High-Resolution Imaging of Mitochondria and Mitochondrial Nucleoids in Differentiated SH-SY5Y Cells.

Mitochondria are highly dynamic organelles which form intricate networks with complex dynamics. Mitochondrial transport and distribution are essential to ensure proper cell function, especially in cells with an extremely polarised morphology such as neurons. A layer of complexity is added when considering mitochondria have their own genome, packaged into nucleoids. Major mitochondrial morphological transitions, for example mitochondrial division, often occur in conjunction with mitochondrial DNA (mtDNA) replication and changes in the dynamic behaviour of the nucleoids. However, the relationship between mtDNA dynamics and mitochondrial motility in the processes of neurons has been largely overlooked. In this chapter, we describe a method for live imaging of mitochondria and nucleoids in differentiated SH-SY5Y cells by instant structured illumination microscopy (iSIM). We also include a detailed protocol for the differentiation of SH-SY5Y cells into cells with a pronounced neuronal-like morphology and show examples of coordinated mitochondrial and nucleoid motility in the long processes of these cells.


Introduction
Neurons are highly polarised cells with long axons and dendrites. Due to their peculiar architecture, neurons are particularly reliant on long-range intracellular transport mechanisms to e ciently distribute organelles, proteins and other cargoes with spatiotemporal precision [1,2]. This process is crucial to maintain neuronal function, and transport defects have been shown to be involved in several neurodegenerative disorders [3].
Mitochondria are essential neuronal cargoes, whose e cient transport is crucial for maintaining neuronal homeostasis through calcium buffering, metabolite synthesis and provision of ATP at sites with high energy demand, such as synapses [4,5]. Not surprisingly, disruption of mitochondrial transport has been shown to alter neuronal functionality and prime neurons for degeneration [6]. Mitochondria are unique axonal transport cargoes as the majority of them can remain stationary for signi cant amounts of time, which is unusual for most axonal cargoes [7]. Despite the fact that only a fraction of mitochondria are motile in neurons, the importance of their transport is underscored by the increasing evidence that reduced axonal motility of these organelles can be considered a hallmark of ageing and neurodegeneration [8].
Mitochondria are the only organelles harbouring their own genome, a trace of their prokaryotic origin and symbiotic incorporation into eukaryotic cells [9,10]. The mitochondrial genome is distinct from that stored in the nucleus of the cell and encodes 13 essential subunits of respiratory chain complexes, 22 tRNAs and 2 rRNAs necessary for their translation [11,12]. The mitochondrial DNA (mtDNA) is located in the mitochondrial matrix and is condensed in small ~100-nm [13] nucleoprotein complexes called nucleoids, by the mtDNA binding protein mitochondrial transcription factor A (TFAM), a key regulator of mtDNA replication and transcription [14,15]. Live-imaging experiments have shown nucleoids to be highly dynamic structures that are able to redistribute through the mitochondrial network via mitochondrial ssion and fusion [11,16,17].
Repositioning and/or redistribution of the nucleoids during mitochondrial morphological transitions are intimately linked to the replication of mtDNA. Thus, mtDNA replication marks the site of ER-mediated mitochondrial constriction and subsequent DRP1-dependent ssion, thus priming the organelles for division [18,19]. On the other hand, mitochondrial fusion aids in the redistribution of mtDNA throughout the mitochondrial network, ensuring mitochondrial integrity and functionality are retained throughout the network [20].
To date, many models for the study of mitochondrial dynamics in neurons have been established in cultured cells and in model organisms. Model organisms are extremely bene cial for in vivo studies of axonal transport in a physiological context [21-31], with the potential added advantage of observing changes that occur during organismal ageing [32][33][34][35][36]. On the other hand, in vitro models for tra cking studies, such as primary neuronal cultures and iPSCs, can be more easily manipulated and afford greater exibility for pharmacological and genetic manipulation [21, [37][38][39][40][41][42][43][44][45][46][47]. These models, however, have not yet been used extensively to study nucleoid dynamics, except in some instances [48]. Axonal tra cking of nucleoids has rarely been investigated and how mitochondrial dynamics affects the distribution of nucleoids in neuronal cells of human origin is currently unknown.
