| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,
,
,1,1,,2
* Neural Regeneration Unit, Institute of Reconstructive Neurobiology, University Bonn Life and Brain Center and Hertie Foundation, University Bonn, Bonn, Germany;
Institute of Multiple Sclerosis Research, University Göttingen and Hertie-Foundation, Göttingen Germany;
Neuroimmunology Unit, European Neuroscience Institute Göttingen, Waldweg, Göttingen, Germany; and
Yale University, Department of Molecular Biophysics and Biochemistry, Sterling Hall of Medicine, New Haven, Connecticut, USA
2Correspondence: Neural Regeneration Unit, Institute of Reconstructive Neurobiology, University Bonn Life and Brain Center, University Bonn, Sigmund-Freud-Str. 25, 53127 Bonn, Germany. E-mail: hneuman1{at}uni-bonn.de
ABSTRACT
Axonal transport of mitochondria and synaptic vesicle precursors via kinesin motor proteins is essential to keep integrity of axons and synapses. Disturbance of axonal transport is an early sign of neuroinflammatory and neurodegenerative diseases. Treatment of cultured neurons by the inflammatory cytokine tumor necrosis factor-
(TNF) stimulated phosphorylation of c-Jun N-terminal kinase (JNK) in neurites. TNF treatment induced dissociation of the heavy chain kinesin family-5B (KIF5B) protein from tubulin in axons but not cell bodies as determined by lifetime-based Förster resonance energy transfer (FRET) analysis. Dissociation of KIF5B from tubulin after TNF treatment was dependent on JNK activity. Furthermore, TNF inhibited axonal transport of mitochondria and synaptophysin by reducing the mobile fraction via JNK. Thus, TNF produced by activated glial cells in inflammatory or degenerative neurological diseases acts on neurites by acting on the kinesin-tubulin complex and inhibits axonal mitochondria and synaptophysin transport via JNK.—Stagi, M., Gorlovoy, P., Larionov, S., Takahashi, K., and Neumann, H. Unloading kinesin transported cargoes from the tubulin track via the inflammatory c-Jun N-terminal kinase pathway.
Key Words: neuroinflammation • tumor necrosis factor- • synaptophysin • mitochondria
ABERRANT ACCUMULATION OF AXONAL cargos such as ß-amyloid precursor protein (APP) and synaptic proteins is a characteristic pathological sign of a wide variety of neurodegenerative and neuroinflammatory diseases (1)
. Particularly, accumulation of APP has been observed in axons in multiple sclerosis (2)
, microbial infections (3)
, traumatic injury (4)
, and ischemia (5)
. Recently, it was suggested that axonal transport impairment is involved in the early development of neurodegenerative diseases (6)
. Axonal swellings and accumulation of abnormal amounts of microtubule-associated and molecular motor proteins have been observed early in the pathogenesis of Alzheimer’s disease (7)
. Molecular motor proteins transport axonal cargoes to and from presynaptic terminals along microtubule tracks (8)
. Transport of axonal cargoes is a highly regulated process. Several studies indicate that c-Jun N-terminal kinase (JNK) interacting proteins (JIP) serve as scaffolding proteins and associate with the kinesin motor protein family and cargo receptors (9
10
11)
. Pioneering work by the Goldstein laboratory described that in Drosophila melanogaster a homolog of the mammalian scaffolding protein JIP3-linked conventional kinesin-I to an unknown class of vesicles (9)
. Data in Caenorhabditis elegans showed that unc-16, a homolog of JIP3, physically interacted with JNK kinases (10)
and that mutations of unc-16 resulted in mislocalization of synaptic vesicle markers (10)
. Also JIP1 and JIP2 have been identified to associate with and to be cargoes of kinesin-I (11)
. Furthermore, JIP1 connects the kinesin light chain to the synaptic vesicle protein APP and could enable protein phosphorylation by JNK (12)
. APP is thought to be a transmembrane cargo receptor of synaptic precursor vesicles (13)
. The scaffolding JIPs are also involved in the organization and facilitation of JNK signaling implicated in multiple processes, allowing a cell to mount an appropriate response to extracellular stress (14
, 15)
. The finding that multiple types of JNK scaffolding proteins interact with kinesin-I and that JNK scaffolding proteins are involved in JNK signaling raises the question whether JNK signaling regulates axonal transport. Recently, we demonstrated that NO produced by activated microglia stimulated phosphorylation of JNK in axons and inhibited anterograde axonal transport of synaptic vesicle precursors (16)
.
Now, we analyzed the effect of tumor necrosis factor- (TNF), a proinflammatory cytokine that induces cellular stress response through the JNK signaling cascade (17)
. TNF-stimulated phosphorylation of JNK in neurites and broke apart the molecular interaction between heavy chain kinesin family-5B (KIF5B) protein and tubulin as determined by lifetime-based Förster resonance energy transfer (FRET) analysis. Furthermore, TNF inhibited axonal transport of mitochondria and synaptophysin via JNK.
