Visualization of the membrane surface and cytoskeleton of oligodendrocyte progenitor cell growth cones using a combination of scanning ion conductance and four times expansion microscopy

2.1 Combined SICM and ExM imaging workflow

Fixed OPCs were first imaged with a brightfield microscope at low magnification to create a large overview tile scan (Figure 1A), in which the growth cones imaged by SICM could be tagged (Figure 1B). This would later on lead to easier recognition of the same growth cone in different images with varying magnifications. In the next step, the surfaces of the OPC growth cones were imaged with SICM (Figure 1C). Afterwards, the cytoskeleton was immunofluorescently labeled. The growth cone membrane was recorded via SICM before labeling as it required membrane permeabilization, which might have altered the membrane topography. Images of the immunofluorescently labeled OPCs were acquired at low magnification (Figure 1D) and compared to the brightfield overview tile scans to re-identify the same growth cones. Cells were then expanded using a 4× ExM protocol (Asano et al. 2018) and subsequently imaged again using the same objective for retrieval and determination of the expansion factor (Figure 1E). The same sample regions, pre- and post-expansion, were compared and gauged to determine expansion factors for each expanded gel. A 4× expansion protocol was chosen as it promised to result in lateral resolutions between 100 and 200 nm in combination with our widefield microscope. This would match the resolutions obtainable with the SICM as capillaries with opening radii in the range of 40–60 nm were used. Lateral SICM resolutions approximately equal 3 times the inner capillary radius (Rheinlaender and Schäffer 2009). Capillaries with radii smaller than 40 nm can be produced (Steinbock et al. 2013). However, smaller capillaries tend to clog faster and as our SICM image acquisition times ranged between 2 and 8 h depending on the step and scanning field size, smaller capillaries would not have remained unclogged throughout the entire image acquisition risking sample-probe contact. Furthermore, these lateral resolutions should be sufficient to distinguish filopodia and lamellipodia. Finally, the same expanded growth cones previously recorded via SICM were located in the expanded gels utilizing the overview scans and imaged using a 60× magnification objective (Figure 1F). Resolutions were determined by calculating the full width at half maximum (FWHM) of a Gaussian fitted line profile through a small but still distinguishable structure within the fluorescence image. To obtain overlays, ExM images were scaled in accordance to the expansion factor of the gel as well as horizontally flipped (Figure 1G). This was necessary, as SICM images were recorded on an upright setup facing the cells from above, whereas ExM images were acquired with an inverted microscope facing the cells from below. Since SICM and ExM images were acquired on different microscopes with further sample preparations in between the two acquisitions, the respective fields of view differed in terms of their acquisition angle. Thus, the images had to be slightly rotated to match one another.

Figure 1:

Experimental workflow. Fixed samples are first imaged as a large brightfield tile scan with a 20× objective to create an overview map (A) in which regions of interest (ROIs) containing cells with OPC morphology can be marked (B). Growth cones within these ROIs are then imaged with SICM to capture membrane topographies (C). After fluorescently labeling the cytoskeleton, micrographs are obtained with a 20× objective (D). After sample expansion, the same regions of the sample are imaged again (E). Expansion factors are determined by measuring distances in images of the same sample region acquired with the same objective before and after expansion. A 60× objective is used to capture fluorescence images of OPC growth cones (F). Finally, expansion factors are used to scale the ExM images, which are then flipped to match the SICM images. ExM and SICM images are manually correlated (G).

2.2 SICM recordings of OPC growth cones

Figure 2 shows brightfield images of the investigated cells (Figure 2a), three- (Figure 2b) and two-dimensional (Figure 2c) projections of three growth cones imaged via SICM as well as height profiles through their widest and longest parts (Figure 2d). The height profiles along the direction of growth start at values of about 1–2 µm and then gradually decrease, with the exception of some protrusions, to about 500 nm. The height profiles along the widest sides of the growth cones are generally flatter, with the exception of growth cone A, ranging mostly around 500 nm. The shape of growth cone A slightly differs from these observations as the flat lamellipodium appears at the right side instead of the very tip of the growth cone.

