13.3
Impact Factor

PDF

Theranostics 2025; 15(2):632-655. doi:10.7150/thno.103449 This issue Cite

Research Paper

Orchestrating the frontline: HDAC3-miKO recruits macrophage reinforcements for accelerated myelin debris clearance after stroke

Jiaying Li^1†, Chenran Wang^1†, Yue Zhang^1†, Yichen Huang¹, Ziyu Shi¹, Yuwen Zhang², Yana Wang¹, Shuning Chen¹, Yiwen Yuan¹, He Wang², Leilei Mao¹ , Yanqin Gao¹

1. State Key Laboratory of Medical Neurobiology, MOE Frontiers Center for Brain Science, and Institutes of Brain Science, Fudan University, Shanghai, China.
2. Institute of Science and Technology for Brain-Inspired Intelligence, Fudan University, Shanghai, China.
^†These authors contributed equally to this work.

✉ Corresponding authors: Email: yqgaoedu.cn (Y Gao) Tel: +8621 5423 7778, or llmaoedu.cn (L Mao).

Received 2024-7-21; Accepted 2024-11-1; Published 2025-1-1

Abstract

Rational: White matter has emerged as a key therapeutic target in ischemic stroke due to its role in sensorimotor and cognitive outcomes. Our recent findings have preliminarily revealed a potential link between microglial HDAC3 and white matter injury following stroke. However, the mechanisms by which microglial HDAC3 mediates these effects remain unclear.

Methods: We generated microglia-specific HDAC3 knockout mice (HDAC3-miKO). DTI, electrophysiological technique and transmission electron microscopy were used to assess HDAC3-miKO's effects on white matter. RNA sequencing, flow cytometry, immunofluorescence staining and ex vivo phagocytosis assay were conducted to investigate the mechanism by which HDAC3-miKO ameliorated white matter injury. Macrophage depletion and reconstitution experiments further confirmed the involvement of macrophage CCR2 in the enhanced white matter repair and sensorimotor function in HDAC3-miKO mice.

Results: HDAC3-miKO promoted post-stroke oligodendrogenesis and long-term histological and functional integrity of white matter without affecting early-stage white matter integrity. In the acute phase, HDAC3-deficient microglia showed enhanced chemotaxis, recruiting macrophages to the infarct core probably by CCL2/CCL7, where dMBP-labelled myelin debris surged and coincided with their infiltration. Infiltrated macrophages outperformed resident microglia in myelin phagocytosis, potentially serving as true pioneers in myelin debris clearance. Although macrophage phagocytosis potential was similar between HDAC3-miKO and WT mice, increased macrophage numbers in HDAC3-miKO accelerated myelin debris clearance. Reconstitution with CCR2-KO macrophages in HDAC3-miKO mice slowed this clearance, reversing HDAC3-miKO's beneficial effects.

Conclusions: Our study demonstrates that HDAC3-deficient microglia promote post-stroke remyelination by recruiting macrophages to accelerate myelin debris clearance, underscoring the essential role of infiltrated macrophages in HDAC3-miKO-induced beneficial outcomes. These findings advance our understanding of microglial HDAC3's role and suggest therapeutic potential for targeting microglial HDAC3 in ischemic stroke.

Keywords: macrophages, myelin debris, chemotaxis, HDAC3-miKO, white matter repair

1. Introduction

Stroke remains a leading cause of death and disability worldwide. Unfortunately, previous therapeutic strategies directly targeting neuronal survival ended in complete failure [1], while an increasing number of studies suggested white matter as a promising target for ischemic stroke [2-6] due to its close association with sensorimotor and cognitive impairment [7-11]. Indeed, while more vulnerable to ischemia compared to gray matter [12], the white matter injury process (i.e. demyelination and axonal breakup) remains reversible to some extents, unlike the irretrievable neuronal loss after ischemic stroke [13]. It is also clear that the demyelinated axons are unable to efficiently perform neuronal signal transduction and function after ischemic stroke even when the neuronal somas are protected [14]. Furthermore, disrupting the remyelination process after stroke also reduces neuronal survival and functional recovery after ischemic stroke [15]. Of note, the clearance of myelin debris from the site of demyelination is known to be one of the prerequisites for remyelination [16]. However, the pattern and destiny of myelin debris after ischemic stroke remain largely unknown. Of note, in the necrotic core of the subacute infarct, myelin debris was found to be present in CD68⁺ microglia/macrophages [17]. Although less frequently discussed in the field of ischemic stroke, the important role of microglia/macrophages in the removal of myelin debris has been well recognized in various demyelination diseases, including multiple sclerosis or spinal cord injury [18-21].