MATERIALS AND METHODS
Neuronal cultures
The neuronal cultures were prepared from hippocampi of C57BL/6 mice embryos (E15 and E16), as described previously (18)
. Hippocampi were isolated, dispersed mechanically, and seeded in four-well chamber culture dishes (Nunc GmbH&Co. KG, Wiesbaden, Germany) or biomembrane optical dishes (Helmut Sauer, Nuernberg, Germany) with a density of 150,000 cells/ml. The dishes were pretreated with poly-L-ornithine (0.01 mg/ml, Sigma, Munich, Germany) and laminin (10 µg/ml, Sigma). The cells were cultured in eagle basal medium (BME)-based neuronal medium (BME, Gibco BRL, Invitrogen, Karlsruhe, Germany) supplemented with 2% B-27 supplement (Gibco BRL, Invitrogen), 1% glucose (Glc) (45% Glc, Sigma), and 1% fetal calf serum (FCS; PAN Biotech GmbH, Germany). Cells were cultured for 5–10 days to obtain morphologically mature neurons. Neurons were treated with TNF (20 ng/ml, murine recombinant TNF-; R&D Systems, Wiesbaden, Germany), JNK inhibitor (100 nM JNK inhibitor II, SP600125; Calbiochem, San Diego, CA) or TNF receptor I-IgG fusion protein (20 µg/ml; gift from Dr. Werner Lesslauer, Roche), as indicated in the text.
Immunocytochemistry of TNF receptors I and II
Hippocampal neurons were cultured for at least 1 wk, fixed in a 4% aqueous solution of paraformaldehyde and then incubated with a rat monoclonal antibody (mAb) specific for TNFRI or TNFRII (1:200, HyCult Biotech, Uden, The Netherlands), followed by FITC-conjugated goat secondary antibody (Ab) against rat IgG (1:400, Dianova, Hamburg, Germany). After washing, cells were incubated with an axon-specific mouse mAb against tau (tau-1, 1:200, PC1C6, Chemicon, Temecula, CA) and Cy3-conjugated goat secondary Ab against mouse IgG (1:400, Dianova, Hamburg, Germany). As a negative control, rat IgG (1:200, 1 µg/ml, Sigma) was applied, followed by FITC-conjugated goat secondary Ab against rat IgG (1:400, Dianova). Optical sections along the z axis were scanned with 40 x objective on a confocal laser-scanning microscope (Olympus, Hamburg, Germany). For quantification, random fields were selected from neuronal cultures and analyzed under fluorescence microscopy counting the number of axonal processes colabeled with antibodies directed against tau and TNF receptor I or tau and TNF receptor II.
Immunocytochemistry of total and phosphorylated JNK
Hippocampal neurons cultured for 1 wk were fixed in a 4% aqueous solution of paraformaldehyde. For total JNK staining, rabbit anti total JNK antibodies (1:100, Cell Signaling Technology, Beverly, MA), followed by Cy3-conjugated goat anti-rabbit IgG (1:400, Dianova). For phosphorylated JNK staining, rabbit anti-phosphorylated JNK (1:100, Cell Signaling Technology), followed by Cy3-conjugated goat anti-rabbit IgG (1:400, Dianova) were used. Subsequently, cells were incubated with mouse monoclonal anti tau (tau-1, 1:200, PC1C6, Chemicon) and FITC-conjugated goat anti-mouse IgG (1:400, Dianova). As a negative control, rabbit IgG (Sigma) was applied, followed by Cy3-conjugated goat anti IgG (1:400, Dianova). Optical sections along the z axis were scanned with 40 x objective with a confocal laser-scanning microscope (Olympus). Neurons were treated for 20 min with TNF as indicated in the text and figures. For quantification, random fields were selected from neuronal cultures either untreated or treated with TNF for 20 min and analyzed under fluorescence microscopy counting the number of axonal processes and cell bodies labeled by tau, showing expression of phosphorylated JNK. Data from independent experiments were presented as mean ± SEM.
Immunoprecipitation and Western blot analysis
Protein extracts of cultured hippocampal neurons were collected by lysis buffer (0.5% Triton X-100, 50 mM Tris pH 7.5, 150 mM NaCl) with protease and phosphatase inhibitor mix (Sigma). The cells were either untreated or treated for 20 min with TNF (20 ng/ml). The total protein lysates were analyzed by Western blot analysis with mouse monoclonal IgG anti-ß-tubulin III Ab (Sigma), as well as with rabbit monoclonal IgG anti-SAPK/JNK and rabbit polyclonal IgG antiphospho-SAPK/JNK antibodies (both from Cell Signaling Technology. For immunoprecipitation, protein lysates were incubated with mouse monoclonal IgG anti ß-tubulin III (2 mg/reaction) at 4° for 2 h. Subsequently, 20 ml of protein G sepharose beads (GE Healthcare, Little Chalfont, UK) were added. The precipitated proteins were analyzed by Western blot analysis with mouse monoclonal anti-KIF5B Ab (Abcam, Cambridge, UK), as well as with rabbit monoclonal IgG anti-SAPK/JNK and rabbit polyclonal IgG antiphospho-SAPK/JNK.
