Abstract
Myelinating oligodendrocytes die in human disease and early in aging. Despite this, the mechanisms that underly oligodendrocyte death are not resolved and it is also not clear whether these mechanisms change as oligodendrocyte lineage cells are undergoing differentiation and maturation. Here, we used a combination of intravital imaging, single-cell ablation, and cuprizone-mediated demyelination, in both female and male mice, to discover that oligodendrocyte maturation dictates the dynamics and mechanisms of cell death. After single-cell phototoxic damage, oligodendrocyte precursor cells underwent programmed cell death within hours, differentiating oligodendrocytes died over several days, while mature oligodendrocytes took weeks to die. Importantly cells at each maturation stage all eventually died but did so with drastically different temporal dynamics and morphological features. Consistent with this, cuprizone treatment initiated a caspase-3–dependent form of rapid cell death in differentiating oligodendrocytes, while mature oligodendrocytes never activated this executioner caspase. Instead, mature oligodendrocytes exhibited delayed cell death which was marked by DNA damage and disruption in poly-ADP-ribose subcellular localization. Thus, oligodendrocyte maturation plays a key role in determining the mechanism of death a cell undergoes in response to the same insult. This means that oligodendrocyte maturation is important to consider when designing strategies for preventing cell death and preserving myelin while also enhancing the survival of new oligodendrocytes in demyelinating conditions.
Significance Statement
Oligodendrocyte death and axonal demyelination are hallmarks of disease and aging, contributing to cognitive decline in both contexts. However, little is known about the molecular mechanisms that lead to the degeneration. Here, we used intravital imaging coupled with models of demyelination to identify a shift in the dynamics and mechanism of oligodendrocyte death. This shift was directly related to the maturation state of the individual cell as differentiating oligodendrocytes died in days via apoptotic caspase-3–dependent mechanisms while mature oligodendrocytes took weeks to die through a caspase-3–independent mechanism instead marked by DNA damage and disruption in PAR subcellular localization. Thus, oligodendrocyte maturation directly impacts the mechanisms underlying cell death and demyelination.
Introduction
The cognitive impairment that occurs in aging and neurodegeneration can be partially attributed to the death of myelinating oligodendrocytes (Meier-Ruge et al., 1992; Tang et al., 1997; Kohama et al., 2012). Several factors initiate myelin damage and oligodendrocyte death in these pathological contexts. First, established and long-lived oligodendrocytes (Tripathi et al., 2017) accumulate oxidative and DNA damage (Al-Mashhadi et al., 2015; Tse and Herrup, 2017). In parallel a proinflammatory environment can cause increased cell damage and death to oligodendrocytes (Clarke et al., 2018; Edler et al., 2021; Lee and Kim, 2022). Simultaneously, these same microenvironmental and age-related contexts limit the ability of OPCs to generate new mature myelinating oligodendrocytes, resulting in failed remyelination (Bartzokis, 2004; Peters and Kemper, 2012; Franklin and Ffrench-Constant, 2017; Hill et al., 2018; Hughes et al., 2018; Neumann et al., 2019; Chapman and Hill, 2020). These factors initiate a feed-forward degenerative cascade that exacerbates myelin and oligodendrocyte loss.
Multiple models are available to study oligodendrocyte death, with cuprizone-mediated demyelination being the most widely used (Hooijmans et al., 2019). However, despite it having existed for decades, the precise mechanisms that govern oligodendrocyte death in this model are not fully understood (Matsushima and Morell, 2001; Gudi et al., 2014; Praet et al., 2014; Zirngibl et al., 2022). Moreover, while it is generally accepted that oligodendrocyte precursor cells (OPCs) are not directly impacted by cuprizone treatment, whether cuprizone has similar impacts on cells at different stages of oligodendrocyte maturation has not been directly addressed. This is important as oligodendrocyte maturation has been shown to impact cell vulnerability in the context of developmental hypoxia/ischemia (Back et al., 1998, 2001) and cuprizone could conceivably similarly impact the ability of new oligodendrocytes to successfully mature and remyelinate in addition to causing the death of mature myelinating oligodendrocytes. Recent work has shown that oligodendrocyte death can be surprisingly slow compared with other cell death events (Pohl et al., 2011; Oluich et al., 2012; Bacmeister et al., 2020; Orthmann-Murphy et al., 2020; Snaidero et al., 2020; Chapman et al., 2023) further raising the question of what the underlying cell death mechanisms are in cuprizone and other models and whether similar mechanisms are involved in human disease and aging.
Here, we describe how oligodendrocyte maturation shifts the form of death that cells within the oligodendrocyte lineage undergo in response to single-cell DNA damage via two-photon apoptotic targeted ablation (2Phatal) and cuprizone-mediated demyelination. Using 2Phatal, we uncovered a shift in the dynamics of cell death that occurs during the differentiation of OPCs into newly formed and mature oligodendrocytes. We go on to show that a similar effect occurs in the cuprizone model, suggesting that the observed shift represents an intrinsic change in how oligodendrocytes respond to different noxious stimuli. Differentiating oligodendrocytes, encompassing both premyelinating and newly formed oligodendrocytes, underwent a caspase-3–mediated form of cell death while mature oligodendrocytes died through a distinct, caspase-independent, mechanism characterized by DNA damage and disruption in poly-ADP-ribose (PAR) localization. Importantly, this was not a change in cell vulnerability to these insults but instead revealed a switch in the morphological and temporal progression of how the cells at each maturation stage died. Previous work has described DNA damage in oligodendrocytes in human aging and multiple sclerosis (Haider et al., 2011; Al-Mashhadi et al., 2015) suggesting that these cell death mechanisms could be playing a role in human pathology. Taken together, these observations demonstrate a switch in the mechanisms of death individual oligodendrocytes undergo as they differentiate and mature. This suggests that distinct therapeutic strategies may be required to protect both the genesis and integration of differentiating oligodendrocytes as well as the preservation of previously established mature oligodendrocytes.
Materials and Methods
Animals
All animal procedures were submitted to and approved by the Institutional Animal Care and Use Committee at Dartmouth College. The following mouse strains were purchased from the Jackson Laboratory and crossed to generate the triple transgenic mice used in this study: Cnp-mEGFP (Deng et al., 2014; JAX #026105), Cspg4-creER (Zhu et al., 2011; JAX #008538), floxed tdTomato Ai9 (Madisen et al., 2010; JAX #007909). Cspg4-creER:tdTomato mice were injected with 50 μl of tamoxifen (20 mg/ml in corn oil, Sigma-Aldrich #T5648) 2 d prior to the start of experiments, to induce recombination of floxed tdTomato in OPCs. All mouse strains used have a C57bL/6 background. Mice were housed in a 12 h light/dark cycle in a temperature (22°C) and humidity-controlled (30–70% relative humidity) animal vivarium with food and water provided ad libitum. Mice used for fixed tissue cuprizone experiments were 6–8-week-old male, wild-type C57bL/6 mice purchased from the Jackson Laboratory. Mice used for 2Phatal and cuprizone intravital experiments were both male and female aged 6–8 weeks at the start of the experiments.