As a member of class I histone deacetylases (HDACs) family, HDAC3, which composes a co-regulator complex with NCoR1 (nuclear receptor corepressor 1) and SMRT (silencing mediator of retinoic acid and thyroid hormone receptor), plays an important role in chromatin modification and thereby the transcriptional regulation of genes [22, 23]. In the recent decade, HDAC3 has been reported to be implicated in neurodevelopment and various neurological diseases [24]. Specifically, in the developing brain, inhibiting HDAC3 promoted the myelination process of Schwann cells and therefore facilitated peripheral myelin growth [25-27]. In demyelination diseases, HDAC3 inhibitors were also found to promote the remyelination of oligodendrocytes [28, 29]. Notably, according to our previous study, in the context of ischemic stroke, HDAC3 was upregulated exclusively in microglia but not in other neural cell types, suggesting microglial HDAC3 as an indispensable molecule in stroke pathology[30]. Indeed, our further investigation revealed that deletion of microglial HDAC3 decreased the SMI32/MBP ratio 35 days after stroke, which was highly correlated with the ameliorated sensorimotor behavioral outcomes [30]. These results provided preliminary evidence demonstrating the association between microglial HDAC3 and de-/re-myelination process after ischemic stroke. However, the mechanism by which microglial HDAC3 regulated this essential process was still elusive.

In this study, we demonstrated that HDAC3-miKO played a reparative but not protective role in post-stroke white matter. Indeed, HDAC3-miKO promoted post-stroke oligodendrogenesis and long-term histological and functional integrity of white matter. By RNA-seq, we found that HDAC3-deficient microglia displayed enhanced chemotaxis as manifested in the increased expression of chemotaxis factors including monocyte chemoattractant CCL2 and CCL7 in the acute phase, recruiting macrophage reinforcements to the infarct core, which was coincided with the occurrence of myelin debris that was exclusively confined to the infarct core. While no difference was found in macrophages between HDAC3-miKO and WT mice in terms of their myelin phagocytosis potential, the clearance of myelin debris was accelerated in HDAC3-miKO due to the increased number of macrophages. Correspondingly, inhibiting CCR2, the receptor for CCL2/CCL7 hindered the accelerated clearance of myelin debris and reversed the beneficial effects of HDAC3-miKO. In conclusion, our study uncovered the entanglement between brain-resident microglia and the infiltrated macrophages in terms of the post-stroke white matter for the first time, advancing the understanding of the role of microglial HDAC3 and highlighting the therapeutic potential of targeting microglial HDAC3 in ischemic stroke.

2. Results

2.1. HDAC3-miKO occupies a reparative but not a protective role in post-stroke white matter

To define the impact of microglial HDAC3 on post-stroke white matter, we used our previously developed HDAC3 conditional knockout mouse models, which were bred from HDAC3^loxp/loxp and CX3CR1^CreER mice (Figure S1A, left panel) [30]. As our previous study demonstrated [30], these transgenic mice could achieve that microglia-specific knockout of HDAC3 (HDAC3-miKO) without deleting HDAC3 in CX3CR1-expressing macrophages at least one month after tamoxifen administration (Figure S1A, right panel). After one month of tamoxifen injection, we used transient focal cerebral ischemia (tFCI) to induce a comparable decrease of cerebral blood flow in HDAC3-miKO mice and WT mice during the ischemia stage, as detected by laser Doppler flowmetry (Figure S1B). Immunofluorescence staining revealed that HDAC3, which was broadly expressed in both sham and tFCI brain, existed in the nuclear of Iba1⁺ microglia/macrophages (Figure 1A). Three days after tFCI, the expression of HDAC3 in microglia/macrophage was elevated, while HDAC3-miKO showed significantly decreased HDAC3 expression in Iba1⁺ microglia/macrophage (Figure 1B&C). To confirm the specificity of HDAC3 knockout in microglia and exclude the effects of HDAC3-miKO on HDAC3 expression in infiltrated macrophages after tFCI, we further utilized FACS to isolate macrophages (CD11b⁺CD45^hiGr1^-CD11c^-) and microglia (CD11b⁺CD45^int) from the tFCI brains (Figure 1D) according to the previous study [31]. As expected in Figure S1A, our results revealed that microglia (Figure 1E) showed significantly decreased RNA level of Hdac3 in HDAC3-miKO while Hdac3 expression was unchanged in infiltrated macrophages (Figure 1F). Of note, HDAC3-miKO did not show any difference in NeuN⁺ number in the peri-infarct striatum, suggesting that HDAC3-miKO did not directly protect the brain tissue at the onset of ischemic stroke (Figure S1C&D).