Western blot analysis was performed using the NuPAGE electrophoresis system and 10% Bis-Tris gels (Invitrogen). Secondary antibodies were horseradish peroxidase-conjugated goat anti-mouse polyclonal IgG and IgM (Chemicon) or anti-rabbit IgG (Cell Signaling Technology. The Enhanced Chemiluminescence (ECL) Advance Western blotting Detection Kit (GE Healthcare) was utilized. For quantification, optical density (OD) of the respective bands of three blots per experimental group was determined by ImageJ software (NIH) and normalized to the mean values of the untreated bands.
Immunocytochemistry of KIF5B and ß-tubulin III
Hippocampal neurons were cultured for 5–10 days, fixed in a 4% aqueous solution of paraformaldehyde for 60 min followed by PBS plus 0.1% TritonX-100 for 5 min and then incubated with a mouse monoclonal anti KIF5B (1:200, Abcam), followed by FITC-conjugated goat anti-mouse IgG (1:400, Fab-fragments directed against Fab, Dianova, Jackson ImmunoResearch Laboratories, West Grove, PA). After washing cells, we incubated them with a mouse monoclonal anti ß-tubulin III (1:400; Sigma) and Cy3-conjugated goat anti-mouse IgG (1:400; Fab-fragments directed against Fab, Dianova). As a negative control for FRET, neurons were immunolabeled with mouse monoclonal anti KIF5B (1:200, Abcam), followed by FITC-conjugated anti-mouse IgG (1:400, Fab-fragments directed against Fab, Dianova). Then neurons were colabeled with Cy3-conjugated mouse monoclonal anti neurofilament (1:400, Sigma). Classic confocal laser-scanning microscopy (Leica Microsystems, Bensheim, Germany) was performed with an oil objective 40x. To exclude any cross-reactivity, the sequential colabeling procedure by omitting the second primary Ab (antitubulin) was applied, and no fluorescence signaling of the Cy3-conjugated secondary Fab-fragment directed against Fab of IgG was detected. Neurons were treated with 20 ng/ml TNF for 20 min or preincubated for 5 min and cotreated with JNK inhibitor (SP600125; 100 nM final) or TNF receptor I-IgG fusion protein (20 µg/ml; gift from Dr. Werner Lesslauer, Roche), as indicated in the text and figures.
Fluorescence lifetime-based FRET analysis
Direct protein-protein interactions can be visualized via Förster resonance energy transfer (FRET). FRET is a nonradiative, dipole coupling process, whereby energy from an excited donor (fluorophore) is transferred to an acceptor (fluorophore), which is in very close proximity. The efficiency of this process may be used to estimate the proximity of the two fluorophores (19)
. We decided to use one of the most common FRET pairs (FITC//Cy3), well known for quantum efficiency and stability. All FLIM measurements were performed with a 63 x oil or 40 x objective lens on a specific multiphoton microscopy system consisting of a time-domain FLIM setup (Becker & Hickl, Berlin, Germany), an upgraded TSC-SP2 AOBS laser scanning confocal microscope (Leica Microsystems) and a Ti:Sapphire Mira900 two-photon laser pumped by a Verdi-V8 laser (Coherent) in the mode-locked femtosecond-pulsed regime. The laser was tuned to 900 nm for proper two-photon excitation of FITC (20)
. An emission filter wheel was placed between the output port of the scanning-head and the time domain-FLIM detector, an MCP-PMT (R3809U-50 by Hamamatsu Photonics). A 520 ± 15 nm filter was used for FITC detection. The time-resolved fluorescence decays were reconstructed by time correlated single-photon counting (TCSPC). The acquisition board (SPC830) and software (SPCImage) were from Becker & Hickl. The gate/rate that we obtained was between 103 and 104 photons/second with peak rates of 105 photons/second. Acquisition time was fixed to 900 s at an excitation power of 0.5–0.6 W. Using this system, 200 counts are required for a reliable monoexponential decay fit and 1,000 counts can give 
5% of statistical error in the lifetime estimations (21)
. Lifetime showed two Gaussian populations indicating FRETing and non-FRETing populations. Accordingly, a threshold (1.9 ns) between the FRETing and non-FRETing populations was set for coding the images. All lifetime images obtained were coded in pseudocolors with lifetime ranges of 1.0 to 1.9 ns in red (shorter lifetimes indicating FRET) and lifetime ranges of 1.9 to 4.0 ns in green (long lifetimes indicating no FRET). Furthermore, lifetime values are presented as histograms. To quantify the percentage of neurites showing a lifetime signal reflecting FRET random images were analyzed. Neurons were treated with TNF (20 ng/ml for 20 min), TNF plus JNK inhibitor (SP600125; 100 nM), as indicated in the text and figures.
