1. Introduction
Mitochondria are cellular organelles crucial for intracellular energy production, calcium homeostasis, apoptosis and various other processes [
1]. In order to meet these variable demands, they adapt to cellular requirements, which is possible through different mitochondrial dynamics processes. Two of these processes are fission and fusion. Mitochondrial fission is important for mitochondrial turnover and transport. The division occurs following pre-constriction at ER–mitochondrial contact sites and depends on the recruitment of Drp1 by various adaptor proteins, which form a ring structure around the mitochondria and result in the separation of the membranes [
2,
3,
4]. Mitochondrial fusion allows for the exchange of mitochondrial content and is the result of subsequent outer and inner membrane fusion, executed by Mfn1/2 and OPA1, respectively [
5,
6]. The balance between these processes is crucial to maintain a healthy mitochondrial pool.
Research into mitochondrial dynamics relies on imaging techniques to visualize mitochondrial morphology, which can provide insight into the structural alterations occurring from changes in mitochondrial dynamics processes. Fluorescent probes are frequently used to visualize mitochondrial morphology, for instance through dyes like MitoTracker [
7]. A commonly applied approach to study mitochondrial dynamics is the direct assessment of mitochondrial morphology changes, often achieved through the quantification of changes in mitochondrial shape between various conditions [
6,
8,
9,
10]. Alternatively, photoactivable or photoconvertable fluorescent proteins can be used to track the diffusion of signals across mitochondria as a measure of fusion and fission events [
11,
12,
13]. Several factors are important for the reliable assessment of mitochondrial morphology, including the application of unbiased imaging and utilization of an accurate quantification method [
14]. Unlike the imaging of fixed cells, live imaging allows the observation of changes in real time and the visualization/tracking of mitochondrial motility, fission and fusion events [
15].
While live cell imaging is the preferred approach to study mitochondria, there are still significant challenges. For instance, fluorescent live cell imaging could induce ROS production in cells, which is a major cause of phototoxicity [
16]. Mitochondrial dynamics are specifically affected by phototoxicity [
17,
18], and this should be taken into consideration when analyzing mitochondrial morphology with live cell imaging. In this study, we describe a cell type-specific pearl-necklace-shaped mitochondrial phenotype, induced in response to the laser excitation of red mitochondrial dyes during live cell imaging. Our data contributes new information highlighting the importance of considering experimental conditions for live cell imaging to prevent imaging artifacts that can have a major impact on the obtained results. Furthermore, we applied serial electron tomography to show the effect of phototoxicity in high resolution and show the benefit of this approach in understanding structural changes in mitochondria.
2. Materials and Methods
2.1. Cell Culture
The cell lines were obtained from other scientists at Maastricht University. Normal human dermal fibroblasts (nHDFs, Lonza, Basel, Switzerland, CC-2511), c0388 [
19], c0407 [
19], c2244 [
19], HeLa cells (ATCC, Manassas, VA, USA, CCL-2) and HepG2 cells (ATCC, HB-8065) were cultured in DMEM medium (Gibco, Waltham, MA, USA, 41966029) supplemented with 10% fetal bovine serum (FBS, Takara, San Jose, CA, USA, 631106) and 0.1% PenStrep and maintained at 37 °C in a humidified incubator with 5% CO2. For HUVEC cells (Corning, Corning, NY, USA, 354151) EGM-2 endothelial cell growth medium (Lonza, CC-3162) was used and cells were kept under the same conditions as mentioned before. Mesoangioblasts (mabs, M10 [
20]) were cultured in IMDM (Gibco, 12440053) supplemented with sodium pyruvate, glutamine, non-essential amino acids, insulin transferase selenium X, 0.2% 2-mercaptoethanol, 10% FBS, 5 ng/mL FGF2 and 0.1% gentamycin. These cells were incubated at 37 °C in a humidified incubator with 5% CO
2 and 4% O
2. Cells were split at 70% confluency and were seeded 24 h before transfection onto 8-well µ-Slides (Ibidi, Gräfelfing, Germany, 80826), with densities of 1000 (fibroblasts and mabs) or 5000 cells (HeLa, HepG2 and HUVEC) per well. For starvation experiments, cells were placed on DMEM medium without glucose and FBS for 2 h to induce mild starvation. Extreme starvation was achieved by incubating the cells without growth medium on HBSS for 2 h prior to imaging.
