP+-treated cells but the recovery percentage was not affected by the neurotoxin (Table 1). These results show that tubulin mobility is significantly reduced by MPP+ both along neurites and in growth cones and suggest that this effect is the result of a significant reduction of tubulin dynamics.

Table 1.   FRAP analysis in NGF-differentiated PC12 cells expressing YFP-tubulin
TreatmentMT area measuredt1/2, s
(mean ± SE)
% Recovery
(mean ± SE)
  1. (*) p < 0.05 versus the same MT area measured in untreated cells.

  2. FRAP, fluorescence recovery after photobleaching; MPP+, 1-methyl-4-phenylpiridinium; MT, microtubule; NGF, nerve growth factor; ROI, region of interest.

UntreatedGrowth cone66.5 ± 5.950 ± 912
Growth cone (fixed ROI)50.4 ± 3.675 ± 16.65
Growth cone (middle)49.8 ± 5.657.8 ± 75
Growth cone (edge)59.4 ± 6.755.5 ± 5.75
Neurite (fixed ROI)58.1 ± 633 ± 4.521
45 μM MPP+Growth cone105.3 ± 15.3 (*)60 ± 7.516
Growth cone (fixed ROI)92.5 ± 17.5 (*)58.5 ± 194
Growth cone (middle)94 ± 15.4 (*)51.5 ± 6.84
Growth cone (edge)109.8 ± 19 (*)54 ± 7.24
Neurite (fixed ROI)106.1 ± 20.1 (*)33.4 ± 722

To evaluate in greater detail dynamics of MTs in live cells following MPP+ treatment, we needed to clearly identify individual MT tips, but it is difficult in cells expressing YFP-tubulin, because MTs are often bundled and we cannot resolve individual growing end. EB3 is a MT-associated protein that binds specifically to the plus end of growing MTs. We used the fluorescent protein EB3-GFP that has been demonstrated to be a convenient suitable fusion protein for live studies of MT dynamics in transfected neuronal cells (Stepanova et al. 2003). Live cell imaging analysis on PC12 cells expressing EB3-GFP shows growing MTs as comet-like dashes exhibiting anterograde and retrograde movements in the proximal and distal regions of neurites and, in some cases, extending into filopodia-like protrusions at growth cones (Fig. 5a, Movies S3 and S4). In control cells, anterograde growth rate was 5.94 ± 0.24 μm/min, whereas MTs moved toward cell body with a mean velocity of 5.58 ± 0.24 μm/min. After 6 h of incubation with 45 μM MPP+, both anterograde and retrograde velocities of EB3 comets were significantly reduced (4.8 ± 0.94 μm/min for anterograde movement and 4.5 ± 0.12 μm/min for retrograde movement). After 24 h of treatment with the neurotoxin, the effects on MT growth were the same; anterograde growth rate (4.68 ± 0.18 μm/min) and retrograde mean velocity (4.74 ± 0.18 μm/min) were significantly reduced as compared with control cells. FRAP data strongly suggest that MT polymerization along neurites as well as at growth cones occurs at a lesser extent in treated cells with respect to the controls, and live imaging of EB3, which allows to visualize individual MTs, confirms that MPP+ reduces MT growth rate.

Figure 5.

 1-Methyl-4-phenylpiridinium (MPP+) alters microtubule (MT) dynamics. PC12 cells were transfected with end-binding protein 3-green fluorescent protein (EB3-GFP) and differentiated 3 days with nerve growth factor (NGF). (a) Frames from typical time-lapse series of control cells and of cells exposed 24 h to MPP+. Gray arrows indicate the starting positions and white arrows show the actual position of EB3 comets moving toward growth cone, whereas gray arrowheads indicate the starting positions and white arrowheads show the actual position of comets moving toward cell body. Elapsed time is shown. Scale bar: 10 μm. Graphs representing the velocities of EB3 comets (b) and the direction of EB3 movement (c), in control cells or in cells treated 6 or 24 h with 45 μM MPP+. (d) Schematic representation of our hypothesis. (b) The brightest region and (d) the darkest part of the comet. (e) White bars indicate the percentage of MTs that are growing toward cell body, whereas black bars indicate the percentage of MTs that are transported toward the soma. *p < 0.05, **p < 0.005 versus control.