Figure 2:

SICM recordings of OPC growth cones. a) Brightfield images of cells imaged via SICM. Scale bars represent 50 µm. A white square marks the scanned growth cone. 3D (b) and 2D (c) plots of the topographies as well as height profiles (d) through the length (blue) and width (yellow) of the growth cones. Ab-c) 7.00 × 7.00 µm scan of an OPC growth cone imaged with a step size of 100 nm. Bb–c) 9.50 × 9.55 µm scan of an OPC growth cone imaged with a step size of 50 nm. Cb–c) 4.23 × 5.00 µm scan of an OPC growth cone imaged with a step size of 30 nm. Width of membrane bulges (orange) and potential filopodia (green) are highlighted.

The flattest parts display heights of 200 nm, which suggest that these could be lamellipodia. Growth cones A and B exhibit what appear to be filopodia with diameters ranging between 150 and 330 nm. These exemplary growth cone recordings correspond to our current knowledge of growth cones containing a thicker microtubule-rich domain at the center, which then flattens out into an actin-rich peripheral domain (Thomason et al. 2020).

2.3 Expansion microscopy images of OPC growth cones

An exemplary fluorescence image of tubulin-labeled cells is displayed pre-expansion in Figure 3Aa and post-expansion in Figure 3Ab. Images were captured with the same optical magnification. The expansion factor of the respective expansion gel was determined to be 3.72 ± 0.02 (n = 10). Figure 3Ba shows an OPC and one of its growth cones (Figure 3Bb) from the same expanded culture at higher magnification, enabling the visualization of fibrous microtubules. A resolution of 105 nm was estimated by measuring the width of the thinnest distinguishable microtubules within this growth cone via the full width at half maximum of an intensity profile (Figure 3C).

Figure 3:

Expansion microscopy (ExM) of oligodendrocyte precursor cell (OPC) cultures. Widefield fluorescence microscopy images of tubulin-labeled OPCs pre- (Aa) and post-expansion (Ab). The expansion factor is 3.72 ± 0.02 (n = 10). Images captured using a 20× objective. Ba) ExM image of an expanded OPC captured with a 20× objective. Bb) Left growth cone of the OPC displayed in (Ba) captured using a 60× objective. The resulting resolution was determined via the full width at half maximum (FWHM) of the Gaussian fitted intensity profile (C) of a small but distinguishable microtubule (Bb) to be approximately 105 nm. Scale bars represent 100 µm (Aa), 27 µm (Ab), 50 µm (Ba), and 10 µm (Bb).

This is within the range of what we expected to be resolved laterally and roughly matches the sizes of structures resolved in the SICM recordings (Figure 2b). The smallest structure, which was measured within the SICM images, was a membrane protrusion with a width of 130 nm (Figure 2Cb).

2.4 Correlation of SICM and ExM images of OPC growth cones

In order to find out, whether prominent structures of the cytosceleton can be matched to defined irregularities of the growth cone surfaces as visualized by SICM recordings, overlays from successive SICM and ExM images were assembled. Figure 4A shows an SICM scan from a growth cone displaying several peripheral membrane protrusions emerging from a higher central tube-like structure at the right side of the image. The ExM image shows a bundle of microtubules entering the growth cone from the right. The superimposed SICM and ExM images shown in Figure 4C confirm that in fact microtubules are mainly located at the higher central domain whereas the flatter peripheral regions, most likely lamellipodia and filopodia, are void of labeled microtubules.

Figure 4:

Correlation of membrane topography and tubulin labeling of an OPC growth cone. (A) Growth cone membrane surface imaged with a SICM step size of 50 nm. (B) Expansion microscopical image of the tubulin cytoskeleton. The image was horizontally flipped to match the SICM image as SICM images show the growth cone from above while the ExM images show the growth cone from below. (C) Overlay of SICM and ExM images of the same growth cone. Grey dotted squares indicate the rotation angle of the SICM image. Colored triangles indicate identical positions. Scale bars correspond to 9.5 µm.