2.2. Transient Knockdown
An esiRNA-mediated knockdown was created for DNM1L (Drp1), MFF, MIEF1, MIEF2, FIS1 and a non-targeting control (GFP). T7 promotor sequences containing primers were used to amplify around 600 bp of the respective transcripts with standard PCR amplification using BIOTAQ DNA Polymerase (Meridian Bioscience, Cincinnati, OH, USA, BIO-21040) and cDNA as a template. cDNA was made with the qScript cDNA Supermix kit (Quantabio, 95048-500), with 400 ng RNA as the input. dsRNA was synthesized from the PCR products using the MEGAscript Kit (Ambion, Waltham, MA, USA, AM1334) according to manufacturer protocol. For esiRNA production, dsRNA was incubated with Shortcut RNase III (NEB, Ipswich, MA, USA, M0245S) according to manufacturer instructions, with 0.75 units of enzyme per 1 µg dsRNA. To induce knockdown, cells were transfected with esiRNAs. In short, esiRNAs were combined with RNAiMAX (Invitrogen, Waltham, MA, USA, 13778075) in OptiMEM medium (Gibco, Waltham, MA, USA, 11058021) with a concentration of 50 ng esiRNA/mL culture medium. Further experiments were performed 72 h after transfection.
2.3. Knockdown Efficiency
nHDF cells were seeded in a 6-well plate with a density of 10,000 cells per well and transfected with esiRNAs. Following 72 h of incubation, RNA was isolated with the RNeasy Mini Kit (Qiagen, Venlo, The Netherlands, 74104) and the RNAse-Free DNase Set (Qiagen, 79254) according to manufacturer instructions. cDNA was synthesized using the qScript cDNA Supermix kit (Quantabio, Beverly, MA, USA, 95048-500), with 400 ng RNA as input. qPCR was performed using 1x SensiMixTM SYBR® (Bioline Meridian, Memphis, TN, USA, QT615-05 2× stock) and a final primer concentration of 0.5 μM on the LightCycler® 480 II (Roche, Basel, Switzerland, 0501524300). The knockdown efficiency was calculated using the ΔΔCt between the knockdown and control samples, using TBP for normalization. The significance of the relative differences was determined using an independent-sample T-test (R version 4.1.3).
2.4. Live Staining and Fixation
Following 72 h of knockdown, mitochondria were incubated with the respective culture medium containing MitoTracker Red FM (300 nM, Invitrogen, M22425), MitoTracker Red CMXRos (50 nM, Invitrogen, M7512), MitoTracker Green FM (300 nM, Invitrogen, M7514) or TMRM (100 nM, Invitrogen, T668) for 30 min at 37 °C. Cells were washed three times in culture medium and were placed in culture medium without phenol red (Gibco, 21063029), to decrease background signal during imaging. For experiments performed on fixed cells, the cells were stained with 50 nM MitoTracker Red CMXRos for 30 min at 37 °C, followed by washing with culture medium and fixation using 3.7% paraformaldehyde for 30 min. After fixation, the cells were washed in PBS before imaging.
2.5. Microscopy
Unless mentioned otherwise, imaging was performed with a Leica TCS SPE confocal system (DMI4000B microscope, Leica LAS-AF software, v. 2.6.3.8173) with a PMT detector and a 63× oil objective (NA 1.3). Excitation lasers with wavelengths of 488 and 561 nm were used to visualize the mitochondria. Images were made with 30% laser power, and single-plane images were acquired. For time-lapsed images, an image of the same plane was made every three seconds and cells were continuously exposed to the respective laser. For the microscope comparison, a FEI CorrSight, equipped with both a widefield and a spinning disk module was used. Here, a 40× air objective (NA 0.9) and an excitation bandwidth of 490/20 nm or 572/35 nm, and an emission DAPI/FITC/TxRed Tripleband HC Filter Set was used for the widefield module. For the spinning disk module, excitation lasers with wavelengths of 488 and 561 nm were used. FEI MAPS software (v. 3.7), in combination with Live Acquisition (LA) software (v. 2.7.0.16), was used to control the microscope and capture single-plane time-lapsed images, with an exposure time of 50 ms (widefield module) or 200 ms (spinning disk module).