We have also analyzed the orientation of the movement of EB3 comets. In control cells, almost 90% of comets moved toward the growth cone whereas only 10% of MTs had retrograde movement. After 6 h of exposure to MPP+, the percentage of EB3 comets with anterograde movement was significantly decreased to 69% while retrograde movement increased concomitantly to 31%, and after 24 h of treatment with the neurotoxin the effects on MT movement orientation were exacerbated; we observed 48% of comets anterograde advancing and 52% moving toward cell body. Since the average number of analyzed MTs/cell does not vary in treated cells with respect to controls, we assume that these data reflect the change in the ratio of anterograde versus retrograde comets. The movement of EB3 comets toward the soma could be result from the transport of the entire MT whose plus end remains oriented toward the growth cone, or to the growth of MT toward cell body as typically occurs in dendrites. To discriminate between these two possibilities, we decided to analyze the shape of the comets to determine their orientation. We divided each backward moving EB3 comet into two parts, perpendicularly to its main axis, and measured the fluorescence intensity of both of them. Because EB3 proteins accumulate at MT growing ends, we assume that if the most fluorescent part of the comet precedes the darkest region when the comet moves, MT is growing; otherwise, if the darkest region of the comet precedes, the brightest region when the comet moves, MT is transported (Fig. 5d). The greater part of comets retrograde moving (80%) was growing toward cell body in control cells, whereas 20% of comets represent MT transport. After the incubation with the neurotoxin, we observed an increased number of comets moving toward cell body, as mentioned above, and among these a significantly increased fraction accounts for MT growing toward the cell body (Fig. 5e), suggesting that MPP+ causes a reorganization of MTs rather than an alteration of MT transport. Interestingly, a similar event has been recently described in a Tau model of neurodegeneration (Shemesh et al. 2008).

Taken together, these data show that MPP+ reduces MT growth and induces a reorientation of MTs already after 6 h of treatment, and by comparing these results with the above reported alterations induced by MPP+ on mitochondrial transport, that become noticeable after 24 h of exposure to the neurotoxin, we highlight that MT dysfunction precedes axonal transport impairment.

MPP+ causes growth cone collapse and retraction through dynamic MT destabilization

Neuropathological studies of idiopathic PD patients (Bernheimer et al. 1973) and MPTP-treated mice (Cochiolo et al. 2000) show that loss of synaptic terminals in the striatum precedes loss of dopaminergic neurons in the substantia nigra, highlighting a mechanism of neuronal degeneration known as ‘dying back’. To investigate if such mechanism starts as early as MT alterations become evident, we analyzed MT organization and stability in growth cones of differentiated PC12 cells exposed to increasing concentrations of MPP+. MT network was studied by double immunofluorescence with antibodies against Tyr and Glu tubulin on Triton X-100-extracted cells to remove unassembled tubulin (Fig. 6a–c).

Figure 6.

 1-Methyl-4-phenylpiridinium (MPP+) causes growth cone collapse and retraction. PC12 cells differentiated for 5 days with nerve growth factor (NGF) were extracted with Triton X-100 and immunostained for Tyr and Glu tubulin, to investigate microtubule (MT) organization and stability at growth cones. Confocal images of ‘spread’ (a), ‘collapsed’ (b), and ‘retracted’ (c) growth cones are shown. Scale bar: 10 μm. The percentage of ‘spread’ (black bar) and ‘collapsed’ plus ‘retracted’ growth cones (white bar) was scored in control and MPP+-treated cells (d). Neurite length was evaluated in control and MPP+-treated cells (e). *p < 0.05, **p < 0.01 versus control.