To determine, whether the flatter filopodia- and lamellipodia-like regions of the growth cones contain actin, double stainings were performed (Figure 5). Both growth cones show a dense staining for actin in the peripheral membrane regions, whereas tubulin is concentrated in the elevated central regions. While an approximate match between prominent structures in the SICM and ExM images is clearly possible, the superimposed images shown in Figure 5C and F do not perfectly align. This can be explained by the inherent differences of 2- and 3-dimensional data sets. Fluorescence signals from higher or lower planes than the focal plane are lost as background fluorescence. Furthermore, during the sample preparation processes necessary for ExM loosely attached filopodia or lamellipodia might have shifted slightly.

Figure 5:

Correlated SICM and ExM images of OPC growth cones. (A, D) SICM images of the growth cone membranes. SICM step sizes: 100 nm (A) and 30 nm (D). (B, E) Expansion microscopical images of the actin (magenta) and tubulin (green) cytoskeletons. Images horizontally flipped to match the SICM images as SICM images show the growth cone from above while the ExM images show the growth cone from below. (C, F) Overlays of SICM and ExM images of the same growth cones. Grey dotted squares indicate the rotation of the SICM with respect to the ExM images. Colored triangles point to identical positions. Scale bars represent 7 µm (B, C) or 5 µm (E, F).

4.1 Cell culture

Cortices from 0- to 2-day-old Wistar-Hannover rats were dissected and collected in 10 mL tubes filled with ice cold Minimum Essential Medium (MEM; Gibco™, Thermo Fisher Scientific, 11095080) supplemented with 10 μg/mL desoxyribonuclease I (Sigma-Aldrich, Merck, DN25), 2.5 mg/mL trypsin (Sigma-Aldrich, Merck, T4549), 25 U/mL penicillin and 25 μg/mL streptomycin (1 % P/S; Sigma-Aldrich, Merck, P0781). Dissected cortices were incubated at 37 °C for 5 min on a thermal shaker (ThermoMixer, Eppendorf) at 1000 rpm. The enzymatic digestion was stopped by addition of 10 % (v/v) heat inactivated fetal calf serum (FCS; Gibco™, Thermo Fisher Scientific, 10270106). To complete tissue dissociation, the resulting tissue suspension was gently triturated 10 times with a 1000 µL pipette. Afterwards, the suspension was left undisturbed for 5 min to let any left-over tissue debris settle at the bottom of the reaction tube. The supernatant was aspirated and centrifuged for 10 min at 180 g and 20 °C. In the following, the pellet was resuspended in Dulbecco’s Modified Eagle Medium (DMEM)/Ham’s F12 nutrient mixture (F12) (Gibco™, Thermo Fisher Scientific, 31330095) containing 10 % (v/v) FCS, 1 % P/S, and 1 mM sodium pyruvate (Gibco™, Thermo Fisher Scientific, 11360070). The resuspended pellet was then seeded into T75 flasks (Sarsted, 83.3911), which had been coated with a 5 μg/mL poly-d-lysine (Sigma-Aldrich, Merck, P6407) solution, and incubated at 37 °C and 5 % CO₂ in a humidified atmosphere (Heraeus, B5060 incubator). One T75 flask eventually contained cells from two cortices. The flask medium was exchanged for fresh medium 3 h after seeding and afterwards every 2 days.

After 7–10 days, flasks were shaken at 37 °C and 180 rpm (Kühner, ES-W) to separate different cell types in accordance to their adherence to the flask bottom. After 3 h of shaking, the flask medium, which mostly contained microglia, was discarded and replaced with fresh medium. After another 18 h of shaking, the medium was removed again. Fresh medium (DMEM/F12, 1 % P/S, 10 % FCS) was added to the flasks before they were put back into the incubator. Cells could be harvested again after another 7–10 days up to two more times.