While our previous study has preliminarily indicated the potential role of HDAC3-miKO in white matter injury by MBP/SMI32 immunostaining [30], further investigation is still needed to determine how exactly HDAC3-miKO exerts its role in post-stroke white matter, either protective or reparative. Since diffusion tensor imaging (DTI) is able to imprint white matter tract integrity in detail and therefore provide useful insight into the de-/re-myelination process after stroke [32], we thus first used DTI to detect the white matter structural integrity in vivo on days 3 and 14 and ex vivo on day 35 (Figure 1G). For DTI, the value of fractional anisotropy (FA) reflects the fine structure of white matter. Increase of FA is coincident with white matter reorganization in the recovery region of cerebral tissue [33]. The value of radial diffusivity (RD) indicates the extent of demyelination [34]. Of interest, while no difference was detected on both day 3 and day 14 in two tFCI groups, a significant increase in FA values and a significant decrease in RD values were observed in both external capsule (EC) and internal capsule (IC) in the miKO-tFCI group on day 35 (Figure 1H&I), especially in some specific planes (Figure S2). This time-dependent difference indicated that HDAC3-miKO exerted reparative but not protective effects on the tFCI-induced white matter injury.

Consistent with the DTI results, HDAC3-miKO showed long-term sensorimotor improvement as revealed by foot fault test (Figure 1J&K). Of note, consistent with previous studies highlighting the close association between white matter injury and stroke outcomes [32], our behavior results were also highly correlated with the FA and RD values in both EC and IC of DTI on day 35 (Figure 1L), further supporting the pro-repair effect of HDAC3-miKO on post-stroke white matter as our previous preliminary evidence has indicated [30]. We also observed the trend of improved behavioral performance as early as 1 day after tFCI. This may be due to factors beyond white matter injury, potentially involving early-stage changes in gray matter or microglial activation as reported in our previous study [30].

2.2. HDAC3-miKO promotes oligodendrogenesis and improves long-term histological and functional integrity of white matter after stroke

Since the repair of white matter injury is marked by successful regeneration of oligodendrocytes, we further used immunostaining of BrdU, a marker for newly generated cells, and APC, a marker for mature oligodendrocytes, to investigate whether HDAC3-miKO promotes oligodendrogenesis. We administered BrdU from 3 days to 6 days after tFCI at the peak of proliferation of oligodendrocyte precursor cells (OPCs) and conducted immunofluorescence 35 days after tFCI to label newly generated oligodendrocytes (Figure 2A) according to a previous study [35]. Apparently, stroke spontaneously induced robust generation of new cells in the penumbral EC and STR, with oligodendrocytes accounting for about a half (Figure 2B). Deficiency of microglial HDAC3 promoted the overall cell proliferation in the STR as indicated by the improved BrdU⁺ cells (Figure 2C). Specifically, in agreement with our findings by DTI, miKO-tFCI did show increased APC⁺BrdU⁺ cells in the STR (Figure 2D) and proportion of APC⁺BrdU⁺ cells to total APC⁺ cells in both the EC and the STR (Figure 2E), further suggesting improved remyelination in miKO-tFCI mice.

Indeed, as demonstrated by TEM, we detected the ultrastructure changes of myelinated axons 35 days after tFCI (Figure 2F) and revealed that miKO-tFCI significantly reversed the decreased number or proportion of myelinated axons (Figure 2G&H). G-ratio directly reflects the thickness of axon-enwrapping myelin sheath with lower g-ratio indicating thicker myelin sheath, serving as a highly reliable index for assessing axonal myelination [36]. Our results revealed that tFCI induced higher g-ratio compared to Sham, while HDAC3-miKO remarkably diminished g-ratio, suggesting the restoration of structural integrity of white matter (Figure 2I-K).

 Figure 1 

HDAC3-miKO occupies a reparative but not a protective role in post-stroke white matter. (A) Representative images of Iba1/HDAC3 immunostaining in the peri-infarct striatum (STR) 3 days after tFCI. White arrows indicate HDAC3+ microglia/macrophages. Yellow arrows indicate HDAC3- microglia/macrophages. 3D rendering was performed on the cell indicated by the white boxes to depict HDAC3 expression in Iba1+ cells. (B) Quantification of the mean fluorescence intensity (MFI) of HDAC3 in each Iba1+ cell. (C) Quantification of HDAC3 MFI in all Iba1+ cells in each animal. Each dot indicates an animal. n = 3-4 mice/group. (D) Gating strategy for flow-sorted microglia (CD11b⁺CD45^int) and infiltrated macrophages (CD11b⁺CD45⁺Gr1^-CD11c^-). (E&F) RNA expression level of Hdac3 in flow-sorted microglia (E, n = 6/group) or macrophages (F, n = 4 for WT-tFCI; n = 6 for miKO-tFCI). (G) DEC map presented from rostral planes to caudal planes was used to visualize ex vivo DTI on day 35 after tFCI. (H&I) FA or RD value on day 3 and day 14 in EC area (H) and IC area (I). n = 5/group. (J&K) The foot fault rate for fore paws (J) and hind paws (K). n = 8 for WT-Sham; n = 5 for miKO-Sham; n = 12 for WT-/miKO-tFCI. (L) Correlation matrix between fore/hind paw fault rates and FA/RD in EC/IC on day 3/14/35. The color and the area of circles indicate the (absolute) value of correlation coefficients. n = 5 for WT-tFCI; n = 6 for miKO-Sham. All data are presented as means±SEM. Data were analyzed using (B-C) one-way ANOVA followed by Bonferroni's post hoc, (E) unpaired two-tailed Student's t test, (F) Mann-Whitney test, and (H-K) two-way ANOVA followed by Bonferroni's post hoc, or (L) Pearson correlation. *p < 0.05, **p < 0.01, ***p < 0.001, ns: no significance, as indicated.

