The interleukin-4/PPARγ signaling axis promotes oligodendrocyte differentiation and remyelination after brain injury

The repair of white matter damage is of paramount importance for functional recovery after brain injuries. Here, we report that interleukin-4 (IL-4) promotes oligodendrocyte regeneration and remyelination. IL-4 receptor expression was detected in a variety of glial cells after ischemic brain injury, including oligodendrocyte lineage cells. IL-4 deficiency in knockout mice resulted in greater deterioration of white matter over 14 d after stroke. Consistent with these findings, intranasal delivery of IL-4 nanoparticles after stroke improved white matter integrity and attenuated long-term sensorimotor and cognitive deficits in wild-type mice, as revealed by histological immunostaining, electron microscopy, diffusion tensor imaging, and electrophysiology. The selective effect of IL-4 on remyelination was verified in an ex vivo organotypic model of demyelination. By leveraging primary oligodendrocyte progenitor cells (OPCs), microglia-depleted mice, and conditional OPC-specific peroxisome proliferator-activated receptor gamma (PPARγ) knockout mice, we discovered a direct salutary effect of IL-4 on oligodendrocyte differentiation that was mediated by the PPARγ axis. Our findings reveal a new regenerative role of IL-4 in the central nervous system (CNS), which lies beyond its known immunoregulatory functions on microglia/macrophages or peripheral lymphocytes. Therefore, intranasal IL-4 delivery may represent a novel therapeutic strategy to improve white matter integrity in stroke and other brain injuries.

Introduction matter components. However, the potential effects of IL-4 on white matter integrity after ischemic stroke remain elusive.
In the current study, we discovered that IL-4 is essential for white matter repair after stroke. Intranasal delivery of IL-4 nanoparticles after stroke robustly promoted oligodendrogenesis, enhanced white matter integrity, and improved long-term functional recovery. Complementary studies in cell cultures and animals revealed that IL-4-mediated oligodendrogenesis was achievable even in the absence of microglia/macrophages in vitro or after microglia/macrophage significant depletion in vivo. Furthermore, IL-4 directly promoted OPC differentiation into mature oligodendrocytes in a peroxisome proliferator-activated receptor gamma (PPARγ)-dependent manner. These collective findings suggest that activation of the IL-4/ PPARγ signaling cascade may represent a novel recovery-enhancing strategy to promote longterm outcomes after ischemic stroke.
In light of these results, we evaluated whether IL-4 influenced white matter integrity after ischemic stroke. WT and IL-4 KO mice were subjected to transient focal ischemia induced by 60-min MCAO. Dual staining for myelin basic protein (MBP, a major myelin protein) and SMI32 (a marker of demyelinated axons) was performed to assess the lesion in white matter (Fig 1F-1H). The overall immunofluorescence intensity of MBP staining decreased in the infarct border along three white matter-rich brain regions (the external capsule [EC], the cortex, and the striatum) (see "total" values in panel of the right of Fig 1F) at 14 d after MCAO, indicating a significant loss of myelin protein after stroke. Additionally, the overall immunofluorescence of SMI32 increased in these three regions, resulting in a significant increase in the SMI32/MBP ratio in WT mice after stroke (Fig 1G). IL-4-deficient mice displayed exacerbation of the loss in MBP ( Fig 1F) and a further increase in the SMI32/MBP ratio compared to WT mice (Fig 1G), suggesting that IL-4 is important in maintaining axonal myelination or promoting axonal remyelination after stroke.
Stroke impacts both gray and white matter. In order to confirm the effect of IL-4 in white matter, we explored the remyelination capacity of IL-4 in a model of lysophosphatidylcholine/lysolecithin (LPC)-mediated demyelination using ex vivo organotypic cerebellar slice cultures. This toxicant-based experimental model of demyelination, although not attempting to fully mimic a disease with complex etiology and pathogenesis, has proven extremely useful for understanding the biology of remyelination [23]. Slices collected from WT and IL-4 KO mice exhibited comparable extents of demyelination, as revealed by the loss of MBP staining along neurofilamentheavy (NFH + ) axonal fibers, at 3 d after LPC (Fig 2A). Partial remyelination was achieved by 7 d following LPC in WT brain slices, but IL-4 deficiency significantly delayed remyelination (Fig 2A and 2B), as shown by lower MBP staining intensity ( Fig 2C) and less MBP + axonal coverage ( Fig 2D). IL-4 supplementation significantly enhanced remyelination in WT brain slices and rescued remyelination capacities in IL-4 KO brain slices (Fig 2A, 2C and 2D).