Transfection of neurons to label mitochondria
The pMH4-HSYN-Mito-EYFP vector was designed to fluorescently label mitochondria. The plasmid pMH4-HSYN-Mito-EYFP encodes a fusion of the EYFP [enhanced yellow fluorescent protein (YFP)] gene and the mitochondrial targeting sequence from subunit VIII of human cytochrome c oxidase under the human synapsin I promotor (the p(synapsin) was kindly provided by Dr. S. Kügler, Dept. of Neurology, University of Göttingen). The plasmid expressing c-terminally tagged synaptophysin enhanced GFP (EGFP) under the human synapsin I promoter was previously described (16)
. Plasmids were purified using EndoFree Maxi Kit (Qiagen, Cologne, Germany). Transfection was performed during cell seeding by the Effectene Transfection Kit (Qiagen) with 0.5–1 µg DNA per 1 x 106 cells, 4–8 µl enhancer, 12 µl Effectene, and 85 µl buffer for a period of 30 min, as described previously (22)
. Neurons were treated for 20 min, 30 min, 60 min, and 360 min with TNF (20 ng/ml) or TNF for 20 min plus JNK inhibitor (SP600125; 100 nM), as indicated in the text and figures.
In vivo imaging by tracking mitochondria and synaptophysin FRAP analysis
The medium of the neuronal cultures was removed and substituted with imaging buffer containing 142 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl2, 1 mM NaH2PO4, 25 mM HEPES, 5 mM Glc, 0.8 mM MgCl2 in ddH2O, pH = 7.4. The cell culture dishes were placed on the heated stage of a confocal microscope Leica LSC (Leica) and kept during the analysis at 35–37°C. The neurons were kept in the image buffer not longer than 60 min. Mitochondria were tracked by time lapse imaging (one frame every 20 s) by confocal microscopy. Velocity, direction, and mobile/immobile fraction of mitochondria were determined. Analysis was done by software View5d (image plug-in kindly provided by Dr. Reiner Heintzmann, MPI of biophysical Chemistry, Göttingen, www.gwdg.de/rheintz/View5D/) combined with MatLab®.
FRAP of synaptophysin-GFP was performed as described previously (16)
. Samples of the 20 µm x 20 µm bleached area were imaged every 10–60 s over 10–15 min after photobleaching. Collected data were fitted to the equation: F(t) = span (1-e-kt)+bottom). The bottom was defined as the intensity of the first image after photobleaching. The span was defined as the difference between the final fluorescence intensity and the bottom. The half-time value t 
was calculated accordingly. The t and k values show the velocity of the movement and are linked by the following equation: t = ln(2)/k. The final fluorescence intensity F(
) was less then the prebleaching fluorescence intensity since a proportion of the fluorescence molecules were not mobile. The mobile fraction was defined as the span of the fluorescence recovery after photobleaching. All data were normalized in respect to the intensity of the bottom (value=0) and the prebleaching intensities (value=1) resulting in a relative FRAP.
Statistical analysis
For statistical analysis, t test and Mann-Whitney U test were applied by statistical Packages for the Social Sciences (SPSS) software, as indicated in the text and figures. Data of FRAP and FLIM were analyzed by SPCImage (Becker & Hickl, GmbH), MatLab/Mathworks software and Excel Software (Microsoft). Statistical analysis of FRAP was performed using two-tailed t test between the groups. Statistical analysis of mitochondria movement analysis was performed with tracker software View5d combined with MatLab/Mathworks.
RESULTS
Phosphorylated JNK in neurites after treatment with TNF
Recently, our group observed an increase of phosphorylated JNK in axons after stimulation with microglial-derived NO (16)
. Because the inflammatory cytokine TNF stimulates JNK in most cell types, we analyzed expression of TNF receptors and phosphorylated JNK in axons after treatment with TNF. Cultured hippocampal neurons were labeled with antibodies to TNF receptor I or II. Analysis by confocal microscopy demonstrated that neurites identified by tau immunolabeling showed immunostaining for TNF receptors I and II (Fig. 1
). In total, 82.0% ± 7.2% (mean±SEM) of tau-positive neurites were labeled with antibodies against TNF receptor I and 88.1% ± 1.3% of tau-positive neurites were labeled with antibodies directed against TNF receptor II. Then, cultured hippocampal neurons were treated with TNF (20 ng/ml) for 20 min, and JNK was analyzed by immunocytochemistry with specific antibodies directed against total JNK and phosphorylated JNK. Expression of total JNK was detected in all neurons and neurites, while the expression of the phosphorylated JNK, indicating JNK activity, was more restricted. In untreated neurons, phosphorylated JNK was not detected in neurites (Fig. 2
A, B). After treatment of neurons with TNF for 20 min, strong immunolabeling of almost all neurites by phosphorylated JNK was detected (Fig. 2B
). Quantification of tau-positive neurites showed that after TNF treatment, the majority (84%±13%; mean±SEM) expressed phosphorylated JNK, while in untreated neurons, a minority of tau-positive neurites (11%±7%) showed immunolabeling for phosphorylated JNK (Fig. 2C
). The acute TNF treatment did not affect neuronal survival since 24 h after TNF application, the number of ß-tubulin-III-positive neurons was still unchanged (relative value after TNF treatment 99±3% compared to untreated, mean ±SEM).