 Figure 2 

HDAC3-miKO promotes oligodendrogenesis and improves long-term histological and functional integrity of white matter after stroke. (A) Time line for BrdU injection and representative large images of BrdU/APC immunofluorescence. The yellow boxes indicated the locations where the BrdU/APC immunofluorescence images (B) were taken in the peri-infarct region of STR or EC on day 35 after tFCI. (B) Representative group-wise images of BrdU/APC immunofluorescence. White arrows indicated BrdU⁺APC⁺ newly-generated oligodendrocytes. (C-E) Quantification of the number of BrdU⁺ cells (C), APC⁺BrdU⁺ cells (D), and the percentage of APC⁺BrdU⁺ cells to the total APC⁺ cells (E). n = 5 for WT-tFCI; n = 6 for miKO-tFCI. (F) Representative electron micrographs. Red arrows indicated myelinated axons and blue axons indicated unmyelinated axons. (G-K) Quantification of the number of myelinated axons per 100 μm2 (G), the percentage of myelinated axons to the total axons (H), scatter plot of g-ratio (I), group-wise comparison of g-ratio in different scales (J), and frequency histogram of g-ratio (K). n = 5/group. (L) Representative images of Caspr/Nav1.6 immunofluorescence staining in the peri-infarct EC 35 days after tFCI. The right sub-image indicated an intact NOR, demonstrating the length of paranode gap and paranode length. (M-O) Quantification of the length of paranode gap (M), the number of NOR (N), and the paranode length (O). n = 6 for WT-Sham/-tFCI; n = 5 for miKO-Sham/-tFCI. (P) Schematic diagram indicating the stimulating site (Sti) and recording site (Rec) for CAPs recording in the EC at Bregma -1.59 mm and group-wise visualization of CAPs demonstrating N1 or N2 amplitudes. (Q&R) Group-wise comparison of N1 (Q) and N2 (R) amplitudes under different stimulus intensity. n = 5 for WT-Sham; n = 6 for miKO-Sham; n = 7 for WT-/miKO-tFCI. All data are presented as means±SEM. Data were analyzed using (C-E) unpaired two-tailed Student's t test, Mann-Whitney test, or (G&H, J, M-O) one-way ANOVA followed by Bonferroni's post hoc or (Q&R) two-way ANOVA followed by Bonferroni's post hoc. *p < 0.05, ***p < 0.001, WT -tFCI vs. WT-Sham or as indicated; ^#p < 0.05, miKO-tFCI vs. WT-tFCI, ns: no significance, as indicated.

Lengthening of the node of Ranvier (NOR) due to myelin retraction or even the loss of NOR are well known to contribute to altering the function of myelinated axons in stroke pathology [37]. Thus, we employed Caspr/Nav1.6 immunostaining to detect the change of NOR (Figure 2L). Consistent with previous studies [37], the pathological condition or HDAC3-miKO still kept the paranode structure intact, as indicated by the unchanged length of paranode gap (Figure 2M). However, tFCI reduced both number of NOR or the paranode length, while miKO-tFCI increased number of NOR and paranode length to some extents (Figure 2N&O), suggesting the improved function of myelinated axons in miKO-tFCI. Indeed, the performance in the foot fault test was correlated with both the NOR number (Figure S3A&B) and the paranode length (Figure S3C&D).

To verify whether the aforementioned structural change contributes to functional integrity of white matter, we performed electrophysiological recordings by detecting composite action potentials (CAPs) in the CC 35 days after ischemic stroke. CAPs recordings reflect the combined conduction of both unmyelinated and myelinated fibers in the CC. The amplitudes of the faster action potential N1 and slower action potential N2 represent the conduction potential of the myelinated and unmyelinated fibers, representatively (Figure 2P). We demonstrated that HDAC3-miKO prevented the stroke-induced decrease in N1 as well as N2 amplitude almost significantly (Figure 2Q&R). Correlation analysis revealed that N1 amplitude was associated with the forepaw fault rate on day 35 (Figure S3E), further suggesting that the axonal conductive speed was closely related to behavioral performance.

Collectively, HDAC3-miKO promoted oligodendrogenesis of the post-stroke brain and restored structural and functional integrity of white matter long-term after tFCI.

