Multi-color live-cell STED nanoscopy of mitochondria with a gentle inner membrane stain

Significance Mitochondrial nanoscopic imaging has developed from proof-of-principle demonstrations to a viable approach for structural and functional research. The next technological leap requires a palette of orthogonal dyes to label various mitochondrial components and unveil their interactive dynamics in 3D space and extended time, at sub-100 nm resolution. We report a mitochondrial inner membrane fluorescent marker, PK Mito Orange (PKMO), featuring markedly reduced phototoxicity for time-lapse imaging, and compatibility with not only commercial stimulated emission depletion (STED) nanoscopes, but also green and far-red fluorophores such as the genetically encoded calcium indicator (GCaMP) and (silicon rhodamine) SiR. Gentle PKMO labeling enables 3D reconstruction of crista structures in live mitochondria, analyzing crista morphology in live primary cells and genetically edited cell lines, and multiplexed recording of mitochondrial dynamics and interactions.


Transfection of cells
For expression of fusion proteins, HeLa cells were transfected 24-48 h prior to imaging using jetPRIME ® transfection reagent (Polyplus-transfection SA, Illkirch-Graffenstaden, France). Transfection was carried out according to the manufacturer protocols using 1-2 µg plasmid DNA. COS-7 cells were transfected 48 h prior to imaging using lipofectamine ® 3000 reagent (Thermo Fisher Scientific). Transfection was carried out according to the manufacturer's protocols using 2.5 µg plasmid DNA.

Viability-based phototoxicity assay of HeLa cells
HeLa cells were grown to 80-90 % confluency in a 96-well plate (costar 3599; Corning, NY, USA) before treatment. To achieve the same brightness after staining, HeLa cells were incubated with 1000 nM PKMO and 650 nM PKMO 0.9 respectively in DMEM (Gibco 11965092; Thermo Fisher Scientific) for 10 min at 37 °C with 5% CO2. Then, the cells were washed with PBS three times and maintained in fresh medium for subsequent imaging analysis. 96-well plates were analyzed using a high-content imaging system ImageXpress Micro XLS (Molecular Devices, San Jose, CA, USA) equipped with a 20 ×/0.4 NA air objective and live-cell imaging device. The average light intensity on cells was measured to be 2.6 W/cm 2 (568 nm). The cells in each well were irradiated with the maximum light intensity at different time points. Time points were selected as 5, 7.5, 9, 12.5, 20, and 26.7 min (each time point with three parallel repeats). After illumination, the cells were directly used for subsequent imaging analysis. The cells were washed with PBS once and a 100 µL cell viability assay solution containing 1 µM Calcein AM (C2012-0.1ml; Beyotime, Shanghai, China) was added to each sample including controls (without dye incubation). Images of two channels (mitochondria: Cy3 channel, 568 nm; Calcein AM: FITC channel, 488 nm) were recorded on ImageXpress Micro XLS. The cell viability (%) was calculated according to the following equation: cell viability % = B / A *100%, where A = the number of total cells before illumination, and B = the number of Calcein AM-positive cells after illumination. More than 500 cells were counted at each time point.
Comparison of the effects of PKMO and PKMO 0.9 on mitochondrial membrane potential HeLa cells were co-stained with PKMO (250 nM) /Rho123 (300 nM) or with PKMO 0.9 (250 nM)/ Rho123 (300 nM) in DMEM for 1 h. Control cells were stained with Rho123 (300 nM) in DMEM for 1 h. Glass slides coated with polymer films containing 1 μM Rho123 were used as an in vitro control. All samples were recorded by time-lapse 2D STED nanoscopy (λex = 488 nm and 561 nm. STED was performed at λSTED = 775 nm). Each line was scanned 2 times and the signal was accumulated with dwell times offset to 10 μs. Fluorescence signal decrease was evaluated using Fiji. We measured the mean grey value for each frame after background subtraction. (Process > Math > Subtract; Image > Stacks > Measure Stack). Bleaching curves of individual measurements were normalized and averaged from four independent experiments.

