Mitochondrial Membrane Potential‐dependent Endoplasmic Reticulum Fragmentation is an Important Step in Neuritic Degeneration

Summary

Background

Neuritic degeneration is an important early pathological step in neurodegeneration.

Aim

The purpose of this study was to explore the mechanisms connecting neuritic degeneration to the functional and morphological remodeling of endoplasmic reticulum (ER) and mitochondria.

Methods

Here, we set up neuritic degeneration models by neurite cutting‐induced neural degeneration in human‐induced pluripotent stem cell‐derived neurons.

Results

We found that neuritic ER becomes fragmented and forms complexes with mitochondria, which induces IP3R‐dependent mitochondrial Ca²⁺ elevation and dysfunction during neuritic degeneration. Furthermore, mitochondrial membrane potential is required for ER fragmentation and mitochondrial Ca²⁺ elevation during neuritic degeneration. Mechanically, tightening of the ER–mitochondria associations by expression of a short “synthetic linker” and ER Ca²⁺ releasing together could promote mitochondrial Ca²⁺ elevation, dysfunction, and reactive oxygen species generation.

Conclusion

Our study reveals a dynamic remodeling of the ER–mitochondria interface underlying neuritic degeneration.

Introduction

Neuritic degeneration is a common and early feature of many neurological conditions including disorders of the central nervous system such as multiple sclerosis, Alzheimer disease, and Parkinson disease 1, 2, 3, as well as peripheral nervous systems disorders such as Charcot–Marie–Tooth type 2A (CMT2A) and hereditary spastic paraplegias 4, 5. Degeneration can be triggered by divergent stimuli including mechanical, metabolic, infectious, toxic, hereditary, and inflammatory stresses 6.

Although the mechanisms underlying neuritic degeneration are incompletely understood, abnormalities in endoplasmic reticulum (ER) and mitochondrial functions have been implicated. Mitochondria were suggested to be a central sensor for neuritic degenerative stimuli 6. Specifically, the opening of mitochondrial permeability transition pore (mPTP) was shown to be a key event during neuritic degeneration 7, 8, 9. Systemic antioxidant treatment and upregulation of endogenous antioxidant mechanisms can protect neurites from degeneration, which indicates a role for reactive oxygen species (ROS) in neuritic degeneration 10. Meanwhile, Ca²⁺ release from neuritic ER stores was recently shown to be involved in degeneration of neurites 11. ER and mitochondria form junctions stabilized by protein tethers, including the Mmm1/Mdm10/Mdm12/Mdm34 complex in yeast and inositol 1, 4, 5‐trisphosphate receptor (IP3R)‐containing complexes, mitofusin 2 (MFN2), and likely other proteins in mammalian cells 12, 13, 14, 15. This ER–mitochondria coupling is required for local propagation of IP3R and ryanodine receptor (RyR)‐mediated Ca²⁺ signals to the mitochondria to control metabolism and mPTP opening and to shape cytoplasmic Ca²⁺ signals 15, 16. Imaging local Ca²⁺ dynamics at the ER–mitochondrial interface provided direct evidence for the existence of high Ca²⁺ microdomains between ER and mitochondria 17, 18. Recently, this ER–mitochondria coupling was reported to contribute to mitochondrial fission by ER‐associated inverted formin 2 (INF2), which induced actin assembly at ER–mitochondria contacts 19, 20.

In neurites, ER was reported to be continuous under resting conditions and to lose continuity in response to N‐methyl‐D‐aspartic acid (NMDA) receptor stimulation, potassium, or cardiac arrest 21, 22, 23, 24, 25. However, whether ER remodeling and ER–mitochondria coupling are involved in neuritic degeneration remains unknown. Here, we showed that in response to neuritic degeneration, ER becomes fragmented and tightly associates with mitochondria, in a mitochondrial membrane potential (ΔΨ_m)‐dependent manner. This tight coupling of ER and mitochondria, together with ER Ca²⁺ release, is required to promote mPTP opening and mitochondrial dysfunction for neuritic degeneration.

Materials and Methods

Ethical Statement

The experiments involving human subjects had been reviewed and approved by the Guangzhou Institutes of Biomedicine and Health Ethical Committee.

Plasmids

Mito‐GFP, Mito‐DsRed, ER‐DsRed, and OMM‐mRFP‐ER plasmids were gifts from Prof. György Hajnóczky and were subcloned into a lentiviral expression vector—pRlenti. pRlenti‐PM‐mRFP‐ER plasmid was constructed by attaching the N‐terminal palmitoylation/myristoylation signal sequence of the Lyn protein (MGCIKSKGKDSAGA) to N‐terminal sequence of mRFP‐ER. OMM‐GFP‐ER and PM‐GFP‐ER plasmids were constructed by replacing mRFP sequence with GFP sequence. pRlenti‐TgBFP was constructed by amplifying TgBFP from pSADdeltaG‐BFP (a gift from Edward Callaway, Addgene plasmid # 32639) through PCR and subcloned into pRlenti vector.