Intranasal delivery of IL-4 nanoparticles promotes white matter integrity after stroke
IL-4 liposome nanoparticles were prepared to facilitate the penetration of IL-4 into the ischemic brain. The mean particle size of IL-4 protein-loaded liposome was 122.6 ± 2.4 nm with a and the ratio of SMI32 to MBP immunofluorescence intensity (G) in ipsilesional EC, cortex, striatum, and the total for all three areas quantified. n = 6-8/group. Data are normalized to the intensities of contralateral hemispheres. � p < 0.05, �� p < 0.01, ��� p < 0.001. One-way ANOVA and Bonferroni post hoc. (H) Representative images for MBP (green) and SMI32 (red) double immunostaining in the peri-infarct areas in brains from sham, WT MCAO mice, and IL-4 KO MCAO mice 14 d after MCAO. Scale bar: 50 μm. Data associated with this figure can be found in the supplemental data file (S1 Data). CNS, central nervous system; EC, external capsule; FSC-W, forward scatter width; GFAP, glial fibrillary acidic protein; IL-4R, interleukin-4 receptor α; KO, knockout; MBP, myelin basic protein; MCAO, middle cerebral artery occlusion; MFI, mean fluorescence intensity; OPC, oligodendrocyte progenitor cell; SSC-A, side scatter area; WT, wild-type.  LPC (500 μg/mL) was added in WT or IL-4 KO organotypic cultures for 17 h to induce demyelination. IL-4 (20 ng/mL) or vehicle was added after LPC removal. Culture slices were fixed 3 d and 7 d after LPC treatment. Control cultures were maintained in normal culture media for 7 d. (A) MBP (green) and NFH (red) staining was performed to label myelin and axonal fibers, respectively. Scale bar: 50 μm. (B) MBP and NFH double staining and Imaris 3D reconstruction. Arrows indicate myelination of individual axonal fibers that are magnified in the images to the right. Scale bar: 15 μm. (C) Myelin integrity was measured as the intensity of MBP immunostaining. (D) The myelination index was calculated as the % MBP + myelin coverage along the entire NFH + fibers. n = 5-7 slices/group. � p < 0.05, �� p < 0.01, ��� p < 0.001. One-way ANOVA and Bonferroni post hoc. Data associated with this figure can be found in the supplemental data file (S1 Data). IL-4, interleukin-4; KO, knockout; LPC, lysophosphatidylcholine/lysolecithin; MBP, myelin basic protein; NFH, neurofilament heavy; WT, wild-type. polydispersity index of approximately 0.212 ± 0.004 and a zeta potential of 4.05 ± 1.35 mV, confirming high stability of the nanodispersion. The IL-4 nanoparticles were applied through the nasal route (50 μg/kg body weight), which has been increasingly utilized for CNS drug delivery because of its capacity to bypass the blood-brain barrier and avoid first-pass metabolism [24]. Intranasal administration of IL-4 nanoparticles effectively increased IL-4 protein levels in the corpus callosum (CC) and EC, cerebral cortex, and striatum 12 h after IL-4 application (Fig 3A-3C). IL-4 remained elevated in cortex and striatum until 24 h after delivery (Fig 3B and 3C).
C57BL/6J mice were subjected to 60-min MCAO and randomly assigned to receive IL-4 nanoparticle (50 μg/kg body weight) or control nanoparticle (vehicle) treatment. Nanomaterials were intranasally delivered at 6 h after MCAO and repeated at 1-7 d, 14 d, 21 d, and 28 d after MCAO (Fig 3D). Magnetic resonance imaging (MRI) on these animals was performed serially at 9.4 T on days 2, 14, 21, and 28 after MCAO. T 2 -weighted images showed no significant difference in the brain lesion volume of IL-4-versus vehicle-treated mice (Fig 3E and 3F), indicating comparable gross tissue damage across treatment groups. Histological staining of the neuronal marker NeuN further demonstrated comparable tissue loss 35 d after stroke in IL-4-and vehicle-treated mice (Fig 3E and 3G).
Serial whole-brain in vivo diffusion tensor imaging (DTI) scans were collected 14 d, 21 d, and 28 d after MCAO. The same animals were then killed at 35 d after stroke, and the fixed brains were subjected to ex vivo DTI scanning (Fig 4A and 4B, S2A and S2B Fig). Longitudinal observation of fractional anisotropy (FA) maps and calculation of the ratio of FA(ipsilateral)/ FA(contralateral) demonstrated progressive white matter injury (ratio < 1) with a slow recovery in the EC of vehicle-treated stroke mice at 14-35 d after stroke, which was significantly improved by day 28 following stroke in IL-4-treated mice (Fig 4C). This trend was observed until day 35 in the ex vivo FA values (Fig 4C). Further analysis of the 35 d DTI images revealed that IL-4 treatment enhanced white matter integrity, as revealed by elevated FA values in the white matter-enriched EC and internal capsule (IC) (Fig 4D, 4E and 4G). In addition, IL-4 treatment significantly reduced radial diffusivity (RD, λ ? ) in the EC ( Fig 4F) and IC (Fig 4H), which likely indicates improved myelin integrity [25]. No significant differences between IL-4 and vehicle treatment groups were observed in axial diffusivity (AD, λ || ), a measurement for axonal integrity (S2C and S2D Fig).
To further explore the biological basis of these DTI results, we compared the DTI results with immunohistochemical staining of MBP performed in the same animals. There were remarkable similarities between the FA maps and MBP images (Fig 4D, 4I and 4J). IL-4 treatment significantly increased the MBP + area in the EC (Fig 4D and 4I). There was a significant positive correlation between the MBP + areas and the FA values in the EC (Fig 4J). The quantification of MBP staining confirmed a significant increase in MBP intensity in the ischemic brains of IL-4-treated mice compared to vehicle-treated mice (Fig 4K).

IL-4 post-treatment improves the structural and functional integrity of myelinated fibers after stroke
To determine whether IL-4 can improve axonal myelination after stroke, myelin thickness was measured in the EC using transmission electron microscopy (TEM) (Fig 5A). Frequency histograms revealed much greater reduction of axon diameters in the MCAO + vehicle group compared to the MCAO + IL-4 and sham groups, suggesting higher axonal degeneration in stroke mice without IL-4 treatment (Fig 5B). Scatter plots of the g-ratio (the ratio of the inner axonal diameter to the total outer diameter of myelinated fiber) as a function of axon diameter in sham mice or stroke mice treated with vehicle or IL-4 revealed significant differences in myelin thickness (average g-ratio: sham = 0.72 [n = 183 axons from 3 animals]; MCAO + vehicle = 0.79 [n = 233 axons from 3 animals]; MCAO + IL-4 = 0.73 [n = 176 axons from 3 animals]) ( Fig  5C). Significant increases in the g-ratio were observed in both low-and high-caliber axons at 35 d after MCAO, indicating a reduction in myelin thickness ( Fig 5D). IL-4 post-treatment significantly reduced the g-ratio of the middle-sized (diameter = 400-800 nm) and large (diameter > 800 nm) axonal fibers, suggesting that IL-4 may maintain axonal myelination or promote axonal remyelination after stroke. The g-ratio in small axonal fibers (diameter < 400 nm) displayed a trend toward reduction with IL-4 treatment ( Fig 5D).
Saltatory conduction of action potentials along myelinated axons relies on normal organization of the nodes of Ranvier (NOR), which are rich in sodium and potassium ion channels. Double immunofluorescent analyses of Nav1.6 (a nodal sodium channel protein) and contactin-associated protein (Caspr, a paranodal cell adhesion protein) revealed the NOR in the EC (Fig 5E). Typical NOR are characterized by punctate Nav1.6 staining at nodes between the paranodal staining of Caspr. The total numbers of normal NOR decreased significantly after stroke in the ipsilateral EC compared to the contralateral EC (Fig 5F), and this was accompanied by a decrease in the Caspr + paranode length (Fig 5G). IL-4 treatment significantly increased the number of NOR and the paranode length in the ischemic EC (Fig 5F and 5G), suggesting an improvement in NOR structure. The paranodal gap between pairs of Caspr staining remained unchanged between groups ( Fig 5H).
As IL-4 enhanced the structural integrity of white matter after MCAO, we further evaluated white matter function by measuring the transmission of action potentials in the CC and EC (Fig 5I). Evoked compound action potentials (CAPs) display an early peak representing fast-conducting myelinated axons (N1, Fig 5J) [26]. The amplitude of the N1 segment was significantly reduced 35 d after MCAO (Fig 5K), indicating impaired conduction through myelinated axons. IL-4-treated mice exhibited less reduction in the N1 amplitude after MCAO, suggesting myelin preservation or improved remyelination of axons compared to the vehicle-treated subjects.