Measurement of mitochondrial dehydrogenase activity
HeLa cells were grown to 80-90 % confluency in a 96-well plate (costar 3599, Corning) the day before treatment. Cells were either stained with 250 nM PKMO for different times (0.25, 0.5, 1, 2, 4, 12 h) or stained with different concentrations of PKMO for 14 hours at 37 °C with 5% CO2. For non-illuminated samples, cells were washed with PBS once after the staining procedure and incubated with cultured medium containing 10% Cell Counting Kit-8 reagent (CCK8; C0038; Beyotime) for 3 hours at 37 °C with 5% CO2. The absorbance (450 nm) of each well was recorded using a TECAN Infinite M Nano+ microplate reader (TECAN, Männedorf, Switzerland). For illuminated samples, the cells were seeded in 96-well plates and stained with PKMO (250 nM in DMEM, 30 min). The cells were washed once with PBS and then illuminated for different periods (0.25, 0.5, 1, 2 h) using a green LED (520-530 nm, 50 mW/cm 2 ) Mitochondrial dehydrogenase activity was measured by CCK8 assay as described before.

Real-time respirometry
Oxygen consumption rate (OCR) experiments were performed using a Seahorse XFe24 Extracellular Flux Analyzer (Agilent Technologies, Santa Clara, CA, USA). HeLa cells were seeded at 28,000 cells/ well and grown on the cell culture microplate (100777-004; Agilent Technologies) overnight. Seahorse XF cartridges were hydrated and calibrated by Seahorse XF calibrant (100840-000; Agilent Technologies). Cells were stained with PKMO (250 nM in DMEM) for 0, 2, 5, and 12 h. Negative controls were incubated with 10 μM carbonyl cyanide 3-chlorophenylhydrazone (CCCP) for 12 h. Baseline respiration was measured in XF DMEM Base Medium (103575-100; Agilent Technologies) with 4.5 g/L glucose and 2 mM glutamine after incubation at 37 °C in an incubator without CO2 for 1 h. Periodic oxygen consumption measurements were performed, and OCR was calculated from the slope of change in oxygen concentration over time. Metabolic states were measured after subsequent addition of 1.5 μM oligomycin, 0.25 μM carbonyl cyanide 4 (trifluoromethoxy) phenylhydrazone (FCCP), 0.5 μM Rotenone/Antimycin A from Seahorse XF Cell Mito Stress Test Kit (103015-100; Agilent Technologies, MA, USA).

PKMO response to the treatment of mitochondrial oxidative phosphorylation uncoupler
HeLa cells were seeded in glass-bottom dishes (STGBD-035-1; Standard Imaging, Beijing, China) one day prior to imaging. Cells were stained with DMEM supplemented with 25 nM PKMO or 25 nM TMRE for 30 min. Cells were washed with medium once and maintained in a fresh medium for subsequent confocal imaging. Time-lapse confocal images were recorded at room temperature for 30 min after adding FCCP (10 µM) to the culture medium. The mitochondrial fluorescence intensities before and 30 min after FCCP addition were measured using Fiji (Process > Math > Subtract background; Image > Adjust > Threshold; Analysis > Analysis Particles > Multi-measure).

PKMO labeling for live-cell imaging of cancer cells, primary cells, and tissue. General remarks on PKMO labeling
Like for other mitochondria-specific probes, the progression and intensity of PKMO labeling are dependent on the individual cell line and the culture condition. For optimal results, we recommend optimizing PKMO concentration and staining duration for each cell line. If long-term time-lapse imaging is desired, labeling density should be kept as low as possible to reduce potential phototoxic effects. For initial testing, we recommend staining cells using 250 nM PKMO for 15-20 min. If the signal is too weak, leading to noisy data or low intramitochondrial contrast, we recommend extending the staining time up to 1 hour. Similarly, the PKMO concentration can be increased up to 600 nM. Some cell lines, such as HeLa cells can show significant heterogeneity in PKMO labeling degree. This effect can be reduced by extending the staining duration at lower PKMO concentrations. Moreover, we observed that a washing step of 30 to 60 min following the staining procedure could significantly increase the intramitochondrial contrast while reducing unspecific ER labeling. For multi-color labeling, we recommend sequential staining if the used dyes require different incubation times. For a summary on PKMO concentration and labeling duration, please see Supplementary Tables S2 and S3.
Labeling of cancer cells COS-7 cells were seeded in glass-bottom dishes (Standard Imaging) two days prior to imaging. COS-7 cells were stained with DMEM supplemented with 250 nM PKMO at 37 °C for 15 min. U-2 OS cells and HeLa cells were seeded in glass-bottom dishes (ibidi GmbH, Germany) one day prior to imaging. U-2 OS cells were stained with McCoy´s medium supplemented with 250 nM PKMO at 37 °C for 20 min. For single-color recordings, HeLa cells were stained with DMEM containing 150-250 nM PKMO at 37 °C for 40-45 min. Following the staining procedure, cells were washed three times with culture medium and incubated at 37 °C for 30-60 min to remove the unbound dye. The cells were imaged at room temperature in HEPES buffered DMEM (HDMEM) containing 4.5 g/l glucose, l-glutamine, and 25 mM HEPES (Thermo Fisher Scientific).