Surgical procedures
All intravital imaging was done using chronic cranial window surgical preparations. Briefly, animals were anesthetized with a combination of ketamine (100 mg/kg) and xylazine (10 mg/kg), given intraperitoneally. The skin over the skull was shaved, sterilized, and removed. A ∼3 mm craniotomy was performed using a high-speed drill, and the skull was replaced by a #0 cover glass. For 2Phatal, Hoechst 33342 nuclear dye (Thermo Fisher Scientific #H3570) was diluted to 50 µg/ml in sterile PBS and applied to the pial surface prior to placing the cover glass. A nut was attached to the skull using cyanoacrylate glue to facilitate repeated imaging. The remaining surface of the exposed skull was covered with dental cement. Analgesia was provided by subcutaneous injection of carprofen (50 mg/kg) immediately pre- and postsurgery and at 24 and 48 h.
Imaging
Intravital fluorescence imaging was performed on either a two-photon (Bruker) microscope with an InSight X3 femtosecond pulsed laser (Spectra-Physics) using a 20× water immersion objective (Zeiss NA 1.0) or an upright laser scanning confocal (Leica SP8) microscope using a 20× water immersion objective (Leica NA 1.0). For one-photon fluorophore excitation, 488 nm laser light was used for membrane EGFP (mEGFP), and 552 nm laser light was used for tdTomato. All images with multiple channels were acquired sequentially. For two-photon microscopy, 775 nm laser light was used for Hoechst dye, 920 nm for mEGFP, and 1,040 nm for tdTomato. Z-stacks were acquired using a step size of 1.5 µm. All fixed tissue fluorescence imaging was done on an upright laser scanning confocal (Leica SP8) microscope, equipped with either a 20× air objective (Leica NA 0.75) or a 63× oil immersion objective (Leica NA 1.4). All images with multiple channels were acquired sequentially, from longest to shortest wavelengths. Deconvolution was used on high-magnification images (Leica Lightning software).
2Phatal
To perform two-photon apoptotic targeted ablations (2Phatal), cranial windows were prepared as previously described. Hoechst 33342 was applied topically (0.1 mg ml−1 diluted in PBS) to the pial surface of the brain and allowed to sit for 5 min. The cover glass was then secured with cyanoacrylate glue. The windows were allowed to recover for 24 h to ensure widespread nuclear labeling. Positions were imaged sequentially with 775, 920, and 1,040 nm wavelengths, and the images merged to enable the identification of nuclei from OPCs and oligodendrocytes. To induce 2Phatal, we tuned the laser to 775 nm wavelength, with a dwell time of 100 µs, and an ROI (8 × 8 µm2) was centered on the nucleus of interest. Identical time series parameters were used for each ablation of 125 scans, lasting 3.72 s. Following 2Phatal, the positions were reimaged 8 and 24 h later and then once daily after the first day. Cells to be targeted were identified based on the presence of tdTomato signal and no mEGFP signal for OPCs and the presence of mEGFP signal for oligodendrocytes. One to nine cells were targeted per position based on the density of labeled cells and the nuclear labeling.
Cuprizone-mediated demyelination
For longitudinal imaging experiments, cranial windows were performed on Cnp-mEGFP mice 3 weeks before the start of cuprizone treatment. Baseline images were acquired on Day 0, and then the mice were placed on a 0.2% (w/w) cuprizone (Sigma-Aldrich #C9012) diet mixed with ground chow. Food was changed every 2–3 d. Mice were returned to an unadulterated ground chow diet after 6 weeks. For fixed tissue cuprizone experiments, 6-week-old C57bL/6 male mice were placed on either a 0.2% (w/w) cuprizone diet mixed with ground chow or an unadulterated ground chow diet. Four mice from both the control and cuprizone groups were euthanized 1, 3, and 5 weeks after the start of the treatment.
Tissue processing
For tissue dissection, mice were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg), given intraperitoneally, and transcardiac perfusion was performed with ∼30 ml of 4% PFA in PBS. The brain was then dissected and placed in 4% PFA overnight at 4°C for postfixation. After ∼24 h, the tissue was transferred to PBS and stored at 4°C. In addition, 75 µm coronal sections were cut using a vibratome. The tissue was then placed in a glycerol-based cryogenic storage solution at −20°C.
Immunofluorescence staining
Slices were washed three times in PBS for 10 min to remove the cryogenic storage solution. Antigen retrieval was then performed. Slices were placed in a preheated 98°C solution of 10 mM Tris, 1 mM EDTA, and 0.05% Tween 20 in PBS buffer at pH 9 for 3 min. They were then allowed to cool to room temperature in the solution. The slices were then transferred to the primary antibody solution in a blocking buffer. The blocking buffer contained 0.3% Triton X-100 and 1% bovine serum albumin (BSA) and was stored at 4°C overnight. Slices were washed three times in PBS for 10 min and transferred to the secondary antibody solution in a blocking buffer for 1 h at room temperature. Slices were then washed again for 10 min three times in PBS. Slices were incubated in a 1:2,000 Hoechst solution in 0.3% Triton X-100 in PBS for 20 min before being mounted using ProLong diamond mounting media. For PAR labeling, slices were incubated in 10% goat serum and 0.3% Triton X-100 for 2 h at room temperature after antigen retrieval, before overnight incubation in the primary antibody solution in PBS at 4°C. Slices were washed three times in PBS for 10 min and incubated for 1 h at room temperature in the secondary antibody solution in PBS. Final washes, Hoechst incubation, and mounting were performed as above.
The following primary antibodies were used for immunofluorescence staining: mouse anti-CNPase (BioLegend, 1:500, catalog #836404), guinea pig anti-CNPase (Synaptic Systems, 1:500, catalog #355004), rabbit anti-BCAS1 (Synaptic Systems, 1:750, catalog #445003), guinea pig anti-BCAS1 (Synaptic Systems, 1:750, catalog #455004), mouse anti-carbonic anhydrase II (Santa Cruz Biotechnology, 1:500, catalog #48351), goat anti-PDGFRa (R&D Systems, 1:1,000, catalog #AF1062), rabbit anti-cleaved caspase-3 (Cell Signaling Technology, 1:700, catalog #9661), and rabbit anti-gammaH2AX (Cell Signaling Technology, 1:500, catalog #9718). The secondary antibodies were conjugated to Alexa Fluor 488, 555, or 647 as necessary (Thermo Fisher Scientific, 1:500). For PAR labeling, anti–PAR binding reagent (Sigma-Aldrich, 1:500, catalog #MABE1031) was used. Aliquots were only used once, as freeze-thaw cycles severely reduced labeling efficacy.