IL-4-afforded white matter protection is associated with superior longterm functional recovery after stroke
Sensorimotor and cognitive functions were assessed by investigators blinded to group assignments. As shown in Fig 6A and 6B, IL-4-treated and vehicle-treated stroke mice showed comparable reductions in latencies to contact and remove adhesive tape from the paw at 3 d after stroke, suggesting similar initial deficits in dexterity across groups. However, at later time points starting from 7-10 d after stroke, IL-4 treatment reduced the time spent touching and removing tapes in stroke mice, culminating in improved performance in the adhesive removal test (Fig 6A and 6B). Long-term sensorimotor performance in the rotarod test after MCAO was also improved in IL-4-treated mice ( Fig 6C). Furthermore, IL-4-treated mice exhibited long-term improvements in cognitive function in the Morris water maze test, as manifested by a reduction in the latency to find the hidden platform (improved spatial learning, Fig 6D and  6E) and increased time spent in the target quadrant when the platform was removed (improved memory, Fig 6D and 6F). There was no difference in swimming speed among different treatment groups, indicating similar gross motor skills across groups ( Fig 6G).
Next, we measured the potential correlation between white mater integrity and behavioral performance after stroke in vehicle or IL-4-treated mice. The results revealed significant negative correlations between MBP intensity and the latency to touch ( Fig 6H) and remove ( Fig 6I) the adhesive tape at 35 d after stroke, confirming an association between histological observations of white matter integrity and long-term functional recovery of behavioral endpoints.

IL-4 treatment promotes oligodendrogenesis partially independent of microglia/macrophages
Successful differentiation of OPCs into myelinating oligodendrocytes is important for remyelination [11]. Thus, we compared the regeneration of myelin-producing oligodendrocytes in MCAO mice with or without IL-4 treatment. 5-bromo-2 0 -deoxyuridine (BrdU) was injected at 3-6 d after MCAO to label newly generated cells. Mature oligodendrocytes were recognized by immunostaining with anti-adenomatous polyposis coli (APC) (or CC1) antibodies [27]. Compared to vehicle-treated mice, a significantly higher ratio of BrdU + APC + newly generated oligodendrocytes was observed in IL-4-treated mice 35 d after MCAO (Fig 7A and 7B). The total numbers of BrdU + cells and the total numbers of APC + cells were also significantly increased in IL-4-treated mice compared to vehicle-treated mice (Fig 7C and 7D). It has been reported that higher concentrations of the APC antibody may detect some populations of astrocytes in certain pathological conditions [27]. However, no APC signal was detected in glial fibrillary acidic protein (GFAP) + astrocytes in the ischemic brain in our preclinical model of stroke (S3 Fig).
IL-4 is known to promote microglia M2 polarization, which in turn enhances OPC differentiation and remyelination [28]. We observed in our model that IL-4 treatment robustly promoted microglia/macrophage polarization toward an anti-inflammatory phenotype (CD206 + Iba1 + ) at 7 and 14 d after stroke (S4 Fig). Microglia depletion through genetic targeting or pharmacological treatment is a widely used approach to evaluate the contribution of this population to various biological processes [29]. Therefore, we chose PLX5622, an inhibitor of colony-stimulating factor 1 receptor, to deplete microglia/macrophages and identify their role in IL-4-afforded oligodendrogenesis ( Fig 7E) [30]. Consistent with previous reports [30], PLX5622 achieved approximately 90% depletion efficacy within a short period. Dietary intake of PLX5622 for 7 d dramatically reduced the number of microglia in the normal brain Significant depletion of microglia/macrophages did not change the number of BrdU + APC + cells or total BrdU + cells compared to microglia/macrophage-competent stroke mice (Fig 7F-7H). IL-4 treatment increased the numbers of BrdU + APC + cells or total BrdU + cells after stroke in mice (H) Quantification of the number of BrdU + cells in the peri-infarct areas. n = 7-8/group. � p < 0.05, ��� p < 0.001. One-way ANOVA followed by Bonferroni post hoc. Data associated with this figure can be found in the supplemental data file (S1 Data). APC, adenomatous polyposis coli; BrdU, 5-bromo-2 0 -deoxyuridine; CTX, cortex; EC, external capsule; IL-4, interleukin-4; MCAO, middle cerebral artery occlusion; STR, striatum. fed with the control diet. IL-4 treatment exhibited a trend (p = 0.086) toward lower capacities to increase the number of BrdU + APC + newly generated oligodendrocytes in microgliadepleted mice versus microglia-competent mice, supporting a potential role of microglia in IL-4-afforded oligodendrogenesis. Importantly, IL-4 treatment still resulted in a significant enhancement of BrdU + APC + oligodendrocyte regeneration in microglia/macrophagedepleted stroke mice (Fig 7F and 7G). Considering the expression of IL-4R on oligodendrocyte linage cells, these results indicate that IL-4 may exert microglia/macrophage-independent effects on oligodendrocyte regeneration after demyelination.

Intranasal IL-4 treatment exerts minimal effects on adaptive immune cell populations
Consistent with a previous study in a model of multiple sclerosis [31], we found that repeated intranasal IL-4 administration did not change T-lymphocyte subpopulations or B lymphocytes in the blood (S6A-S6C Fig

IL-4 directly stimulates OPC differentiation in primary OPC cultures
Consistent with a direct effect of IL-4 on oligodendrocytes, in vitro studies showed that IL-4 at a concentration range of 5-100 ng/mL directly stimulated the differentiation of neuron-glia antigen 2 (NG2) + OPCs into MBP + mature oligodendrocytes without the aid of microglia ( Fig  8A and 8B). IL-4 at concentrations of 10 or 20 ng/mL induced OPC differentiation that was comparable to the effects of a T3 cocktail typically used to induce OPC differentiation in vitro. The effect of IL-4 on OPC differentiation was indeed IL-4R-dependent, because specific IL-4Rα neutralizing antibodies (IL-4RAbs) blocked the IL-4-induced increase in mRNA expression of the mature oligodendrocyte marker mbp (Fig 8C). IL-4 at a concentration range of 5-100 ng/mL had no effect on OPC survival (S7A and S7B

