Transcranial alternating current stimulation at 10 Hz promotes oligodendrogenesis and reduces g-ratio after cuprizone-induced demyelination

TL;DR

Applying 10 Hz transcranial alternating current stimulation helps repair myelin in mice with cuprizone-induced brain damage, improving memory and myelin thickness.

Contribution

This study shows tACS at 10 Hz promotes oligodendrogenesis and myelin repair in a non-invasive, translatable way for MS.

Findings

tACS at 10 Hz accelerates oligodendrocyte maturation and increases axonal myelin thickness.

Mice treated with tACS showed improved spatial memory compared to controls.

Global myelin content was not affected, but g-ratio decreased, indicating better myelination.

Abstract

Multiple sclerosis (MS) is characterized by demyelination and incomplete remyelination. While numerous disease modifying treatments are available, none of these have been shown to aid in remyelination. In this study, we examined the remyelinating potential of transcranial alternating current stimulation (tACS) at 10 Hz in the cuprizone (CPZ) model. Our experiments show that 1 week of tACS during the recovery phase of the CPZ model could accelerate the maturation of newly formed oligodendrocytes and increase the relative myelin thickness of axons without changing the gross myelin content. Cognitively, mice treated with tACS showed specific improvements in spatial memory performance. Overall, these findings indicate that tACS can promote oligodendrogenesis and myelin repair in vivo in a non-invasive manner, highlighting its potential as a translatable strategy for people with MS.

•TACS 10 Hz boosts oligodendrogenesis during remyelination•TACS 10 Hz lowers axonal g-ratio, indicating thicker myelin sheaths after CPZ•Global myelin content remains unchanged•CPZ-tACS mice perform akin to healthy mice in the NOL in contrast to CPZ-sham mice

TACS 10 Hz boosts oligodendrogenesis during remyelination

TACS 10 Hz lowers axonal g-ratio, indicating thicker myelin sheaths after CPZ

Global myelin content remains unchanged

CPZ-tACS mice perform akin to healthy mice in the NOL in contrast to CPZ-sham mice

Behavioral neuroscience; Biomedical engineering; Neuroscience

Figures

Study designMice were either given CPZ 0.25% for 6 consecutive weeks followed by normal chow (NC) for a week or NC for 7 weeks. Mice were randomized (R) into four groups (*n* = 16 per group) consisting of CPZ + tACS 10 Hz; CPZ + Sham; NC + tACS 10 Hz and NC + Sham. Cognitive tasks were performed at baseline, after 5 weeks of CPZ or NC and 1 week after tACS or sham stimulation. At day 14, electrode holders were stereotaxically implanted on the skulls of the mice. Mice were subjected to 5 consecutive days of tACS or sham stimulation while given 5-ethynyl-2′-deoxyuridine (EdU, A10044, Invitrogen)

tACS 10 Hz reduces g-ratio in the corpus callosum, particularly in small and medium-diameter axons(A) Representative TEM images of the corpus callosum from mice receiving sham or tACS 10 Hz stimulation under normal chow (NC) or cuprizone (CPZ) diet. Scale bars, 2 μm.(B–D) Quantification of g-ratio (B), myelin thickness (C), and axon diameter (D) in the corpus callosum. g-ratio stratified by axon diameter (small < 0.5 μm, medium ≥ 0.5 and ≤ 1.0 μm, large > 1.0 μm). In the CPZ group, tACS significantly reduced the g-ratio in small- and medium-diameter axons. No significant differences were obser

Regions of interest analyzed for LFB staining identified by black square(A) Representative images of LFB staining in the rostral corpus callosum in three different conditions (NC + sham; CPZ + sham; CPZ + tACS 10 Hz). Scale bars, 400 μm.(B and C) (B) Graph representing the area of myelination using LFB staining in the rostral corpus callosum, caudal corpus callosum, and dentate gyrus (C). Sample size (*n*) for CPZ-sham = 8; CPZ-tACS 10 Hz = 11.

Cognitive tests(A) Graph representing the discrimination index of different conditions after stimulation. DI in CPZ + sham after 1 week of remyelination did not differ from the chance level.(B) Spontaneous alternations in the different groups were all significantly different compared to a chance level of 50%. The red line represents the chance level (0 for the discrimination index; 50% for spontaneous alternations). Flat horizontal lines above bars indicate results of one-sample *t* tests against chance level, while brackets indicate post hoc Sidak’s multiple comparisons. Data are shown as mea

Full text

Introduction

The hallmark features of multiple sclerosis (MS) include focal inflammatory demyelinating lesions and the inability to fully remyelinate existing injury.1 These lesions cause significant delays in conduction speed leading to a variety of symptoms. In addition, failure of remyelination leads to subsequent axonal degeneration which further aggravates the disability caused by the disease.2^,^3 Current disease-modifying therapies targeting the immune system do not halt disease progression and cognitive dysfunction. Consequently, strategies tackling the failure of remyelination are increasingly studied.4 Several pharmacological approaches have shown remyelinating potential in preclinical models. For example, clemastine promotes oligodendrocyte precursor cell (OPC) differentiation by blocking muscarinic receptors and improved conduction in a trial of chronic optic neuropathy in MS patients, although MRI measures showed only modest remyelination.5^,^6^,^7^,^8^,^9 Opicinumab (anti-LINGO-1) enhanced oligodendrocyte (OL) maturation in animal models but failed to meet primary endpoints in a phase 2 trial in MS.10 Thus, alternative or combined strategies are needed to achieve remyelination in patients with MS.

One of those strategies promoting the differentiation of OL and subsequent myelin formation is modulation of neuronal activation.11^,^12 This was discovered using tetrodotoxin to inhibit action potential firing in a neuron-glia co-culture and in vivo which reduced myelination.13 Recently, more elegant techniques such as optogenetics and Designer Receptors Activated Only by Designer Drugs (DREADDS) were designed to modulate the activity of specific subsets of neurons and were shown to enhance myelin formation in healthy mice as well as in models of demyelination.14^,^15^,^16 Moreover, distinct processes in myelin (re)generation can be influenced by different patterns of electrical activity. Nagy and colleagues17 were able to demonstrate that OPCs react differently to distinct stimulation frequencies at the axon-OPC synapse.