Intravital imaging quantification
Changes in soma size over time of differentiating oligodendrocytes, in Cnp-mEGFP:Cspg4-creER:tdTomato mice (Fig. 1c), were quantified using cytoplasmic tdTomato signal. Images were acquired as z-stacks. The slice with the largest soma area was used and the soma was outlined in Fiji. Processes were excluded from the soma area. For the quantification of mEGFP fluorescence in Figure 1d, analysis was performed on the same z-slice used for the soma area quantification. The average mEGFP fluorescence of the soma was obtained in Fiji. This was normalized to adjacent background fluorescence intensity in the mEGFP channel. Three background measurements were obtained immediately adjacent to the cell soma. These measurements were averaged, and this average was used to normalize the soma fluorescence value. For the quantification of cell death, single cells were followed throughout the imaging time series, and the day of death was designated as the first day when the cell soma was absent. Raw data for mature oligodendrocytes were reanalyzed from our previous publication (Chapman et al., 2023) with regard to the day of death in both 2Phatal and cuprizone. New analyses were conducted to determine soma size and mEGFP fluorescence intensity for these cells.
Identifying actively differentiating oligodendrocytes in vivo. a, In vivo images of oligodendrocytes and OPCs in Layer 1 somatosensory cortex of a Cnp-mEGFP:Cspg4-creER:tdTomato transgenic mouse. Oligodendrocytes are labeled with membrane-tethered EGFP (yellow arrowhead), and OPCs are labeled with cytosolic tdTomato (white arrowhead). A differentiating oligodendrocyte is identified by faint EGFP fluorescence and a large soma (yellow arrow). b, Representative time series of an OPC differentiating into an oligodendrocyte (arrowhead). EGFP fluorescence signal appears during the differentiation process. c, Trace of normalized EGFP fluorescence over time from the representative time series in (b; green). Trace of soma area of the differentiating cell in (b) over time (magenta). d, Scatter plot of the soma area as a function of normalized EGFP fluorescence of oligodendrocytes prior to 2Phatal photobleaching (n = 50 cells, nine animals). The soma area and normalized fluorescence were obtained using only the Cnp-mEGFP signal. Unbiased k-means clustering identified two populations of cells. The X's label the centroids of the two populations, differentiating oligodendrocytes (orange, n = 26 cells, five mice) and mature oligodendrocytes (green, n = 24 cells, four mice).
Fixed tissue quantification
All fixed tissue analysis was performed using Fiji. Cells were counted, and then the volume was calculated to enable the calculation of cell densities. All densities reported were obtained from the cerebral cortex. Experimenter blinding was used for all quantification analyses. For each animal used, two to three coronal sections were imaged, each two to three times, for a total of 6–9 images per mouse per staining condition, except for Figure 4 where five sections per animal were imaged each two to three times for a total of 15 images quantified. The densities of each of these images were then averaged to produce a single-cell density value used for the reported quantification.
Statistical analyses
All statistical tests were performed in GraphPad Prism or MATLAB, and the tests used are indicated in the figure legends and Results section. Statistical significance is indicated as p values <0.05 with * indicating p ≤ 0.05, ** indicating p ≤ 0.01, *** indicating p ≤ 0.001, and **** indicating p ≤ 0.0001.
Results
Oligodendrocyte maturation alters the temporal dynamics of programmed cell death
2Phatal enables an on-demand initiation of programmed cell death in single cells in vivo (Hill et al., 2017). We used this precision to target all the cell types across the oligodendrocyte lineage and investigated their cell death dynamics. Before proceeding with this approach, we first needed to precisely identify the maturation stage of the cells under investigation. We used Cnp-mEGFP:Cspg4-creER:tdTomato mice that label all myelinating oligodendrocytes with mEGFP along with OPCs and their progeny with tdTomato after tamoxifen-induced cre recombination (Fig. 1a; Deng et al., 2014; Hill et al., 2018; Chapman et al., 2023). In addition to labeling OPCs and mature myelinating oligodendrocytes, this mouse line also allows the visualization of the cells that are actively differentiating. To identify and define this population, we visualized the differentiation of single OPCs with longitudinal intravital imaging and characterized changes in cell morphology and fluorescence intensities (Fig. 1b). As expected, during active differentiation, the Cnp-mEGFP signal increased over the course of several days before becoming stable. The cell somas also showed transient and consistent increases in area during differentiation (Fig. 1c). Therefore, plotting the soma area as a function of mEGFP fluorescence intensity revealed two oligodendrocyte cell populations, which could be separated using unbiased k-means clustering (Fig. 1d). In this study, we termed these two cell stages as differentiating oligodendrocytes and mature oligodendrocytes. This allowed us to separate our analysis into two oligodendrocyte stages along with OPCs, which were identified by their distinct morphology and lack of mEGFP fluorescence.
2Phatal works by causing DNA damage in single cells through the photobleaching of nuclear dye in the targeted cell (Hill et al., 2017; Chapman et al., 2023). Hoechst 33342 dye was applied to the cortical surface of the brain during cranial window surgery in Cnp-mEGFP:Cspg4-creER:tdTomato mice. This enabled us to identify nuclei of OPCs, differentiating oligodendrocytes, and mature oligodendrocytes (Fig. 2). Nuclear dye labeling in these cell populations showed differences in initial fluorescence intensity with mature oligodendrocytes labeling the brightest (Fig. 2c). After single-cell nuclear dye photobleaching, all targeted OPCs underwent an apoptotic-like cell death within 24 h with nuclear fragmentation and the formation of apoptotic bodies, as previously described (Hill et al., 2017). In contrast, mature oligodendrocytes degenerated over an extended period of ∼45 d (Chapman et al., 2023) without clear morphological characteristics of classical apoptosis, instead exhibiting consistent soma condensation and sheath retraction. On the other hand, differentiating oligodendrocytes demonstrated features and temporal dynamics more like OPCs with many dying in 24 h while others remained for several days before eventually disappearing (Fig. 2b). While all the cells died, there were significant differences across these three groups regarding the temporal dynamics of cell death progression (Fig. 2d, one-way ANOVA, Tukey's multiple-comparisons test). Importantly, the nuclear labeling did not correlate with their ultimate death progression as the brightest nuclei (with the greatest potential for DNA damage from 2Phatal photobleaching) were the mature oligodendrocytes (Fig. 2c,d). Thus, the shift in cell death dynamics occurs during differentiation resulting in a population of differentiating oligodendrocytes that display the characteristics of OPCs but take longer to die from similar DNA damage protocols.