For the phenotype quantification, between 70 and 200 cells were randomly selected and assessed using a Leica DMI4000B microscope. For each cell, the presence/absence of the pearl phenotype was determined. When the pearl-necklace phenotype was only present in one to two mitochondrial extremities, the cell was classified as containing an ‘intermediate’ phenotype. On the other hand, when the pearl-necklace phenotype was present in more than two extremities within one cell, this cell was classified as ‘yes’ for containing the pearl-necklace phenotype, and a complete absence was classified as ‘no’. Differences between gene knockdowns were determined with an independent sample T-test (R version 4.1.3).
2.6. Electron Tomography
For electron tomography, the pearl-necklace phenotype was induced in nHDFs cultured on a µ-Slide with a grid (Ibidi, 80826-G500). This was performed with an FEI CorrSight, using the widefield module and a 20× objective (NA 0.8). Cells were exposed to a 561 nm laser for one minute and the presence of the pearl-necklace phenotype was confirmed. To prevent the reversal of the phenotype, the cells were fixed with 2.5% Glutaraldehyde in 0.1 M phosphate buffer for 24 h at 4 °C. Then, the cells were washed with 0.1 M cacodylate buffer and postfixed with 1% osmium tetroxide in the same buffer containing 1.5% potassium ferricyanide for 1 h in the dark at 4 °C. Samples were dehydrated in ethanol, infiltrated with Epon resin for 2 days, embedded in the same resin and polymerized at 60 °C for 48 h. Ultrathin serial sections with a thickness of 150 nm were cut using a Leica Ultracut UCT ultramicrotome (Leica Microsystems Vienna) and mounted on Formvar-coated copper slot grids. Before staining with 2% uranyl acetate in water and lead citrate, 10 nm BSA-gold fiducials were adsorbed to the grid.
Overview images of all sections were made with a FEI Tecnai G2 Spirit BioTWIN electron microscope at 2900× magnification using the EMmesh software [
21]. Mitochondria that presented the pearl-necklace phenotype, both on EM and the corresponding fluorescent images, and were present in multiple of the serial sections were selected for tomography. For the selected positions, 4800x magnified images were made, before single-axis tilt series were recorded with a FEI Tecnai G2 Spirit BioTWIN. For each tilt series, 121 images were recorded between −65 and 65 degrees, with 1 degree increments. Images were recorded with an Eagle 4 k × 4 k CCD camera at binning 2 (Thermo Fisher Scientific, Waltham, MA, USA) and a 6800× magnification, resulting in a pixel size of 3.25 nm. Tilt series were reconstructed, tomograms from serial sections were aligned with IMOD and 3D reconstructions of pearl-necklace phenotype mitochondria were segmented and visualized with AMIRA.
4. Discussion
The visualization of mitochondria through various live cell imaging techniques is crucial in mitochondrial dynamics research, providing insight into the structural alterations that occur due to changes in mitochondrial fission, fusion and transport. In this study, we observed a fibroblast-specific pearl-necklace (PN)-like phenotype, which was not observed in other cell types including HeLa cells, HUVEC cells and mesoangioblasts. This phenotype appears irrespective of the fission factor knocked down or the genetic background of the fibroblast. The pearl-necklace-like phenotype was previously observed in fibroblasts obtained from patients with mutations in fission factors
DNM1L (Drp1) and
MFF imaged with confocal microscopy; however, the origin was not investigated in these studies [
23,
24]. Starvation experiments were conducted to reveal whether the PN phenotype was the result of increased fusion or decreased fission. Mild starvation increases mitochondrial fusion through HDAC6-dependent deacetylation of Mfn1 [
25]. Extreme starvation on the other hand results in a hyperfused network by inhibiting the phosphorylation of Drp1, which prevents recruitment to the mitochondria and thus mimics Drp1 depletion [
25]. Interestingly, mild starvation resulted in a hyperfused network but no PN phenotype, while extreme starvation did result in the PN phenotype. These results indicate that the PN phenotype is caused by decreased fission activity and not the result of increased fusion.