Cell Culture, Lentivirus Production, and Infection

HEK 293T cells and fibroblasts were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum and streptomycin (50 μg/mL) and penicillin (50 U/mL). All cultures were maintained in a humidified incubator containing 5% CO₂ at 37°C.

For lentivirus production, HEK 293T cells were plated onto 10‐cm dishes and cotransfected using the Ca²⁺ phosphate method with target plasmids and packaging vectors PMD2.G and PSPAX2. The viruses were harvested by centrifuging at 50,000 g for 2.5 h and then used to infect cells.

Induced Pluripotent Stem Cell (iPSC) Generation and Neuron Rosettes Induction

Normal human iPSCs 90D was reprogrammed from ATCC IMR90 cells 26. Human iPSCs were cultured in mTesR1 medium (Stem Cell). All iPSC colonies were transfered by EDTA onto Matrigel (Becton Dickinson, Los Angeles, CA, USA). Human iPSCs were differentiated to neuron rosettes according to the protocol as described previously 27. Briefly, iPSCs were derived for embryoid bodies (EB) formation, and after 4 days, the EB medium was changed to N2 medium for 3 days. Then, EBs were attached to Matrigel for another 7 days in N2 medium. At day 14, neural tube‐like rosettes were observed and then detached and cultured in suspension in N2B27 medium. Finally, these neuron rosettes formed neurospheres, which could be passaged for more than 10 passages in N2B27 medium.

Neuron Differentiation, Virus Infection, and Neuritic Degeneration Induction

For neuron differentiation, neurospheres were plated onto Matrigel‐coated glass coverslips at the density of 20–40 neurospheres per well on 6‐well plate and cultured in N2B27 medium, supplemented with 10 ng/mL BDNF (BioVision 4004) and 1 μM cAMP (Sigma A6885, St.Louis, MO, USA). After 4–6 days of differentiation, lentiviruses were used to infect neurons without the use of polybrene. Three days after infection, neurons derived from neurospheres (about 3.0 × 10⁶ neurons per well on 6‐well plate) were detected or cut near the soma with laser for axotomy by PALM MicroBeam system (Zeiss, Jena, Germany). For neurite ablation in vitro, we used a modified microdissection method for human neurons based on a previous report 28. In brief, neurons were cultured in Nunc^™ Glass Bottom Dishes (Thermo Scientific, Waltham, MA, USA). During the dissection procedure, neurons were imaged to live with a 20 × LD plan objective (N.A. = 0.4), the laser was then focused on the edge of the neurospheres at the proper power level (set the value of power range from 40 to 50 in software). As a result, the neurites separated from the soma. After cutting, the neurons were further cultured for detection.

For analysis, the colocalization of ER and mitochondria and neural ER and mitochondria is marked by ER‐DsRed and Mito‐GFP viruses, respectively. The fluorescence was detected and analyzed by Zeiss confocal microscope. For detection of the time course of mitochondria and ER remodeling, neurites were cut near the soma with laser as above and immediately detected under confocal microscope. The profiles of ER and mitochondria were analyzed by ZEN software.

Transmission Electron Microscopy (TEM) Analysis

Neurospheres were plated onto Matrigel‐coated glass coverslips to neural induction. Neurites were cut near the soma with laser, and after 4 h, cell slices were fixed in 1.5% paraformaldehyde and 1.5% glutaraldehyde in 0.1 M Sörensen phosphate buffer (0.1 M NaH₂PO₄/Na₂HPO₄, pH 7.2) for 1 h followed by washing 3 times in Sörensen phosphate buffer. Tissue was postfixed in 1% osmium tetroxide in Sörensen phosphate buffer for 1 h, washed 3 times, and dehydrated in ethanol with increasing concentration: 25, 50, 75, and 96% for 2 × 10 min, respectively, and 100% for 2 × 15 min. Prior to embedding, the slices were placed in 100% acetone for 2 × 20 min and then in a mixture of acetone and Epon resin poly/bed‐812 (1:1, Polysciences) overnight. Next day, the specimen was transferred to pure resin for at least 4 h before embedding in new pure resin and polymerization at 60°C for 48 h. The embedded specimen was sectioned in an ultratome (Super Nova, Reichert Jung) at 50 nm and mounted on slot copper grids previously covered with a thin film of pioloform. Grids were stained in 4% uranyl acetate for 30 min at 40°C and 0.5% lead citrate for 2 min at room temperature and observed with a Philips CM 10 electron microscope. Sections from 4 control and 4 cut treated slices were analyzed.

Immunofluorescence

Cells were fixed with 4% paraformaldehyde (pH 7.4) for 30 min at room temperature. The cells were incubated with primary antibodies in PBS containing 10% goat serum, 0.3% Triton X‐100 overnight at 4°C, washed, and then incubated with secondary antibodies for 1 h. Fluorescence images were acquired on Leica or Zeiss confocal microscope. The primary antibody was beta‐III TUBLINA (Novus MAB1195, 1:1000).