Oligodendrocyte maturation alters the temporal dynamics of induced programmed cell death. a, Representative in vivo images in a Cnp-mEGFP:Cspg4-creER:tdTomato transgenic mouse. Examples of an OPC (Box 1), differentiating oligodendrocyte (Box 2), and mature oligodendrocyte (Box 3) are shown. The cropped boxes are single z sections through the center of the cell soma to indicate the dim mEGFP signal in the differentiating oligodendrocyte (blue arrows). b, Representative time series of cells within the oligodendrocyte lineage targeted with 2Phatal. From top to bottom: OPC, differentiating oligodendrocyte (arrowhead identifies the cell soma), differentiating oligodendrocyte, and a mature oligodendrocyte. c, Initial nuclear dye fluorescence intensity before 2Phatal photobleaching of OPCs (magenta, n = 34 cells, four mice), differentiating oligodendrocytes (orange, n = 26 cells, four mice), and mature oligodendrocytes (n = 82 cells, six mice, one-way ANOVA with Tukey's correction for multiple comparisons; error bars indicate SEM; F(2,139) = 8.084). d, Average day of death after 2Phatal of OPCs (magenta, n = 34 cells, four mice), differentiating oligodendrocytes (orange, n = 26 cells, four mice), and mature oligodendrocytes (Chapman et al., 2023; green, n = 24 cells, four mice; one-way ANOVA with Tukey's correction for multiple comparisons, error bars are SEM; F(2,86) = 19.38).
Oligodendrocyte maturation defines the cell death dynamics to cuprizone-mediated demyelination
We next determined whether oligodendrocyte maturation impacted the degeneration of single cells in the cuprizone model of demyelination (Praet et al., 2014). Cnp-mEGFP mice were placed on a diet of 0.2% (w/w) cuprizone for 6 weeks, and we performed intravital imaging of Layer 1 of the somatosensory cortex. As expected, this regimen produced widespread cortical demyelination after 6 weeks (Fig. 3a). We identified differentiating and myelinating oligodendrocytes using the criteria described above (Fig. 3c,d, n = 40 cells, three mice). Mature oligodendrocytes degenerated over an extended period of time, 34 d on average (n = 32 cells, three mice), as previously described by our lab (Chapman et al., 2023). Like 2Phatal, however, differentiating oligodendrocytes underwent cell death more rapidly than myelinating oligodendrocytes, with an average degeneration time of 14 d (Fig. 3e, n = 8 cells, three mice, unpaired t test).
Differentiating but not mature oligodendrocytes die rapidly during cuprizone treatment. a, In vivo timelapse images showing cuprizone-induced demyelination and oligodendrocyte loss in Layer 1 somatosensory cortex over 6 weeks, dwc = days with cuprizone. b, Representative time series showing mature oligodendrocyte death (top), differentiating oligodendrocyte death (middle), and the death of an oligodendrocyte that appears after the start of cuprizone treatment (bottom). c, Scatter plot of the soma area as a function of normalized EGFP fluorescence of oligodendrocytes prior to the start of cuprizone (n = 40 cells, three animals). Unbiased k-means clustering identified two populations of cells. The X's label the centroids of the two populations, differentiating oligodendrocytes (orange, n = 8 cells, three mice) and mature oligodendrocytes (green, n = 32 cells, three mice). d, Scatter plot made by plotting the ratio of normalized fluorescence to the soma area as a function of time to death of each cell, during cuprizone demyelination. Unbiased k-means clustering identified two populations of cells. The X's label the centroids of the two populations. Differentiating oligodendrocytes (orange, n = 9 cells, three mice) and mature oligodendrocytes (green, n = 31 cells, three mice). e, Average time to cell death of differentiating oligodendrocytes (orange, n = 8 cells, three mice) and mature oligodendrocytes (green, n = 32 cells, three mice). The two populations were defined based on the k-means clustering in (c; unpaired t test; error bars indicate SEM; F(31,7) = 2.151).
We did occasionally observe new oligodendrocyte formation over the course of the cuprizone experiment (Fig. 3b, bottom). These new oligodendrocytes behaved like differentiating cells identified at the start of the experiment, dying more rapidly than previously established mature oligodendrocytes, further indicating that the maturation state dictates the temporal dynamics of cell death between differentiating and mature oligodendrocytes.
Caspase-dependent cell death is limited to differentiating oligodendrocytes
Given the differences in cell death dynamics between differentiating and mature oligodendrocytes revealed with intravital imaging, we next investigated the differences in defined molecular markers of cell death across the oligodendrocyte lineage. To do this, we performed another cuprizone experiment in 6-week-old C57bl6/j mice. Mice were placed on a 0.2% (w/w) cuprizone or control diet. Animals were euthanized after 1, 3, and 5 weeks. The brains were then dissected, fixed, and sliced to enable immunofluorescence staining. All imaging for these experiments was done in the somatosensory cortex.
It was not possible to use the criteria that defined the oligodendrocyte lineage cell types in our intravital imaging without Cnp-mEGFP expression. This meant we had to define the same populations based on molecular markers. We used platelet-derived growth factor receptor α (PDGFRα, OPCs), breast carcinoma–amplified sequence 1 (BCAS1, differentiating oligodendrocytes), 2′,3′-cyclic nucleotide 3′-phosphodiesterase (CNP, differentiating and mature oligodendrocytes), and carbonic anhydrase II (CAII, mature oligodendrocytes) to define the specific populations (Fig. 4a,b; Cammer, 1984; Butt et al., 1995; Fard et al., 2017). Using these markers, we were able to distinguish between differentiating and mature oligodendrocytes, despite both populations expressing CNP (Fig. 4b). We also observed that the soma area was larger in the differentiating populations compared with that in the mature oligodendrocyte population, in agreement with our intravital data (Fig. 4b). However, this characteristic was not used as a defining criterion in our fixed tissue analysis.
Cuprizone causes failed integration of new oligodendrocytes and delayed death of mature oligodendrocytes. a, Diagram showing the stages of differentiation of the oligodendrocyte linage and the markers used to identify each cell type. The lines represent the relative gene expression of each marker in time. b, Representative images of cells within the oligodendrocyte lineage. From left to right: premyelinating oligodendrocyte (BCAS1+, CNP−, CAII−), newly formed oligodendrocyte (BCAS1+, CNP+, CAII−), and a mature oligodendrocyte (BCAS1−, CNP+, CAII+). c, Representative images and quantification of oligodendrocyte and myelin loss during cuprizone. Myelin and oligodendrocyte densities increase over time in the control condition (n = 4 mice), while myelin and oligodendrocyte densities decrease in the cuprizone condition (n = 4 mice). d, Representative images of CNP and CAII immunostaining in control (top) and cuprizone (bottom) tissue from Week 3 animals. Zoomed-in images showing differentiating oligodendrocytes (CNP+ CAII−, arrowheads) and mature oligodendrocytes (CNP+ CAII+, arrows). e, Quantification of the density of differentiating oligodendrocytes (CNP+ CAII−) and mature oligodendrocytes (CNP+ CAII+) in control (gray, n = 4 mice) and cuprizone (n = 4 mice) over time (two-way ANOVA with Tukey's correction for multiple comparisons; error bars indicate SEM; across time F(2,36) = 4.687; control to cuprizone comparisons F(3,36) = 1,022).