However, these cell-type-specific approaches such as optogenetics are not readily translatable to people with MS (pwMS). Therefore, it is important to test whether clinically feasible, non-invasive brain stimulation (NIBS) approaches, such as transcranial electrical stimulation (tES) or transcranial magnetic stimulation (TMS), can provide symptomatic relief in MS. In addition to being non-invasive, these techniques have a limited side effect profile compared to experimental pharmacological agents.18 A recent consensus paper reported level B evidence for TMS with regards to tackling spasticity in MS.19 Moreover, bilateral tACS at 6 Hz over the prefrontal cortex demonstrated improvements in information processing speed in pwMS.20 Other symptoms that have been successfully managed with NIBS in MS include lower urinary tract symptoms, depression and cognition, albeit with less robust evidence.21 While these findings highlight the potential cognitive benefits of tACS in pwMS, the mechanisms remain unclear, and the effects are mainly temporary.

In contrast to the emerging evidence of NIBS-induced symptomatic relief, the potential remyelination effects of repeated NIBS have been investigated in the cuprizone (CPZ) model. Anodal tDCS was associated with a recovery of the visual evoked potential,22 and increased number of OLIG2+ cells,23 whereas rTMS was associated with an increased MBP via western blot analysis, and increased the number of myelin internodes in Nguyen et al.24 Importantly, none of these studies addresses the question whether the repeated administration of tACS—which shows promising albeit short-lived effects in cognition during stimulation—has potential remyelinating effects.

We selected 10 Hz stimulation within the alpha frequency band because alpha oscillations are often reduced in pwMS and are strongly associated with cognitive performance, particularly information-processing speed.25^,^26^,^27 Moreover, different frequencies seem to have distinct effects on oligodendrogenesis and 10 Hz lies in between the pro-proliferative 5 Hz and pro-differentiating 25 Hz frequency.17 In the present study, we aimed to investigate whether transcranial alternating current stimulation (tACS) at 10 Hz can promote oligodendrogenesis and enhance remyelination in the CPZ mouse model. To this end, we assessed (i) oligodendrocyte proliferation and differentiation using EdU and immunofluorescent labeling for OLIG2 and CC1, (ii) myelin ultrastructure using transmission electron microscopy (TEM), and (iii) overall myelin content and cognitive performance through luxol fast blue staining and behavioral testing. We hypothesized that tACS 10 Hz would enhance oligodendrocyte maturation and increase relative myelin thickness in demyelinated mice as well as restore cognitive impairment.

Results

tACS 10 Hz promotes oligodendrogenesis in the rostral corpus callosum

We induced demyelination and subsequent remyelination by feeding mice a cuprizone diet for six weeks, followed by one week of regular chow, to evaluate the therapeutic impact of tACS (see Figure 1). To assess whether tACS 10 Hz expanded the OL lineage pool during remyelination, we quantified the density of OLIG2^+^ cells, representing the total pool of oligodendrocytes. We found a significant increase of OLIG2^+^ cells following CPZ diet (F(1,28) = 12.05, p = 0.0017; see Figures 2A and 2B and Table S1), whereas tACS had no additional effect (F(1,28) = 0.03002, p = 0.8637). The interaction between diet and treatment was also not significant (F(1, 28) = 0.01959, p = 0.8897).Figure 1. Study designMice were either given CPZ 0.25% for 6 consecutive weeks followed by normal chow (NC) for a week or NC for 7 weeks. Mice were randomized (R) into four groups (n = 16 per group) consisting of CPZ + tACS 10 Hz; CPZ + Sham; NC + tACS 10 Hz and NC + Sham. Cognitive tasks were performed at baseline, after 5 weeks of CPZ or NC and 1 week after tACS or sham stimulation. At day 14, electrode holders were stereotaxically implanted on the skulls of the mice. Mice were subjected to 5 consecutive days of tACS or sham stimulation while given 5-ethynyl-2′-deoxyuridine (EdU, A10044, Invitrogen) 0.2 mg/mL in their drinking water. Following the cognitive tasks performed after the last tACS/sham, mice were sacrificed to harvest brain tissue for further histological analysis.Figure 2. Oligodendrogenesis following 6 weeks of CPZ exposure and 1 week of withdrawal with either 10 Hz tACS or sham treatment(A) Example of CC1^+^, EdU^+^, OLIG2^+^ cells and colocalization of each cell type in CPZ + tACS 10 Hz and CPZ + sham group. Scale bars, 10 μm. White arrows represent newly differentiated mature OLs.(B) Density of OLIG2+ cells per mm^2^ in each group.(C) Density of OLIG2^+^CC1^+^ cells per mm^2^ across treatment groups.(D) Density of OLIG2^+^CC1^−^ cells per mm^2^ across treatment groups.(E) Density of EdU^+^OLIG2^+^ cells per mm^2^ across treatment groups.(F) Density of EdU^+^OLIG2^+^CC1^+^ cells per mm^2^ across treatment groups.(G) Density of EdU^+^OLIG2^+^CC1^−^ cells per mm^2^ across treatment groups.(H) Ratio of EdU^+^OLIG2^+^ cells to total OLIG2^+^ cells. Values represent mean with SEM. Data were analyzed with two-way ANOVA and post hoc Sidak’s test. Sample size (n) for NC + sham = 5; NC + tACS 10 Hz = 6; CPZ + sham = 11; for CPZ + tACS 10 Hz = 10; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

We next evaluated the effects of tACS 10 Hz on differentiation of OLs. OLIG2^+^ cells that colocalized with CC1^+^ were considered mature OLs. The density of mature OLs (OLIG2^+^CC1^+^) was not altered by CPZ (F(1,28) = 2.352, p = 0.1364) nor by stimulation (F(1,28) = 0.9936, p = 0.3274; see Figure 2C) and the interaction was also not statistically significant (F(1, 28) = 0.07597, p = 0.7849). On the other hand, the density of immature OLs (OLIG2^+^CC1^−^) significantly increased after CPZ (F(1,28) = 13.78, p = 0.0009; see Figure 2D). Further, tACS 10 Hz tended to increase the density of immature OLs (F(1,28) = 2.923, p = 0.0984) but this did not reach statistical significance. The interaction between these two factors was also not statistically significant (F(1, 28) = 0.02372, p = 0.8787).

