Introduction
Understanding the transmission of an infectious agent from one cell to another was a challenge of the last century. The involvement of cell-surface receptors has been shown, but other routes have also been described. Tunneling nanotubes (TNTs) form one such path. TNTs have been described in various cell types, including neuronal and immune cells. They are filamentous-actin-containing membranous structures with a diameter of 50 to 800 nm, not always linked to the substrate, and forming bridges that connect remote cells [
1‐
6]. For instance, TNTs physically connect T cells, presenting a new pathway for HIV-1 transmission [
7]. In such cells, the tip of the TNT is an active zone of actin cytoskeleton reorganization and contains ezrin, Exo70, myosin 10 and N-WASP, suggesting a regulation at the cellular level [
8,
9]. Extrinsic factors such as arachidonic acid in endothelial cells [
10], HIV-1 infection in macrophages [
11], oxidative stress [
12] and prion-like proteins (e.g., Huntingtin fibrils, TDP-43) in neuronal cells [
6,
13,
14] have been shown to trigger TNT formation.
Many protein aggregates have prion-like properties: they can act as self-propagating templates. They disrupt cellular proteostasis, eventually leading to neurodegenerative disorders such as Alzheimer’s disease (AD), Parkinson’s disease (PD), amyotrophic lateral sclerosis (ALS), or transmissible spongiform encephalopathies (TSEs) [
15‐
17]. The exact mechanisms of the cell-to-cell spreading of pathological species are still subject to intense investigation. Among others, the role of TNTs in such propagation has been suggested in Huntington’s disease, Parkinson’s disease and ALS/fronto-temporal dementia [
18]. Regarding Alzheimer’s disease, the amyloid Aβ peptide has been shown to traffic through TNTs and to induce cytotoxicity [
12]. The role of TNTs in aggregated Tau spreading has not yet been documented.
In the present work, using two different cellular models (CAD neuronal cells and rat primary embryonic cortical neurons), we demonstrate that extracellular Tau species acts as an extrinsic factor leading to increased formation of TNTs, which in turn facilitate the intercellular spread of pathological Tau.
Materials and methods
Ethics statement- Animals were provided by Janvier Laboratories and had access to food and water ad libitum. Animal experiments were performed in compliance with and with the approval of the local ethics committee (agreement CEEA 062010R), standards for the care and use of laboratory animals, and the French and European Community guidelines.
Cell culture
Primary Embryonic Neuronal Culture- Rat primary embryonic cortical neurons (primary neurons) were prepared from 17–18-day-old Wistar rat embryos as follows. The brain and meninges were removed. The cortex was dissected out and mechanically dissociated in culture medium by trituration with a polished Pasteur pipette. Once dissociated and after blue trypan counting, cells were plated in Ibidi μ-Dishes (Biovalley) or Lab-Tek four-well chamber slides (Becton Dickinson) coated with poly-D-lysine (0.5 mg/mL) and laminin (10 μg/ml). For dissociation, plating, and maintenance, we used Neurobasal medium supplemented with 2 % B27 and containing 200 mM glutamine and 1 % antibiotic-antimycotic agent (Invitrogen). Primary neurons at 7 days in vitro (DIV7) were infected with lentiviral vectors (LVs) encoding GFP/mCherry actin, tubulin or human wild type Tau (hTau1N4R containing a V5 tag; V5-hTau1N4R).
Cell lines- Mouse neuronal CAD cells (mouse catecholaminergic neuronal cell line, Cath.a-differentiated) were cultured in Opti-MEM (Invitrogen) with 10 % fetal bovine serum, penicillin/streptomycin (1 %) and L-glutamine (1 %). Neuronal CAD cells were plated overnight in poly-D-lysine (0.5 mg/mL) coated Ibidi μ-Dishes for live imaging or Lab-Tek four-well chamber slides for immunostaining. Neuronal CAD cells were infected with LVs encoding GFP-actin, mCherry-tubulin or human wild-type Tau (hTau1N4R containing a V5 tag; V5-hTau1N4R).
Viral vectors- The procedures to produce the lentiviral vectors (LVs) and to control their viral titers and the absence of competent retroviruses have been described previously [
19]. All viral batches were produced in appropriate areas in compliance with institutional protocols for genetically modified organisms according to the “Comité Scientifique du Haut Conseil des Biotechnologies” (Identification Number 1285).