PPARγ is essential for IL-4-induced OPC differentiation and maturation
Microarray analyses were performed to further elucidate the mechanisms underlying the promotion of oligodendrogenesis by IL-4. OPCs were treated with PBS or IL-4 (20 ng/mL) for 3 d. Total RNA extraction from cell pellets was used for the Affymetrix Clariom S rat array. Principal component analysis (PCA) showed clustering of samples in a treatment-dependent manner ( Fig 8D). In total, 996 differentially expressed genes (DEGs) were identified (false discovery rate [FDR] < 0.05; and fold change > 2), out of which 495 were up-regulated and 501 were down-regulated in IL-4-treated OPCs relative to PBS-treated OPCs (Fig 8E, S1 and S2 Tables). The gene ontology (GO) enrichment analysis was then performed using the Metascape analysis [32] (http://metascape.org). The top 20 enriched GO terms in three categories (biological processes, cellular components, and molecular functions) were identified in the upregulated DEGs by IL-4 treatment (S8A Fig). A network plot further captured the relationships between the main regulated biological processes, which were related to cell adhesion and motility, cytokine production, cell responses to external stimuli, cellular component organization, and cell death (Fig 8F). The enriched GO terms among the down-regulated DEGs in IL-4-treated OPCs represent biological processes such as developmental growth, small molecule biosynthetic process, vessel development, and responses to nutrients and peptides (S8B Fig). These results indicated changes in cell mobilization, growth, differentiation, and defense in IL-4-treated OPCs. A Kyoto encyclopedia of Genes and Genomes (KEGG) pathway analysis  8G). The PPAR transcription factor assay confirmed that PPARγ activity, but not PPARα or PPARδ activities, was up-regulated in OPCs upon IL-4 treatment ( Fig 8H); these effects were abolished when IL-4Rα was blocked with a neutralizing antibody ( Fig 8I). Furthermore, immunostaining (Fig 8J and 8K) and western blotting (Fig 8L-8N, and S9 Fig) demonstrated that the PPARγ inhibitor GW9662 blocked IL-4-induced expression of mature oligodendrocyte markers (MBP or myelin oligodendrocyte glycoprotein [MOG]), suggesting that PPARγ activation is essential for the OPC differentiation induced by IL-4.
Finally, OPC-specific PPARγ KO (PPARγ-OPC KO) mice were used to confirm the importance of PPARγ signaling in IL-4-enhanced oligodendrogenesis in an in vivo stroke model ( Fig 9A). To confirm the KO of PPARγ specifically in OPCs, PDGFRα + CD45 − OPCs, GLAST + CD45 − astrocytes, CD45 + immune cells, and other CD45 − GLAST − PDGFRα − CNS cells were sorted from the brains collected from PPARγ-OPC KO mice at 9 d after 4-hydroxytamoxifen injections. The DNA extracts were subjected to PCR using primers flanking the loxp site and primers detecting the Cre cDNA. The loxp band was abolished in OPCs because of the targeted deletion after Cre-induced recombination, while remaining intact in astrocytes, immune cells, and other CNS cells sorted from the same brain ( Fig 9B). The expression of Cre was detected in the OPCs of the PPARγ-OPC KO mice ( Fig 9B). Deletion of PPARγ specifically in OPCs exerted no impact on IL-4R expression on O4 + oligodendrocytes (S10 Fig). Vehicle or IL-4 treatment in PPARγ-OPC KO mice resulted in similar neuronal tissue loss (Fig 9C and 9D) and white matter lesion (Fig 9C, 9E and 9F) 35 d after stroke. The lack of PPARγ expression in OPCs did not significantly impair oligodendrocyte replacement after stroke, as shown by comparable numbers of BrdU + APC + cells in the ischemic brain from PPARγ-OPC KO mice and control mice at 35 d after MCAO (Fig 9G and 9H). However, the absence of PPARγ in OPCs reduced the capacity of IL-4 to increase the number of newly generated BrdU + APC + oligodendrocytes (Fig 9G and 9H). Consistent with a reduced capacity to promote oligodendrogenesis, IL-4 failed to improve long-term sensorimotor (Fig 9I and 9J) or cognitive (Fig 9K-9M) deficits after stroke in the absence of PPARγ. IL-4/PPARγ signaling improves white matter repair after stroke the ischemic brain, which is attributed, at least partly, to its release from injured neurons or from infiltrated immune cells [19]. Cerebral IL-4 levels then return to baseline within 1 wk after stroke [16,19]. In patients, serum levels of IL-4 are known to change following an acute stroke [33]. Furthermore, IL-4 gene polymorphisms are related to susceptibility to ischemic stroke and subsequent functional outcomes [34,35]. Finally, a recent study reported the expression of IL-4R surrounding axonal filaments in postmortem brain tissue from patients with multiple sclerosis [31]. Although these clinical and experimental data are correlative, the present series of in vivo and in vitro studies firmly establish a causal link between IL-4 repletion and white matter integrity in preclinical stroke models. Deficiency of endogenous IL-4 An improvement in white matter integrity might be achieved in two ways: preservation of preexisting white matter structures or enhancement of white matter repair. Our data favor the latter as the predominant mechanism of action for IL-4. First, longitudinal DTI revealed similar reductions in FA values shortly after stroke in vehicle or IL-4-treated mice, indicating comparable degrees of initial white matter injury during the early/acute phase. A significant improvement in white matter microstructure was only observed in IL-4-treated animals in later stages of recovery (28 and 35 d after stroke). These results strongly suggest a delayed effect of IL-4 on white matter repair, rather than preservation shortly after injury. Second, electron microscopy (EM) and histological studies revealed structural signs of remyelination in the ischemic brain, including thinner myelin segments than before stroke, and reduced paranode length with Caspr/Nav1.6 staining [36]. Each of these structural alterations were significantly improved by IL-4 treatment, indicating superior remyelination. Third, IL-4 treatment increased the number of new, mature oligodendrocytes (APC + BrdU + ) in the ischemic brain 35 d after stroke. Such improvements in oligodendrogenesis are critical for successful remyelination. Finally, the LPC model of demyelination-remyelination unequivocally confirmed the white matter-specific effects of IL-4 in remyelination. As expected, the IL-4-mediated reconstitution of myelin structure culminated in functional improvements in neural conduction and behavioral performance, despite no changes in the degree of gray matter loss. Although IL-4 is important for white matter repair after brain injury, the IL-4 KO mice do not display deficits in myelin coverage in organotypic brain slice cultures under physiological conditions. This observation is consistent with the view that different mechanisms control myelin sheath formation during normal development and remyelination [36].