The removed flask medium was centrifuged at 180 g for 5 min to obtain the desired OPCs. The centrifuged pellet was resuspended in DMEM/F12, which was supplemented with a customized version of B27 supplement (Lesslich et al. 2022) as well as 10 ng/mL platelet-derived growth factor (PDGF; HumanKine, Proteintech, HZ-1215) and 10 ng/mL fibroblast growth factor 2 (FGF-2; HumanKine, Proteintech, HZ-1285). Cells were seeded in glass bottom dishes (ibidi, 81218-200) aiming for a density of 30,000 cells/cm². To enable better cell adhesion, dishes had been pre-coated with a 5 μg/mL poly-d-lysine solution. The dish medium (DMEM/F12, customized B27, 10 ng/mL PDGF, 10 ng/mL FGF-2) was replaced with fresh medium 3 h after seeding and following every 24 h. Dishes with OPCs were maintained at 37 °C and 5 % CO₂ in a humidified incubator. This procedure provided a higher OPC yield than the previous protocol used in our laboratory (Feldhaus et al. 2004).

4.2 Sample preparation

3–6 days after seeding, OPCs were fixed with a mixture of 4 % (w/v) formaldehyde (Roth, 0335.3) and 1 % glutaraldehyde (Fluka™, 49629) in phosphate buffered saline (PBS; Gibco™, Thermo Fisher Scientific, 14190144). In the following, samples were washed twice with PBS and incubated for 5 min at room temperature (RT) with a quenching solution containing 100 mM glycine (Fischer Chemical™, Fischer Scientific, 10070150) and 10 mM ethanolamine (Merck Chemicals, 8008491000) to reduce the autofluorescence caused by the aldehyde fixatives. Afterwards, samples were washed again twice with PBS. A paint marker was used to draw a cross from the outside of the dish onto the glass bottom. This marking aided with orientation and made it easier to locate the same cell and growth cone at different microscope setups.

After acquiring brightfield tile scans of the samples, all cells were identified as OPCs, which displayed a circular soma and had exactly two processes with process tips further than 15 µm apart from each other. These cells had previously been identified as OPCs in similar cultures (Feldhaus et al. 2004) using anti-A2B5 immunostaining and were clearly morphologically distinguishable from the contaminating GFAP-positive astrocytes and the multipolar or amoeboid microglia. The identified OPCs were then imaged via SICM, and afterwards fluorescently labeled: For membrane permeabilization and unspecific binding site blocking, samples were incubated twice for 5 min at RT with PBS additionally containing 0.1 % (v/v) Triton X-100 (Boehringer, 759704) and 3 % (v/v) FCS. They were washed twice with PBS afterwards. Then, the cells were incubated for 1 h at 37 °C with a primary antibody solution containing mouse anti-β-tubulin (Proteintech, 66240-1-Ig) and rabbit anti-β-actin (Proteintech, 20536-1-AP) antibodies each diluted 1:1000 in PBS. In the following, samples were washed again twice with PBS before being incubated at 37 °C in the dark with a PBS-based secondary antibody solution containing AlexaFluor488 goat anti-mouse IgG (Invitrogen, A11001) and AlexaFluor594 goat anti-rabbit IgG (Invitrogen, A11012) antibodies at a dilution of 1:250. After 4 h, 10 μg/mL Hoechst33258 dye (Sigma-Aldrich, B2883) were added to the samples to label the cell nuclei. Cells incubated for another 20 min before the samples were washed again twice with PBS.

4.3 Widefield microscopy

Before samples were imaged at the SICM, images were captured in brightfield mode at an Olympus IX83 P2ZF microscope with IX3 LED illumination and a 20× objective (air, NA 0.45, WD 7200 µm, LUCPlanFLN, Olympus). Using the Olympus cellSens Dimension software (version 3.2, build 23706) a large tile scan of the whole sample was captured. This overview was then used to locate single cells and growth cones and to help identify the same cells again later on.

After SICM images had been captured and samples were fluorescently labeled, more overview tile scans and images were captured to later on enable expansion factor determination and easier identification of the same growth cones previously recorded with the SICM. These images were captured with the same Olympus IX83 microscope using the 20× objective, 385 nm excitation light and 442 nm emission light for the Hoechst33258 labeled nuclei, 465 nm excitation light and 519 nm emission light for the AlexaFluor488 labeled beta-tubulin, and 559 nm excitation light and 575 nm emission light for the AlexaFluor594 labeled beta-actin.