2.3. RNA-seq reveals increased chemotaxis in HDAC3-deficient microglia

Of note, due to their early-stage heterogeneous nature, microglia act as a double-edge sword for the later remyelination process after stroke. On one hand, microglia secrete inflammatory factors that hinder oligodendrocyte function [38]. On the other hand, pro-regenerative microglia could also promote remyelination by removing myelin debris and releasing regenerative factors [39]. Therefore, it is reasonable that the early-stage HDAC3-deficient microglia may have exerted determinant effects on the stroke-induced white matter injury. In this case, we isolated CD45^intCD11b⁺ microglia on day 3 after tFCI and performed bulk RNA-seq to investigate the mechanism by which HDAC3-miKO modulated post-stroke white matter.

In our previous study, we have identified massive DEGs between groups (4345 DEGs in WT-tFCI vs. WT-Sham, 1310 DEGs in miKO-tFCI vs. WT-tFCI, and 2288 DEGs in miKO-Sham vs. WT-Sham), suggesting a considerable difference induced by either tFCI or HDAC3-miKO [30]. Then we used the DEGs generated from miKO-tFCI vs. WT-tFCI for the downstream gene set enrichment analysis (GSEA) on the basis of GO database. Surprisingly, in addition to the downregulated proliferation-associated terms including “sister chromatid segregation”, “mitotic sister chromatid segregation” and “nuclear chromosome segregation” that have been carefully explored in our previous study [30], the GSEA results also uncovered the simultaneously upregulated chemotaxis-associated terms, including “chemotaxis”, “myeloid leukocyte migration”, “granulocyte migration”, “leukocyte chemotaxis”, and “leukocyte migration” (Figure 3A&B). Correspondingly, depleting microglial HDAC3 further elevated the expression of numerous stroke-induced upregulated genes associated with leukocyte chemotaxis (Figure 3C). Among these genes, we noticed that C-C chemokines including Ccl2, Ccl3, Ccl5, Ccl7 and Ccl22 showed remarkable upregulation in HDAC3-deficient microglia, as validated by qPCR (Figure 3D). These chemokines are engaged in recruitment of multiple types of immune cells, suggesting the possibility that HDAC3-deficient microglia directly called for reinforcements of peripheral immune cells while their own expansion was disrupted.

 Figure 3 

RNA-seq reveals increased chemotaxis in HDAC3-deficient microglia. (A) Ridge plot visualizing GSEA results (miKO-tFCI vs. WT-tFCI). (B) GSEA results showed significantly upregulated GO pathways of “chemotaxis”, “granulocyte migration”, “leukocyte migration” and “myeloid leukocyte migration” (miKO-tFCI vs. WT-tFCI). (C) Heatmap showed expression level of genes associated with the term “Leukocyte Chemotaxis” in microglia. Each column represents a biological replicate. (D) qPCR validation of representative C-C chemokine genes whose expression significantly changed in miKO-tFCI vs. WT-tFCI. n = 6 for WT-tFCI; n = 5 for miKO-tFCI. All data are presented as means±SEM. Data were analyzed using unpaired two-tailed Student's t test. *p < 0.05, ***p < 0.001, as indicated.

2.4. HDAC3-deficient microglia recruit macrophage reinforcements to the infarct core

Therefore, to determine the certain type of infiltrated immune cells that were augmented in HDAC3-miKO mice, we proceed to perform flow cytometry to analyze the infiltrated cells 3 days after tFCI using the gating strategy of T cells (CD11b^-CD45^hiCD3⁺), B cells (CD11b^-CD45^hiCD19⁺), dendritic cells (CD11b⁺CD45^hiCD11c⁺Gr1^-), neutrophils (CD11b⁺CD45^hiCD11c^-Gr1⁺), microglia (CD11b⁺CD45^int), macrophages (CD11b⁺CD45^hi) and inflammatory macrophages (CD11b⁺ CD45^hi Ly6C⁺) (Figure 4A). Consistent with our previous study[30], HDAC3 deficiency further decreased microglial number after tFCI by over a half (Figure 4B). In addition, tFCI remarkably induced the recruitment of macrophages (including inflammatory macrophages) and DCs but not neutrophils, T cells or B cells, in agreement with existing reports (Figure 4C-H) [40]. In our study, neither tFCI nor HDAC3-miKO induced remarkable immune response in the peripheral blood or spleen (Figure S4). Notably, among all these immune cells, HDAC3-miKO recruited significantly more macrophages compared to WT, while the numbers of other infiltrated cells remained unchanged (Figure 4C). However, the number of inflammatory macrophages did not show significant change as the overall macrophages did (Figure 4D). Another interesting fact showed that HDAC3-miKO also facilitated recruitment of B cells even into the Sham-induced brains and also into the tFCI-induced brain to a greater extent 3 days after tFCI (Figure 4H) when infiltration of B cells was not observed. More investigations involving more time points regarding this phenomenon were needed in the future study. Considering that B cells did not infiltrate into the brain and remained a small population even in HDAC3-miKO group at this early stage, in this study, we still aimed at macrophages, which have been widely reported to exert an important but controversial role in stroke pathology [41].