Discussion
Although our T2 scanning did not show significant reductions in brain lesion size under conditions of IL-4 repletion, there was a trend toward lesion volume reduction at 14-35 d after stroke. As the IL-4 treatment was initiated 6 h after stroke, it remains possible that IL-4 may provide some degree of early protection [16]. It is also possible that some NeuN + neurons remaining within the gray matter of mice without IL-4 might not be physiologically functional, although they are still histologically present, perhaps because their communication via white matter tracts is significantly impeded.
A previous study using mixed glial cultures demonstrated that the addition of IL-4 promotes the differentiation of OPCs into MBP-expressing oligodendrocytes in vitro, an effect mainly attributed to IL-4-mediated anti-inflammatory effects [37]. An additional conclusion emerging from our studies is that IL-4 enhances white matter repair at least partially independent of its well-established anti-inflammatory effects on microglia/macrophages. In IL-4-treated stroke mice, we observed an anti-inflammatory phenotypic change in microglia/ macrophages, which is known to promote remyelination after CNS injury [28]. However, dramatic depletion of microglia/macrophages in the ischemic brain did not abolish the capacity of IL-4 to enhance oligodendrocyte replacement, supporting the involvement of microglia/ macrophage-independent mechanisms. In vitro studies then confirmed that IL-4 can also directly target white matter components and promotes OPC differentiation into mature oligodendrocytes in the absence of microglia/macrophages. Previous fate mapping studies revealed that OPCs are responsible for generating new oligodendrocytes during remyelination [38,39]. However, the differentiation of OPCs is frequently impaired in lesioned brains, mainly because of nonpermissive environmental cues [40][41][42]. We found that IL-4 treatment may overcome these obstacles for OPC differentiation in the ischemic brain, both directly by promoting OPC differentiation and indirectly through immunomodulation. Notably, endogenous oligodendrocyte differentiation has been observed until as late as 3 mo after CNS injuries [39], and this prolonged effort may provide a sufficiently wide therapeutic window that allows delayed IL-4 treatment to restore white matter and encourage long-term stroke recovery.
In the present study, we report a previously undefined mechanism of IL-4, namely that it directly induces OPC differentiation by activating PPARγ signaling. PPARγ is a master transcription factor known to promote OPC differentiation and maturation though manifold mechanisms involving enhanced myelin lipid synthesis, reduced oxidative stress, and promotion of mitochondrial function [43,44]. Our previous work has shown that rosiglitazone, a PPARγ agonist, enhanced OPC proliferation and increased the number of newly generated mature oligodendrocytes after MCAO [45]. In addition, PPARγ may interact with histone deacetylases to modulate cell differentiation [46]. We demonstrated here that PPARγ inhibition abolished the effect of IL-4 on OPC differentiation in vitro and that OPC-specific KO of PPARγ in vivo reduced IL-4-enhanced oligodendrocyte replacement in the ischemic brain. These collective results indicate that IL-4-induced oligodendrocyte differentiation/maturation relies at least partially on the activation of PPARγ activity in OPCs. Precisely how IL-4/IL-4R engages PPARγ activity is not yet known, but several studies suggest that IL-4 may activate PPARγ in macrophages and other cell types through signal transducer and activator of transcription 6 (STAT6) signaling [37,47]. Whether this pathway is also operational in OPCs lies beyond the scope of the present study but warrants further investigation.
Aside from the cell-specific effects of IL-4 on OPCs, we note that IL-4R is widely distributed in many types of CNS cells as well as infiltrated immune cells. Several lines of evidence from our study support the capacity of IL-4 to target multiple types of cells in addition to oligodendrocytes. First, we did not observe significant differences in the density of BrdU + APC + cells between WT and PPARγ-OPC KO mice following MCAO, suggesting that endogenous IL-4 is able to promote white matter integrity by targeting other types of cells. Second, although PPARγ inhibition eliminated the effect of IL-4 on OPC differentiation in cultures, the absence of PPARγ in OPCs reduced, but did not abolish, the capacity of IL-4 to increase the number of newly generated oligodendrocytes in vivo. These results suggest that PPARγ expression in OPCs is an important but not exclusive mechanism for IL-4-induced oligodendrogenesis in preclinical stroke. Indeed, we observed that IL-4 treatment robustly promoted microglia/macrophage polarization toward an anti-inflammatory M2phenotype, as mentioned above. IL-4-stimulated microglia have been shown to release IL-4 induced 1 (IL4I1), which in turn augments remyelination [48]. Thus, direct and indirect effects on microglia/macrophages may contribute to the white matter repair afforded by IL-4. It has also been reported that the effects of IL-4 on adaptive immune responses contribute partially to recovery from neurotrauma [49]. In our hands, however, repeated intranasal IL-4 administration did not change T-lymphocyte subpopulations or B lymphocytes in blood and did not alter the composition of infiltrated lymphocytes. These data dispute the involvement of adaptive immune responses in IL-4-afforded neuroprotection in our stroke model.
Aside from its roles in immunoregulation, IL-4 can also directly target other types of CNS cells. For example, a recent study reported that the activation of IL-4R signaling in neurons may increase axonal outgrowth and ameliorate axonal pathology in a model of multiple sclerosis [31]. In this context, it is important to note that we did not observe improvements in the DTI measurement for axonal integrity (AD, λ || ). In other disease models, multiple factors including inflammation, cellular infiltration, and gliosis may contribute to AD at chronic stages of disease. Therefore, the measurement of λ || appears to lack sufficient resolution to distinguish axonal damage/repair versus confounding cellular responses over the course of progressive demyelination [50]. On the other hand, our EM data demonstrated an increase in axon diameter in IL-4-treated mice compared to vehicle-treated controls, supporting a beneficial role of IL-4 in maintaining axonal structure. These structural data were supported by our electrophysiological studies of axon conductance. Preservation of axonal structure and electrical function might provide an important feedback signal for OPC differentiation and guide remyelination [51,52], preventing Wallerian degeneration of neurons after ischemia and enhancing white matter repair. In addition to these mechanisms, astrocytes are known to produce trophic factors in response to IL-4 stimulation, including brain-derived neurotrophic factor (BDNF) and nerve growth factor (NGF), both of which are also critical for brain function and brain repair [53].
A series of in vitro studies assessed the effective concentrations of IL-4 in various types of CNS cells. For example, IL-4 regulates microglia activity at concentrations of 1-20 ng/mL [19,54]. IL-4 inhibits inflammatory responses and boosts trophic factor release in astrocytes at 5-10 ng/mL [53]. IL-4 protects neurons against excitotoxicity and stimulates axonal outgrowth in vitro at 50 ng/mL [31]. In the present study, IL-4 directly stimulated OPC differentiation within a concentration range of 5-100 ng/mL. These collective studies reveal similar effective IL-4 concentrations (in nanoscales) in glial and neuronal cells of the CNS. Indeed, it is precisely the multitargeting capacities of low-dose IL-4 that may render it an excellent therapy for CNS injuries characterized by a wide array of pathological mechanisms.
The ultimate implementation of preclinical research into clinical treatments requires extensive information on safety, optimal dose, route, and schedule of administration, among other variables. In the present study, IL-4 stimulated OPC differentiation in cultures at a concentration range of 5-100 ng/mL, and the optimal concentration was 10-20 ng/mL. Although there is neither a mathematical nor empirical method to accurately convert in vitro concentrations to in vivo intranasal dosage, these in vitro data support the capacity of IL-4 to directly target OPCs at low concentrations. Previous preclinical studies have similarly documented therapeutic effects of low doses of IL-4 in ischemic stroke and other neurological diseases. For example, subcutaneous IL-4 application at 2 μg/kg/d [19] or intracerebroventricular application of IL-4 at 60 ng/d [16] significantly improved long-term outcomes after stroke. Vogelaar and colleagues reported that intrathecal or intranasal (1 μg/d) IL-4 treatment ameliorated clinical signs and reversed disease progression in a model of multiple sclerosis [31]. Those studies, however, did not measure IL-4 concentrations in the brain after treatment. Here, we delivered IL-4 intranasally at a dose of 50 μg/kg/d (about 1.25 μg/d based on 0.025 kg average body weight), which significantly elevated IL-4 levels in key parts of the brain. This dose is similar to the effective intranasal dose for Vogelaar's study and improved long-term functional recovery in our stroke model. Repeated dosing regimens or controlled continuous release was employed for all in vivo IL-4 treatment because of its rapid clearance and inactivation. The use of liposome-entrapped peptides has been reported to protect the cytokine and enhance its tissue delivery [55,56]. Consistent with those findings, we demonstrated that IL-4 liposome nanoparticle delivery elevated IL-4 levels in the brain for at least 24 h. Further characterization of the optimal regimen of IL-4 treatment for stroke would accelerate its clinical translation.
Despite growing awareness of the importance of white matter restoration in poststroke rehabilitation, there is no effective therapy to preserve white matter function or to improve its repair in humans. Our study has uncovered a novel role of IL-4 in poststroke white matter recovery-a role that lies beyond its well-known immunoregulatory functions. Furthermore, we have shown that IL-4 promotes oligodendrogenesis and white matter repair in a PPARγdependent manner. Therefore, it seems reasonable to continue to test IL-4 as a therapeutic strategy for functional recovery after ischemic stroke. In particular, the intranasal route enables CNS IL-4 delivery, which is expected to significantly reduce the risk of peripheral side effects and expedite the eventual translation of IL-4 treatment for stroke victims.