EdU was administered between day 1 and day 5 of the stimulation in the remyelination phase of the CPZ model. This allowed us to investigate the effect of tACS 10 Hz on the proliferative activity of OLs during the stimulation. CPZ had a strong effect on the density of OLs that had proliferated during the stimulation period (EdU^+^OLIG2^+^) (F(1,28) = 34.79, p < 0.0001; see Figure 2E), and tACS tended to increase the number of these OLs, with a borderline significant stimulation effect (F(1,28) = 4.124, p = 0.0519) and interaction effect between diet and stimulation (F(1,28) = 3.785, p = 0.0618). If we analyze the proportion of OLIG2^+^ cells expressing EdU to the total number of OLIG2^+^ cells (EdU^+^OLIG2/OLIG2^+^; see Figure 2H), we find main effects for diet (F(1, 28) = 45.56, p < 0.0001), stimulation (F(1, 28) = 7.237, p = 0.0119) and an interaction effect for both factors (F(1, 28) = 5.878, p = 0.0220). Post hoc comparisons reveal a significant increase in the ratio EdU^+^OLIG2^+^cells/total OLIG2^+^ cells after CPZ (NC + sham: 8.01% vs. CPZ + sham: 18.06%, p = 0.0170) and further effects of tACS 10 Hz (CPZ + sham: 18.06% vs. CPZ + tACS 10 Hz: 29.94%, p = 0.0005) representing an increased conversion of proliferative immature OLs into mature OLs in the CPZ-tACS 10 Hz group. Within the group receiving NC, we did not find differences following tACS 10 Hz (NC + sham: 8.01% vs. NC + tACS 10Hz: 8.63%, p = 0.9979).

Next, we divided these cells in immature OLs (EdU^+^OLIG2^+^CC1^−^ cells; see Figure 2G) and differentiated mature OLs (EdU^+^OLIG2^+^CC1^+^; see Figure 2F) that had proliferated during the stimulation period. EdU^+^OLIG2^+^CC1^+^ cells were significantly impacted by both diet (F(1,28) = 34.04, p < 0.0001) and stimulation (F(1,28) = 10.73, p = 0.0028). In addition, we found an interaction effect between both variables (F(1,28) = 9.133, p = 0.0053). Sidak’s post hoc analysis revealed that tACS 10 Hz increased the number of EdU^+^OLIG2^+^CC1^+^ more than 2-fold in the CPZ-group (mean EdU^+^OLIG2^+^CC1^+^ cells per mm^2^ CPZ + sham = 260.8, mean CPZ + tACS 10 Hz = 583.4, p < 0.0001; see Figure 2F) indicating that tACS 10Hz further enhances the increase in differentiation of proliferative OLs after CPZ intoxication and withdrawal. Diet significantly influenced the number of EdU^+^OLIG2^+^CC1^−^ cells (F(1,28) = 17.23, p = 0.0003), whereas tACS did not significantly affect this population (F(1,28) = 0.1597, p = 0.6925; see Figure 2G) nor did we find an interaction effect (F(1, 28) = 0.2273, p = 0.6372).

As CPZ typically increases the microglia-density in the rostral corpus callosum, we investigated whether tACS 10 Hz reduced microgliosis. As expected, CPZ induced an increase in IBA1^+^ cells both in the tACS and sham group (F(1,31) = 17.61, p = 0.0002). However, tACS 10 Hz did not reduce the number of IBA1^+^ cells (F(1,31) = 0.2225, p = 0.6404; see Figure S1).

tACS 10 Hz promotes the restoration of myelin architecture following CPZ-induced demyelination

A key metric reflecting relative myelin thickness in relation to axonal diameter is the g-ratio. Mean g-ratios for individual axons per mouse were assessed for (re)myelination in the rostral part of the corpus callosum (n = 3 for NC + sham and NC + tACS 10 Hz; n = 4 for CPZ + sham and CPZ+tACS 10 Hz). Both diet (F (1, 10) = 33.56, p = 0.0002, η^2^ = 0.77) and stimulation (F (1, 10) = 14.61, p = 0.0034, η^2^ = 0.59) had a significant effect on g-ratio, however, no significant interaction was found between diet and stimulation (F (1, 10) = 0.9114, p = 0.3622, η^2^ = 0.08, see Figure 3B). Sidak’s post-hoc analysis revealed significantly lower g-ratios in the tACS-treated mice (mean g-ratio = 0.7381) compared to sham-treated mice (mean g-ratio = 0.7788, p = 0,0134, d = 2.58) that received CPZ. We did not observe significantly lower g-ratios in the tACS-treated mice (mean g-ratio = 0.6970) compared to sham-treated mice (mean = 0.7214, p = 0.2391, d = 1.55) that received normal chow (NC). Moreover, tACS 10 Hz produced a left-ward shift in the cumulative distribution plot of g-ratios in the CPZ-treated mice (D = 0.220, p < 0.0001, see Figure 3G), providing additional evidence for thicker myelin sheets in the rostral corpus callosum following tACS 10 Hz in CPZ-treated mice. In mice receiving NC, we did not observe differences in cumulative g-ratio distributions (D = 0.2857, p = 0.9627).Figure 3tACS 10 Hz reduces g-ratio in the corpus callosum, particularly in small and medium-diameter axons(A) Representative TEM images of the corpus callosum from mice receiving sham or tACS 10 Hz stimulation under normal chow (NC) or cuprizone (CPZ) diet. Scale bars, 2 μm.(B–D) Quantification of g-ratio (B), myelin thickness (C), and axon diameter (D) in the corpus callosum. g-ratio stratified by axon diameter (small < 0.5 μm, medium ≥ 0.5 and ≤ 1.0 μm, large > 1.0 μm). In the CPZ group, tACS significantly reduced the g-ratio in small- and medium-diameter axons. No significant differences were observed large axons. Two-way ANOVA with Sidak-adjusted post hoc comparisons ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, ns, not significant.(F–G) Scatterplots of g-ratio vs. axon diameter for individual axons in CPZ (F) and NC (G) conditions.(H–I) Cumulative distribution of g-ratio values in CPZ (H) and NC (I) groups. A significant leftward shift in g-ratio distribution was observed in CPZ + tACS vs. CPZ + sham (Kolmogorov-Smirnov test, ∗∗∗∗p < 0.0001), indicating increase in relative myelin thickness. No shift was observed in NC groups.