Assessment of Neuritic Beading, Fragmented Neurites, and ER Fragmentation

To assess neuritic beading and fragmented neurites, 90D neurons were cut and then cultured for 2–6 h. For the cutting experiments with 1 μM Xestospongin C (Xes) (Sigma X2628), 1 μM Ru360 (Merck 557440, Darmstadt, Germany), 1 μM Thapsigargin (TG) (Sigma T9033) or 1 μM Carbonylcyanide‐p‐trifluoromethoxyphenylhydrazone (FCCP) (Sigma C2920) and 1 μg/mL oligomycin A (Oligo) (Sigma 7535) (FCCP & Oligo) treatment, these drugs were added 30 min before cutting. For the TG treatment experiments with ER‐DsRed, OMM‐ER, or PM‐ER expression, neurons were infected with virus for linker expression and then treated with TG for 24 h. Neurons were observed under a phase‐contrast microscope by a modification of the method previously reported 29. More than 100 neurons in duplicate wells were assessed blindly in 3 independent trials. The ratio of neurites with neuritic beads or fragmented neurites was calculated as a percentage of total cells.

To assess neuritic ER fragmentation, ER‐DsRed is used to mark neuritic ER. More than 100 neurites in duplicate wells were assessed blindly in 4 independent trials. The ratio of neurons with fragmented neuritic ER was calculated as a percentage of total neurites.

Measurement of Mitochondrial Ca²⁺ and Intracellular Ca²⁺

Rhod‐2‐acetyl ester (Rhod2) (Invitrogen R‐1244, Carlsbad, CA, USA), an indicator of mitochondrial Ca²⁺, was used to measure mitochondrial Ca²⁺. Cells were first plated on glass coverslips and then incubated with 1 μM Rhod2 for 30 min at 37°C. Cells were further incubated for 30 min after being washed in indicator‐free medium. After loading Rhod2, cells were returned to growth conditions for an additional 18 h to eliminate the residual cytosolic fraction of the Rhod2 probe. Fluo‐4 acetoxymethyl (Fluo‐4) (Invitrogen F‐14201) was used to measure intracellular Ca²⁺. Cells were incubated with 1 μM Fluo‐4 for 30 min at 37°C and then washed twice with medium before imaging. The fluorescence was detected by Zeiss confocal microscope.

Mitochondrial Rhod2 fluorescent intensity and neuritic Ca²⁺ blebs were analyzed by a modification of the method previously reported 30, 31. Neuritic Ca²⁺ vacuoles with higher Ca²⁺ concentration than the rest of neurites were recognized as Ca²⁺ blebs in Fluo‐4 images, more than 100 neurites in duplicate wells were assessed blindly in more than two independent trials. The fraction of neurons with neuritic Ca²⁺ blebs was calculated as a percentage of total neurons.

Measurement of mPTP Opening, ΔΨ_m, and ROS Level

For measuring mPTP opening, cells were washed twice with Hank buffer, and then stained with 1 μg/μL calcein acetoxymethyl ester (calcein) (Invitrogen C3100MP) in the presence of 1 μM CoCl₂ for 25 min at 37°C and washed twice with Hank buffer before imaging.

For measuring ΔΨ_m, cells were stained with 50 nM 1,1′,3,3,3′,3′‐hexamethylindodicarbocyanine iodide (Dilc(5)) (Molecular Probes M34151) for 25 min in 37°C and then washed twice with serum‐free medium before imaging.

For measuring ROS level, cells were incubated with 1 μM carboxy‐DCFH‐DA (DCFH‐DA) (Beyotime S0033) in B27‐free neural basal media for 25 min at 37°C and washed out before imaging; more than 100 neurites in duplicate wells were assessed blindly in more than two independent trials.

The fluorescence was detected and analyzed by Zeiss confocal microscope.

Measurement of Mitochondrial Motility and 3D Reconstruction

Mito‐GFP was used to mark mitochondria. Neurons co‐expressing Mito‐GFP and ER‐DsRed/OMM‐mRFP‐ER/PM‐mRFP‐ER were transferred to Nunc^™ Glass Bottom Dishes and maintained at 37°C. Mitochondrial movements were recorded for 10 min at 4/s using Zeiss confocal microscope.

Image analysis was performed with ImageJ software developed by Wayne Rasband (NIH, Bethesda, MD, USA; http://rsb.info.nih.gov/ij/) extended with custom plugins. Calculations of mitochondrial transport parameters were analyzed from kymographs using the multiple kymograph plug‐in described previously 32. Briefly, this plug‐in calculates the velocity by measuring the distance between the position of individual mitochondria at the start and end of time‐lapse recordings and dividing by the time elapsed. This yields an overall velocity of transport that includes anterograde and retrograde movements and stationary periods; mitochondria were subsequently classified as motile (velocity >0.1 μm/s) or stationary (velocity <0.1 μm/s).