In this chapter, we describe a protocol to image mitochondrial and nucleoid dynamics in differentiated SH-SY5Y cells. This cell line is a clone of the SK-N-SH neuroblastoma cell line originally isolated from a human metastatic bone marrow [49] and can be differentiated to produce cells that express neuronal markers and have mature neuronal morphology [50,51]. The simplicity and e ciency of the differentiation protocol, combined with the human origin of this cell line, the ease of maintenance and mass expansion at relative low cost, make SH-SY5Y cells an advantageous alternative to iPSCs and primary neuronal cultures for routine studies of intracellular tra cking. This system can be easily established in most laboratories and the ndings obtained with this system could then be validated in primary neuronal cultures or in model organisms.

Methods
The SH-SY5Y cell line was shown to be composed of two cellular subpopulations: substrate adherent epithelial-like (S-type) and neuroblastic (N-type) cells, with the latter undergoing differentiation to produce neuronal-like cells [50,52]. Currently, a number of differentiation protocols exist which utilise various neuralising agents that can drive differentiation toward a more cholinergic, adrenergic, or dopaminergic phenotypes [53,54]. Due to the versatility of the differentiation protocols, SH-SY5Y cells have been extensively used as disease models for different neurological disorders. They have been largely employed to model several aspects of Parkinson's disease biology [53,55,56], but they were also exploited to study Alzheimer's disease-related insults [57] and neurodevelopment in the context of psychiatric disorders [58].
Retinoic Acid (RA)-mediated differentiation protocols have been shown to induce differentiation into cells with higher choline acetyltransferase activity [59][60][61], while other studies have focused on a more dopaminergic phenotype [62]. RA has also been often used in combination with other agents, such as 12-O-tetradecanoylphorbol-13-acetate (TPA), for a more e cient induction of cells with a dopaminergic phenotype, for example increased expression of tyrosine hydroxylase (TH), the dopamine transporter (DAT), and dopamine receptor subtypes 2 and 3 [63][64][65]. If desirable, treating SH-SY5Y cells with siRNA targeting the splicing factor PTBP1 prior to differentiation could be used to potentiate neurite outgrowth, as reported for SH-SY5Y cybrids [66].
For the purpose of studying tra cking, the combination of RA treatment with decreasing amounts of serum, followed by the addition of neurotrophic factors such as BDNF, has been shown to be particularly effective in promoting neurite extension during neuronal differentiation [67][68][69], with the end point of differentiation de ned as commitment to a neuronal phenotype with long extended processes [70]. SH-SY5Y cells expressing a truncated form of spastin have been used to study the tra cking of synaptophysin-positive vesicles [71]. Furthermore, mitochondrial tra cking was found to be perturbed in SH-SY5Y cells differentiated with RA and BDNF for 8 days in vitro and overexpressing the A53T αsynuclein mutant [68], indicating the suitability of the differentiation protocol to study tra cking in neuronal processes.
We differentiated SH-SY5Y cells by adapting a protocol published by Forster et al [69], which requires the addition of neuralising agents such as EC23 (a synthetic analog of all-trans retinoic acid [72][73][74]) and BDNF to sustain neurite outgrowth ( Figure 1A). Coating matrices, seeding densities and time in vitro were optimised to ensure the morphology of the cells was suitable for neurite tracking, i.e. cultures sparse enough to visualise long processes derived from a speci c cell body but without compromising their survival. Morphological assessments via phase contrast image acquisition ( Figure 1B) and staining of plasma membrane compartments ( Figure 1C), immunocytochemistry (ICC) and biochemical analysis ( Figure 2) validate our differentiation protocol. Figure 1B shows the progression of typical SH-SY5Y cultures throughout the differentiation protocol, with a noticeable increase in neurite extension, accompanied by clustering of cellular somata, already visible around DIV7 and with a clear network of neuronal-like processes around DIV14 (see Note 2). Figure 2 show substantial upregulation of neuronal markers (neuro laments, MAP2, Figure 2A-C), including markers for catecholaminergic neurons (tyrosine hydroxylase, Figure 2D) in differentiated cells compared to undifferentiated SH-SY5Y. A signi cant increase in the number of axonal-like, neuro lament positive neurites can be observed, which were also signi cantly longer than the sparser processes extended by undifferentiated cells (Figure 2B-C).