Labeling of primary cells
Primary brown adipocytes (pBAcs) were seeded in glass-bottom dishes (Standard Imaging) 3 days before imaging. pBAcs and primary hippocampal neurons were stained with DMEM containing 250 nM PKMO at 37 °C for 15 min. After removing the staining solution, the cells were washed with medium once and maintained in fresh medium for subsequent STED imaging. CMs were seeded in laminin-coated glassbottom dishes 3 days before measurements. CMs were stained with 500 nM PKMO in DMEM at 37 °C for 15 min before imaging experiments. Islets tissues isolated from Ins1-Cre+/+; GCaMP6ffl/fl mice (7) were stained with Krebs-Ringer bicarbonate buffer (KRBB) solution containing 125 mM NaCl, 5.9 mM KCl, 2.4 mM CaCl2, 1.2 mM MgCl2, 1 mM L-Glutamine, 25 mM HEPES, 3 mM glucose, 0.1% (v/v) bovine serum albumin, and 600 nM PKMO at 37 °C for 30 min. After removing the staining solution, the islets were then washed with KRBB solution three times and maintained in a fresh medium for the STED imaging.

Sample preparation for multi-color imaging
Labeling of cristae and mtDNA in HeLa cells For labeling of cristae and mtDNA, HeLa cells were incubated with DMEM containing 150 nM PKMO at 37° C for 45 min. Cells were washed twice and incubated with DMEM containing 0.5 µl/ml Quant-iT PicoGreen reagent (Thermo Fisher Scientific) at 37 °C for 30 min. The culture medium was replaced and the cells were incubated for 30 min to remove unbound dye. Cells were recorded in HDMEM at room temperature.

Labeling of cristae and SNAP/Halo fusion proteins
For labeling of MIC10-SNAP and cristae, HeLa cells were incubated with DMEM containing 200 nM PKMO at 37 °C for 40 min. Afterward, MIC10 was labeled with 1 µM SNAP-Cell 647-SiR (New England BioLabs Inc.) at 37 °C for 60 min. Halo-KDEL was labeled by co-staining with 300 nM PKMO and 500 nM 647-SiR-CA at 37 °C for 45 min. Following the staining procedure, cells were washed three times with culture medium and incubated at 37 °C for approximately 60 min to remove the unbound dye. The cells were imaged in HDMEM at room temperature. COS-7 cells expressing TOM20-Halo were incubated with DMEM containing 500 nM 647-SiR-CA (synthesized according to literature protocol (12) ) at 37 °C for 10 min. After removing the staining solution and three washing steps with PBS, the cells were then stained with DMEM containing 250 nM PKMO for 15 min at 37 °C. After removing the staining solution, the cells were then washed with DMEM once and maintained in fresh DMEM for following STED imaging at room temperature.
Labeling of cristae, tubulin, and mtDNA. For labeling of cristae together with the cytoskeleton and mtDNA, HeLa cells were incubated with DMEM containing, 350 nM PKMO, 200 nM 4-610CP-CTX (13), and 0.2 µl/ml Quant-iT PicoGreen reagent (Thermo Fisher Scientific) at 37 °C for 40 min. Cells were washed three times with culture medium and incubated at 37 °C for 60 min to remove unbound dye. Cells were recorded in HDMEM at room temperature.
For a summary of used fluorophore concentrations and labeling duration, please also see Supplementary  Tables S2 and S3.