Antibodies- As part of this work, various primary antibodies were used: mouse anti-α acetylated Tubulin (Sigma; 1:200 for immunocytochemistry); rabbit polyclonal antibody to V5 (Merck Millipore; 1:10,000 for immunocytochemistry); rabbit polyclonal antibody against the C-terminal part of Tau (C-ter, raised in-house; 1:800 for immunocytochemistry and 1:10,000 for biochemistry) [
20]; rabbit polyclonal antibody M19G, which recognizes the N-terminal part of Tau (N-ter, raised in-house; 1:10,000 for biochemistry) [
21]; and rabbit polyclonal raised against human anti-myosin 10, which detects myosin 10 from multiple species, including mouse and rat (Sigma; 1:200 for immunocytochemistry). These antibodies were visualized using appropriated secondary antibodies coupled to Alexa 488 or 647 (Life Technologies; 1:1000 for Alexa 488 and 1:500 for Alexa 647).
Immunofluorescence- Neuronal CAD cells and primary neurons were washed with pre-warmed PBS, fixed with 4 % paraformaldehyde (PFA) for 20 min at room temperature, permeabilized with 0.2 % Triton X-100 for 20 min at room temperature and blocked for 45 min at room temperature using blocking solution (Bovine Serum Albumin (BSA) 2 % in PBS). Cells were then incubated overnight at 4 °C with primary antibody diluted in blocking solution before being carefully washed and incubated for 30 min at room temperature with the appropriate Alexa Fluor-conjugated secondary antibody. Cells were washed and mounted with VectaShield/4′,6-diamidino-2-phenylindole (DAPI, Vector Laboratories) to label nuclei.
Tubulin tracker staining for live imaging- Neuronal CAD cells were plated overnight at 100,000 per 35 mm glass-bottomed culture μ-dish (Biovalley, France), washed with pre-warmed PBS and incubated 30 min at 37 °C with tubulin Tracker green (Life Technology; 1:1000 dilution) diluted in HBSS buffer and rinsed 3 times with pre-warmed PBS before imaging.
Fluorescence imaging- Immunofluorescence and short time-lapse acquisitions were performed using an inverted confocal microscope (LSM 710, Zeiss, Jena, Germany) with a 40× oil-immersion lens (NA 1.3 with an optical resolution of 176 nm) with an optical resolution of X nm. DAPI, Alexa 488/GFP, mCherry and Alexa 647 were imaged using UV, Argon 488 nm, DPSS 561 nm and Helium/Neon lasers 633 nm. Images were processed with ZEN software. Long time-lapse acquisitions were performed using an inverted Yokogawa Spinning Disk confocal microscope with a 63× oil-immersion lens (NA 1.4 with with an optical resolution of 164 nm) and epifluorescence microscope (Eclipse Ti-E, Nikon, Tokyo, Japan) with a 40× air-immersion lens (NA 0.9 with an optical resolution of 338 nm). Images were processed with ZEN software and NIS software. To reduce noise, the signal was subjected to line averaging to integrate the signal collected over four lines. The confocal pinhole was adjusted to facilitate a minimum field depth. A focal plane was collected for each specimen. Cells were maintained at 37 °C and 5 % CO
2 during real time acquisitions. All setups, using similar illumination and recording conditions (detector frequency, gain, and laser intensity), were applied to non-treated primary neurons to avoid misinterpretation due to non-specific labeling (Additional file
1: Figure S1 d-e).