SH-SY5Y cells stained with the SYBR Gold nucleotide dye at low concentration [75] display a discrete punctuated pattern that colocalises with the mitochondrial network ( Figure 3A), indicative of speci c nucleoid staining. Overexpression of the uorescently tagged nucleoid protein TFAM, a major component of mammalian nucleoids and routinely used as nucleoid marker, results in extensive co-localisation between TFAM and the SYBR signals ( Figure 3B), thus con rming SYBR Gold as a suitable dye for the visualisation of nucleoids in these cells. In the neurites of differentiated cells, mitochondria and nucleoids exhibit robust motility with most mitochondria containing at least one nucleoid when observed by live-cell iSIM imaging (Figures 4 and 5 and Movies 1 and 2). Differentiated SH-SY5Y cells have been recently used for imaging the movement of synaptic vesicles by single-molecule localization microscopy [76]. Our protocol shows the suitability of this system for studying long-range co-transport of mitochondria and nucleoids by live super-resolution imaging. The method also lends itself to studying the partitioning and distribution of nucleoids during mitochondrial dynamics in the neurites of the cells. Figures 4 and 5 show that nucleoids can be distributed differently into daughter mitochondria during ssion in the processes, and mitochondria devoid of nucleoids can arise after mitochondrial division, as recently shown in Cos-7 cells and mouse cardiomyocytes [77].

Differentiation of SH-SY5Y cells into neuronal-like cells
1. On the rst day of the differentiation protocol, coat the wells of an 8 well Ibidi chamber with PDL matrix by adding 4 μL of 10 mg/mL PDL to 2 mL autoclaved MilliQ water and incubate at 37° C for at least 2 hours. We coat each Ibidi well (1 cm 2 ) with 250 μL of matrix solution (see Note 3).
2. Wash the wells 3x times with PBS and allow them to dry out prior to seeding cells. Ensuring the PDL has dried out will improve cell adhesion.
3. Add 1 mL of Trypsin-EDTA (approximately 0.014 ml/cm 2 ) to a T75 ask of undifferentiated SH-SY5Y cells and leave for 3 minutes to detach the cells. 7. The following day change the standard medium to 300 μL of Phase I medium containing the retinoic acid receptor agonist EC23. This starts the differentiation process and is referred to as DIV 1 ( Figure  1A).
. After 72 hours, on DIV 4, change medium to 300 μL Phase II containing BDNF to sustain neurite outgrowth. This is the medium used throughout the rest of the experiment and is changed every 3 to 4 days (see Note 5).
9. Assess cell differentiation daily by bright eld or phase contrast microscopy (Fig. 1B). At key timepoints ( Figure 1A), markers of cell differentiation can be veri ed by ICC, western blotting and CellMask staining (Fig. 2 and see Note 6).

Transfection of SH-SY5Y cells
If transfection is required to visualise uorescent organelles or proteins, we suggest performing it 48 hours before imaging (for instance, DIV 12 for imaging DIV 14). 3. During incubation, set the microscope temperature to 37°C and CO 2 to 5% (see Note 8).
4. After incubation, remove the dyes and wash wells 3x times with PBS to remove excess dye, which might otherwise aspeci cally stain other non-mitochondrial compartments (see Note 9).
5. Once nished with the washes, add fresh Phase II media to the cells.
. Go to the microscope and image the cells.