|
|
Association between ß-tubulin-III, KIF5B, and JNK
Recently, a molecular interaction was observed between JNK, JNK-interacting protein 3 (JIP3), and kinesin-1 (23)
. Therefore, we performed Western blot analysis to study whether JNK forms a complex together with the motor protein kinesin heavy chain isoform KIF5B. Protein lysates were obtained from cultured hippocampal neurons either untreated or treated for 20 min with TNF. Treatment with TNF increased the amount of phosphorylated JNK in the cultured neuronal cells (Fig. 3
A, B). Then the proteins of the cultured neurons were immunoprecipitated with antibodies directed against ß-tubulin-III. Immunodetection after Western blot analysis demonstrated that ß-tubulin-III specific antibodies coimmunoprecipitated KIF5B together with JNK (Fig. 3C
). Thus, JNK formed a complex together with KIF5B and ß-tubulin-III. Treatment with TNF (20 ng/ml) for 20 min reduced the amount of KIF5B and JNK bound to ß-tubulin-III in immunoprecipitated lysates of cultured neurons (Fig. 3D
), suggesting that TNF affects the complex between ß-tubulin-III, KIF5B, and JNK.
|
Dissociation of KIF5B from ß-tubulin-III after TNF treatment
Since JNK, KIF5B, and ß-tubulin-III formed a complex and TNF treatment induced phosphorylation of JNK in neurites, we analyzed the effect of TNF on the molecular interaction between KIF5B and ß-tubulin-III at a subcellular level. Lifetime-based Förster resonance energy transfer (FRET) analysis was performed to allow subcellular localization of the molecular interaction between KIF5B and ß-tubulin-III. Cultured hippocampal neurons were double-immunolabeled with monoclonal antibodies directed against KIF5B marked with the fluorophore FITC (KIF5B-FITC) and monoclonal antibodies directed against ß-tubulin-III marked with the fluorophore Cy3 (ß-tubulin-III-Cy3). As a negative control, neurons were double-immunolabeled with monoclonal antibodies KIF5B followed by FITC and monoclonal antineurofilament, then Cy3 (neurofilament-Cy3). Furthermore, neurons were solely labeled with antibodies KIF5B-FITC to determine the regular lifetime as a control. First, confocal fluorescence intensity-based images showed staining with ß-tubulin-III-specific antibodies in all neurites and colocalization of KIF5B-FITC and ß-tubulin-III-Cy3 (Fig. 4
). As a control, confocal images of neurons double-labeled with neurofilament-Cy3 and KIF5B-FITC were collected. Then, fluorescence lifetime images were obtained from KIF5B-FITC by two-photon excitation and time correlated single-photon counting. An emission filter for KIF5B-FITC detection was placed between the output port of the scanning head and the time domain fluorescence lifetime imaging detector. Lifetimes from 1.0 to 1.9 ns were coded in red, representing expected signals from FRET (Fig. 5
). Lifetimes from 1.9 to 4.0 ns were coded in green, representing the expected lifetime of KIF5B-FITC in the absence of acceptor (Fig. 5)
.
|
|
Both the lifetime of KIF5B-FITC in the absence of acceptor or in the presence of an irrelevant acceptor (neurofilament-Cy3) showed a normal distribution with 2.2 ± 0.13 ns (mean±SD) and 2.25 ± 0.14 ns, indicative of a lack of FRET (Fig. 5D
and Fig. 6
). A strong FRET interaction was observed in untreated neurons colabeled with ß-tubulin-III-Cy3 and KIF5B-FITC (Fig. 5A
), showing a shifted distribution toward shorter lifetimes with a peak value of 1.5 ± 0.18 ns (Fig. 6)
. Pretreatment with TNF (20 ng/ml) for 20 min abolished the FRET interaction specifically in neurites but not cell bodies, allowing us to recover lifetimes similar to negative controls in neurites (Fig. 5B
).