Ethics statement
All animal procedures were approved by the University of Pittsburgh Institutional Animal Care and Use Committee (approval number: 18032546), performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and reported in accordance with the Animal Research: Reporting In Vivo Experiments (ARRIVE) guidelines. All efforts were made to minimize animal suffering and the number of animals used. All animals were housed in a temperature-and humidity-controlled facility with a 12-h light/dark cycle. Food and water were available ad libitum.

Murine models of transient cerebral ischemia
Transient cerebral ischemia was induced by intraluminal occlusion of the left middle cerebral artery (MCA) for 60 min, as described previously [57]. Sham-operated animals underwent the same anesthesia and exposure of arteries without MCAO. Briefly, mice were anesthetized with 3% isoflurane in a 30% O 2 /68.5% N 2 O mixture until they were unresponsive in the tail-pinch test. Mice were then fitted with a nose cone blowing 1.5% isoflurane in a 30% O 2 /68.5% N 2 O mixture under spontaneous breathing for anesthesia maintenance. An 8-0 monofilament with a silicon-coated tip was introduced into the common carotid artery, advanced to the origin of the MCA, and left in place to limit MCA blood flow for 60 min. Rectal temperature was controlled at 37.0 ± 0.5˚C using a temperature-regulated heating pad during surgery. Regional cerebral blood flow was measured in all stroke animals using laser Doppler flowmetry. Animals that did not display at least 70% reduction in regional cerebral blood flow during MCAO compared to pre-ischemia levels were excluded from further experimentation. Surgeries and quantification were performed by investigators blinded to animal genotypes and experimental grouping. Animals were randomly assigned to sham or cerebral ischemia groups and received randomized treatments using a lottery drawing box. A grand total of 287 mice (64 sham-operated and 223 ischemic mice) were used in this study, including 26 mice that were excluded from further assessments, either because of death after ischemia or failure of ischemia induction. The mortality during MCAO surgery was 7.6% in WT male mice (12/158), 14.3% in IL-4 KO male mice (1/7), 11.1% in microglia/macrophage-depleted mice (3/27), and 12.5% in PPARγ-OPC KO mice (3/24). There was no significant difference in mortality rate across groups.

Intranasal IL-4 nanoparticle preparation and administration
Lecithin (150 mg) and cholesterol (15 mg) were dissolved in 5 mL of a chloroform-methanol mixed solution (3:1, volume:volume). A lipid layer was formed after evaporating the mixed solvent under vacuum distillation. Recombinant IL-4 protein (1 mg, Peprotech, Rocky Hill, NJ, USA) was dissolved in 5 mL of water and then added to the lipid layer for hydration for about 30 min. The mixture was homogenized by ultrasonic probes for 15 min in an ice bath and then filtered through a 0.45-μm membrane. The free IL-4 protein was removed by ultrafiltration with centrifugal filter units (100-KDa cutoff). The particle size and zeta potential of IL-4 protein-loaded liposomes were determined by dynamic light scattering (DLS). The concentration of IL-4 protein was determined by the bicinchoninic protein determination kit (Thermo Fisher Scientific, Pittsburgh, PA, USA).
Animals were randomly assigned to receive vehicle or IL-4 (50 μg/kg body weight, intranasal) nanoparticle treatment starting 6 h after MCAO and repeated at 1-7 d, 14 d, 21 d, and 28 d postsurgery. Briefly, under isoflurane anesthesia, five 2-μL drops (total 10 μL) of IL-4 nanoparticles were applied alternately into each nostril with a 2-min interval between drops. Control animals received an equal volume of vehicle control using the same anesthesia and delivery regimen.

BrdU injections
In order to label proliferating cells, animals were intraperitoneally injected with the thymidine analogue BrdU (50 mg/kg, B9285, Sigma, St. Louis, MO, USA) twice a day (with an interval of at least 8 h) for 4 consecutive days, beginning at 3 d after MCAO.

Behavioral tests
Behavioral tests were performed by an individual blinded to experimental groups. Sensorimotor functions were measured 3-35 d after MCAO.
The rotarod test. Briefly, mice were forced to run on a rotating drum (IITC Life Science, Woodland Hills, CA, USA) with speeds starting at 4 rpm and accelerating to 40 rpm within 400 s. Three consecutive trials were conducted for each mouse with an interval of 15 min. The time at which a mouse fell off the drum was recorded as the latency to fall. Data were expressed as mean values from three trials.
The adhesive removal test. The adhesive removal test was performed to assess tactile responses and sensorimotor asymmetries. Two 2 × 3-mm adhesive tapes were applied to lesioned forepaws. Tactile responses were measured by recording the time to initially contact the ipsilateral paws, as well as the time to remove the adhesive tape, with a maximum observation period of 120 s.
Morris water maze test. Cognitive function was analyzed with the Morris water maze test 23-28 d after MCAO, as described previously [16]. In the learning test, a square platform (11 × 11 cm 2 ) was submerged 2 cm beneath the water surface in a circular pool (diameter = 109 cm) of opaque water. Mice were placed into the pool from one of the four locations and allowed to locate the hidden platform for 60 s. Each mouse was trained on 3 trials (with randomly assigned starting positions) per day to locate the platform for three consecutive days before MCAO. At the end of each trial, the mouse was placed on the platform or allowed to stay on the platform for 30 s with prominent spatial cues displayed around the room. The time spent to reach the platform (learning phase) was recorded. Three trials were performed on each day. The memory test was performed on day 28. The platform was removed, and a single 60-s probe trial was conducted. Time spent in the goal quadrant where the platform was previously located was recorded.

Measurement of tissue loss and MBP loss
Animals were euthanized and perfused with saline followed by 4% PFA (Sigma-Aldrich; St. Louis, MO, USA) in PBS. Brains were cryoprotected in 30% sucrose in PBS. Coronal brain sections (25 μm) were sliced on a freezing microtome, and six equally spaced coronal brain sections encompassing the MCA territory were stained with a rabbit anti-NeuN antibody (ABN78, 1:500; Millipore, Burlington, MA, USA) or rabbit anti-MBP antibody (MBP ab40390, 1:500, Abcam, Cambridge, MA, USA). Loss of NeuN + or MBP + tissue was analyzed by a blinded observer using NIH Image J software. The infarct volume with edema correction was calculated as the volume of the contralateral hemisphere minus the noninfarcted volume of the ipsilateral hemisphere.