Demyelination was determined by quantifying the density of CNP+ oligodendrocytes in cuprizone and control conditions (Fig. 4c). As expected, there was substantial oligodendrocyte loss and demyelination after 5 weeks of cuprizone. The oligodendrocyte density in the control condition increased slightly over the 5 weeks, which is consistent with oligodendrogenesis in 8-week-old mice (Fig. 4c). Using the CNP and CAII markers, we next wanted to understand how both differentiating and mature oligodendrocyte populations were contributing to these changes in oligodendrocyte density (Fig. 4d). There was no difference in differentiating oligodendrocyte density between control and cuprizone conditions at any time point (Fig. 4e). This suggests that cuprizone does not affect the rate of OPC differentiation into oligodendrocytes. However, unlike the control condition, where differentiating OPCs were able to integrate, mature, and increase the density of CNP+ CAII+ oligodendrocytes over time, differentiating oligodendrocytes in cuprizone were not able to properly integrate and mature, thus resulting in a net decrease in the production of new mature CNP+ CAII+ oligodendrocytes (Fig. 4e).
Previous work has described the appearance of cleaved caspase-3 (CC3)–positive cells in the cuprizone model (Hesse et al., 2010). Interestingly, that work identified two distinct populations of degenerating oligodendrocytes, based on nuclear fragmentation with and without CC3 labeling. We sought to understand the role of CC3 in the cuprizone-induced cell death progression of cell populations across the oligodendrocyte lineage. Cuprizone and control tissues were stained for CC3, CNP, and CAII (Fig. 5a). Control tissue showed consistently low densities of CC3+ cells at each time point. Tissue from cuprizone-treated mice displayed significantly higher densities of CC3+ cells than controls at all three time points. In agreement with previous works (Hesse et al., 2010; Zirngibl et al., 2022), CC3+ density was highest after 7 d of cuprizone and sharply fell at Weeks 3 and 5, although CC3+ densities continued to be higher in cuprizone than controls at these later time points (Fig. 5c, n = 4 mice, two-way ANOVA, Tukey's multiple-comparisons test).
Differentiating but not mature oligodendrocytes die via a caspase-3–dependent form of cell death in cuprizone. a, Representative images of a premyelinating oligodendrocyte stained for CNP, CC3, and CAII (arrowhead). The lack of CAII combined with a dim CNP signal identifies this cell as a premyelinating oligodendrocyte. b, Representative images of a newly formed oligodendrocyte stained for CNP, CC3, and CAII (white arrowheads). The cell has visible CNP+ myelin sheaths (white arrowheads), but it lacks CAII, identifying this cell as a newly formed oligodendrocyte. c, Quantification of CC3+ cells in the control (gray, n = 4 mice) and cuprizone (green, n = 4 mice) conditions (two-way ANOVA with Tukey's correction for multiple comparisons; error bars indicate SEM; across time F(2,18) = 26.93; control to cuprizone comparisons F(1,18) = 125.6). d, Comparison of the density of CC3+ differentiating oligodendrocytes (CNP+ CAII−) and CC3+ mature oligodendrocytes (CNP+ CAII+) in control (gray, n = 4 mice) and cuprizone (green, n = 4 mice) conditions. Each graph is a different week of the experiment [one-way ANOVA with Tukey's correction for multiple comparisons; error bars indicate SEM; F(3,12) = 8.496 (Week 1), 3.780 (Week 3), 14.92 (Week 5)].
CC3+ cells were then further separated based on CNP (differentiating and mature oligodendrocytes) and CAII (mature oligodendrocytes) labeling. There was no difference in CC3 labeling in mature oligodendrocytes between cuprizone-treated animals and controls at any time point (Fig. 5d, n = 4 mice, one-way ANOVA, Tukey's multiple-comparisons test). In both conditions, CC3+ CNP+ CAII+ densities were either zero or near zero, indicating that mature oligodendrocytes do not contribute to the CC3+ cells observed in cuprizone. In contrast, CNP+ CAII− differentiating oligodendrocytes showed increased CC3 labeling in cuprizone compared with controls at all time points (Fig. 5d, n = 4 mice, one-way ANOVA, Tukey's multiple-comparisons test). These data demonstrate that CC3 activation in cuprizone occurs exclusively in differentiating oligodendrocytes, while mature oligodendrocytes undergo cell death through a caspase-3–independent mechanism.
Cuprizone causes double-strand DNA breaks in mature oligodendrocytes
Oxidative stress and DNA damage in oligodendrocytes have been described in response to cuprizone and aging (Al-Mashhadi et al., 2015; Fischbach et al., 2019). We next determined whether the occurrence of DNA damage differed across the oligodendrocyte lineage. The histone protein H2AX has been shown to be phosphorylated at the ser139 residue in response to double-strand DNA breaks (Rogakou et al., 1998; Fernandez-Capetillo et al., 2004). This modified version of the protein is referred to as γH2AX (Dickey et al., 2009). As before, we stained tissue from control and cuprizone-treated mice for CNP (myelinating oligodendrocytes) and CAII (mature oligodendrocytes), combined with γH2AX (Fig. 6a). As expected, γH2AX signal was confined to the nucleus of positive cells (Fig. 6b). Unlike our previous results with CC3, γH2AX+ cell density was only elevated after 1 week of cuprizone compared with control. There was no significant difference in γH2AX+ cell density between cuprizone and control at Weeks 3 and 5 (Fig. 6c, two-way ANOVA, Tukey's multiple-comparisons test).
Cuprizone induces double-strand DNA breaks in mature but not differentiating oligodendrocytes. a, Representative images of CNP, γH2AX, and CAII staining in a mouse treated with cuprizone. The cells that are positive for γH2AX colabel with CNP and CAII. b, High-magnification images of two mature oligodendrocytes. The top cell nucleus is positive for γH2AX, whereas the bottom cell nucleus is not. c, Quantification of γH2AX+ cells in the control (gray, n = 4 mice) and cuprizone (green, n = 4 mice) conditions (two-way ANOVA with Tukey's correction for multiple comparisons; error bars indicate SEM; across time F(2,18) = 17.19; control to cuprizone comparisons F(1,18) = 31.05). d, Comparison of the density of γH2AX+ differentiating oligodendrocytes (CNP+ CAII−) and γH2AX+ mature oligodendrocytes (CNP+ CAII+) in control (gray, n = 4 mice) and cuprizone (green, n = 4 mice) conditions. Each graph is a different week of the experiment [one-way ANOVA with Tukey's correction for multiple comparisons; error bars indicate SEM; F(3,12) = 10.70 (Week 1), 15.03 (Week 3), 8.558 (Week 5)].