For confocal z‐axis stacks, fluorescence proteins (ER‐DsRed and Mito‐GFP) were used to mark neural ER and mitochondria, respectively. Fifteen images separated by 0.4 μm along the z‐axis were acquired. Three‐dimensional reconstruction and volume rendering of the stacks were performed with Imaries software (Bitplane).

Statistics

Data are shown as mean ± SEM. Statistical significance was determined by t‐test or two‐way ANOVA test.

Results

Neuritic ER Becomes Fragmented and Forms Complexes with Mitochondria During Neuritic Degeneration

To observe the ER and mitochondrial remodeling during neuritic degeneration, we set up neuritic degeneration models. As mouse neurons may not be the best model for explaining human neurodegeneration, we selected human neurons. Considering the difficulty of acquiring human primary neuron, here we take advantage of human iPSC‐derived neurons to study neurodegeneration. We differentiated human iPSCs 90D to neurons. During the process, neurites derived from neurospheres were cut near the soma with a laser beam (axotomy) and then further cultured for 6 h. Consistent with previous reports 29, we observed a much higher fraction of neurites with focal bead‐like swelling (0.90 ± 0.03 in cutting; 0.05 ± 0.01 in control), and fragmented neurites, which are markers for neuritic degeneration (Figures 1A,B and S1A–C). We then marked ER and mitochondria in iPSC‐derived neurons using ER‐DsRed and Mito‐GFP to observe the remodeling of the organelles during neuritic degeneration induced by cutting. In 2 h after cutting, neuritic ER became fragmented, compared with the continuous morphology in control (Figure 1C). Fluorescence profile analysis (Figure 1D) and 3D reconstruction (Figure 1E) both showed that the fragmented ER colocalizes with mitochondria. We further examined ultrastructure of ER and mitochondria by TEM. The contact between ER and mitochondria was much narrower in cutting neurites (14.27 ± 2.36 nm) than that in controls (38.71 ± 3.98 nm) (Figure 1F,G). We then analyzed time course of the remodeling of ER and mitochondria after cutting, and we found that ER became condensed and fragmented in 30 min and then fragmented ER tightly associated with mitochondria in 45 min, during which neurites did not become swollen (Figure 1H). Cellular contents such as TUBLIN have been reported to accumulate at beads in degenerated neurites 33. To exclude the possibility of fragmented ER simply resulting from this bead accumulation, we stained βIII TUBLIN in ER‐DsRed expressing neurons in 30 min after cutting and showed the fragmentation of neuritic ER is much earlier than the accumulation of cellular contents (Figure S1D). Collectively, we observed that ER becomes fragmented and colocalizes with mitochondria during neuritic degeneration, implicating the tightened ER–mitochondria association.

Figure 1.

Neuritic ER becomes fragmented and forms complex with mitochondria during neuritic degeneration. (A and B) Fraction of neurons with neuritic beads in 90D neurons after cutting for 6 h. The boxed areas are enlarged in the insets. The neurons with neuritic beads are indicated by arrows. ***, P < 0.001. Scale bars: 25 μm. (C and D) Distal neuritic ER becomes fragmented and colocalizes with mitochondria in 90D neurons after cutting for 2 h. 90D neurons were infected with ER‐DsRed and Mito‐GFP virus after differentiation. In the image panels, the top panel is a merge of the two lower panels. The left panels indicate before cut, and the right panels indicate 2 h after cut. Scale bars: 10 μm. (D) Analysis of fluorescence intensity performed on neurites presented in (C) using the ZEN software from Zeiss. Fluorescence intensity profiles of ER and mitochondria are plotted from soma to distal direction along neurites. (E) 3D reconstruction of neurite. ER‐DsRed and Mito‐GFP by confocal microscopy show their colocalization in 90D neurons after cutting for 2 h. Boxed areas are enlarged in the right panels. Scale bars: 50 μm. (F and G) TEM analysis of ER in neurites after cutting for 2 h. ER profiles are indicated by arrows. Scale bar: 500 nm. (G) Qualification of the distance of ER /mitochondria linkers in cut neurites and control neurites (n ≥ 9), ***, P < 0.001. (H) Time course of neuritic ER and mitochondria remodeling. 90D neurons were marked with ER‐DsRed and Mito‐GFP and cut by laser to induce degeneration. Boxed areas in the left magnification panel are enlarged middle. Scale bars: 10 μm.