Since g-ratio depends on axon diameter and myelin thickness, we examined these metrics individually. A significant effect of diet was found on axon diameter (F (1, 10) = 14.03, p = 0.0038, see Figure 3C), with CPZ-treated animals exhibiting larger axon diameters (mean = 0.6127 μm) compared to NC (mean = 0.4841). Stimulation had no significant effect (F (1, 10) = 0.8306, p = 0.3835) on axon diameter (F (1, 10) = 0.02112, p = 0.8873). Another parameter influencing the g-ratio is myelin thickness. We observed no significant effect of stimulation (F (1, 10) = 0.5480, p = 0.4762) nor for diet (F (1, 10) = 2.935, p = 0.1174), nor a significant interaction effect (F (1, 10) = 0.8177, p = 0.3871, see Figure 3D). These analyses reveal that tACS 10 Hz increases relative myelin thickness (i.e., myelin thickness in relation to axon diameter) rather than absolute myelin thickness.

Next, we quantified whether the changes in g-ratio following 10 Hz tACS in the CPZ group were mainly driven by small (0–0.5 μm), medium (0.5–1 μm), or large (>1 μm) axons. A two-way ANOVA with stimulation group (CPZ-tACS 10 Hz vs. CPZ-sham) and axon size (small vs. medium vs. large) as fixed factors was performed on the mean g-ratio per mouse. The analysis revealed a significant main effect of stimulation (F(1,18) = 11.15, p = 0.0037), with lower g-ratios in tACS-treated mice, and a strong main effect of axon size (F(2,18) = 102.3, p < 0.001). The interaction between stimulation and axon size was also significant (F(2,18) = 3.921, p = 0.0386). Post-hoc Sidak-adjusted comparisons showed that the reduction in g-ratio after tACS was significant for small axons (mean difference = 0.053, p = 0.0036), but was absent for medium axons (mean difference = 0.029, p = 0.1518), and large axons (mean difference = −0.0016, p = 0.992; see Figure 3E and Table S2).

To assess whether tACS influences the relationship between axon diameter and g-ratio, we performed a simple linear regression. In CPZ mice, both sham and tACS groups showed a positive correlation between axon diameter and g-ratio (CPZ-sham: slope = 0.1493, p < 0.0001; CPZ-tACS: slope = 0.1916, p < 0.0001; see Figure 3F). The slope was steeper in the tACS group compared to the sham group (0.1916 vs. 0.1493; F(1,460) = 4.104, p = 0.0434), indicating that g-ratios in the tACS group are lower in smaller axons and as axon diameter increases the g-ratios approach the values of the sham group. In NC mice, slopes were not different between tACS and sham (NC-sham: slope = 0.2644; NC-tACS: slope = 0.2677; F(1,671) = 0.02473, p = 0.8751; see Figure 3G), revealing no effect of tACS 10 Hz in healthy mice.

The distribution of g-ratios in CPZ mice also showed a remarkable leftward shift in the tACS group compared to the sham group (Kolmogorov-Smirnov D = 0.2220, p < 0.0001; see Figure 3H), consistent with increased myelin thickness per given axon diameter after tACS. In contrast, no shift was evidenced in NC mice between sham and tACS (Kolmogorov-Smirnov D = 0.2857, p = 0.9627; see Figure 3I).

tACS 10 Hz does not induce changes in the overall myelin content

To evaluate whether the observed effect of tACS 10 Hz on axonal g-ratios would translate to changes in myelination histochemically, we performed LFB staining. We analyzed three different regions of interest, namely the rostral part of the corpus callosum, the caudal part of the corpus callosum, and the ipsilateral (relative to the placement of the cathode) dendate gyrus (see Table S4). In neither of the three regions, we observed changes in global myelination following tACS 10 Hz. A two-way ANOVA (see Figure 4C) analyzing factors “brain region” and stimulation reveal differences in brain region (F(2, 49) = 4.589, p = 0.0149) but not in stimulation (F(1, 49) = 0.3703, p = 0.5457) nor in the interaction (F(2, 49) = 0.6401, p = 0.5316) between the two factors.Figure 4. Regions of interest analyzed for LFB staining identified by black square(A) Representative images of LFB staining in the rostral corpus callosum in three different conditions (NC + sham; CPZ + sham; CPZ + tACS 10 Hz). Scale bars, 400 μm.(B and C) (B) Graph representing the area of myelination using LFB staining in the rostral corpus callosum, caudal corpus callosum, and dentate gyrus (C). Sample size (n) for CPZ-sham = 8; CPZ-tACS 10 Hz = 11.

Cognitive performances following tACS 10 Hz

Effect of tACS on cognitive performance in CPZ-Treated mice in the NOL task

In the NOL task, sham mice receiving CPZ did not exhibit a preference for the novel object location as the discrimination index did not differ from the chance level of 0 (t(14) = 1.400, p = 0.1832; see Figure 5A). Conversely, both sham and tACS stimulated mice that received NC and CPZ-tACS mice performed significantly better than the chance level, indicating performances in the CPZ-tACS group akin to healthy controls (t(15) = 3.966, p = 0.0048). Our two-way ANOVA (see Table S5), however, revealed no interaction (F(1,46) = 2.228, p = 0.1423) between the diet factor (CPZ vs. NC) and the stimulation factor (Sham vs. tACS 10 Hz). Furthermore, both the diet (F(1,46) = 0.006262, p = 0.9373) and stimulation (F(1,46) = 1.674, p = 0.2022) had no significant effect on cognitive performance, indicating that neither CPZ nor tACS 10 Hz influenced cognitive performances under the used experimental conditions.Figure 5. Cognitive tests(A) Graph representing the discrimination index of different conditions after stimulation. DI in CPZ + sham after 1 week of remyelination did not differ from the chance level.(B) Spontaneous alternations in the different groups were all significantly different compared to a chance level of 50%. The red line represents the chance level (0 for the discrimination index; 50% for spontaneous alternations). Flat horizontal lines above bars indicate results of one-sample t tests against chance level, while brackets indicate post hoc Sidak’s multiple comparisons. Data are shown as mean ± SEM. Data were analyzed with one-sample t test and two-way ANOVA with Sidak’s post hoc analysis. Sample size (n) for for NC + sham = 11; for NC + tACS 10 Hz = 11; for CPZ + sham = 15; for CPZ + tACS 10 Hz = 16. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

tACS 10 Hz does not affect spatial working memory in the Y-maze test

We found no spatial working memory deficits following CPZ exposure, as a one-sample t test revealed that all groups performed significantly better than the chance level (50%, p values ranging from 0.0041 to <0.0001; see Figure 5B). Similarly, we found no significant effect of tACS in the CPZ-treated and mice receiving NC (two-way ANOVA F(1, 48) = 0.004980, p = 0.9440; see Figure 5B and Table S6). Next, we analyzed the number of arm entries, a reflection of locomotor activity. Our results reveal that there were no significant differences in arm entries as result of diet, stimulation, or interaction between the two factors (two-way ANOVA F(1, 49) = 0.5214, p = 0.4737; see Figure S2).