We next separated the populations of differentiating oligodendrocytes (CNP+ CAII−) and mature oligodendrocytes (CNP+ CAII+) to determine if there was a difference with respect to γH2AX. We did not observe a single γH2AX+ CNP+ oligodendrocyte in the control condition at any time point. This is perhaps not surprising as the occurrence of double-strand breaks in the healthy brain should be low. However, we also identified only six total CNP+ CAII− γH2AX+ oligodendrocytes across all three time points in the cuprizone condition (Fig. 6d). The observed γH2AX+ cells in the cuprizone condition were almost exclusively from the mature oligodendrocyte (CNP+ CAII+) cell population (Fig. 6d, n = 4 mice, one-way ANOVA, Tukey's multiple-comparisons test). These data suggest a transient spike in cuprizone-induced DNA damage specifically in mature and not differentiating oligodendrocytes over the first week of cuprizone treatment, which rapidly subsides. Importantly, this occurs well in advance of actual oligodendrocyte degeneration and demyelination (Fig. 4).
Cuprizone disrupts PAR subcellular localization in mature but not differentiating oligodendrocytes
The DNA damage response protein poly-ADP-ribose polymerase 1 (PARP1) has been implicated in oligodendrocyte degeneration: its product, PAR, is upregulated in the cytoplasm of oligodendrocytes in MS lesions and cuprizone-treated mice, and blocking PARP1 activity is partially protective against cuprizone-induced demyelination (Veto et al., 2010). However, it is unclear if differentiating and mature oligodendrocytes are equally affected; therefore, we next determined whether PAR differed across oligodendrocyte maturation stages in response to cuprizone demyelination. To do so, we stained the control and cuprizone tissues for PAR, CNP (myelinating oligodendrocytes), and CAII (mature oligodendrocytes, Fig. 7a) and examined PAR subcellular localization. We found that PAR widely labeled cortical cell nuclei at baseline (Fig. 7a), as expected given the roles for PARylation in gene expression (Ke et al., 2019). In control animals, PAR was confined to the nucleus in most differentiating oligodendrocytes (CNP+ CAII−) and mature oligodendrocytes (CNP+ CAII+, Fig. 7a–c). In contrast, we observed a reduction in the density of mature oligodendrocytes with nuclear PAR in cuprizone-treated mice (Fig. 7c), along with an increase in the density of mature oligodendrocytes with cytoplasmic PAR (Fig. 7d,e) or lacking PAR signal (Fig. 7f,g, n = 4, one-way ANOVA, Tukey's multiple-comparisons test). These cuprizone-induced changes in subcellular localization did not occur in differentiating oligodendrocytes (Fig. 7c,e,g). We observed a disruption in PAR subcellular localization in mature oligodendrocytes at all cuprizone timepoints, with the most marked changes occurring after 3 weeks of treatment. These data indicate that cuprizone treatment alters PAR in mature but not differentiating oligodendrocytes and suggest that PAR redistribution may play a role in mature oligodendrocyte death.
Cuprizone disrupts PAR subcellular localization in mature but not differentiating oligodendrocytes. a, Representative maximum projection images of CNP and PAR labeling in control (left) and cuprizone-treated (right) mice. PAR widely labels the cortical nuclei at baseline and is confined to the nucleus of CNP+ cells in controls. In cuprizone-treated mice, some CNP cells display cytoplasmic PAR labeling or lack PAR reactivity. b, Single z plane of a CNP+ CAII+ cell with nuclear PAR in a control mouse. c, Quantification of CNP+ CAII− and CNP+ CAII+ cells with nuclear PAR in control (gray, n = 4 mice) and cuprizone (green, n = 4 mice) conditions. Each graph is a different week of the experiment [one-way ANOVA, F(3,12) = 112.5 (Week 1); 496.0 (Week 3); 362.8 (Week 5), with Tukey's correction for multiple comparisons; error bars indicate SEM]. d, Single z plane of a CNP+ CAII+ cell with cytoplasmic PAR in a cuprizone-treated mouse. e, Quantification of CNP+ CAII− and CNP+ CAII+ cells with cytoplasmic PAR in control (gray, n = 4 mice) and cuprizone (green, n = 4 mice) conditions. Each graph is a different week of the experiment [one-way ANOVA, F(3,12) = 3.174 (Week 1); 24.04 (Week 3); 13.36 (Week 5), with Tukey's correction for multiple comparisons; error bars indicate SEM]. f, Single z plane of a CNP+ CAII+ cell lacking PAR reactivity in a cuprizone mouse. g, Quantification of CNP+ CAII− and CNP+ CAII+ cells lacking PAR reactivity in control (gray, n = 4 mice) and cuprizone (green, n = 4 mice) conditions. Each graph is a different week of the experiment [one-way ANOVA, F(3,12) = 31.72 (Week 1); 13.93 (Week 3); 8.424 (Week 5), with Tukey's correction for multiple comparisons; error bars indicate SEM].
Cortical OPCs do not undergo caspase-3–dependent cell death in adulthood or cuprizone
Finally, we determined whether cuprizone had any effect on OPC cell death dynamics. OPCs are generally thought to not be directly affected in the cuprizone model of demyelination (Praet et al., 2014). Many studies have sought to describe the remyelinating response of OPCs during and after oligodendrocyte degeneration in the context of cuprizone. We were interested in understanding whether the switch toward differentiation was required for OPCs to become susceptible to death. To this end, we stained our tissue for PDGFRα (OPCs), BCAS1 (differentiating oligodendrocytes), and CC3 (Fig. 8). Quantification of 180 randomly selected PDGFRα+ cells from four control mice showed zero overlap in PDGFRα and BCAS1 labeling indicating that we were not including any differentiating oligodendrocytes in our OPC analysis (Fig. 8a,b). We were unable to identify any PDGFRa+ CC3+-double positive OPCs in tissue from either the cuprizone or control conditions (Fig. 8c,d, n = 4 mice, one-way ANOVA, Tukey's multiple-comparisons test). This suggests that cuprizone does not directly induce apoptosis in OPCs. The complete lack of CC3+ OPCs in both conditions calls into question whether direct OPC apoptosis plays a substantial role in OPC population homeostasis in the adult cortex.