Mitochondrial Ca²⁺ Elevation and Dysfunction During Neuritic Degeneration

To evaluate the effects of enhanced physical coupling between organelles on Ca²⁺ signaling during neuritic degeneration, we conducted imaging of mitochondrial Ca²⁺ by Rhod2. In 2 h after cutting, neuritic mitochondrial Ca²⁺ concentration significantly increased (Figure 2A,B). The distance of ER–mitochondria association decreased during neuritic degeneration, which provides possibility that the increase of mitochondrial Ca²⁺ concentration results from Ca²⁺ transfer from ER to mitochondria 15. To test this, Xes, an inhibitor of IP3R, which is present at ER–mitochondrial contacts and to mediate Ca²⁺ channeling to mitochondria 13, was used in our neuritic degeneration system. Rhod2 and Fluo‐4 costaining showed that the elevation of mitochondrial Ca²⁺ but not neuritic Ca²⁺ concentration during neuritic degeneration can be prevented by Xes (Figure 2C), which indicates the elevation of mitochondrial Ca²⁺ concentration is dependent on IP3R. Furthermore, inhibition of mitochondrial Ca²⁺ overload by Xes, Ru360 (blocking the mitochondrial uniporter) or TG (emptying ER Ca²⁺ into the cytoplasm before cutting) could inhibit neuritic degeneration, which was marked by neuritic beads formation or neuritic fragmentation (Figure S2A,B). In addition, Ca²⁺ blebs formation (where cytosolic Ca²⁺ concentration elevates), marker of neuritic degeneration 31, was also observed during neuritic degeneration. The regions of Ca²⁺ blebs in degenerated neurites colocalized with mitochondria or fragmented ER before neuritic swellings (Figure S3A–D). Thus, these data suggest an IP3R‐dependent mitochondrial Ca²⁺ overload resulting from the tightened ER–mitochondrial coupling during neuritic degeneration. It was reported that axotomy induces Ca²⁺ in the axon above normal range 34, and our analysis revealed the rules of Ca²⁺ signaling at subcellular level during neuritic degeneration.

Figure 2.

Mitochondrial Ca²⁺ elevation and dysfunction during neuritic degeneration. (A and B) Mitochondrial Ca²⁺ concentration increases during neuritic degeneration: (A) Images of 90D neurons stained with Rhod2 to monitor neurite mitochondrial Ca²⁺ concentration after cutting for 2 h and (B) Quantification of the Rhod2 fluorescent intensity. Fluorescent intensity of control was normalized to 1 (n ≥ 15), ***, P < 0.001. (C) The time course of the fluorescence intensity of Fluo‐4 or Rhod2 in neurites after cutting, the left panel indicates control group, the right panel indicates the group with Xes treatment. Fluorescence intensity of Fluo‐4 and Rhod2 at initiating time was normalized to 1, respectively. (D and E) mPTP opens during neuritic degeneration: (D) Images of 90D neurons after cutting for 2 h and then stained with calcein in the presence of CoCl₂. Scale bars: 10 μm. (E) Quantification of calcein fluorescence. Fluorescent intensity of control was normalized to 1 (n ≥ 10), ***, P < 0.001. (F and G) ΔΨ_m dissipation during neuritic degeneration: (F) The far‐red fluorescent dye Dilc(5) was used to monitor ΔΨ_m of 90D neurons after cutting for 2 h. Scale bars: 10 μm. (G) Quantification of the Dilc(5) fluorescent intensity. Fluorescent intensity of control was normalized to 1 (n ≥ 10), ***, P < 0.001. (H‐J) Mitochondrial motility arrests during neuritic degeneration. (H) Kymographs of neuritic mitochondria in 90D neurons after cutting for 2 h. The relative velocity of neuritic mitochondria is shown in (I) (mitochondrial velocity of control neurites was normalized to 1) and the fraction of stationary or mobile mitochondria of total neuritic mitochondria is shown in (J) (n ≥ 9), ***, P < 0.001.

As mitochondrial Ca²⁺ overload causes ΔΨ_m dissipation 35, and indeed, the inhibition of mPTP opening is reported to prevent axon degeneration 7, we then tested these parameters during neuritic degeneration. We performed calcein assay in the presence of CoCl₂ to monitor mPTP state, a procedure by which the cytosolic calcein signal was quenched by Co²⁺. The results show that mPTP is opening during cutting‐induced neuritic degeneration, compared with a closed mPTP state in control (Figure 2D,E). We then used the far‐red fluorescent dye Dilc(5) to monitor ΔΨ_m and showed the ΔΨ_m dissipation (cutting 0.44 ± 0.04, control normalized to 1) during neuritic degeneration (Figure 2F,G).

Mitochondrial motility is important for neuritic energetics and signal regulation 36, 37, 38, 39, 40, 41. We then quantified mitochondrial motility and showed mitochondrial motility arrests during cutting‐induced neuritic degeneration (Figure 2H–J). Together, our data strongly suggest that mitochondrial Ca²⁺ overload by the tightened ER–mitochondrial coupling and consequent mPTP opening are involved in neuritic degeneration.

ΔΨ_m is Required for Neuritic Beading, ER Fragmentation, and Mitochondrial Ca²⁺ Elevation during Neuritic Degeneration

Recently, ER–mitochondria contacts and ER‐associated formin INF2 were reported to play an active role in mitochondrial fission 19, 20. We then asked whether ER morphology is modulated by ER–mitochondria coupling or mitochondria. We hypothesized that ΔΨ_m is involved in ER fragmentation as it is required to drive mitochondrial Ca²⁺ uptake 42. To test this, we used uncouplers FCCP & Oligo to dissipate ΔΨ_m in neurites treated with cutting. Firstly, we analyzed neuritic beads formation and neuritic fragmentation during neuritic degeneration. Uncoupling alone did not induce any neuritic beads formation and neuritic fragmentation. The fraction of neurons with neuritic beads after cutting was 0.90 ± 0.03, whereas it decreased to 0.43 ± 0.04 with FCCP & Oligo. FCCP & Oligo also significantly prevented the fragmentation of neurites (Figures 3A,B and S4). These results show that uncoupling significantly inhibits neuritic beads formation and neuritic fragmentation during neuritic degeneration.