Live-cell imaging of cancer cells, primary cells, and tissue.
For detailed summary of imaging parameters please see Supplementary Table S3. In brief, COS-7 cells, primary cells, and tissues were recorded using Facility Line or STEDYCON STED microscopes (Abberior Instruments GmbH, Göttingen, Germany) equipped with an Olympus UPlanXAPO 60x oil, NA1.42 objective (Olympus, Tokyo, Japan) or CFI Plan Apochromat Lambda D 100x oil, NA1.45 objective (Nikon, Tokyo, Japan). Pixel sizes of 20-30 nm were used for STED nanoscopy. PKMO was excited at 561 nm wavelength and STED was performed using a pulsed depletion laser at 775 nm wavelength with gating of 1-7 ns and dwell times of 10 μs. For dual-color STED imaging of COS-7 cells, SiR was excited at 640 nm wavelength. STED was performed at 775 nm wavelength with gating set to 0.75-8.75 ns. Dwell times of 10 μs were used. Rho123 was excited at 485 nm and recorded in the confocal mode. The fluorescence signal was usually accumulated over 2-5-line steps. STED nanoscopy of HeLa and U-2 OS cells was carried out using an Expert Line dual-color STED 775 QUAD scanning microscope (Abberior Instruments GmbH). The microscope was equipped with a UPlanSApo 100×/1.40 Oil [infinity]/0.17/FN26.5 objective (Olympus). In brief, PKMO was excited at 561 nm and SiR was excited at 640 nm wavelength. Depletion was performed at a 775 nm wavelength. Imaging parameters were adjusted based on the individual samples. We typically used pixel sizes of 20 -30 nm and dwell times of 5-7 µs. In the STED mode, each line was scanned 6 to 9 times and the signal was accumulated. PicoGreen, mEGFP and Rho123 were excited at 485 nm wavelength and were recorded in the confocal mode. The pinhole was set to 0.7 -1.0 AU.