Human Tau1N4R purification
Full-length human Tau1N4R cDNA was cloned in the pET14b vector. hTau1N4R was expressed in
E. coli BL21 DE3 CodonPlus cells (Stratagene). Cells were grown in LB medium to an optical density at 600 nm of 0.8 absorbance units. hTau1N4R expression was induced with 0.5 mM IPTG for 3 h. The cells were then harvested by centrifugation (4000 g, 10 min). The bacterial pellets were resuspended in lysis buffer (20 mM MES pH 6.8, 500 mM NaCl, 1 mM EGTA, 0.2 mM MgCl
2, 5 mM dithiothreitol, 1 mM PMSF + 1 tablet of Complete (Roche)) per liter and lysed by sonication. Cell extracts were clarified by centrifugation at 14,000 g, 30 min. The lysate was heated to 80 °C for 20 min and centrifuged at 14,000 g for 30 min. The supernatant was dialyzed against 100 volumes of buffer A (20 mM MES pH 6.8, 50 mM NaCl, 1 mM EDTA, 1 mM MgCl
2, 2 mM DTT, 0.1 mM PMSF) at 4 °C. The dialyzed protein mixture was loaded on an SP Sepharose column (60 ml bed volume). Proteins were separated with a linear gradient of 0 to 100 % buffer B (20 mM MES pH 6.8, 1 M NaCl, 1 mM EGTA, 1 mM MgCl
2, 2 mM DTT, 0.1 mM PMSF). Fractions were analyzed via SDS-PAGE stained with Coomassie blue. Fractions containing hTau were pooled and dialyzed against 100 volumes of PBS buffer containing 1 mM DTT. The hTau concentration was determined spectrophotometrically using an extinction coefficient at 280 nm of 7450 M
−1.cm
−1. Pure hTau1N4R at a concentration of 50 to 100 μM. Fibrillar samples were sonicated for 5 min on ice in 2-ml Eppendorf tubes in a VialTweeter powered by an ultrasonic processor UIS250v (250 W, 2 4 kHz; Hielscher Ultrasonic, Teltow, Germany) set at 75 % amplitude, 0.5 s pulses. Aliquots were stored at −80 °C. Length distribution of sonicated Tau fibrils has been obtained by measuring the length of 227 fibrils in negatively stained TEM samples (15 to 85 nm with a mode size of 55 nm) (Additional file
2: Figure S2).
Human Tau1N4R assembly and labeling
Fibrillation of hTau1N4R was achieved at 40 μM in the presence of 10 μM heparin by shaking 0.5 ml solution aliquots at 37 °C in an Eppendorf Thermomixer set at 600 rpm for 4 days. Fibrils were spun for 20 min at 20 °C and 16,000 rpm. The amount of fibrillar material was estimated by subtraction of the soluble fraction remaining after centrifugation from the initial concentration. The pelleted material was resuspended in PBS at an equivalent monomeric hTau1N4R concentration of 100 μM.
Labeling of hTau1N4R fibrils was achieved by the addition of 2 molar equivalents of lysine-reactive ATTO 488, ATTO 568 or ATTO 647 (Life Technologies #A20003) for 1 h at room temperature. The unreacted fluorophore was removed by two cycles of centrifugation at 15,000 g for 10 min and resuspension of the pellet in PBS.
Sup35NM purification and assembly
Purification and assembly of Sup35NM were performed as described in Krzewska et al. [
22]. Labeling of Sup35NM fibrils was performed identically to that of Tau fibrils.
Electron microscopy
The nature of hTau1N4R and Sup35NM assemblies was assessed using a JEOL 1400 transmission electron microscope following adsorption onto carbon-coated 200-mesh grids and negative staining with 1 % uranyl acetate. The images were recorded with a Gatan Orius CCD camera (Gatan).
TNTs activation- For TNT activation experiments, neuronal CAD cells and primary neurons were incubated for 5 min at 37 °C with 1 μM recombinant hTau1N4R fibrils labeled with ATTO 647 or unlabeled. Rinses with pre-warmed PBS were performed (3×) before immunostaining or real-time imaging.
Antibody saturation- Before incubation with cells, C-ter antibodies were incubated (24 h/4 °C/orbital agitation) with blocking solution containing either saturating concentrations of recombinant hTau1N4R fibrils (100×) or BSA (100×) as a control. Cells were immunolabeled as described above.