It is well known that infiltrated macrophages share very similar signatures and functions with brain-resident microglia under pathological conditions [42]. However, in recent years, owing to the emerging techniques that can well distinguish macrophages from microglia, growing studies have highlighted the different roles between macrophages and microglia. Given that the pathological environment differed according to the distance to the infarct core, understanding the distribution of macrophage reinforcements may help us further investigate their role. In this case, we employed immunofluorescence staining of P2RY12, a marker for bona fide microglia, and F4/80, a marker for activated microglia and infiltrated macrophages, to determine the macrophages distribution in the contralateral hemisphere, peri-infarct region and infarct core. As expected, Iba1⁺P2RY12^- infiltrated macrophages were not observed in the contralateral hemisphere (Figure S5). In the peri-infarct STR, Iba1⁺P2RY12⁺ brain-resident microglia predominated the region in terms of the overall Iba1⁺ cells in both two groups, and miKO-tFCI mice had significantly fewer Iba1⁺ microglia/macrophages, mainly attributed to a significant reduction in the number of microglia (Figure 4I-K), supporting our previous study that HDAC3-miKO inhibited microglial proliferation [30]. Of note, the absolute number of infiltrated macrophages in the peri-infarct region did not show any significant change (Figure 4K). Unlike the peri-infarct region, in the infarct core, macrophages far outnumbered microglia, in agreement with previous studies indicating robust loss of microglia and the repopulation by infiltrated macrophages in the infarct core (Figure 4L-M) [43]. Interestingly, in this region, HDAC3-miKO induced significant increase in the number of infiltrated macrophages while the number of microglia remained unchanged (Figure 4M). Taken together, these findings implied that HDAC3-miKO resulted in more macrophages infiltration into the infarct core but not into the penumbra area.

2.5. Spatiotemporal pattern of myelin debris distribution after stroke

The removal of myelin debris is an essential step in the remyelination process. Mononuclear phagocytic cells, including monocyte-derived macrophages and microglia, are actively implicated in the myelin debris clearance [21, 44-46]. While the clearance of myelin debris has been well discussed in demyelinating diseases [21, 46], this process and the associated mechanism specific to stroke pathology are poorly understood. Recognizing the spatiotemporal pattern of myelin debris distribution after stroke is a necessary prerequisite to understanding of myelin debris clearance. Therefore, we performed immunofluorescence staining of NeuN and dMBP (a marker for degenerated myelin) on Sham and on day 1, 3, 7, 14 and 35 after tFCI (Figure 5A). We demonstrated that in the context of stroke, myelin debris was only present in the infarct core where NeuN⁺ neurons vanished on day 1, 3 and 7 (Figure 5B). Quantification of the volume or the area fraction of dMBP⁺ myelin debris exhibited that as the stroke pathology proceeded, dMBP signal augmented and remained high on both day 3 and day 7 (Figure 5C&D).

 Figure 4 

HDAC3-deficient microglia recruit macrophage reinforcements to the infarct core. (A) Macrophages (CD11b⁺CD45^hiLy6C⁺), neutrophils (CD11b⁺CD45^hiCD11c^-Gr1⁺), dendritic cells (DCs) (CD11b⁺CD45hiCD11c⁺Gr1^-), T cells (CD11b^-CD45^hiCD3⁺), B cells (CD11b^-CD45^hiCD19⁺) and microglia (CD11b⁺CD45^int). (B-H) The percentage of these cells to single cells, respectively. n = 8/group. (I) Representative large image of Iba1/NeuN immunostaining, which indicated the infarct core and penumbra region photographed for Iba1/F4/80/P2RY12 staining (J-M). (J&K) Representative images of Iba1/P2RY12 immunostaining (J) and quantification of the number of Iba1⁺P2RY12^- cells and Iba1⁺P2RY12⁺ cells (K) in the ischemic penumbra 3 d after tFCI. The right panel showed the proportion of Iba1⁺P2RY12^- cells or Iba1⁺P2RY12⁺ cells to the overall Iba1⁺ cells. (L&M) Representative images of F4/80/P2RY12 immunostaining (L) and quantification of the number of F4/80⁺P2RY12^- cells and P2RY12⁺ cells (M) in the ischemic core 3 d after tFCI. The right panel showed the proportion of F4/80⁺P2RY12^- cells or P2RY12⁺ cells to the overall F480⁺ or P2RY12⁺ cells. n = 4 for WT-tFCI; n = 5 for miKO-tFCI. All data are presented as means±SEM. Data were analyzed using (B-E) one-way ANOVA followed by Bonferroni's post hoc, (F-H) Kruskal-Wallis test followed by Dunn's post hoc test or (K&M) unpaired two-tailed Student's t test. *p < 0.05, **p < 0.01, ***p < 0.001, ns: no significance, as indicated (B-H). ***p < 0.001 for comparison of Iba1⁺P2RY12⁺ cell number between miKO-tFCI and WT-tFCI. ^##p < 0.01 for comparison of the overall Iba1+ cells (K). ***p < 0.001 for comparison of F4/80⁺P2RY12^- cell number between miKO-tFCI and WT-tFCI. ^##p < 0.01 for comparison of the overall F480⁺ or P2RY12⁺ cells (M).