7. Focus on the cells by using the transmitted light (through eyepiece) and scan the well to nd an appropriate area which contains a good number of processes that can be traced back to their cell bodies (see Note 10).
. Switch to confocal mode (through microscope software) and observe the MitoTracker Deep Red signal on the computer via the 640 nm laser. Fine tune the focus and laser settings so that the mitochondria are clearly visible in the region of interest. Do the same for the SYBR Gold signal with the 488 nm laser (see Note 11 and Note 12).
9. Image the region of interest at a rate of 1fps for 1 minute (see Note 13). This will be saved as a ".nd2" le that contains two channels and 61 timeframes.

Analysis of mitochondria and nucleoids transport
1. Open the ".nd2" le with Fiji/ImageJ. Adjust the intensity settings with the "Brightness/Contrast" tool, if needed (Image> Adjust> Brightness/Contrast).
2. Inspect the images for any potential sample drifts that may have occurred during imaging. Align image frames using the StackReg plugin to correct the drift, if necessary.
3. Identify process you wish to analyse, select it with the "Segmented" or "Freehand" (right click to change in Line Selection Tool) and use the Straighten plugin to straighten it. When the pop-up window appears, adjust the "Line width" to ensure it includes the whole width of your process and select "Process Entire Stack". This will open a new window which shows the whole movie of the straightened process; the direction from which you select the process will result in its orientation in this window.
4. The straightened segment can now be used to create a kymograph with the Velocity Measurement Tool. Once the plugin has been launched, retrace the straightened process ("Straight line", press Shift while selecting to keep line straight) and press the "k" button which will have appeared on the toolbar (see Note 14).

Notes
1. Phase contrast imaging was performed on live samples at a constant temperature of 37°C and 5% CO 2 . We feel the use of live cells rather than xed sample gives a more accurate representation of the culture as a whole, especially in preparation for the tra cking experiments in live cells.
2. Formation of clear neuronal-like neurites was observed between DIV 10 and DIV 17, without a discernible difference in neurite length between DIV 14 and DIV 21. With this protocol, we found that cultures at DIV 21 started to appear less healthy, with a reduced number of processes and clusters of somata. They also tended to detach more easily from the glass surface during media changes. Thus, DIV 10-17 was determined to be the best interval for imaging experiments. Within this interval, we recommend using cells at the same DIV throughout an experimental series. We did not observe a decrease in the differentiation e ciency of SH-SY5Y cells cultured for many passages (> 40 passages in vitro).
3. We also tested coating the wells with 10 µg/mL laminin [78] but no signi cant differences in cell morphology and neurite extension were noticed when compared to PDL.
4. We use Ibidi μ-Slide 8-well Glass Bottom slides for live imaging experiments. In these chambers, the area of a well is 1 cm 2 , therefore the seeding density will be 15'000 cells/well. We tested a range of seeding densities to obtain cultures where the cells are sparse enough for the purpose of mitochondria and nucleoid tracking within single processes and to be able to unequivocally assign a process to the cell body it emanates from. At the same time, the cells need to be seeded at a su cient density as to not compromise cell survival. In 8-well Ibidi slides, densities lower than 10'000 cells/cm 2 lead to poor growth and occasional cell death after few days of culture, while densities above 20'000 cells/cm 2 result in cultures that are too compact for clear imaging of single processes. We recommend testing a range of seeding concentrations to nd the optimal density for imaging studies.
5. Because retinoic acid is used to reduce cell proliferation and induce neuronal differentiation, we tested leaving the cells in Phase I media for an additional 72 hours (i.e., throughout DIV 1-6) to prolong the incubation in EC23. Only at this point (DIV 7), we changed to Phase II. We expected this alternative protocol to bring forward the point of optimal differentiation. Surprisingly, we found that the culture reached full con uency by DIV 6 with many cells detaching during media changes. As a result, these cultures badly withstood transfection, with the majority of cells detaching prior to imaging.