4.4 Scanning ion conductance microscopy

SICM imaging was conducted with an in-house built setup (Gesper et al. 2017). Nano-capillaries (access resistances: 50–100 MΩ; estimated opening radii: 40–60 nm) were pulled from borosilicate glass (World Precision Instruments) with a P2000 laser puller (Sutter Instrument). A Python 2.7 software written by Dr. Patrick Happel recorded the data and controlled the instrument. We used approach velocities in the range of 25–100 nm/ms and retraction distances of 5–10 μm to ensure contact-free scanning of the growth cones. Pixel sizes ranged from 30 to 100 nm. We used thresholds of 0.5–2 %, bias voltages of 150–300 mV and filtered the current with a low-pass filter of 0.3, 0.7, or 1 kHz. A 150 mM NaCl solution was used as electrode and bath solution. A chlorinated silver wire served as an Ag|AgCl measuring electrode and a chlorinated silver pellet was used as an Ag|AgCl counter electrode. Cells were imaged using backstep mode scanning (Happel and Dietzel 2009) to enable contact-free scanning.

4.5 Expansion microscopy

After imaging the fluorescently labeled cells with the same Olympus IX83 P2ZF microscope, the coverslip glass bottoms of the Petri dishes containing the OPCs were separated from the rest of the dishes by gently pushing them downwards with forceps. The cells were then treated according to the four times protein retention expansion microscopy (proExM) protocol published by Asano et al. (2018). This protocol has been demonstrated to yield expansion with nanoscale isotropy (Chen et al. 2015). As soon as the samples had been expanded, they were placed into single specimen plates with a No. 1.5 glass coverslip bottom and a viewing area of 96 mm × 67 mm (Mattek, P384G-1.5-10872-C) to accommodate the expanded gels. They were imaged with the same Olympus IX83 P2ZF microscope with increased exposure times due to the decreased fluorescence intensity caused by the expansion (Wassie et al. 2019). Using the identical 20× objective enabled expansion factor determination by comparison with the images of the unexpanded samples. Finally, the samples were imaged with a 60× water immersion objective (NA 1.20, WD 280 µm, UPlanSApo, Olympus) to further increase the resolution.

4.6 Data processing and analysis

Raw data used to generate the displayed figures was made available in an open access data repository (doi: 10.5281/zenodo.7931003).

SICM images were cropped to the region of interest and filtered using a median filter with a width of 2 or 3 pixels. Furthermore, the lowest z value of the data set was subtracted from all z values to set the origin of the z axis to zero. All SICM data processing and analysis was conducted with a self-developed software written in Python 3.10 (pySICM Analysis, version 0.1.1, doi: 10.5281/zenodo.7930581).

Light microscopy images were processed and analyzed using FIJI/ImageJ 1.53t (Schindelin et al. 2012). Any quantitative measurements such as expansion factor or resolution determinations were conducted on unprocessed raw data. Images displayed within this manuscript have been contrast adjusted for better visualization. Also, images shown in Figures 3, 4, and 5 have been post processed with FIJI’s ‘Subtract Background…’ function (radius: 100 px).

ExM images of the same growth cones but illuminated with different wavelengths to excite fluorophores either labeling tubulin or actin were manually corrected for gel drift, cropped and merged in FIJI.

Expansion factors were determined by measuring the distances between or lengths of the same sample structures e.g. cell processes in pre- and post-expansion images of the same sample region captured with the 20× objective. Per expanded sample 10 distances were measured pre- and post-expansion. Post-expansion distances were divided by pre-expansion distances, the mean and standard error were calculated and the mean value regarded as the expansion factor.

Resolutions were determined by plotting the gray scale values along a line through a small but distinguishable structure within the image. Plotted values were then fitted with a Gaussian fit and the full width at half maximum (FWHM) was determined as the resolution.

SICM and ExM images were aligned manually by first cropping, scaling and flipping the ExM images to match the SICM recording. SICM and ExM images were then rotated with respect to each other. Images were not sheared or stretched.