Figure 3.

ΔΨ_m is required for neuritic beading, ER fragmentation, and mitochondrial Ca²⁺ elevation during neuritic degeneration. (A and B) FCCP & Oligo prevents neuritic beads formation during neuritic degeneration: (A) Images of 90D neurons after cutting for 6 h simultaneously with or without FCCP & Oligo. The boxed areas are enlarged in the insets, the neurons with neuritic beads are indicated by arrow. Scale bars: 25 μm. (B) Fraction of neurons with neuritic beads, ***, P < 0.001. (C and D) FCCP & Oligo prevents neuritic ER fragmentation during neuritic degeneration. Images of ER‐DsRed in 90D neurons after cutting for 2 h (C) simultaneously with or without FCCP & Oligo treatment. The neurites with fragmented ER are indicated by arrow. Scale bars: 10 μm. Fraction of neurons with fragmented ER is shown in (D), ***, P < 0.001. (E and F) FCCP & Oligo prevents mitochondrial Ca²⁺ elevation during neuritic degeneration. Images of 90D neurons expressing Mito‐GFP stained with Rhod2 to monitor neuritic mitochondrial Ca²⁺ concentration after cutting for 2 h (E) simultaneously with or without FCCP & Oligo. Scale bars: 10 μm. Quantification of the Rhod2 fluorescent intensity is shown in (F). Fluorescent intensity of untreated neurons was normalized to 1 (n ≥ 15), ***, P < 0.001.

Then, we detected ER morphology during neuritic degeneration induced by cutting with or without FCCP & Oligo treatment. Uncoupling alone did not affect ER continuity in normal neurites. The fraction of neurons with neuritic fragmented ER after cutting was 0.88 ± 0.03, whereas it decreased to 0.42 ± 0.03 with FCCP & Oligo (Figure 3C,D). These results indicate that uncoupling prevents neuritic ER fragmentation during neuritic degeneration.

To verify whether uncoupler prevents mitochondrial Ca²⁺ elevation during neuritic degeneration, we used Rhod2 staining to measure mitochondrial Ca²⁺ concentration during neuritic degeneration induced by cutting with or without FCCP & Oligo treatment. Uncoupling did not affect mitochondrial Ca²⁺ concentration in normal neurites and dramatically prevented the Ca²⁺ elevation during cutting neuritic degeneration (Figure 3E,F). These data indicate that ΔΨ_m is required for neuritic beading, ER fragmentation and mitochondrial Ca²⁺ elevation during neuritic degeneration.

ER–Mitochondrial Linkage and ER Ca²⁺ Release Together Promote Neuritic Degeneration

We have observed ΔΨ_m‐dependent ER fragmentation, tight mitochondrial association and consequent mitochondrial dysfunction during neuritic degeneration. Then, we asked whether tightening the ER–mitochondria coupling could promote neuritic degeneration. To answer this question, we used a synthetic linker encoding fluorescence protein (FP) between the outer mitochondrial membrane (OMM) targeting sequence of mAKAP1 and the ER targeting sequence of yUBC6 (OMM‐FP‐ER), which decreases the ER–mitochondria distance from 24 ± 3 to 6 ± 1 nm 15. As a control, the above construct was also prepared with FP fused to the plasma membrane targeting sequence of the Lyn protein and the ER targeting sequence of yUBC6 (PM‐FP‐ER) (Figure 4A). Cells expressing OMM‐mRFP‐ER but not PM‐mRFP‐ER displayed colocalization of ER and mitochondria (Figure S5).

Figure 4.