Cortical PDGFRα+ OPCs do not undergo caspase-dependent cell death in adulthood or with cuprizone. a, Representative images of PDGFRα and BCAS1 staining in cortical tissue from a control mouse showing no overlap between cells that are labeled with these two antibodies. b, High-magnification images showing an OPC (PDGFRα+, BCAS1−, CNP−), a premyelinating oligodendrocyte (PDGFRα−, BCAS1+, CNP−), and a myelinating oligodendrocyte (PDGFRα−, BCAS1−, CNP+). There is no overlap between PDGFRα and BCAS1. c, Representative images of PDGFRα and CC3 staining in tissue from a cuprizone-treated mouse. d, Quantification of the density of CC3+ PDGFRα+ cells in control (gray, n = 4 mice) and cuprizone (green, n = 4 mice) conditions over time (one-way ANOVA with Tukey's correction for multiple comparisons; error bars indicate SEM; F(2,18) = 0.5016). e, Representative images of PDGFRα and γH2AX staining in tissue from a cuprizone-treated mouse. f, Quantification of the density of γH2AX+ PDGFRα+ cells in control (gray, n = 4 mice) and cuprizone (green, n = 4 mice) conditions over time (one-way ANOVA with Tukey's correction for multiple comparisons; error bars indicate SEM).
In addition to CC3, we also investigated whether cortical OPCs sustain DNA damage during cuprizone treatment, like mature oligodendrocytes. The previous staining was repeated with γH2AX (Fig. 8e). There was no difference in the density of γH2AX+ OPCs between control and cuprizone conditions (Fig. 8f, n = 4 mice, one-way ANOVA, Tukey's multiple-comparisons test) with only a single γH2AX+ OPC identified at both Weeks 3 and 5 in the cuprizone condition.
Discussion
When an OPC differentiates into an oligodendrocyte, there are broad morphology changes driven by shifts in gene expression, chromatin remodeling, and epigenetic regulation (Emery, 2010; Liu and Casaccia, 2010; Liu et al., 2016; Marques et al., 2019; Stadelmann et al., 2019; Tiane et al., 2019; Samudyata et al., 2020). As a result, cells within this lineage could respond differently to cytotoxic stress. Indeed, the field has known for decades that cuprizone is selectively toxic to oligodendrocytes and not OPCs, and differences in cell death vulnerability across the lineage have been noted in culture and developmental hypoxia–ischemia models (Back et al., 1998, 2001). In addition, the fact that the cuprizone demyelination model takes weeks to achieve full demyelination offered a hint into the slow degeneration dynamics of mature oligodendrocytes.
We sought to better characterize the cell death of oligodendrocytes as they differentiate and fully integrate, becoming mature oligodendrocytes. First, we showed that the maturation state of individual cells dictates the length of time they take to die in response to 2Phatal. OPCs died within 24 h, whereas mature oligodendrocytes died 45 d after 2Phatal. Differentiating oligodendrocytes, a category including premyelinating and newly formed oligodendrocytes, died with dynamics between these two populations. This demonstrates that the shift in mechanistic response to this kind of insult occurs throughout the maturation process. Second, we found that differentiating and mature oligodendrocytes go through molecularly distinct forms of cell death during cuprizone treatment. Differentiating cells used a caspase-dependent form of cell death, indicating an apoptotic-like mechanism, whereas mature oligodendrocytes were rarely labeled with CC3. Finally, we showed direct evidence of double-strand DNA breaks in mature oligodendrocytes treated with cuprizone accompanied by redistribution of PAR, neither of which were observed in differentiating oligodendrocytes. While these findings do not directly identify the mechanisms utilized by these two populations, the realization that there is a shift from a caspase-dependent to a caspase-independent mechanism clearly identifies a change in the overall mechanism (Bonora et al., 2015). Taken together, these findings highlight the differences in how cells within the oligodendrocyte lineage respond to cytotoxic stress, and when in maturation, that change occurs (summarized in Fig. 9).
Summary of findings and proposed effects of oligodendrocyte maturation on cell death mechanisms. a, The different cell stages of the oligodendrocyte lineage are indicated including the molecular markers that were used in this study. Underneath these markers, the response of these cells to cuprizone treatment is summarized. PDGFRa labeled OPCs did not show increased death during cuprizone treatment. In contrast, dying differentiating oligodendrocytes (CNP+ CAII−) were often found with CC3 labeling, whereas mature myelinating oligodendrocytes (CNP+ CAII+) were never labeled with the CC3 antibody but instead exhibited signs of DNA damage and disruption in PAR localization. b, Additional descriptions of the differences in the morphological, temporal, and molecular events detected between differentiating and mature oligodendrocytes in the cortex of mice subjected to 2Phatal or cuprizone treatment. Differentiating oligodendrocytes died on average within ∼1 week, and this death can be identified by CC3 labeling resembling morphological apoptosis. During cuprizone treatment, new oligodendrocytes that are generated from OPCs die rapidly through apoptosis. New oligodendrocytes generated after single-cell demyelination with 2Phatal can successfully mature and remyelinate. Mature oligodendrocytes at baseline have PAR labeling in the nucleus. At 1 week during cuprizone treatment, many mature oligodendrocytes were labeled with γH2AX indicating DNA damage, and this also coincided with some cells showing PAR labeling in the cytoplasmic compartment and loss in the nucleus. At Week 3 during cuprizone treatment, few cells were labeled with γH2AX while PAR was still found in the cytoplasm or completely absent from the mature oligodendrocytes. Finally, PAR was absent from many of the mature oligodendrocytes that had condensed nuclei and small cell soma, both indicative of eventual death. Whether mature oligodendrocytes are dying through parthanatos is still an open question however these signs suggest that possibility.
Outside of cuprizone, there are examples of differential susceptibility for cell populations as they mature that often results in discrete developmental windows for altered cell survival under varying pathological contexts (Back et al., 1998; Grinspan et al., 1998; Gerstner et al., 2008). This is true not only for the oligodendrocyte lineage but also in the context of neuronal maturation and survival (Blanquie et al., 2017; Wilkens et al., 2022). We again emphasize that our finding is related, but unique, as we are not reporting differences in susceptibility or vulnerability, but instead the type of death that occurs in response to similar cytotoxic damage.
The accumulation of DNA damage is a hallmark of aging (Al-Mashhadi et al., 2015; Schumacher et al., 2021). Oligodendrocytes represent a long-lived cell type with a unique metabolic load (Bartzokis, 2004). This makes them particularly susceptible to age-related stress and degeneration. Importantly, while the precise mechanisms that govern cuprizone demyelination are not fully characterized, our data, and that of others, have shown that DNA damage likely plays a substantial role (Fischbach et al., 2019). Indeed, DNA damage is likely the sole driving force in the 2Phatal model. Spontaneous oligodendrocyte and myelin degeneration in aging is a slow process, whereby individual sheaths are lost over time (Peters, 2002; Hill et al., 2018; Chapman et al., 2023). This contrasts with what would be expected during a single apoptotic event. It therefore seems likely that oligodendrocytes degenerate via a caspase-independent mechanism in aging. Understanding the precise molecular events that precede this spontaneous degeneration will be critical in the future.