Strikingly, the PN phenotype can only be observed in live cells, but not in fixed cells, and is reversible over time when light exposure is stopped, indicating that it is a dynamic adaptation related to the imaging process. Furthermore, we observed that the time it takes for the mitochondrial network to form the PN phenotype differs between various microscopes. The PN phenotype was most severe when using LSCM and appeared only after 20 s using widefield microscopy and SDCM. LSCM generally uses a high laser intensity to visualize structures, and it is a relatively slow imaging technique. The signal is received pixel for pixel by the use of a pinhole, increasing the signal-to-noise ratio and allowing the imaging of a specific cell layer. However, the consequence of this technique is that the cell is exposed for a longer time to the fluorescent laser at a high intensity [
26]. Contrastingly, widefield microscopy does not use a pinhole system and results in images with more background signal, especially in thicker samples. Contrastingly, the imaging is significantly faster as the field of view does not need to be scanned pixel by pixel [
26]. While the total laser exposure on the cells is similar to an LSCM, the exposure is shorter and less focused on a single cell layer, decreasing the exposure per mitochondrion. Spinning disk confocal microscopy (SDCM), has the advantage of an increased signal-to-noise ratio due to the use of pinholes, while simultaneously being a fast imaging method through the use of multiple spinning disks containing pinholes [
26]. This results in a decrease in required exposure time compared to LSCM. The correlation between the level of laser exposure and the severity of the PN phenotype indicates that it is a dynamic adaptation of the mitochondrial morphology over imaging time, caused by light exposure.
The main cause of the negative cellular effect of laser exposure during live cell imaging is phototoxicity. Advanced phototoxicity is visible on a whole-cell level and is characterized by changes in cell morphology, cell motility and cell death [
16,
27]. While these cellular changes are generally used to determine phototoxicity, more subtle changes can be observed earlier, which also significantly affect the cell. These changes include a slowed down cell cycle and changes in cytoplasmic calcium, as well as effects on mitochondrial membrane potential and dynamics [
16,
17]. Considering the direct link between laser exposure and PN-phenotype severity, the onset is most likely the result of phototoxicity. The main cause of phototoxicity is the production of reactive oxygen species (ROS) as the result of a reaction between fluorophores and oxygen following laser exposure [
16,
27]. Generally, there are not sufficient ROS scavengers in the cells to handle this additional ROS production [
16], resulting in prolonged exposure to the produced ROS, which induces DNA and mitochondrial damage and causes cellular stress [
28]. Under physiological circumstances, mitochondria undergo increased amounts of fusion under light stress to increase OXPHOS capacity and lower the load of damaged compartments [
28]. Furthermore, mitochondria undergo fission to split off damaged mitochondrial parts, which can subsequently be recycled through mitophagy, to decrease the level of damage [
28]. Since mitochondrial fission is substantially decreased in the Drp1 knockdown cells, this damage mitigation is largely prevented. The buildup of mitochondrial damage eventually results in apoptosis, and the observed phenotype might be a cellular response initiated to prevent apoptosis. For instance, the constriction sites observed in the PN phenotype are possibly caused by the initiation of fission at many sites simultaneously in order to increase the chances of a successful fission event. In case the laser exposure is halted, the additional production of ROS is ceased and the present scavengers can remove the access build up, allowing the cells to recover, which is in line with the recovery of the PN phenotype following the termination of laser exposure as observed in this study. Aside from causing damage, ROS are also able to modulate calcium signaling [
29], which is an important factor related to the constriction of the inner mitochondrial membrane [
30,
31]. Moreover, ROS have been shown to result in the acidification of cells [
32], which can temporarily cause mitochondrial elongation [
33]. However, the potential role of calcium imbalance or acidification on the pearl-necklace phenotype needs to be determined.