Discussion

In this study, we demonstrate that five daily sessions of tACS 10 Hz significantly increase the proportion of OLs that proliferated during the stimulation period and promoted their differentiation into mature OLs during the recovery phase of the CPZ model. Furthermore, we show that tACS 10 Hz can increase the myelin thickness of individual axons in the CPZ model, more specifically small-sized axons. Despite observing thicker myelin sheaths following tACS, we were not able to detect changes in overall myelin content in several brain regions. Moreover, we found that tACS-treated mice displayed a preference for the novel object location in contrast to sham-treated mice in the CPZ group. Nevertheless, the total exploration time did not differ between the two groups.

Our findings align with several studies that have found that neuromodulation is an effective strategy to instruct OPCs to proliferate, differentiate and mature into myelin-producing OLs. Neuronal activity causes the release of several neurotransmitters at the OPC-neuron synapse including glutamate and GABA.28^,^29 These neurotransmitters play an important role in the distinct processes of oligodendrogenesis. Following demyelination, de novo synapses are generated between demyelinated axons and OPCs.30 Through optogenetic 20 Hz stimulation of motor neurons in the premotor cortex (30 s on/2 min off, for 10 min per day on 7 consecutive days), Gibson and colleagues14 could enhance the formation of EdU-positive OPCs and their differentiation into mature OLs in healthy mice. Similarly, Piscopo and colleagues31 observed that decreasing the activity of inhibitory parvalbumin interneurons in the anterior cingulate cortex through optogenetic 1 Hz stimulation (30 min per day over 4 weeks) contributed to an increase in EdU-positive OLs. Under demyelinating conditions, Ortiz and colleagues15 used a repeated 20 Hz optogenetic stimulation paradigm (3-h stimulation of callosal axons for 7 days) and were able to induce an increase in the total number of mature OLs. This observation contrasts with our study in which the total pool of mature OLs remained unchanged. However, within that pool of mature OLs, we found an increase in EdU-positive mature OLs in the tACS group compared to sham stimulation in CPZ mice. This suggests a dynamic turnover of OLs that may be beneficial in repairing myelin. Similarly, Nguyen et al.32 found that low intensity rTMS, consisting of 10 bursts of three pulses at 50 Hz and repeated at 5 Hz for a 2 s period over a total period of 4 weeks did not yield an increase of the pool of Ols in the CPZ model. Their stimulation paradigm, however, increased the remyelination capacity of new OLs as shown by genetic lineage tracing models and calculation of number of internodes per new OL. Both our study and the study from Nguyen et al.32 were performed on CPZ mice whereas Ortiz et al.15 used a focal demyelinating model induced by lysophosphatidylcholine (LPC). This may explain the observed differences in the total OL pool. In the CPZ model, OLs tend to be maximally recruited after 1 week of remyelination and do not further increase in the weeks thereafter.33 We speculate that areas of demyelination in the CPZ model only attract a limited number of OPCs and that tACS 10 Hz is responsible for the promotion of a dynamic turnover of OLs, prioritizing the maturation of OLs that proliferated during the stimulation period rather than increasing the whole pool.

Importantly, our TEM analysis reveals that tACS can lower the g-ratio of axons, indicative of increased relative myelin thickness. Our observation of a lower g-ratio could be caused by different mechanisms. It is well known that in humans remyelination results in thinner myelin sheaths.34 This is similar to remyelination in the CPZ model when remyelination is induced by resident OPCs. Yet, when OPCs are recruited from the subventricular zone, remyelination tends to result in normal myelin thickness.35^,^36 Here, it is important to note that our study compared the g-ratio of spontaneous remyelination, which typically shows thinner myelin sheaths, with spontaneous remyelination and tACS. Given that our intervention increased relative myelin thickness rather than the number of remyelinated axons, the observed lower g-ratio in our study can be interpreted via non-mutually exclusive pathways: tACS could induce the recruitment of SVZ-derived progenitors or it could enhance sheath growth by newly formed oligodendrocytes beyond what is typically observed.

Comparable to our findings, Ortiz and colleagues15 observed a decrease in g-ratio after seven days of optogenetic stimulation of callosal axons compared to sham stimulation in the LPC model. However, their observation of increased number of myelinated axons in the corpus callosum contrasts with our findings. In the rTMS study of Nguyen et al.32 no decrease was seen in g-ratio in the corpus callosum after four weeks of stimulation compared to sham stimulation, yet they also noted an increase in number of myelinated axons. Using a multisensory stimulation paradigm at 40 Hz for three weeks, Rodrigues-Amorim and colleagues37 similarly observed a decrease of g-ratio in CPZ mice that were stimulated compared to their sham counterparts. However, while our histochemical results did not reveal overall remyelination, their paradigm resulted in increased MOG expression and myelin staining in the corpus callosum.

Several factors seem to play a role in neuronal-activity dependent remyelination. It was previously shown that OPC behavior is modulated in a frequency-dependent manner.17 In our study, we focused on 10 Hz stimulation due to its involvement in cognitive deficits in MS,25 however, other frequencies may exert even greater effects on OPC proliferation as 300 Hz in healthy mice exerted the highest effect on OPC proliferation. Similarly, stimulation at 5 Hz was most effective in OPC differentiation into mature OLs.17 Another important variable seems to be the duration of the stimulation. We chose 1 week during the remyelination phase of the CPZ because we intended to specifically investigate the effects of tACS 10 Hz on remyelination. Longer stimulation would likely not have been feasible due to the rapid reversal of demyelination following CPZ withdrawal in the acute CPZ model. While we compare with other studies exploring the effects of neuronal activation on (re)myelination, it is important to note that most used paradigms with stimulation durations up to 4 weeks.14^,^15^,^31^,^32^,^37^,^38 It is therefore, not possible to conclude whether the studies in demyelinating animal models reduced demyelination, promoted remyelination, or affected both processes. Optimizing the factors that can further promote oligodendrogenesis and remyelination is essential and may result in more robust remyelination.