Immunohistochemistry and image analysis
Coronal brain sections were subjected to immunofluorescence staining. Briefly, floating brain sections were blocked with 5% donkey serum in 0.3% Triton X-100 in PBS (PBST) for 1 h at room temperature, followed by overnight incubation with primary antibodies at 4˚C. After three washes in 0.3% PBST, sections were incubated with the appropriate secondary antibodies for 1 h at room temperature. The process was repeated once for double staining. Image analyses were performed on one or two randomly selected microscopic fields in the peri-infarct areas of the EC, two in the cortex, and two in the striatum of each section; two sections covering the infarct area were assessed for each mouse brain. The recorded images were loaded onto Image J (NIH) and were manually quantified by 2 observers blinded to grouping. Positively stained cells were electronically labeled with the software to avoid duplicated counting. The infarct area was identified as the region in which the majority of DAPI-stained nuclei were shrunken. The border of the infarct was determined by the loss of NeuN staining and the accumulation of Iba1 + microglia/macrophage or BrdU + APC + oligodendrocytes on adjacent sections. We defined the peri-infarct area as the tissue that covers a radial distance of 200-300 μm from the border of the infarct.

Flow cytometry and ImageStream
Animals were euthanized and perfused with cold saline. Brains were dissected, and the ipsilateral (left) and contralateral (right) hemispheres were collected. Brains were dissociated into a single-cell suspension using a gentleMACS Dissociator (Miltenyl Biotec) following the manufacturer's instructions. The suspension was passed through a 70-μm cell strainer (Fisher Scientific, Pittsburgh, PA, USA) and resuspended in 30% Percoll (GE Health). Single-cell suspensions were separated from myelin and debris by centrifugation (500g, 30 min, 18˚C) on a 30%-70% Percoll gradient. Cells at the interface were collected and washed with Hank's balanced salt solution (HBSS; Sigma-Aldrich, St. Louis, MO, USA) containing 1% fetal bovine serum (Sigma-Aldrich, St. Louis, MO, USA) and 2 mM EDTA (Sigma-Aldrich, St. Louis, MO, USA) before staining. Blood cells were prepared as described previously [57]. Single-cell samples were first incubated with antibodies to surface antigens for 30 min on ice at 4˚C in the dark. After two washes, cells were fixed and permeabilized with an intracellular staining kit (Thermo Fisher eBioscience, Pittsburgh, PA, USA) according to the manufacturer's protocol.

CAP measurements
CAPs were measured in the EC. Mice were decapitated after CO 2 euthanasia, and brains were removed. Coronal slices (350 μm thick) were prepared −1.06 mm from bregma and transferred to an incubation chamber containing pregassed (95% O 2 /5% CO 2 ) artificial cerebrospinal fluid (aCSF; 126 mmol/L NaCl, 2.5 mmol/L KCl, 1 mmol/L Na 2 H 2 PO 4 , 2.5 mmol/L CaCl 2 , 26 mmol/L NaHCO 3 , 1.3 mmol/L MgCl 2 , 10 mol/L glucose [pH 7.4]) for 30 min at 34˚C and then for 1 h at room temperature (25˚C). Brain slices were then perfused with aCSF at a constant rate (3-4 mL/min) at 25˚C. For recording, a bipolar stimulating electrode (intertip distance, 100 μm) was placed across the CC at about 0.9 mm lateral to the midline. A glass extracellular recording pipette (5-8 MO tip resistance when filled with aCSF) was placed in the EC, 0.25, 0.5, 0.75, and 1 mm lateral from the stimulating electrode. The recording at 0.75 mm from the stimulating electrode was reported. The CAP signal was digitized (Digidata 1500B plus humsilencer; Molecular Devices, San Jose, CA, USA), amplified (×1 K), and recorded by the Axoclamp 700B amplifier (Molecular Devices, San Jose, CA, USA). The signal was then analyzed by pClamp 10 software (Molecular Devices, San Jose, CA, USA). The input-output curves were generated by increasing the stimulating intensity by 0.25-mA intervals from baseline, up to 2 mA. The amplitude of the N1 component of the CAP (representing myelinated fibers) was calculated as the variance from the first peak to the first trough.

Transmission electron microscopy
TEM was used to measure myelin thickness in the CC/EC area. Mice were perfused with icecold saline, followed by 4% PFA and 2.5% glutaraldehyde in 0.1 mol/L PBS buffer. The CC/EC tissue near the site of ischemia was microdissected into 1-mm 3 blocks. These specimens were immersion-fixed for 24 h in 2% glutaraldehyde. Tissue was then rinsed in PBS and fixed in 1% osmium tetroxide in 0.1 M PBS for 45 min. Samples were dehydrated in increasing concentrations of acetone and embedded in Araldite resin. Ultrathin 60-nm sections were cut on a Leica UCT ultramicrotome with a diamond knife (Diatome, Wetzlar, Germany), stained with uranyl acetate and lead citrate, and viewed using a JEM1400 TEM (JEOL, Akishima, Japan). Two sections from each animal were analyzed. Three to 5 images (600 μm 2 each) were acquired in randomly selected areas within the CC/EC from each section at a magnification of 50,000× and analyzed with Image J by an investigator blinded to experimental groups. At least 50 random axons per animal were analyzed by tracing the axonal circumference and the whole fiber circumference in a blinded fashion. G-ratios were calculated as the ratio of the inner axonal diameter to the total outer diameter (axonal diameter + total myelin sheath thickness).

MRI
For in vivo MRI, mice were initially anesthetized with 3% isoflurane and then positioned on an animal cradle with a stereotaxic head holder. Anesthesia was maintained between 1% and 1.5% isoflurane throughout the experiment. Respiration and temperature were monitored continuously, and temperature was maintained by a warm animal heating system (SA Instruments, Stony Brook, NY, USA). MRI was performed on a 9.4 T/30-cm AVIII HD spectrometer (Bruker Biospin, Billerica, MA, USA) equipped with a 12-cm high-performance gradient set and using an 86-mm quadrature RF transmit volume coil and a 2-channel receive surface RF coil and Paravision 6.0.1. A T 2 -weighted RARE sequence was used to visualize the infarct, with the following setup: repetition time (TR)/echo time (TE) = 4,000/40 ms, field of view (FOV) = 20 × 20 mm, acquisition matrix = 256 × 256, 16 slices with a slice thickness of 0.5 mm, 4 averages, and a RARE factor = 8. The DTI data were collected using an echo planar imaging (EPI)-DTI imaging sequence with the same geometry as the T 2 -weighted scan, using the following setup: TR/TE = 2,300/22 ms, FOV = 20 × 20 mm, acquisition matrix = 128 × 128, 16 slices with a slice thickness of 0.5 mm, 8 averages, 2 segments, 5 A0 images and 30 noncolinear diffusion images, Δ/δ = 10/3 ms, and a b-value = 1,000 s/mm 2 .
DTI data were analyzed with DSI Studio software (http://dsi-studio.labsolver.org/). ROIs were drawn manually in a blinded manner encompassing the EC and IC in the ipsilesional and contralesional hemispheres to determine FA, AD, and RD. DEC, FA, and RD maps were generated by DSI Studio software.