Phototoxicity as the result of ROS production is generally expected to be increased when exciting cells with a shorter wavelength, due to the high energy levels [
34]. If this were solely the case for the PN phenotype, it should result in an increase in the phenotype when cells are exposed to a 490 nm light source compared to a 572 nm light source, while the opposite was observed. In fact, the formation of the PN phenotype does not appear to be wavelength-dependent. If wavelength were a determining factor, cells stained with MitoTracker Green FM and excited a 572 nm light should result in the PN phenotype, while cells stained with MitoTracker Red FM and excited with 490 nm light should not. Instead, the PN phenotype is dye-dependent, with dyes that have excitation peaks in the orange/red spectrum presenting with the phenotype (
Figure 4). Furthermore, the level of excitation of these dyes is a factor in the extent to which the PN phenotype is formed. Red dyes excited with 490 nm light results in the phenotype to a lesser extent than 572 nm, likely through the less-efficient excitation of red dyes at 490 nm. Aside from the excitation wavelength, a difference between the dyes used is their dependence on the mitochondrial membrane potential (MMP) for import into the mitochondria. Where MitoTracker red FM and TMRM are highly dependent on the MMP, MitoTracker CMX-ROS is only partially dependent, and the import of MitoTracker Green FM is the least dependent on the MMP [
35]. Previous research has shown distinct phototoxicity in cells stained with red MitoTrackers, which was absent upon staining with various green dyes (Rhodamine123, MitoTracker Green, JC-1) [
36,
37]. Both Rhodamine123 and JC-1 are also highly dependent on the MMP [
35], suggesting that the difference between red and green/orange dyes is not the result of MMP dependency. Additionally, the damage in this previous study was less severe in the presence of supplemented antioxidants, indicating that the cause of the damage was indeed ROS production [
37]. This suggests that there might be a chemical reaction resulting in ROS production that is specific to red dyes. MitoTrackers have been suggested to artificially alter the mitochondria during imaging. For example, MitoTracker Orange has been noted to inhibit Complex I, and similar effects have been observed for MitoTracker Red and Deep Red [
35]. Additionally, TMRM has been suggested to have an inhibitory effect on the electron transport chain. While changes to mitochondrial morphology have not been described, changes in mitochondrial function are known to result in morphological changes. This study clearly shows morphological changes as the result of MitoTracker presence, which can result in the misinterpretation of data.
Serial electron tomography was performed to identify the effect of the PN phenotype on the high-resolution mitochondrial structure. We showed that the application of correlative fluorescent and electron microscopy, in combination with serial electron tomography is a feasible technique, even for a phenotype which quickly reverses to the original state. The application of this approach resulted in the observation that the double mitochondrial membrane remained intact in the PN phenotype. While there are reports that the inner membrane can divide separately from the outer membrane, an event that often precedes fission, this does not occur in the PN phenotype [
31,
38]. Cristae structure also appears to be altered in the most expanded PN phenotype regions, where the cristae are not present in the mitochondrial center. This is in line with a recent study that observed the stacking of cristae along the mitochondrial edges, with a hollow space in the middle, as the result of phototoxicity [
18]. Additionally, the presence of ER and filaments along the constricted mitochondrial regions indicate that the constrictions are the result of ER-mediated mitochondrial pre-constriction, which normally happens before mitochondrial fission [
3,
38]. This would be in line with the previously mentioned hypothesis that fission is initiated at many sites simultaneously in these cells to mitigate the damaging effect of the induced phototoxicity. However, due to Drp1 depletion, fission cannot properly take place, resulting in the observed PN structures.
While phototoxicity occurs in all cell types, the presence of the PN phenotype is restricted to fibroblasts. While the reason behind the cell type specificity remains unknown, it could be caused by a difference in sensitivity to phototoxicity of the cell types tested. Although this could imply that other cell types can present the PN phenotype under more severe phototoxic conditions, we did not observe the phenotype even after extensive exposure of HeLa cells, to the extent that the mitochondria were no longer visible due to photobleaching. Alternatively, fibroblasts might have a specific mechanism to regulate mitochondrial stress in order to postpone the induction of apoptosis. Since dermal fibroblasts are often exposed to outside stressors, such as UV, additional stress-regulating mechanisms might be in place. In this case, the fibroblast-specific expression of proteins could elucidate a potential mechanism.