Comparison of PKMO and SNAP-cell SiR for cristae imaging
HeLa cells stably expressing COX8A-SNAP (11) were labeled with SNAP-cell SiR (1 µM in DMEM, 37 °C, 40 min). HeLa wild type cells were labeled with PKMO (250 nM in DMEM, 37 °C, 40 min). Afterward, cells were counterstained with DMEM supplemented with 1 µM Rho123 (Thermo Fisher Scientific) for 20 min at 37 °C. For controls, cells were stained only with Rho123. Cells were recorded by time-lapse 2D STED nanoscopy. For each individual cell, a single field of view (FOV) of 10 µm x 10 µm was recorded over 20 frames (total time: 317 seconds). Samples stained with PKMO/Rho123 were excited at λex = 485 nm and 561 nm. Samples stained with SiR/Rho123 were excited at λex = 485 nm and 640 nm. STED was performed at λSTED = 775 nm. Overall, 15-line steps were performed (9-line accumulations in the STED mode (PKMO/SiR), 3-line accumulation in the confocal mode (PKMO/SiR), and 3-line accumulations in the confocal mode (Rho123). The pinhole was set to 0.8 AU. The pixel size was set to 30 nm. The pixel dwell time was set to 7 µs. For laser power settings, please see Supplementary Table S3. Fluorescence signal decrease was evaluated using Fiji. We measured the mean grey value for each frame (Analyze > Set Measurements > Area & Mean Grey Value > OK; Image > Stacks > Measure Stack). Bleaching curves of individual measurements were normalized and averaged. The brightness of the PKMO staining was quantified in Fiji. To this end, the mitochondrial networks of the measured cells were manually selected on confocal overview images (80 x 80 µm FOV, 100 nm pixel size, 7 µs dwell time, 1x line accumulation) and the maximum grey values were estimated (Analyze > Set Measurements > Area, Min & max gray value > OK; Image > Analyze > Measure). Transmission electron microscopy COS-7 cells were grown on Aclar ® film (Electron Microscopy Sciences, Hatfield, PA) to a confluency of approximately 60-70%. Fixation was performed by immersion with pre-warmed 2.5% glutaraldehyde in 0.1M phosphate buffer (pH 7.4) at room temperature for 1h. After washing 3 times with 0.1 M phosphate buffer (pH 7.4), samples were post-fixed with 2% osmium tetroxide (w/v) and 1.5% potassium ferricyanide (w/v) in the same buffer at 4°C for 2 h and then washed 3 times. Following en bloc staining with 2% uranyl acetate (w/v) performed at 4°C overnight, the samples were dehydrated and embedded in fresh resin, polymerized at 65°C for 24 h. Ultrathin (70 nm) sections were obtained by a Leica UC7 ultramicrotome (Leica Microsystems, Wetzlar, Germany) and recorded on a JEOL Jem-1400 transmission electronmicroscope (JEOL Ltd., Akishima, Japan) using a XAROSA CMOS camera (EMSIS GmbH, Münster, Germany) Preparation of HeLa and U-2 OS cells was performed as described previously (8). In brief, cells were grown on Aclar ® film (Plano GmbH, Wetzlar, Germany) to a confluency of approximately 70%. Fixation was performed by immersion with pre-warmed (37 °C) 2% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) at room temperature. For complete fixation, samples were stored at 4°C overnight. Following postfixation with 1% osmium tetroxide and pre-embedding staining with 1% uranyl acetate, samples were dehydrated and resin-embedded. Ultrathin sections (70 nm) were recorded on a Talos L120C transmission microscope (Thermo Fisher Scientific, Hilsboro, Oregon, USA) at 11,000-13,500× magnification using a Ceta 4k × 4k CMOS camera (Thermo Fisher Scientific).
Primary neurons were grown on Aclar ® discs (Electron Microscopy Sciences) to a confluency of approximately 70-80%. Fixation was performed by immersion with pre-warmed 3% glutaraldehyde in 0.1M phosphate buffer (pH 7.4) at 4°C for 2 h. After washing 3 times with 0.1 M phosphate buffer (pH 7.4), samples were post-fixed with 1% osmium tetroxide (w/v) in 0.24 M phosphate buffer (pH 7.4) at 4°C for 2 h and then washed 3 times. The samples were dehydrated and embedded in fresh resin, polymerized at 60°C for 24 h. Ultrathin (60-70 nm) sections were obtained by a Leica UC7 ultramicrotome (Leica Microsystems). Thin sections laid on copper mesh were stained with heavy metals, uranyl acetate, and lead citrate for contrast. A Hitachi-H7800 transmission electron microscope (Hitachi High-Technologies Corporation, Chiyoda, Japan) was used to observe the hippocampal ultrastructure.

Apoptosis by expression of mEGFP-BAX
Oligonucleotides (g) and (h) were used for site directed mutagenesis PCR of pEGFP-C3 (Clontech Laboratories Inc., Mountain View, CA, USA) to produce the expression plasmid pmEGFP-C3. BAX was amplified by PCR using oligonucleotides (i) and (j) and incorporated into pmEGFP-C3 using HindIII and EcoRI restriction sites.
HeLa cells were seeded in glass-bottom dishes (Ibidi GmbH) and were stained with 250 nM PKMO (DMEM, 37 °C, 45 min) the next day. Afterward, cells were transfected with pmEGFP-C3-BAX using jetPRIME ® transfection reagent (Polyplus). Q-VD-Oph hydrate (APExBIO Technology LLC, Houston, TX, USA) was added to the medium at a concentration of 20 µM immediately after transfection in order to prevent detachment of apoptotic cells. Cells were analyzed by 2D STED nanoscopy 4-6 h after transfection. Cells were recorded in HDMEM at room temperature.
Image processing STED nanoscopy STED nanoscopy images of COS-7 cells, primary cells, and tissues were deconvoluted using Huygens software (Scientific Volume Imaging B.V., Hilversum, The Netherlands). STED nanoscopy images of HeLa and U-2 OS cells and 3D STED images of COS 7 cells were deconvoluted using the Richardson-Lucy algorithm in the Imspector software (Abberior Instruments GmbH; version 0.14.11616). 3D rendering and reconstruction of 3D STED data were performed using Imaris (Bitplane, Belfast, UK). Resolution estimation was performed by Gaussian fitting of fluorescence intensity line profiles. The full width at half maxima (FWHM) was estimated via Analysis >> Fitting >> Nonlinear curve fit >> Gaussian fit in the Origin Pro 2020b software (OriginLab Corporation, Northampton MA, USA) or using Matlab (The Math Works, Natick, MA, USA). All images that were used to analyze the resolution in supplementary information were raw data without background subtraction. For time-lapse recordings, photobleaching was compensated using the bleach correction feature in Fiji/ImageJ (version 1.53f51).