PCR
In order to verify cre-induced deletion of flanked PPARγ exons, a pair of primers (forward: 5 0 -TGTAATGGAAGGGCAAAAGG; reverse: 5 0 -TGGCTTCCAGTGCATAAGTT) that flank the downstream loxp site were used to perform PCR with genomic DNA. Program for PCR was 98˚C for 1 min; (98˚C for 15 s; 60˚C for 15 s; 72˚C for 15 s) × 38, and then 72˚C for 1 min. A pair of primers that detect the Cre cDNA (forward: 5 0 -GCGGTCTGGCA GTAAAAACTATC; reverse: 5 0 -GTGAAACAGCATTGCTGTCACTT) were used as a positive control to confirm the absence of PCR band in OPC was not due to failed PCR protocol itself.

Microarray
Microarray experiment was performed in the Genomics Research Core in the University of Pittsburgh. Total RNA was prepared from IL-4-or PBS-treated OPCs using Trizol (Qiagen) according to manufacturer's instruction. Preparation of cDNA was performed using the WT Plus reagent kit (Thermo Fisher) according to the manufacturer's instructions. Reverse transcription was performed using 100 ng total RNA and dNTP-T7 random primers. Second-strand synthesis and in vitro transcription created amplified cRNA. cDNA from a second reverse transcription reaction was fragmented and end-labeled with biotin for hybridization to the Affymetrix Clariom S rat array. Following overnight (16 h) hybridization at 45˚C with rotational mixing at 60 rpm, arrays were processed on a GeneChip 450 Fluidics Station using manufacturer-specified protocols. To remove unbound sample, arrays were first washed with nonstringent wash buffer A. The GeneChips were then stained for 10 min in stain cocktail 1. Buffer A was again used to wash off excess stain. Signal amplification was achieved by 10-min incubation with stain cocktail 2 followed by a second 10-min incubation with stain cocktail 1. The chip was washed with high-stringency buffer B and filled with Array Holding Buffer before being removed from the fluidics station and scanned using the GeneArray 3000 scanner. First-level image analysis was performed using Affymetrix Expression Console. Four biologic replicates were performed for each experimental condition. Data resulted in 23,188 detected probes that were used for subsequent analysis using Transcriptome Analysis Console (Affymetrix) and Metascape analysis [32] (http://metascape.org). The GO network plot is visualized using Cytoscape, in which terms with a similarity > 0.3 are connected by edges. Each node represents an enriched term and is colored by its cluster ID.

Primary OPC and oligodendrocyte cultures
Primary OPC cultures were isolated with the differential attachment method. Briefly, brains of mixed-sex Sprague Dawley rat pups (postnatal day 1-2) were triturated after removing the meninges and blood vessels. The brain tissues were dissociated with Trypsin (0.01%) at 37 C for 15 min. After washing with ice-cold DMEM, the cells were passed through a 70-μm filter and plated onto poly-D-lysine-coated T-flasks filled with culture media (DMEM/F12 containing 10% heat-inactivated fetal bovine serum, 2 mM L-glutamine, 1 mM sodium pyruvate, 100 μM nonessential amino acids, 50 U/mL penicillin, and 50 μg/mL streptomycin) [45]. Cells were grown to confluence (12)(13)(14)) in a humidified incubator at 37˚C and with 5% CO 2 . Microglia were removed by shaking the mixed glia-containing flasks for 1 h at 180 rpm. After removing microglia, flasks were subjected to shaking at 200 rpm overnight to separate OPCs from the astrocyte layer. OPCs were maintained for 3-5 d in a serum-free basal defined medium (BDM: DMEM, 0.1% bovine serum albumin, 50 μg/ mL human apo-transferrin, 50 μg/mL insulin, 30 nM sodium selenite, 10 nM D-biotin, 10 nM hydrocortisone) containing 10 ng/mL PDGF and 10 ng/mL bFGF. For oligodendrocyte induction, OPCs were stimulated with T3 and CNTF (10 ng/mL). The medium was exchanged every 2 d.

Western blot
Protein was isolated from in vitro cultures. Western blots were performed using the standard SDS-polyacrylamide gel electrophoresis (PAGE) method. The primary antibodies used in this study include rabbit anti-MBP (Cell signaling Technology, 78896, 1:500, Danvers, MA, USA), mouse MOG (Santa Cruz, Sc-166172, 1:200, Santa Cruz, CA, USA), and mouse anti-β-actin (A2228, 1:20,000; Sigma-Aldrich, St. Louis, MO, USA). In brief, polyvinylidene difluoride (PVDF) membranes were blocked in 5% nonfat milk for 1 h at room temperature and then incubated at 4˚C overnight with primary antibodies. Next, the membrane was incubated with secondary antibodies for 1 h at room temperature (1:10,000, LI-COR Biosciences, Lincoln, NE, USA), and scanned with LI-COR Odyssey Infrared Imaging System 9201-550U (LI-COR Biotechnology, Lincoln, NE, USA). ImageJ software was used for gel analyses.

Myelination assessment and quantification on organotypic cultures
MBP staining intensity was analyzed with Imaris software (Bitplane AG). Morphology reconstructions were carried out with the Imaris FilamentTracer module. Quantitative assessment of myelination was performed in a blinded manner using the Imaris colocalization module. Thresholds for colocalization of red (NFH + axon) and green (MBP + myelin) channels were calculated automatically and maintained across different images from the same batch of staining. A colocalization channel was created and channel statistics were calculated automatically. The myelination index was calculated as the percent of colocalized areas in the total NFH + red pixels above threshold. For each condition, at least 6 slices collected from three independent experiments were analyzed. Two to three fields were imaged from each slice.

Statistical analysis
Sample sizes for animal studies were determined by power analyses based on pilot studies or the literature. Results are presented as mean ± standard error of the mean (SEM). GraphPad Prism software (version 8.1.0, La Jolla, CA, USA) was used for statistical analyses. The Student's t test was used for comparison of two groups for continuous variables with normal distributions. The Mann-Whitney U rank sum test was used for continuous variables with nonnormal distributions. The differences in means among multiple groups were analyzed using one-way ANOVA. Differences in means across groups with repeated measurements over time were analyzed using the repeated-measures ANOVA. When the ANOVA showed significant differences, pairwise comparisons between means were tested by post hoc Bonferroni tests. The correlation analyses between continuous data with normal distribution were performed using Pearson correlation analysis. Spearman rank correlation analyses were used to test correlations between data sets with nonnormal distributions. In all analyses, p < 0.05 was considered statistically significant.