|
To quantify the effect of TNF on the interaction between KIF5B-FITC and ß-tubulin-III-Cy3, we determined the average of the peak lifetimes collected from several regions of interest in neurites (Fig. 6B
). In detail, the averaged peak lifetime value in neurites was 2.3 ± 0.5 ns (mean±SD) after 20 min of TNF treatment, while in untreated neurites the averaged peak lifetime was 1.5 ± 0.18 ns (Fig. 6B
). Thus, treatment of hippocampal neurons with TNF changed the conformation of the KIF5B-tubulin complex, indicating that it broke apart the molecular interaction between KIF5B and ß-tubulin-III in axons (Fig 5B
). To determine whether the interaction between KIF5B-FITC and ß-tubulin-III-Cy3 in all neurons was affected by TNF, we analyzed the percentage of neurites displaying lifetimes lower than 1.9 ns. In detail, in untreated neurons 93 ± 27% (mean±SD) of neurites showed shorter lifetimes compatible with FRET, while in TNF-treated neurons, the percentage of neurites showing FRET between KIF5B-FITC and ß-tubulin-III-Cy3 was reduced to 4 ± 19% (Fig. 6C
). No FRET was observed in the controls by colabeling KIF5B-FITC and neurofilament-Cy3 or by imaging KIF5B-FITC in the absence of acceptors (Fig. 6C
).
Involvement of JNK in dissociation of KIF5B and ß-tubulin-III
The involvement of JNK activity in TNF-induced breakage of the molecular interaction between KIF5B and ß-tubulin-III was analyzed. Hippocampal neurons were treated with TNF for 20 min, and additionally, JNK was blocked with the JNK inhibitor SP600125 (100 nM). Cells were colabeled for KIF5B-FITC and ß-tubulin-III-Cy3. Both confocal fluorescence intensity and lifetime images were obtained. Additional treatment with the JNK inhibitor reduced the effect of TNF on the molecular interaction between KIF5B and ß-tubulin-III (Fig. 5C
and 6)
. In detail, after JNK inhibitor application, the mean lifetime value of KIF5B-FITC of TNF-treated neurons was 1.76 ± 0.15 ns (mean±SD) compared to 2.3 ± 0.5 ns in neurons treated with TNF alone (Fig. 6A and B
). After inhibition of JNK, low lifetimes were again detected in 60 ± 19% (mean±SD) of neurites of TNF plus JNK inhibitor-treated neurons, while reduced lifetimes were observed in 4 ± 19% in neurites treated with TNF alone (Fig. 6)
.
Inhibition of axonal mitochondrial transport by TNF via JNK
The motor protein KIF5B has been shown to be essential for axonal transport of mitochondria (24)
. To study whether TNF disturbs axonal transport of mitochondria, we transfected cultured hippocampal neurons with a mitochondrial targeting sequence tagged with enhanced YFP. To achieve low levels, this mitochondrial marker was expressed under a neuron-specific synapsin I promoter. The neurons showed expression of YFP in mitochondria starting from day 2 or 3 in culture and normal axonal morphology comparable to untransfected neurons. The YFP signal was specifically localized in structures typical of mitochondria (Fig. 7
A). Transfected cultures were placed in a heated chamber and scanned every 20 s by confocal microscopy. Mitochondria showed anterograde and retrograde movement in neurites (Supplemental Movie 1), which are illustrated in the corresponding kymographs (Fig. 7A
). To analyze the movement of mitochondria, we determined the number of moving mitochondria (mobile fraction) and their velocity. In total 20 ± 12% (mean±SD) of mitochondria moved anterogradely and 21 ± 17% retrogradely. After treatment with TNF for 20 min, mitochondria stopped moving. In total, the mobile fraction of mitochondria decreased from 41 ± 14% to 7 ± 4% after treatment with TNF for 20 min. Both anterograde and retrograde movements were substantially reduced to 3 ± 4% and 4 ± 4%, respectively (Fig. 7B
). No change in the velocity in the remaining mobile mitochondria was observed after TNF treatment. Mitochondria moved anterogradely with a velocity of 0.39 ± 0.15 µm/s (mean±SD) and retrogradely with a velocity of 0.39 ± 0.19 µm/s in untreated neurons. In neurons treated 20 min with TNF anterograde and retrograde velocity were 0.36 ± 0.17 µm/s and 0.35 ± 0.19 µm/s, respectively (Fig. 7C
). Inhibition of axonal transport by TNF was mediated via JNK. In detail, after cotreatment of neurons with TNF for 20 min and the generalized JNK inhibitor (SP600125; 100 nM), 45 ± 9% of mitochondria were mobile, very similar to the untreated situation, in which 41 ± 14% of mitochondria were mobile. Both anterograde and retrograde movements were regained in TNF-treated neurites after JNK inhibition to 23 ± 10% and 22 ± 12%, respectively (Fig. 7B
). The inhibitory effect of TNF on axonal movement of mitochondria was not long lasting. Already 360 min after treatment with TNF, neurites again showed an increased percentage of moving mitochondria (Fig. 7B
).