Electron microscopy
Electron microscopy recordings were filtered using a median filter in Fiji/ImageJ (version 1.53f51).

Analysis of mitochondrial network appearance
HeLa cells stably expressing COX8A-mNeonGreen were stained at 37 °C for 30 minutes using DMEM supplemented with 250 nM of PKMO, PKMR, Mitotracker Red (Thermo Fisher Scientific) or TMRE (Thermo Fisher Scientific), respectively. Controls were treated with 20 µM FCCP for 30 min at 37 °C to induce fragmentation of mitochondria. Cells were washed three times using DMEM and recorded 2 hours and 5 hours later using a Facility Line confocal microscope (Abberior Instruments). The cells were recorded at 37°C and 5% CO2. Tile scans (500 µm x 500 µm, 10% overlap) were recorded with a pixel size of 100 nm. The pinhole was set to 1.0 AU. mNeonGreen was excited at λex = 485 nm, the mitochondrial probes were excited at λex = 561 nm. The pixel dwell time was set to 6.5 µs and 2-line accumulations were recorded for each channel. The mNeonGreen channel was utilized for segmentation of individual cells using a machine learning approach. For optimized cell detection, the cell shape was manually determined on about 10% of the images. The manually labeled recordings served as input for an image-to-image translation network (14) or training. After applying the trained network on all recordings (using the mNeonGreen channel) minor corrections were applied manually (like cutting a residual connection between two adjacent cells or joining fragments to cover a whole cell). Overall, the cell shape detection (data not shown) could distinguish the vast majority of cells at high accuracy. Therefore, we used the segmentation data to correlate the fluorescence readout on a single cell level. The structure of the mitochondrial network was detected and analyzed using a custom written script in MATLAB (The MathWorks Inc., Natick, MA, USA). First, the confocal images were deconvolved (Richardson-Lucy) with a 200nm FWHM Gaussian peak PSF and a relative threshold above the background (10% of maximum) was applied to get a binary representation of the mitochondrial network. For the analysis of the variances of the cellular brightness (of mNeonGreen, PKMO, PKMR, Mitotracker Red or TMRE) the average signal intensity was calculated in each color channel on the detected mitochondria for each individual cell. The results were then pooled for each recording and normalized by the mean of all cells of the recording. Therefore, all brightness distributions show the cellular brightness relative to the mean brightness per cell per recording. For the estimation of eccentricity, the detected mitochondrial network was divided into disjoint segments. For each segment, the eccentricity of the ellipse that has the same second-moments as the segment was calculated (MATLAB function regionprops). For the determination of the average mitochondrial network branch length, the skeleton of the detected mitochondrial network per cell was calculated, branch points were eliminated and the average length of the remaining disjoint curves was calculated. .            Compound S2: To a solution of compound S1 (500 mg, 2.39 mmol) in toluene (5 mL) was added 2-Bromoethan-1-ol (CAS: 540-51-2) (896 mg, 7.17 mmol) at room temperature (r.t.), the mixture was heated at 110°C and stirred for 16 h. The reaction mixture was then cooled to r.t slowly and a solid precipitated. The solid was collected by filtration, followed by washing with methyl tert-butyl ether (10 mL), the solid was then dried under vacuum to afford the compound (700 mg, 88%) as a brown solid. Compound S3: A mixture of compound S3 (450 mg, 1.35 mmol), triethyl orthoformate (120 mg, 0.808 mmol), and acetic anhydride (4.5 mL) in a pressure tube was heated at 120°C and stirred for 16 h. The mixture was then cooled to r.t. The solvent was removed under reduced pressure to obtain a residue, which was purified by a silica gel chromatography column, eluting with DCM