 Figure 5 

Spatiotemporal pattern of myelin debris distribution after stroke. (A) Representative large images of dMBP/NeuN immunofluorescence staining on Sham group or on day 1,3,7,14,35 after tFCI. The white dashed lines outlined the infarct core with neuronal loss. The following images (B) were taken from where the yellow boxes indicated. (B) Representative images demonstrating dMBP signals and the 3D rendering of these signals in the infarct core. (C) Quantification of dMBP volume (μm3) per mm3. (D) Quantification of dMBP area fraction (%). n = 4 for WT-/miKO-Sham; n = 5 for WT-/miKO-tFCI. All data are presented as means±SEM. Data were analyzed using (C) one-way ANOVA followed by Bonferroni's post hoc and (D) Kruskal-Wallis test followed by Dunn's post hoc test *p < 0.05, **p < 0.01, ***p < 0.001, ns: no significance, as indicated.

2.6. HDAC3-miKO accelerates myelin debris clearance by boosting macrophage recruitment without altering phagocytosis capability

Based on the above understanding that infiltrated macrophages predominantly occurred in the infarct core where myelin debris surged and that HDAC3-deficient microglia have summoned additional macrophage reinforcements, we hypothesized that HDAC3-miKO could accelerate myelin debris clearance by boosting macrophage recruitment, in this way promoting oligodendrogenesis and improving long-term histological and functional integrity of white matter after stroke.

Therefore, we employed 3D reconstruction of F4/80/dMBP immunostaining to quantify the phagocytosis capacity of myelin debris by F4/80⁺ cells (mostly infiltrated macrophages), as demonstrated as overlap volume ratio. We defined cells with overlap volume ratio > 0 as “Engulfed” macrophages, or otherwise “No contact” macrophages (Figure 6A). Consistent with our previous results (Figure 4M), HDAC3-miKO experienced an increase in the number of macrophages in the infarct core (Figure 6B&C). Among these cells, “Engulfed” cells accounted for the majority and increased significantly in HDAC3-miKO while “No contact” cells were comparable in two tFCI groups (Figure 6D). However, the percentage of both cell types to the total F4/80⁺ cells did not significantly change (Figure 6E), suggesting that HDAC3-miKO did not alter the phagocytosis capacity of macrophages actually. Indeed, no significant difference was observed in the average overlap volume ratio for each cell in two groups (Figure 6F). Moreover, the average cell volume, which has been reported to imply the cellular phagocytosis state, did not significantly change as well in two groups (Figure 6G).

To precisely determine the phagocytosis potential of infiltrated macrophages in two groups, we conducted an ex vivo phagocytosis assay as previous studies described using PKH26 or pHrodo-Red-labelled myelin debris and leukocytes collected from brains 3 days after tFCI (Figure 6H). In details, PKH26 is a lipophilic dye that was able to simply detect the uptake of myelin debris, and pHrodo-Red is a pH-sensing dye that was able to further detect the degradation of myelin debris owing to its dramatic increase in fluorescence in acidic lysosomes. After 4 h ex vivo incubation with myelin debris, CD11b⁺CD45^hi cells (mostly macrophages) in two tFCI groups displayed comparable capacity for both uptake (Figure 6I, J right panels) and degradation (Figure 6K, L right panels), in agreement with the fact that macrophages in HDAC3-miKO remained HDAC3-sufficient (Figure 1F). However, despite the unchanged phagocytosis capacity, macrophages in HDAC3-miKO still internalized more dMBP⁺ myelin debris as a whole 3 days after tFCI (Figure 6M). Given that HDAC3-miKO promoted macrophages infiltration into the infarct core, we inferred that the only reason why myelin debris was more engulfed in macrophage population in HDAC3-miKO was because of the increased number of but not the increased phagocytosis capacity of macrophages. Surprisingly, unlike macrophages, HDAC3-deficient microglia (CD11b⁺CD45^int) showed significantly reduced capacity for both uptake and degradation of myelin debris (Figure 6J&L, left panels). Corresponding to this result, in the penumbra area, HDAC3-deficient microglia showed remarkably smaller proportion contacting with non-degraded intact myelin bundles (Figure S6). A previous study has suggested that contact with intact myelin sheath exacerbated the ischemia outcomes [47]. Therefore, our result indicated that HDAC3-deficient microglia may also directly contribute to the reduced myelin damage after stroke.

To verify whether HDAC3-miKO subsequently affected the clearance of myelin debris, we further evaluated dMBP signal on day 3 and 7 after tFCI (Figure 6N), when dMBP signal remained apparent (Figure 5). Our results demonstrated that although dMBP signal did not significantly change on day 3 in two groups, HDAC3-miKO did induce more efficient clearance of myelin debris later on day 7 after tFCI (Figure 6O&P).