|
Inhibition of axonal synaptophysin transport by TNF via JNK
Recently, we demonstrated that axonal synaptophysin transport was impaired via JNK by NO (16)
. To study whether TNF inhibits synaptophysin transport in axons, we transfected cultured hippocampal neurons with synaptophysin tagged with green fluorescent protein (GFP) and performed fluorescence recovery after photobleaching (FRAP) analysis of a 20-µm-long axonal segment (Fig. 8
). The mobile fraction and the half-time value t of the fluorescence recovery curve were determined as characteristics of the movement. At 20 min and 30 min after treatment of neurons with TNF (20 ng/ml), a significant reduction in the axonal transport of synaptophysin-GFP was detected as determined by FRAP (Fig. 9
). In detail, the relative FRAP intensity at 300 s was reduced from 0.32 ± 0.14 (mean±SD) to 0.11 ± 0.04 after 20 min of treatment with TNF (Fig. 9)
. At 360 min after treatment with TNF, the value returned back to levels comparable to untreated neurons. Furthermore, there was a significant decrease in the mobile fraction of synaptophysin-GFP after treatment with TNF, indicating that TNF reduced the percentage of synaptophysin-GFP transported along the axon. The mobile fraction of synaptophysin-GFP decreased from 39 ± 14% (mean±SD) to 13 ± 4% after 20 min of treatment with TNF (Fig. 9)
. The half-time value t 1/2 of the FRAP, which is correlated to the velocity of synaptophysin-GFP, was unchanged and not modulated by TNF treatment (Fig. 9)
. Inhibition of JNK by the inhibitor SP600125 completely neutralized the effect of TNF on axonal movement of synaptophysin-GFP. In detail, blockade of JNK in TNF-treated axons resulted in a mobile fraction of 46 ± 21%, while axons treated with TNF alone showed a mobile fraction of synaptophysin-GFP of 13 ± 4% (Fig. 9)
.
|
|
DISCUSSION
The results allow several crucial insights into the mechanisms regulating axonal transport. First, we demonstrate that TNF breaks apart the molecular interaction between KIF5B and tubulin via JNK selectively in neurites but not cell bodies. Thus, JNK not only forms a complex with kinesin and JIP but also appears to directly modulate the function of kinesin. Second, our data show that TNF inhibits anterograde and retrograde axonal movement of mitochondria via JNK. Third, we demonstrate that TNF impairs axonal movement of synaptophysin via JNK. Thus, JNK-mediated effects on the kinesin-tubulin complex after TNF application has important functional consequences leading to impairment in axonal transport of mitochondria and synaptic vesicle precursors.
Recently, JNK and its scaffolding binding partner JIP have been implicated in anterograde axonal transport (25)
. Primarily, JNKs have been suggested to sense axonal injury and participate in the initiation of the axonal injury response (11)
. JNKs are rapidly activated within the axons following nerve injury and can be transported on microtubules along the axon via their association with motor proteins of the kinesin family (11)
. The JNK pathway would, therefore, be an attractive candidate as a sensor of the axonal injury response and regulator of axonal transport (25)
. Phosphorylated JNK was detected in neurites in Alzheimer’s disease associated with senile plaques (26
, 27)
. In the case of mechanical damage occurring on nerve injury, JNK was locally phosphorylated at the site of injury (28)
, and retrograde signals involving JNK support nerve survival and regeneration (29
, 30)
.
In this study, we concentrated on the axons of cultured hippocampal neurons and demonstrated that JNK activation has local axonal effects on the motor machinery in response to the proinflammatory cytokine TNF. Because the expression and up-regulation of TNF receptor I and the TNF-mediated activation of JNK is conserved in most cell types (17)
, we expect that our data obtained from cultured hippocampal neurons can be transferred to other neuronal cell types as well. JNKs are linked to the kinesin motor proteins via JIPs (9
, 11)
. The indirect linkage of motors, JNK, and cargo via the JIP scaffolding proteins would allow JNK to act as signal transducers to regulate motor activity. Principally, JNK kinases might phosphorylate either kinesin itself or an associated protein, thereby inactivating the motor and leading to dissociation of kinesin from tubulin. Alternatively, activity of JNK might be indirectly linked to the dissociation of the kinesin-tubulin complex. Western blot analysis of cultured neurons showed that JNK is associated with the kinesin-tubulin complex and that acute TNF treatment reduced the amount of kinesin and JNK bound to microtubules. Furthermore, we demonstrated that lifetime-based FRET analysis can be applied in primary neurons to analyze the interaction between tubulin and motor proteins on a subcellular level (31)
. The FRET data showed that TNF changed the conformation of the kinesin-tubulin complex, which was partially reverted by inhibition of JNK. Principally, the FRET data are consistent with either a change in the conformation of the kinesin-tubulin complex without detachment or a complete detachment of kinesin from the microtubules. However, data of axonal transport analysis with the characteristic increase in the immobile fraction without interference in the velocity of the remaining mobile particles indicate that the kinesin complex dissociates from the microtubules. Particularly, immobile kinesin still binding in an "off" conformation to tubulin would possibly act as an obstacle for the remaining mobile particles, leading to reduced velocity. As this is not the case, it is indicative that the path of moving mitochondria and synaptophysin is not obstructed at all and immobile mitochondria and synaptophysin-containing precursor vesicles are likely to dissociate from tubulin after JNK stimulation by TNF. Interestingly, mitochondria detached from the tubulin track, although relatively large in size compared to the axonal diameter, appear to be no obstacle per se since they were able to rapidly pass each other in the opposite direction without altering their velocity.