Cuprizone has been used to great effect, to understand the consequences of demyelination, mechanisms that govern it, and the ability of OPCs to facilitate repair. Despite this, our understanding of how it works remains a mystery (Praet et al., 2014; Zirngibl et al., 2022). We show that the common oligodendrocyte marker, CNP, captures two separate oligodendrocyte populations that degenerate via distinct mechanisms. This finding provides context for several studies that have shown apparently different responses of mature oligodendrocytes to cuprizone (Li et al., 2008; Hesse et al., 2010; Hagemeier et al., 2013; Ma et al., 2017; Xing et al., 2018). These studies hint at our findings, including the presence of CC3+ cells in the first weeks of cuprizone (Hesse et al., 2010), and taken together with our data, these studies generate some intriguing new hypotheses. Knocking out the master regulator p53, for example, provides partial protection from cuprizone (Li et al., 2008; Ma et al., 2017). One possible explanation for this is that p53 may be required for the death of differentiating but not mature oligodendrocytes since p53 is known to mediate distinct forms of cell death through caspase proteases (Schuler and Green, 2001). In addition, genetic ablation of the BH3 protein, PUMA, has been shown to eliminate the emergence of CC3+ cells in the first 10 d of cuprizone treatment (Hagemeier et al., 2013). This suggests that PUMA may be mediating the caspase-dependent form of cell death that differentiating oligodendrocytes undergo. Consistent with this, knocking out PUMA has been shown to induce hypermyelination in the brain, as well as ectopic myelination in the molecular layer of the cerebellum (Sun et al., 2018). As PUMA is known to mediate apoptosis by directly activating BAX/BAK, triggering procaspase cleavage, it seems likely that the susceptibility of differentiating oligodendrocytes to cuprizone is due to mechanisms involved in homeostatic cell death. The same study found that conditional p53 knock-out in oligodendrocytes did not induce the same ectopic myelination, further supporting the hypothesis that cell death within the oligodendrocyte linage may occur through both p53-dependent and p53-independent mechanisms (Sun et al., 2018).
The DNA damage response protein, PARP1 has been shown to play a role in a specific cell death progression termed parthanatos (Yu et al., 2002; David et al., 2009). This cell death is characterized by redistribution of PAR to the cytoplasm and recruitment of mitochondrial apoptosis-inducing factor to the nucleus followed by programmed cell death, independent of caspase activity (Yu et al., 2002, 2006). Molecular inhibition of PARP1 has been shown to have a cytoprotective effect in some contexts including cuprizone and cell ischemia (Veto et al., 2010; Domercq et al., 2013) while also playing other roles in oligodendrocyte development and regeneration in demyelination models (Wang et al., 2021, 2022). Our data suggest that the potential protective effects of PARP1 molecular inhibition observed previously could be acting to decrease mature oligodendrocyte cell death since PAR disruption in our experiments was observed only in mature and not differentiating oligodendrocytes during cuprizone treatment. However, given that PARP1 and PAR signaling is implicated in oligodendrocyte differentiation (Wang et al., 2021), PARP1 disruption could simultaneously block new oligodendrocyte formation and remyelination. It is important to note that we have not fully established that mature oligodendrocyte death in the cuprizone model is indeed parthanatos; however, many signs are pointing toward a nonapoptotic mechanism such as this.
This discussion also raises the question of whether the rescue of dying mature oligodendrocytes would be a beneficial outcome for the underlying axon. The slow death progression of mature oligodendrocytes means that disrupted myelin is covering the axon for extended periods of time. Indeed, we have observed sheath decompaction preceding sheath degeneration by weeks in both 2Phatal and spontaneous oligodendrocyte death in aging (Hill et al., 2018; Chapman et al., 2023), and other recent works have suggested that the presence of disrupted myelin is more detrimental for the axon than full demyelination (Groh et al., 2023; Schäffner et al., 2023). Therefore, two options are to (1) speed up the death process for mature oligodendrocytes, assuming remyelination by local OPCs could proceed, or (2) rescue cell death of the mature oligodendrocytes while also ensuring that the myelin sheath is functional and providing the proper biophysical and metabolic support needed to the underlying axon. While it is not clear how to directly accomplish either option, they are important to consider for future treatment strategies.
It is well established that programmed cell death is critical for cell population homeostasis in the oligodendrocyte lineage during development (Barres et al., 1992; Trapp et al., 1997), and some other work has shown similar homeostatic mechanisms in the adult mouse cortex with live imaging (Hughes et al., 2013). However, the distinction of when the programmed cell death is occurring specifically identifying OPCs versus differentiating oligodendrocytes is less clear in the adult. While we did not observe an increase in CC3+ OPCs at any time during cuprizone demyelination, we also did not observe a single CC3+ PDGFRα+ OPC in the control tissues. This was surprising, as caspase-3 is generally thought of as the final executioner caspase in the intrinsic apoptosis pathway, and these cells require a mechanism to maintain population homeostasis. It is possible the OPC death proceeds through a caspase-independent form of death; however, another possibility is that OPC population homeostasis in the adult may be achieved almost exclusively as part of oligodendrocyte differentiation instead of at the OPC stage. The fact that we consistently observed CC3+ CNP+ cells in the control cortex but no CC3+ PDGFRα+ cells in these same tissues suggests that OPCs at the progenitor stage are not regularly spontaneously dying in the adult. This is consistent with much of the past live imaging data and the idea that there is insufficient integration of differentiating oligodendrocytes (Trapp et al., 1997; Hughes et al., 2018). This integration success could be altered with changes in axon features/signals that enhance or block myelin formation and oligodendrocyte maturation (Gibson et al., 2014; Hill et al., 2014, 2023; Hughes et al., 2018; Mitew et al., 2018; Hughes and Stockton, 2021). As our data are limited to neocortical OPCs in the young adult brain, it would be interesting to see if apoptotic OPCs can indeed be found in other brain regions in adults.
The occurrence of multiple forms of oligodendrocyte cell death during cuprizone demyelination has been previously described in the literature, using differential caspase staining or nuclear morphology (Hesse et al., 2010; Plemel et al., 2017). This led to the speculation that mature oligodendrocytes may undergo different forms of cell death in response to the same insults (Zirngibl et al., 2022). Here, we provide context as to the molecular and cellular mechanisms that govern these observations. Our use of longitudinal imaging of single cells coupled with fixed tissue analysis of defined cell stages revealed that it is the maturation state of oligodendrocytes that dictates the speed and mechanism with which these cells die. As these observations have been made in the context of multiple, distinct, models of oligodendrocyte degeneration, it seems likely that similar mechanisms may govern the death of oligodendrocytes in human aging and disease.
Footnotes
This work was supported by the National Institutes of Health R01NS122800 and the Esther A. & Joseph Klingenstein Fund and Simons Foundation to R.A.H. and a Gilman Fellowship from the Department of Biological Sciences at Dartmouth to T.W.C. We thank the members of the Hill lab at Dartmouth for helpful discussions and feedback on the project.
The authors declare no competing financial interests.
- Correspondence should be addressed to Robert A. Hill at robert.hill{at}dartmouth.edu.