Our study confirms previous research conducted by Schepers et al.,39 demonstrating that 6 weeks of CPZ followed by 1 week of recovery leads to cognitive deficits in the NOL task. CPZ mice that received sham stimulation failed to show a preference for the novel object location indicative of cognitive impairment in contrast to CPZ mice receiving tACS 10 Hz preferentially exploring the novel object location similar to healthy mice. Importantly, overall group comparisons did not reveal significant differences, indicating merely a subtle cognitive benefit of tACS 10 Hz on spatial memory. In the present study, 6 weeks of CPZ treatment did not cause cognitive deficits in the Y-maze test. This contrasts with the study from Zhou and colleagues40 who used an identical experimental setup treating mice with 6 weeks of CPZ and allowing 1 week of recovery. In their study, mice subjected to the CPZ treatment and given sham showed a significant reduction in spontaneous alternations that was reversed using one week of rTMS. Mercier et al.41 similarly observed significant reductions in alternation behavior after 5 weeks of CPZ. In order to comply with the ARRIVE guidelines,42 environmental enrichment was provided to the mice in our experiments including cardboard houses, nesting material and wooden gnawing blocks. Such environmental enrichment may reduce CPZ-induced cognitive impairment as reported in the Morris water maze after 6 weeks of CPZ.43 Similarly to our findings, Palavra and colleagues44 did not report robust cognitive impairment in both the NOL and the Y-maze after 5 weeks of CPZ intoxication.

In conclusion, we report that tACS 10 Hz promotes a dynamic turnover of OLs rather than an overall increase in the OL density. We specifically observed an increase in EdU^+^ mature OLs reflecting the beneficial effects of tACS 10 Hz on oligodendrogenesis. The myelin thickness relative to the axon diameter (e.g., g-ratio) was significantly increased following tACS 10 Hz in CPZ-treated mice compared to sham-stimulated mice without affecting overall remyelination. In the NOL task, tACS-treated CPZ mice performed comparably to healthy controls while sham-treated CPZ mice did not perform above chance-level, indicative of cognitive impairment. However, group differences were not detected limiting firm conclusions on reversal of cognitive impairment after tACS 10 Hz. Further studies should explore longer stimulations and different frequencies which may lead to more robust findings both in remyelination and cognitive outcomes.

Limitations of the study

Our study has several limitations. To mimic the conditions of NIBS and limiting the effects of anesthesia, we opted to implant two electrode holders on the skull of the mice. This setup allowed for placement of electrodes in awake mice but rendered long stimulation protocols challenging as electrode holder detachment caused suboptimal stimulation quality beyond 1 week of stimulation. An important limitation for translation to pwMS is the inherent characteristics of the CPZ model. In the acute CPZ model, remyelination is not impaired and complete remyelination is expected after 4 weeks of CPZ withdrawal. In contrast, the chronic CPZ model, entailing a CPZ diet of at least 12 weeks, leads to impaired inherent characteristics of remyelination.45 Moreover, the pathology of MS relies on two important features, namely peripheral inflammation and demyelination. The CPZ model only captures the demyelination and remyelination aspect of the disease. Such controlled models are extremely interesting to delve deeper in the features of oligodendrogenesis and remyelination but may not present the full picture. Importantly, while relapsing remitting MS has a predilection for females, we chose to conduct our experiments in male C57BL/6 mice because both demyelination and behavioral deficits might be less pronounced in female mice possibly due to genetic and hormonal factors.46^,^47

An additional limitation of our study is that, although we quantified the density of IBA1^+^ microglia in the rostral corpus callosum and found no detectable effect of tACS on total microglial counts, we did not assess key functional or phenotypic aspects of microglia. Microglia play an important role in myelin debris, an essential process for effective remyelination. Moreover, as these microglia transition to pro-regenerative phenotypes, these cells aid in oligodendrogenesis and myelin repair.48 Furthermore, modulating neuronal activity at microglia-axon contacts can boost remyelination in models of demyelination.49 As our analysis was limited to cell counts, we cannot exclude that tACS affected microglial phenotype or function in ways that contributed to the observed increases in EdU^+^ mature OLs and g-ratio improvement. Future work with NIBS should therefore include detailed assessments of microglial morphology, activation markers, and cytokine/chemokine profiles. Finally, the goal of our work was to investigate the effects of repeated administration of tACS on oligodendrocyte cell dynamics and remyelination. Future experiments could perform transcriptomic profiling to study, e.g., whether neuronal activity dependent pathways have been upregulated.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Thomas Jonathan Scheinok (thomas.jonathan.scheinok@vub.be).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •Behavioral and histological data have been deposited at Zenodo and are publicly available as of the date of publication. Accession numbers are listed in the key resources table.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

This research was funded by 10.13039/501100003130Fonds Wetenschappelijk Onderzoek (10.13039/501100003130FWO-G042821N) and the Belgian 10.13039/501100006401Charcot Foundation. Furthermore, we want to thank Veerle De Punt from the Diabetes Pathology & Therapy lab at the Vrije Universiteit Brussel for her important work in processing the tissues for TEM. We also thank Gino De Smet and Anke De Smet for the technical assistance during the in vivo and ex vivo experiments. Lastly, we thank Brecht Ghesquire and the VUB research core facilities to aid in the acquisition of the confocal images.