ER–mitochondrial linkage and ER Ca²⁺ release together increase neuritic mitochondrial Ca²⁺, elevate mPTP opening, and reduce ΔΨ_m. (A) The structures of OMM‐FP‐ER linker and PM‐FP‐ER linker. OMM‐ER linker tethers mitochondria to ER and PM‐ER linker links plasma membrane and ER. (B and C) Mitochondrial Ca²⁺ concentration in 90D neurons expressing Mito‐GFP, OMM‐GFP‐ER linker or PM‐GFP‐ER linker with and without TG treatment. (B) 90D neurons were stained with Rhod2 to monitor neuritic mitochondrial Ca²⁺ concentration. Scale bars: 10 μm. Quantification of the Rhod2 relative fluorescent intensity is shown in (C). Fluorescent intensity of neurons expressing ER‐DsRed without TG treatment was normalized to 1 (n ≥ 21), **, P < 0.01, ***, P < 0.001. (D and E) mPTP opening in 90D neurons expressing ER‐DsRed, OMM‐mRFP‐ER linker or PM‐mRFP‐ER linker with and without TG treatment. (D) 90D neurons were stained with calcein to monitor mPTP in the presence of CoCl₂. Only OMM‐ER linker expressing neurons with TG treatment show elevated mPTP opening. Scale bars: 10 μm. (E) Quantification of calcein fluorescence intensity in neurites. Fluorescence intensity of neurons expressing ER‐DsRed without TG treatment was normalized to 1 (n ≥ 29), ***, P < 0.001. (F and G) ΔΨ_m in 90D neurons expressing OMM‐ER linker or PM‐ER linker with and without TG treatment. (F) 90D neurons were stained with Dilc(5) to monitor neurite ΔΨ_m. Only OMM‐ER linker expressing neurons with TG treatment show decreased ΔΨ_m. Scale bars: 10 μm. (G) Quantification of Dilc(5) fluorescence intensity in neurites. Fluorescence intensity of neurons expressing ER‐DsRed without TG treatment was normalized to 1 (n ≥ 14), **, P < 0.01, ***, P < 0.001.

To evaluate the effect of the enhanced or decreased ER–mitochondria coupling on Ca²⁺ transport, we conducted imaging of mitochondrial Ca²⁺ by Rhod2. Rhod2 staining showed that neither OMM‐GFP‐ER nor PM‐GFP‐ER expression affects the mitochondrial Ca²⁺. However, in the presence of TG to mobilize ER Ca²⁺ pool, the neuritic mitochondrial Ca²⁺ concentration significantly increased with OMM‐GFP‐ER expression compared with PM‐GFP‐ER or Mito‐GFP expression (Figure 4B,C). These results show that tightened ER–mitochondria association enables mitochondrial uptake during ER Ca²⁺ release.

To verify mitochondrial dysfunction in those conditions, we detected the mPTP opening state and ΔΨ_m by the same methods described above. Our results showed that neither OMM‐mRFP‐ER nor PM‐mRFP‐ER expression affects the mPTP opening state and ΔΨ_m. However, in the presence of TG, the calcein fluorescence, ΔΨ_m, with OMM‐mRFP‐ER expression all significantly decreased compared with PM‐mRFP‐ER or ER‐DsRed expression (Figure 4D–G). These data indicate that ER–mitochondrial linkage and ER Ca²⁺ release are required in tandem to promote mitochondrial dysfunction for neuritic degeneration.

Then, we detected mitochondrial motility and found that neither ER/mitochondria nor ER/plasma membrane tethering affects mitochondrial motility. This suggests that only ER associations with other organelles do not account for the defect of mitochondrial transport in neural degeneration. However, in the presence of TG, both the mitochondrial relative velocity and mobile faction with OMM‐mRFP‐ER expression significantly decreased compared with PM‐mRFP‐ER or ER‐DsRed expression (Figure 5A–C). These data indicate that the mitochondrial transport defect during neuritic degeneration requires ER Ca²⁺ releasing to the closely associated mitochondria. In addition, as ROS has been reported to be a key and early step in neural degeneration 10, 43, 44, 45, we detected ROS and found ROS level significantly increases in neurites expressing OMM‐mRFP‐ER treated with TG, which is consistent with ROS enhancement in neural degeneration. (Figure 5D,E)

Figure 5.

ER–mitochondrial linkage and ER Ca²⁺ release together promote mitochondrial motility arrest, ROS production, and neuritic beading. (A‐C) Mitochondrial motility in 90D neurons expressing OMM‐ER linker or PM‐ER linker with and without TG treatment: (A) Representative kymograph of neuritic mitochondria in a 90D neuron. Only OMM‐ER linker expressing neurons with TG treatment show mitochondrial motility arrest. The relative velocity of neurite mitochondria is shown in (B) (mitochondrial velocity of neurites expressing ER‐DsRed without TG treatment normalized to 1). The fraction of stationary or mobile mitochondria of total neuritic mitochondria is shown in (C) (n ≥ 9), *, P < 0.05, **, P < 0.01, ***, P < 0.001. (D and E) ROS level in 90D neurons expressing OMM‐ER linker or PM‐ER linker with and without TG treatment. (D) 90D neurons were stained with DCFH‐DA to monitor neural ROS level. Only OMM‐ER linker expressing neurons with TG treatment show increased ROS levels. Scale bars: 10 μm. Quantification of the DCFH‐DA fluorescent intensity is shown in (E). Fluorescence intensity of neurons expressing ER‐DsRed without TG treatment was normalized to 1, *, P < 0.05, **, P < 0.01. (F and G) Neuritic bead formation in 90D neurons expressing OMM‐ER linker or PM‐ER linker with and without TG treatment. (F) The neurites of 90D neurons expressing OMM‐ER, PM‐ER, or ER‐DsRed with or without TG treatment, Scale bars: 25 μm. (G) Fraction of neurons with neuritic beads, ***, P < 0.001.