Previously, it was shown that TNF induced perinuclear clustering of mitochondria in fibroblasts via the TNF receptor I (32)
. It was suggested that TNF acted via hyperphosphorylation of kinesin light chain and inhibited the kinesin-mediated transport of mitochondria (33)
. Perinuclear clustering of mitochondria was also evident in mice with targeted disruption of the conventional kinesin heavy chain KIF5B (24)
, indicating that KIF5B is required for plus end-directed transport of mitochondria along microtubules. Our data now link mitochondrial movement and conformational change in KIF5B via TNF-induced JNK activity. Surprisingly, TNF also affected retrograde transport of mitochondria. Very recent evidence suggests that heterozygous mutations in APLIP1 (the drosophila JNK interacting protein-1) affects both anterograde and retrograde axonal transport of mitochondria (34)
. Like our findings, these data indicate that JNK and its associated scaffolding proteins (JIPs) are either directly or indirectly involved in retrograde transport of mitochondria, too. Another article from the Saxton laboratory described that mutations in kinesin heavy chain also impaired dynein-dependent retrograde axonal transport (35)
. Therefore, it might be possible that kinesin-1 acts as a biochemical or biophysical activator of dynein in axons (36)
.
Furthermore, we demonstrated that TNF-induced JNK activity inhibits synaptophysin transport. Synaptophysin is produced at the cell body, integrated into synaptic vesicle precursors, and then anterogradely transported along axons (37)
. Defects in axonal transport of synaptophysin-containing vesicle precursors have been observed in KIF1A mutant mice (38)
. Recently, JNK has been associated with axonal vesicular structures (28)
and JNK activity via NO impaired anterograde axonal transport of synaptic vesicle precursors containing synaptophysin and VAMP2 (16)
. Both signaling pathways, TNF and NO, converge at JNK activation followed by impairment of the axonal transport. Thus, the JNK signaling pathway appears to be widely involved in axonal transport of organelles and vesicular structures. Principally, TNF might stimulate neuronal NO synthase and release of NO, which then act on JNK. However, neuronal NO synthase is localized in the synapses and neuronal cell bodies but not in the axons (39)
, indicating that NO is most probably not involved in TNF-mediated effects on the axonal transport.
Axonal transport impairments are early signs of a variety of neurodegenerative and neuroinflammatory diseases. Several studies observed a close association between activated microglial cells and neuronal injury in Alzheimer’s disease (40)
, multiple sclerosis (41)
, microbial infection (3)
, and brain injury (42)
. Activated microglial cells produce substantial amounts of TNF and are highly motile under normal and disease conditions (43)
. In neuroinflammatory diseases, they are activated and migrate to sites of neuronal lesions, where they bind and ensheathe naked or demyelinated axons (42
, 44)
.
Recently, accumulations of the axonally transported amyloid precursor protein (APP) were transiently observed in T cell-mediated EAE in Lewis rats (45)
, indicating a reversible axonal transport impairment. Interestingly, the APP accumulations were correlated with the number of activated NO producing microglia and macrophages (45)
, suggesting that inflammatory mediators might act on axonal transport. Our data demonstrate that TNF, prevalent in most neuroinflammatory and neurodegenerative diseases, affects the axonal transport system, which is required to maintain function, integrity, and viability of axonal and synaptic structures. Thus, our findings show that the TNF-c-Jun NH2-terminal kinase signaling pathway impairs the kinesin-mediated axonal transport system and opens new avenues for therapeutic intervention.
ACKNOWLEDGMENTS
The Neuroimmunology Group at the European Neuroscience Institute Göttingen was supported by the University of Göttingen and the Hertie-Foundation (IMSF). This project was supported by the Deutsche Forschungsgemeinschaft and the European Union (LSHM-CT-2005–018637). The Neural Regeneration Group at the University Bonn Life and Brain Center is supported by the Hertie-Foundation and Walter-und-Ilse-Rose-Foundation. We thank Alexandra Bohl, Heiko RhöSE., and Christine Frank for excellent technical support of cultures and molecular biology. We thank Alessandro Esposito and Laura Swan for critically reading the manuscript and helpful comments.
FOOTNOTES
1 These authors equally contributed to this work. 
Received for publication June 19, 2006. Accepted for publication July 24, 2006.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  N. Hirokawa and Y. Noda Intracellular Transport and Kinesin Superfamily Proteins, KIFs: Structure, Function, and Dynamics Physiol Rev, July 1, 2008; 88(3): 1089 - 1118. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||