Uptake and transfer of hTau1N4R fibrils- Neuronal CAD cells were plated overnight at 60,000 cells per well in poly-D-lysine-coated Lab-Tek four-well chamber slides (Becton Dickinson). Cells were infected with LVs encoding mCherry-Actin. ATTO 488-hTau1N4R fibrils were diluted at 1 μM in 100 μL of OptiMEM (Gibco). Then, 96 μL of OptiMEM and 4 μL of Lipofectamine-2000 (Invitrogen) were added to the ATTO 488-hTau1N4R fibrils to a final volume of 200 μL for 20 min before the mixture was added to the cells. For primary neurons, 50,000 cells were plated in poly-D-lysine- and laminin-coated Lab-Tek four-well chamber slides (Becton Dickinson). Cells were infected at DIV7 with LVs encoding mCherry-Actin. ATTO 488-hTau1N4R fibrils were diluted at 1 μM in 100 μL of Neurobasal medium (Gibco). Then, 96 μL of Neurobasal medium and 4 μL of Lipofectamine-2000 (Invitrogen) were added to the ATTO 488-hTau1N4R fibrils to a final volume of 200 μL for 20 min before the mixture was added to the cells. Six hours later, cells were rinsed with pre-warmed medium (3×) before immunostaining.
For neuron-to-neuron transfer, CAD cells or primary neurons (100,000 cells/dish) were plated on Ibidi μ-Dishes and infected with LVs encoding GFP-Actin. ATTO 568-hTau1N4R fibrils were diluted at 1 μM and added to the cells. Six hours later, cells were wash with pre-warmed PBS (3×) before acquisition using real time microscopy.
Electrophoresis and immunoblotting- Neuronal CAD cells were rinsed once in PBS and lysed in RIPA buffer (150 mM NaCl, 1 % NP40, 0.5 % sodium deoxycholate, 0.1 % SDS, and 50 mM Tris HCl; pH = 8.0). The positive controls were wild-type mouse hippocampal cell homogenate (CTL1) or neuronal CAD cells infected with LVs encoding hTau1N4R (CTL2). Protein concentrations were determined (PIERCE BCA Protein Assay Kit), and samples were diluted at 1 μg/μL in LDS containing 50 mM DTT. Then, 15 μg of protein was denatured at 100 °C for 10 min, loaded on 4-12 % NuPAGE Novex gels (Invitrogen), and transferred to nitrocellulose membranes. Membranes were blocked in Tris-buffered saline, pH 8.0, 0.05 % Tween 20 with 5 % skim milk or 5 % BSA and incubated with the appropriate primary overnight at 4 °C. Membranes were then rinsed and further incubated with horseradish peroxidase-labeled secondary antibody (goat anti-rabbit IgGs, Sigma), and bands were visualized by chemiluminescence (ECL, Amersham Biosciences) with a LAS3000 imaging system (Fujifilm).
Statistical analysis- Data are presented as the means (± SEM) of experiments performed at least in triplicate and are representative of the results obtained from three independent experiments that produced similar results. Statistical analyses were performed using the Mann–Whitney U-Test (GraphPad prism software) to determine the p-value. The differences were considered significant at * p < 0.05, ** p < 0.01, or *** p < 0.001.
Discussion
Tau protein, a microtubule-associated protein first described in 1975, has long been regarded as a mono-functional intracellular protein that modulates neuronal microtubule dynamics via complex regulation of its phosphorylation state [
42‐
44]. The role of this protein is widely documented in neurodegenerative diseases collectively known as tauopathies [
45]. These diseases, including AD, are characterized by intracellular accumulation of fibrillar material and neuronal loss. Through a set of neuroanatomical and biochemical studies, a spatio-temporal progression of Tau aggregation has been identified, thereby defining pathological stages as reported for example in AD [
46‐
48].
There is now a growing body of evidence that Tau is a multifunctional protein involved in various pathophysiological processes that were previously unsuspected. These new functions related to specific locations (nucleus [
49‐
52], plasma membrane [
53,
54], synapses [
55,
56]…) of Tau protein may help in the design of innovative therapeutic approaches. In addition, Tau has been identified as physiologically secreted into the extracellular space [
57,
58]. Moreover, the presence of Tau in cerebrospinal fluid under pathological conditions has been known for many years [
59], and in stress conditions, it has recently been identified in extracellular compartments such as cell culture media [
60,
61] or interstitial fluids [
62]. These data, combined with the existence of a hierarchical progression of Tau pathology, recently led to the hypothesis that Tau assemblies have prion-like properties. Tau assemblies propagate from affected neuronal cells to naïve neuronal cells [
63‐
65]. To traffic between cells, they must either be secreted naked or be encapsulated into vesicles. Naked Tau assemblies can be secreted into the extracellular environment by exocytosis and taken up by naïve cells by endocytosis. Evidence for the secretion of Tau species within extracellular vesicles e.g., exosomes (~30–100 nm), which originate from the fusion of multivesicular bodies with the plasma membrane, and ectosomes (~100–1000 nm), which form directly from the plasma membrane, has been reported, but the contribution of these systems to the spread of disease remains to be confirmed and defined [
65‐
68].