It is worth noting that two cellular subtypes exist within the SH-SY5Y population: a predominant N-type cell population which becomes non-proliferative upon differentiation, and a small S-type subpopulation that continues to proliferate. Different differentiation methods can result in a different balance of proliferative and non-proliferative cell types [50,52].
For western blotting analysis, 5 µg of protein were run on 4-12% Bis-Tris acrylamide gels and transferred on PVDF membranes for 1-1.5 hours at 35 V. The membranes were incubated with anti-GAPDH (1:5'000, ThermoFisher) and anti-TH (1:200, Abcam) primary antibodies overnight at 4°C. Membranes were then incubated with HRP-conjugated secondary antibodies (1:5'000, Amerhsam) for 1.5 hours at room temperature, before being imaged with a Bio-Rad Chemidoc Imager. Exposure times to detect speci c immunoreactive protein bands were set between 5 and 10 minutes.
For plasma membrane staining with CellMask, cells were incubated with 300 μL/well of culture media supplemented with a 1:1'000 dilution of the commercial CellMask Green stock for 30 minutes. Prior to imaging, the dye was removed, wells were washed 3x times with PBS to remove excess dye, and kept in Phase II media post-wash and during imaging.
7. PEI and Lipofectamine were also tested during the optimisation of the transfection protocol but using NeuroMag resulted in higher transfection rate and healthier cell cultures. The transfection rate is signi cantly reduced in differentiated cells with all reagents, which is a further con rmation of the neuronal identity of these cells. We recommend using lentiviral transduction of differentiated cells for high expression of exogenous genes.
. It is advisable to turn on the temperature and CO 2 control with enough time before the start of the imaging session to allow the system to reach the desired temperature (37°C) and CO 2 (5%).
9. Be gentle while washing off the excess dye with PBS by pipetting into the side of the well as to not mechanically detach the cells from the bottom of the well, especially at higher DIV. From DIV 16-17, the cells appear to come off more easily during media changes. 10. Finding a eld of view where neurites can be traced back to their cell body with ease is essential for analysing anterograde and retrograde tra cking (i.e., away from and towards the cell soma, respectively). Using a cell membrane dye such as CellMask ( Figure 1C) could aid in doing so (see Note 6). 11. In time-lapse iSIM experiments in HeLa cells, SYBR Gold was used at a 1:10'000 dilution of the commercial stock (i.e., 1x) [75]. In our setup, we found that the range of SYBR dilution that yielded speci c and visible puncta varied between 1:100'000 and 1:1'000'000 of the commercial stock (i.e., 0.1x-0.01x). With further dilutions, the SYBR stain was barely detectable, while higher concentrations often produced bright and nonspeci c aggregates. We recommend optimisation of the working dilution prior to live imaging experiments.
12. SYBR Green [17] and PicoGreen [79] have been previously used as mtDNA dye. We therefore tested PicoGreen (Invitrogen, P7581) in addition to SYBR Gold. We observed co-localisation between overexpressed TFAM and PicoGreen-positive puncta at concentrations as low as 1.5 µL/mL, which con rms the suitability of this dye to stain nucleoids in SH-SY5Y cells. SYBR Gold was preferred for quanti cation of tra cking and dynamics due to the higher resolution and more de ned puncta observed. As with SYRB Gold, the DIV of the cells affects the quality of PicoGreen staining, with cultures at higher DIV more commonly showing non-speci c staining in the form of larger extracellular accumulations of dye.
13. To avoid bleaching, nd the lower exposure time and laser power at which the mitochondria and the nucleoids are clearly detectable. We observe minimal phototoxicity with our settings of 30% laser power, 100 ms exposure time for both channels and 1 fps acquisition rate. A high acquisition rate captures more detailed movements which is bene cial when observing the dynamics of small structures such as nucleoids, although it comes at the expense of phototoxicity.
14. Other plugins and kymograph analysis bundles are suitable to produce Kymographs, for instance,    Figure 4, the nucleoid puncta