Finally, we detected neuritic degeneration in those conditions. The ratio of neurites with beads or fragmented among neurites in OMM‐mRFP‐ER expressing neurons increased significantly upon TG treatment (Figures 5F,G and S6A). Similarly, in this condition, the fraction of neurons with neuritic Ca²⁺ blebs was much higher (Figure S6B,C). Next, we used this linker expression system in cutting‐induced neuritic degeneration. The expression of OMM‐mRFP‐ER but not ER‐DsRed or PM‐mRFP‐ER can accelerate beads formation and neuritic fragmentation (Figure S7A,B). Collectively, structural ER‐mitochondrial close tether and ER Ca²⁺ release are both required for triggering neuritic degeneration. The tightened connection between ER and mitochondria itself has no effect on neuritic degeneration until ER Ca²⁺ releasing occurs based on this structure.

Discussion

The scheme in Figure 6 illustrates the novel aspects of ER–mitochondria remodeling uncovered in the present work. The association between ER and mitochondria is due to the presence of tethers that link ER to mitochondria. In response to neuritic degeneration, ER becomes fragmented and tightly associates with mitochondria to form complexes, indicating a novel dynamic regulation of this interorganellar junction. Interestingly, this ER–mitochondria remodeling is dependent on ΔΨ_m. As a result of this remodeling, tightening of the ER–mitochondria association sensitizes mitochondria to Ca²⁺ overload, leading to mPTP opening and ROS production. These results reveal an unexpected dependence of neuritic degeneration on ΔΨ_m‐dependent ER fragmentation and mitochondria linkage.

Figure 6.

Model of neuritic ER fragmentation in neuritic degeneration.

Among the many studies of the intimate relationship between ER and mitochondria, this is the first to report a mitochondria dependent regulation of ER morphology. Recently, ER–mitochondria contacts were reported to play an active role in mitochondrial fission. ER tubules have been shown to wrap around mitochondria to define the sites of mitochondrial fission 19. ER‐localized INF2 induces actin filaments at ER–mitochondria contact drives initial mitochondrial constriction, which allows Drp1‐driven secondary constriction 20. Furthermore, in yeast, ER‐associated mitochondrial division links the distribution of mitochondria and mitochondrial DNA 46. In that report, ER‐associated mitochondrial fission occurred at sites where newly replicated mitochondrial DNA nucleoids were segregated, thereby distributing nucleoids to the tips of newly formed mitochondria. All these studies suggest the existence of a manifold complex regulating of mitochondrial morphology and structure at the ER interface. A fundamental question is whether this structure affects ER morphology. As the morphology of the ER regulates mitochondrial fission and, thereby, bioenergetics, so our data indicate regulation in the opposite direction—an impact of mitochondrial polarization on ER morphology, namely ΔΨ_m‐dependent fragmentation. The molecular mechanisms of this signaling remain to be elucidated in later studies.

Our “synthetic linker” results demonstrate that this structural remodeling of tight ER–mitochondria associations should facilitate local Ca²⁺ coupling to promote neuritic degeneration. Although the role of Ca²⁺ in neuritic degeneration has been long accepted, the spatiotemporal control of local Ca²⁺ remains unknown. Sustained increased intracellular Ca²⁺ levels are associated with the onset of degradative cascades leading to neuritic degeneration 47, 48. On the other hand, chelation of extracellular or depletion of intracellular Ca²⁺ protects against neuritic degeneration 49, 50. Recently, the release of ER Ca²⁺ stores was reported to be a critical step by activating the mPTP leading to neuritic degeneration 11. Here, we set up a “synthetic linker” expression system to enforce or eliminate ER–mitochondria linkage. We show that neither tightening of the ER–mitochondria associations nor ER Ca²⁺ release alone could activate mPTP opening and mitochondrial dysfunction, a crucial event for neuritic degeneration 7, but in tandem, they could. Thus, for initiating neuritic degeneration, functional Ca²⁺ release from ER is not enough; ER is required to tightly associate with mitochondria to strengthen the local Ca²⁺ coupling for the mPTP opening.

Conflict of Interest

The authors declared no conflict of interest.

Supporting information

Figure S1. Neuritic swell and fragmentation appears in neuritic degeneration induced by cutting.

Figure S2. Xes, Ru360 or TG prevents neuritic beads formation and fragmentation during neuritic degeneration.

Figure S3. Neuritic Ca²⁺ blebs formation in degenerated neurites.

Figure S4. ΔΨ_m is required for neuritic fragmentation during neuritic degeneration.

Figure S5. Localization of OMM‐ER or PM‐ER linker in HEK 293T cells. Scale bars: 10 μm.

Figure S6. Neuritic fragmentation and neuritic Ca²⁺ blebs in 90D neurons expressing ER‐DsRed, OMM‐ER linker or PM‐ER linker with or without TG treatment.

Figure S7. Fraction of neurites with neuritic beads (A) or fragmented neurites (B) after cutting.