Author contributions

T.J.S., conceptualization, methodology, validation, formal analysis, investigation, writing – original draft, visualization, and project administration; M.G., formal analysis, investigation, writing – review and editing, supervision, and funding acquisition; M.D., conceptualization, methodology, writing – review and editing, and supervision; G.N., conceptualization, methodology, and writing – review and editing; D.D.B., conceptualization, methodology, validation, formal analysis, resources, writing – original draft, visualization, supervision, and funding acquisition; J.V.S., conceptualization, methodology, validation, formal analysis, resources, writing – original draft, visualization, supervision, and funding acquisition.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesRabbit anti-OLIG2Merck MilliporeCat# AB9610; RRID: AB_570666Mouse anti-CC1AbcamCat# AB16794; RRID: AB_443473Goat anti-IBA1InvitrogenCat# PA5-18039; RRID: 10982846AF488 donkey anti-chickenJackson Immuno ResearchCat# 703-545-155; RRID: AB 2340375AF488 donkey anti-mouseJackson Immuno ResearchCat# 715-545-150; RRID: AB_2340846Cy3 donkey anti-rabbitJackson Immuno ResearchCat# 711-165-152; RRID: AB_2307443Cy5 donkey anti-goatJackson Immuno ResearchCat# 705-175-147; RRID: AB_2340415Chemicals, peptides, and recombinant proteinsCuprizoneSigma-AldrichC90125-ethynyl-2′-deoxyuridine (EdU)InvitrogenCat# A10044Alexa Fluor 647 Click-IT EdU kitInvitrogenCat# C10340Luxol fast blue MBSNThermo FisherCat# 212170250Isoflurane (Vetflurane®)VirbacBE-V380177Sodium pentobarbital (Dolethal®)VetoquinolBE-V171692Heparin (Heparin Leo)Leo PharmaB01AB01Dental acrylics (Dentalon® plus)KulzerN/ATissue adhesive (Superbond®)Sun MedicalN/AExperimental models: Organisms/strainsMouse: C57BL/6J (Male, 6–8 weeks)Janvier LabsN/ASoftware and algorithmsEthovision 17.0Noldushttps://www.noldus.com/ethovision-xtFIJINIHhttps://imagej.net/software/fiji/MyelTracerKaiser et al., 2021https://github.com/HarrisonAllen/MyelTracerRandom.orgRandomness and Integrity Services Ltd.https://www.random.org/lists/OtherTranscranial electric stimulatorSoterix MedicalModel NO:2001Electrode holders (internal ø: 2.1 mm)Dixi MedicalN/AElectrodes (3.5 mm^2^ contact area)Dixi MedicalN/AStereotaxic frameKopf® InstrumentsN/AManual micromotor drill (Volvere Vmax®)NSK NakanishiN/ANylon screws (0-80 x 3/32N)Bilaney Consultants GmbHN/AConfocal microscopeZeissLSM800Electron microscopeThermo FisherTecnai 10VibratomeLeicaVT 1000SCameraBasler AGAcA1920Mouse cagesOriginal research dataTechniplastZenodoSealsafe PlusGM 500 https://doi.org/10.5281/zenodo.16846017

Experimental model and study participant details

A total of 64 male C57BL/6J mice (6–8 weeks old, mean body weight 25.3 ± 0.42 g at study onset, Janvier, France) were used. Mice were randomly assigned to four experimental groups (n = 16 per group): NC + sham, NC + tACS, CPZ + sham, and CPZ + tACS. Mice were group housed (3–4 per cage, Sealsafe PlusGM 500, Techniplast, Italy) in a specific pathogen-free animal facility under standard conditions (light/dark 12h/12h, temperature 20°C–22°C, humidity 45–65%, environmental enrichment including cardboard houses, nesting material and wooden gnawing blocks). We specifically chose male mice because CPZ-induced behavioral changes and weight loss are more reproducible in male mice.47^,^50 Food and water were provided ad libitum. Mice were habituated to handling for at least one week. All experiments were approved by the Ethische Commissie Dierproeven at Vrije Universiteit Brussel (ECD 22-213-9) and performed in accordance with the guidelines set by the Belgian Council for Laboratory Animal Science, the ARRIVE guidelines42 and the European Community Council Directives (2010/63/EU).

Method details

Cuprizone exposure

Widespread demyelination was induced by feeding 0.25% CPZ (Sigma-Aldrich, France) in ground chow for 6 weeks (see Figure 1). Mice were weighed weekly and given diet supplement gel if body weight dropped >10%. After 6 weeks, CPZ was withdrawn, and mice were returned to normal chow (NC) for a 1-week recovery period to allow partial remyelination.

Experimental design

Mice were either given CPZ 0.25% for six consecutive weeks followed by normal chow (NC) for a week or NC for seven weeks. Mice were randomized (R) into four groups (n = 16 per group) consisting of CPZ + tACS 10 Hz; CPZ + Sham; NC + tACS 10 Hz and NC + Sham. Cognitive tasks were performed at baseline, after six weeks of CPZ or NC and one week after tACS or sham stimulation. At day 14 electrode holders were stereotaxically implanted on the skulls of the mice. Mice were subjected to five consecutive days of tACS or sham stimulation while given 5-ethynyl-2′-deoxyuridine (EdU, A10044, Invitrogen) 0.2 mg/mL in their drinking water. Following the cognitive tasks performed after the last tACS/sham, mice were sacrificed to harvest brain tissue for further histological analysis.

Surgery

Mice were first anesthetized with isoflurane (Vetflurane, 1000 mg/kg, Virbac, Belgium) (3.5% for induction, ∼2% for maintenance) and placed in a stereotaxic frame (Kopf instruments). After exposing the skull, two small craniotomies (approximately 1 mm ø) were made above the right prefrontal cortex and above the left hippocampal region with a manual micromotor drill (Volvere Vmax, NSK Nakanishi, Japan). Two nylon screws (0-80 x 3/32N, Bilaney Consultants GmbH, Germany) were implanted in these craniotomies and a tissue adhesive (Superbond, Sun Medical) was applied around the plastic screws to fix them. Two electrode holders (internal ø:2.1 mm, Dixi medical, France) were stereotactically placed on the skull at the following coordinates: ML -1.0 mm AP +1.0 mm and ML +1.6 mm AP -2.6 mm relative to bregma and attached with layer of dental acrylics (Dentalon plus, Kulzer, Germany).

Transcranial alternating current stimulation

Two weeks after surgery, mice underwent five consecutive daily sessions of tACS or sham (20 min each) while awake and freely moving in their home cage. Saline-soaked electrodes (3.5 mm^2^ contact area, Dixi medical, France), were screwed into the electrode holder. The anode was placed in the left frontal electrode holder, whereas the cathode was screwed in the right posterior electrode holder. Stimulation (0.1 mA, 10 Hz) or sham (ramp-up/down with no sustained current) was applied using an animal transcranial electric stimulator (Soterix medical, NO:2101, USA). Mice were randomly assigned to either sham stimulation or tACS 10 Hz using https://www.random.org/lists/.