In both control and treated cells, three types of growth cones with different MT arrangements and stability can be observed: (i) ‘spread’ growth cones (Fig. 6a) that appear as broad structures showing a well-organized network of Tyr MTs, whereas both Tyr and Glu MTs co-localize and sustain the central region as well as the neurite shaft; (ii) ‘collapsed’ growth cones (Fig. 6b) with a narrow shape displaying a well-organized MT lattice, the loss of the distal Tyr MTs and the concomitant enrichment of Glu MTs; and (iii) ‘retracted’ growth cones (Fig 6c) that exhibit MT coiling and sinusoidal bending typical of retracting axons (He et al. 2002) and the overall enrichment of stable MTs. We scored the percentage of growth cone types in control and MPP+-treated cells, and we observed that MPP+ clearly induces a significant and concentration-dependent increase in collapsed and retracted growth cones (Fig. 6d). These results show that changes in growth cone structure occur in response to MPP+, and that they are associated with changes in MT stability. The increasing number of retracted growth cones suggests that some of the neurites are shortening. Therefore, we measured neurite length and we observed a mild but significant and progressive shortening of neurites following MPP+ treatment (Fig. 6e). The observed effects are likely mediated by the primary cytoskeletal components of growth cone including actin filaments and MTs. In fact, in neuronal growth cones MTs act in concert with actin-based cytoskeleton located predominantly at the growth cone periphery, where lamellipodia and filopodia exhibit an intense exploratory activity. Double fluorescent staining of MTs and actin in MPP+ treated cells (Fig. S4) shows that the peripheral region of the three growth cone types, ‘spread’, ‘collapsed’, and ‘retracted’, remains rich in actin filaments. Thus, depolymerization of actin filaments and actin bundle loss might not to be the primary cause of growth cone collapse and retraction induced by MPP+.


The molecular pathways implicated in neurodegenerative disorders are gradually being elucidated and several culprits including accumulation of aberrant or misfolded proteins, mitochondrial injury, oxidative/nitrosative stress, and failure of axonal and dendritic transport have been identified (Bossy-Wetzel et al. 2004). The results reported here demonstrate that, in differentiated PC12 cells, MT dynamics and organization are specifically and precociously targeted during neurodegeneration induced by the parkinsonism-inducing neurotoxin MPTP, long before impairment of axonal transport and mitochondrial damage occur. Although coming from an in vitro model, our results lend support to the concept that MT dysfunctions could play a significant role in triggering early phases of neuronal death in PD.

The underlying mechanisms connecting MT dynamics to axonal transport are unclear and sometimes controversial, but several intracellular pathways converge in regulating MT stability and axonal transport. It has been shown that acetylation enhances MT stability (Hubbert et al. 2002) and also that stabilization of MTs by acetylation speeds up anterograde transport by recruiting kinesin-1 (Reed et al. 2006); in contrast, over-stabilization of MTs through the elevation of tau compromises axonal transport (Stamer et al. 2002). Furthermore, a recent article, combining cell-based and in vitro assays, shows that kinesin-1 targets preferentially Glu tubulin, and that kinesin moves along highly modified MTs at significant lower velocities than along Tyr MTs (Dunn et al. 2008). Thus, the MPP+-induced enrichment in highly modified and stable MTs observed in our model, could easily explain the reduction of anterograde transport consequent to the neurotoxin exposure. Another relevant point connecting MT dysfunction, axonal transport, and neurodegeneration is the polar reorientation of MTs. We observed the significant increase in the percentage of EB3 comets backward moving in MPP+-treated cells consistent with the reorientation of MTs, an event that has been recently described in cultured Aplysia neurons over-expressing human tau (Shemesh et al. 2008). Because MT-associated motor proteins are highly oriented, changes in MT polarity lead to accumulation of organelles and vesicles at the point of MT polar mismatching (Shemesh et al. 2008), providing an alternative explanation for the impairment of axonal transport in tau or MPTP models of neurodegeneration. In particular, in dopaminergic neurons, a disruption of vesicular transport causes the accumulation of dopamine-loaded vesicles in the cell soma, dopamine whose oxidation produces large quantities of reactive oxygen species that may trigger cell death (Hastings et al. 1996), and easily explain how MT dysfunctions, axonal transport, and PD could be related.