Previous studies have reported the presence of misfolded protein assemblies associated with disease within TNTs directly connecting the cytoplasm of distant cells, from immune to neuronal cells [
1]. Indeed, in addition to allowing the long-range intercellular transfer of cytoplasmic macromolecules, plasma membrane components, vesicles and organelles, TNTs have been shown to allow the transfer of pathogens such as HIV, of infectious prion particles, and of huntingtin-Exon1 aggregates from affected donor to naive recipient cells [
7,
8,
13].
Here, we demonstrate that extracellular Tau species (monomers and fibrils) activate the formation of TNTs (Figs.
2 and
3) that subsequently facilitate fibrillar Tau transfer from neuron to neuron (Figs.
4,
5 and
6). Extracellular Tau is found in stress and pathological conditions [
59‐
62] and may likely act as a signal for TNT formation. More interestingly, TNTs facilitate the transfer of Tau assemblies between neurons. While these assemblies are fibrillar in our setup, the Tau species involved in inter-neuronal propagation in vivo remain unclear and subject to debate [
69]. Both oligomeric and fibrillar forms are found during the course of Tau pathology. Thus, one can speculate that both forms are involved in the process, albeit with different kinetics and functions. Nonetheless, our data clearly demonstrate that fibrillar Tau transfer from neuron to neuron through TNTs.
Our finding that Tau is present in TNTs is also of great importance. Indeed, TNTs are ill-defined, and various types of TNT-like structures have been described [
26]. Identification and characterization of the types of tubular bridges that connect different cells (e.g., TNTs, filopodia) requires for specific markers. Our data demonstrate that Tau, together with actin, is a specific constitutive marker of TNTs. Our data further suggest that Tau may contribute to TNT formation and function, thus allowing a better characterization and understanding of these highly dynamic structures. This is in agreement with recent data showing that Tau organizes the actin networks. Indeed, in the absence of Tau, only single actin filaments could be observed, whereas in the presence of Tau, they progressively formed long and thick F-actin bundles [
70,
71]. This actin-binding property of Tau, in combination with data emerging from this study, strongly suggests that Tau might be transported into TNTs via actin. However, Tau overexpression itself is not sufficient to induce TNT formation. Whether myosin 10, also found in neuronal TNTs and acting as an actin motor protein, is required remains to be elucidated. Identifying cellular mechanisms supporting in vivo pathological Tau transfer through TNTs will help us to define therapeutic targets. In fact, the role of TNTs in a pathological context such as inflammation has already been shown between widely spaced dendritic cells in the adult mouse cornea [
28,
72]). Nevertheless, even if in different in vitro and in vivo experimental models [
6,
7,
13,
14,
72], there is now a strong evidence for the role of TNTs in different stress-induced pathological processes including the transport of many pathological proteins and pathogens, it has never been addressed in vivo in the brain and remains to be investigated.
Acknowledgments
We thank Dr. Isabelle Arnal for her critical reading of the manuscript and BiCel and IMAGIF facilities for access to microscopes. This work was supported by grants from ANR (SPREADTAU), Association France Alzheimer, Region Hauts de France StartAIRR grant (TiNTs) and from the program Investissement d’avenir LabEx (laboratory excellence) DISTALZ (Development of Innovative Strategies for a Transdisciplinary approach to ALZheimer’s disease). Our laboratories are also supported by LiCEND (Lille Centre of Excellence in Neurodegenerative Disorders), CNRS, Inserm, Métropole Européenne de Lille, Univ. Lille 2, FEDER, DN2M and the Fondation Bettencourt Schueller. The authors declare no conflict of interest regarding the present work.