Cognitive tests

The NOL task was used in this study to evaluate spatial learning and recognition memory. For this procedure we followed the habituation paradigm and setup as described previously by Van Goethem et al.51 In short, mice were placed in a circular arena (⌀ 40 cm) for 5 min containing two identical objects placed next to each other (i.e., training phase). Following a delay of 1 h, the mice were re-inserted in the arena for 5 min with one of the objects displaced to a new location (i.e., test phase). Trials in which the total exploration did not exceed 15 s were excluded from the statistical analysis. Both the training and test phases were recorded using a Basler camera (Basler AG, Germany) and analyzed with Ethovision 17.0 tracking software (Noldus, Wageningen, The Netherlands) in a blinded fashion. The discrimination index (DI) measured the effectiveness of mice to recognize the new object location, we calculated this as follows:

$$[eqn]$$

Spatial working memory was tested with the 6-min Y maze test as previously described.52 Spontaneous alternations (SA) were calculated as follows:

$$[eqn]$$

Immunofluorescence

At the end of the cognitive tests, mice were sacrificed to harvest brain tissue for histological analysis. Mice were sacrificed by sodium pentobarbital overdose (Dolethal, 200 mg/kg) followed by transcardiac perfusion. Mice were first perfused with heparin (heparin Leo, Leo Pharma, Belgium) 20 UI/ml in phosphate-buffered saline avoid blood clots and then perfused 4% paraformaldehyde in PBS. Brains were post-fixated overnight in 4% paraformaldehyde. Coronal brain sections (40 μm thickness) were cut on a vibratome (Leica, VT 1000S, Belgium). Prior to performing immunofluorescence on free-floating brain slices, EdU was labeled using the Alexa Fluor 647 Click-IT EdU kit (C10340, Invitrogen) according to manufacturer’s guidelines. Slices were then washed three times in Tris-buffer saline (TBS) after which they were incubated overnight at 4°C in 0.3% Triton X in TBS mixed with 10% donkey serum and following primary antibodies: rabbit anti-OLIG2 antibody (Merck Millipore, IgG, polyclonal, AB9610 at a concentration of 1/500), mouse anti-CC1 antibody (Abcam IgG, polyclonal, AB16794 at a concentration of 1/500), goat anti-IBA1 (Invitrogen, IgG, PA5-18039 at a concentration of 1/1000). The following day, slices were incubated for 60 min in a mixture of 0.3% Triton X in TBS and following secondary antibodies: AF488 donkey anti-chicken (Jackson Immuno Research, 703-545-155, concentration 1/500), AF488 donkey anti-mouse (Jackson Immuno Research, 715-545-150, concentration 1/500), Cy3 donkey anti-rabbit (Jackson Immuno Research, 711-165-152, concentration 1/400), Cy5 donkey anti-goat (Jackson Immuno Research, 705-175-147, concentration 1/500). 4′,6-diamidino-2-phenylindole (DAPI) was used at a concentration of 1/1000 to counterstain nuclei.

Confocal z stack images (20 optical slices at 0.5 μm intervals) were acquired on a Zeiss LSM800 microscope (63× oil objective). Oligodendroglial cells (OLIG2^+^ and CC1^+^) were counted manually by two blinded observers (MG and TJS) in the rostral part of the corpus callosum using FIJI (NIH, USA).

Luxol fast blue

LFB staining was performed on 40-μm free-floating vibratome sections, following the protocol described by Geisler et al.53 To enhance LFB penetration we first incubated the slices in ethanol 70% for 24 h. Slices were then mounted on superfrost microscope slides and incubated the slices in LFB (luxol fast blue MBSN, 212170250, Thermo Fisher) overnight at 56°C. The excess stain was removed by rinsing the slices in distilled water and PBS for 3 min each. To remove excess dye from non-myelinated areas, we transferred the slices in lithium carbonate 0.05% for 3 min followed by 70% ethanol for 3 min and two changes of PBS for 5 min each. Slices were dehydrated by immersing them for 1 min in 70% ethanol, 2 min in 80% ethanol, 5 min in 96% ethanol, twice in 100% ethanol for 5 min each and finally twice in xylene for 5 min each.53 For analysis, three different regions of interest (ROIs) were selected based on anatomical relevance and observed demyelination patterns. Specifically, the “remyelination area (%)” was quantified by calculating the proportion of the LFB-positive area within predefined regions of interest relative to the total ROI area using FIJI (ImageJ). ROIs were manually delineated in the rostral and caudal corpus callosum and the ipsilateral dentate gyrus, based on anatomical landmarks. Thresholding parameters were kept constant across all sections and groups to ensure consistency. The resulting LFB-positive fraction was expressed as a percentage of the total ROI area.

Transmission electron microscopy (TEM)

For myelin visualization via TEM, mice were sacrificed as described previously. However, instead of using 4% paraformaldehyde, an electron microscopy fixative containing 0.1 M cacodylate buffer, 1 mM CaCl_2_ and 2.5% glutaraldehyde was used. Brains were postfixed overnight in the same mixture and slices at the region of interest were cut at a thickness of 400-500 μm using a vibratome. These slices were incubated in osmium tetroxide 1% and embedded in Spurr’s resin. Ultrathin sections were stained with uranyl acetate and lead citrate and images were acquired at 8700× magnification on a Tecnai 10 electron microscope. We specifically chose the rostral corpus callosum because this region was targeted by the anodal electrode. Images were analyzed using the software MyelTracer by two blinded experimenters. This semi-automatic software allowed to obtain g-ratios for all examined axons.54 G-ratio was calculated as follows:

$$[eqn]$$

Quantification and statistical analysis

Data are presented as mean ± SEM. Behavioral performance in the NOL and Y-maze tasks was compared to chance levels (0% or 50%) using one-sample t-tests with false-discovery rate correction. Normality of data and residuals was tested using the Shapiro–Wilk test. Group effects were evaluated by two-way ANOVA followed by Sidak’s multiple-comparison tests. In cases where normality assumptions were not met, non-parametric alternatives were used such as aligned rank transform ANOVA with Sidak’s multiple-comparison tests. Statistical significance was defined as p < 0.05. Statistical significance was defined as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.