Taken together, our results indicated that HDAC3-miKO promoted myelin debris clearance exclusively by recruiting more macrophages without affecting their phagocytosis capacity.

2.7. CCR2-KO BMDMs reconstitution reverses the effects of HDAC3-miKO

To investigate the role of macrophage CCR2 in mediating HDAC3-miKO effects after tFCI, we employed a macrophage reconstitution approach (Figure 7A) based on a previous study [48]. In brief, after isolation from bone marrow of WT or CCR2-KO mice (Figure S7A), bone marrow-derived macrophages (BMDMs) were cultured for 5 days for differentiation. Resident macrophages in WT and HDAC3-miKO mice were depleted by intravenous injection of clodronate (CLO) administered for two consecutive days starting 2 days prior to tFCI. Flow cytometry showed reduction of macrophages in blood after CLO injection (Figure 7B). Subsequently, BMDMs were intravenously injected 2 hours after tFCI.

Consistent with our previous reasoning, reconstitution with CCR2-KO BMDMs (miKO-tFCI + CCR2-KO BMDMs) significantly decreased macrophage recruitment to the infarct core in HDAC3-miKO mice compared to those reconstituted with WT BMDMs (miKO-tFCI + WT BMDMs) (Figure 7C&D). Despite similar weight and survival rates across the three tFCI groups (Figure S7B&C), CCR2-KO BMDMs reconstitution reversed the improved outcomes in HDAC3-miKO mice, as revealed by lower Garcia scores and poorer foot-fault test performance (Figure 7E-G).

Given that our prior results implicated recruited macrophages in facilitating myelin debris clearance within the infarct core, we evaluated the dMBP signal on day 7 post-tFCI across the three groups (Figure 7H). As expected, miKO-tFCI + WT BMDMs exhibited significantly reduced dMBP volume and area fraction compared to WT-tFCI + WT BMDMs (Figure 7I&J, blue bars vs. red bars). Notably, this reduction was reversed in miKO-tFCI mice reconstituted with CCR2-KO BMDMs (Figure 7I&J, red bars vs. pink bars).

 Figure 6 

HDAC3-miKO accelerates myelin debris clearance by boosting macrophage recruitment without altering phagocytosis capability. (A) Representative image of F4/80/dMBP immunofluorescence staining, indicating the (a) “Engulfed” state (overlap volume ratio > 0) or (b) “No contact” state (overlap volume ratio = 0) of F4/80⁺ cells. (B) Group-wise representative images of F4/80/dMBP immunofluorescence staining and the associated 3D rendering illustrating the internalization of dMBP⁺ myelin debris by an F4/80⁺ cell. (C) Quantification of F4/80⁺ cell number per mm³. (D&E) Quantification of the number of F4/80⁺ “Engulfed” or “No contact” cells (D), their respective percentage to the total F4/80⁺ cells (E). (F&G) The average overlap volume ratio of all F4/80⁺ cells (F) and the average cell volume of all F4/80⁺ cells (G) in all the FOVs photographed. Each dot represented a mouse, 10-15 F4/80⁺ cells were quantified for each mouse. n = 4 for WT-tFCI; n = 6 for miKO-tFCI (C-G). (H) Schematic diagram showing the method of ex vivo phagocytosis assay. (I) Representative histograms showing PKH26-red fluorescence intensity in CD11b⁺CD45^int microglia or CD11b⁺CD45^hi macrophages, assessed by flow cytometry. (J) Percentage of microglia/macrophages containing PKH26 signal. n = 6 for WT-tFCI; n = 3 for miKO-tFCI. (K) Representative histograms showing pHrodo-red fluorescence intensity in CD11b⁺CD45^int microglia or CD11b⁺CD45^hi macrophages, assessed by flow cytometry. (L) Percentage of microglia/macrophages containing pHrodo-red signal. n = 6/group. (M) Quantification of total engulfed dMBP volume by F4/80⁺ cells in all photographed FOVs. Each dot represented a mouse. n = 4 for WT-tFCI; n = 6 for miKO-tFCI. (N) Representative images of dMBP immunofluorescence staining on day 3/7 after tFCI. (O) Quantification of dMBP volume (μm³) per mm³. (P) Quantification of dMBP area fraction (%).3 d: n = 6 for WT-tFCI; n = 7 for miKO-tFCI. 7 d: n = 5 for WT-tFCI; n = 4 for miKO-tFCI (O&P). All data are presented as means±SEM. Data were analyzed using (C-G, J&L and N-P) unpaired two-tailed Student's t test, Mann-Whitney test or (J&L) one-way ANOVA followed by Bonferroni's post hoc. *p < 0.05, **p < 0.01, ***p < 0.001, ns: no significance, as indicated.