Mitochondrial and microtubular systems are tightly correlated, because energy failure resulting from mitochondrial injury could be reflected in an altered behavior of MTs, and MT dysfunction can lead to mitochondria mispositioning and damaging. Consequently, it is not surprising that drugs specifically affecting either mitochondria or MTs induce dysfunction also in the other system. Annonacin, a natural complex I inhibitor correlated with atypical PD in Guadeluope (Lannuzel et al. 2003), causes MT fragmentation and alteration of the mitochondrial localization (Escobar-Khondiker et al. 2007). On the other side, MT-specific drugs either with stabilizing or with destabilizing action cause both axonal transport failure (Nakata and Hirokawa 2003; Liang et al. 2008) and mitochondrial depolarization and Ca++ release (Mironov et al. 2005). For long time, mitochondrial dysfunction has been implicated in the pathogenic mechanism of MPP+ and rotenone, since they block mitochondrial ATP production by inhibiting complex I (Nicklas et al. 1985; Betarbet et al. 2000). In vivo studies showed that the expression of the yeast rotenone-insensitive complex I subunit NDI1 of Saccharomyces cerevisae resulted in the significant protection of dopaminergic neurons to toxicity induced by MPTP (Barber-Singh et al. 2009). However, a recent article demonstrates that the lack of complex I does not affect the viability of dopaminergic neurons in culture, and does not protect them from the administration of MPP+ and rotenone (Choi et al. 2008), suggesting that there is a possible different intracellular target where PD-inducing neurotoxin actions converge: MTs. Both rotenone and MPP+ directly reduce polymerization of purified tubulin (Marshall and Himes 1978; Cappelletti et al. 2001); furthermore, rotenone induces MT depolymerization (Ren et al. 2005) leading to dopaminergic neuron degeneration, and, as we show in this study, MPP+ affects MT organization and dynamics inducing growth cone collapse and neurite retraction. Ren et al. (2005) have also shown that toxins specific for mitochondria or MTs, as amytal and colchicine, need to be combined to reach the same extent of damage of rotenone, having opened the way to the concept that MTs and mitochondria could act in a synergistic manner to induce PD. Nevertheless, there is another point to consider: which came first the chicken or the egg? In axon undergoing Wallerian degeneration, the first detectable alteration is the MT fragmentation (Zhai et al. 2003); in this study, we show for the first time, that during MPP+-induced neurodegeneration MT dysfunction is noticeable without any sign of mitochondrial injury. Putting together the facts that MT dysfunction is detectable at lower concentrations than mitochondrial damage, and that MT reorientation and reduction of MT dynamics precede axonal transport impairment, one can argue that MPP+ initially causes MT dysfunction. Subsequently, the chain of events could be comparable with that observed in tau neurodegeneration (Thies and Mandelkow 2007): MT dysfunction leads to axonal transport impairment that can be reflected in mitochondria damaging. Altered mitochondria are more sensible to the MPP+, and the block of the respiratory chain results in increased oxidative stress and energy failure. In this way, MT dysfunction and mitochondrial injury reinforce each other and collaborate to induce dopaminergic neuronal death in PD.

Future work shall be done using in vivo models to better understand the relative contribution of microtubular and mitochondrial dysfunctions in PD etiopathogenesis. Interestingly, the effective concentration of MPP+ in the striatum of MPTP-treated mice in the early phase of the treatment (90 min post-injection) has been reported to be around 400 pmoles/mg proteins (Battaglia et al. 2007) that is comparable with that we used in PC12 cells (50 pmol/mg protein). So, this study, showing that reorientation of MTs and reduction of MT dynamics precedes axonal transport impairment and mitochondrial damage, adds a new tessera to the mosaic of neuronal degeneration in PD and could be the starting point to shed light on the very early events underlying neurodegeneration.


We are grateful to Dr Galjart for the gift of EB3-GFP construct and to Dr Rugarli for the pMito-DsRed. We thank Dr Boncoraglio for help with statistics, Dr Surrey for critically reading the article, and Dr Denis Donini for helpful discussion. This work was funded from MIUR (FIRB grant n. RBAUO1FSR9).