Detrimental influence of Arginase-1 in infiltrating macrophages on poststroke functional recovery and inflammatory milieu

Significance

Ischemic stroke remains a leading cause of long-term disability worldwide. While Arginase-1 (Arg1) expressing macrophages are typically associated with anti-inflammatory responses and tissue repair, we reveal an unexpected detrimental influence of Arg1 on poststroke recovery. We demonstrate that Arg1 in infiltrating macrophages alters the inflammatory milieu and negatively impacts functional recovery following stroke. Notably, our research highlights a unique interaction between infiltrating macrophages and resident microglia, where Arg1-expressing macrophages modulate microglial function, affecting synaptic pruning and inflammatory responses in the peri-infarct region. These findings provide crucial insights into the complex immune mechanisms governing stroke recovery and suggest advanced therapeutic strategies. Targeting Arg1 in infiltrating macrophages may modulate the poststroke inflammatory environment, improving long-term outcomes for stroke patients.

Abstract

Poststroke inflammation critically influences functional outcomes following ischemic stroke. Arginase-1 (Arg1) is considered a marker for anti-inflammatory macrophages, associated with the resolution of inflammation and promotion of tissue repair in various pathological conditions. However, its specific role in poststroke recovery remains to be elucidated. This study investigates the functional impact of Arg1 expressed in macrophages on poststroke recovery and inflammatory milieu. We observed a time-dependent increase in Arg1 expression, peaking at 7 d after photothrombotic stroke in mice. Cellular mapping analysis revealed that Arg1 was predominantly expressed in LysM-positive infiltrating macrophages. Using a conditional knockout (cKO) mouse model, we examined the role of Arg1 expressed in infiltrating macrophages. Contrary to its presumed beneficial effects, Arg1 cKO in LysM-positive macrophages significantly improved skilled forelimb motor function recovery after stroke. Mechanistically, Arg1 cKO attenuated fibrotic scar formation, enhanced peri-infarct remyelination, and increased synaptic density while reducing microglial synaptic elimination in the peri-infarct cortex. Gene expression analysis of fluorescence-activated single cell sorting (FACS)-sorted CD45^low microglia revealed decreased transforming growth factor-β (TGF-β) signaling and proinflammatory cytokine activity in peri-infarct microglia from Arg1 cKO animals. In vitro coculture experiments demonstrated that Arg1 activity in macrophages modulates microglial synaptic phagocytosis, providing evidence for macrophage–microglia interaction. These findings present unique insights into the function of Arg1 in central nervous system injury and highlight an interaction between infiltrating macrophages and resident microglia in shaping the poststroke inflammatory milieu. Our study identifies Arg1 in macrophages as a potential therapeutic target for modulating poststroke inflammation and improving functional recovery.

Central nervous system (CNS) injuries following ischemic stroke often result in lifelong functional disabilities (1). While endovascular thrombus removal during the hyperacute stage has significantly improved functional outcomes (2–4), only a small proportion of acute ischemic stroke patients are eligible for mechanical thrombectomy (5, 6). Moreover, a substantial number of treated patients still experience moderate to severe functional impairments (7, 8). Consequently, there is a critical need to develop therapeutic strategies, either as standalone treatments or in combination with endovascular procedures, to mitigate the cellular and molecular effects of ischemic injury during the subacute to chronic stages.

Poststroke inflammation has emerged as a promising target for modulating functional recovery (9–12). In the acute phase of ischemic stroke, damaged tissues release damage-associated molecular patterns and inflammatory cytokines, activating glial cells such as astrocytes and microglia. This process is accompanied by the infiltration of various immune cells from peripheral blood, including neutrophils, lymphocytes, monocytes, and macrophages, all contributing to the poststroke inflammatory milieu (13–15). Among these, macrophages and microglial cells, both of myeloid lineage, undergo dynamic phenotypic changes during poststroke pathological processes. The phenotypic variability of these cells, traditionally categorized as either M1 or M2 at their polarized extremes, plays a crucial role in shaping the inflammatory landscape following stroke (16–18). Recent studies have revealed complex interactions between microglial cells and infiltrating monocytes/macrophages in various CNS injuries (13, 19–21), further complicating our understanding of the inflammatory environment and its regulation.

Arginase-1 (Arg1), a canonical marker for M2 macrophages, is known for its role in resolving proinflammatory microenvironment (22). As an enzyme that converts L-arginine to L-ornithine in the urea cycle (23), Arg1 plays a pivotal role in determining macrophage phenotypic characteristics, particularly given that arginine is also a substrate for inducible nitric oxide synthase, which generates oxidative stress (24, 25). The functional contributions of Arg1 in macrophages have been studied in various inflammatory conditions in peripheral organs outside of the brain. For instance, Arg1 activity is required for suppressing chronic inflammation and promoting wound healing in cutaneous injuries (26). In the kidney, Arg1 is required for macrophage-mediated renal tubule regeneration (27). However, few studies have investigated the functional contribution of Arg1 in macrophages to poststroke inflammation and functional recovery. The present study revealed that Arg1 in macrophages plays a detrimental role in motor functional recovery following a stroke, contradicting the prevailing notion of Arg1 as a marker for macrophages promoting tissue repair (28). Our results indicate that Arg1 may induce excessive fibrosis at the border between the infarction core and peri-infarct spared cortical tissue and increase pathological synaptic elimination by altering macrophage–microglial interaction. These findings suggest that Arg1 may be a promising therapeutic target for enhancing functional recovery through modification of the poststroke inflammatory milieu.

Results

Arg1 Expression Following Photothrombotic Stroke in Mice.

Arg1 protein expression was assessed by immunohistochemistry at different time points following the induction of a photothrombotic infarction in mice. The photothrombosis protocol used in this study generated a localized infarction, affecting the entire thickness of the cerebral cortex while preserving the corpus callosum (Fig. 1A). Arg1-positive cells were observed within the infarcted tissue in a time-dependent manner. Arg1 expression was first noticeable 3 days poststroke (dps) and peaked at 7 dps (Fig. 1 B and C). By 14 dps, the number of Arg1-positive cells decreased almost to the baseline level (Fig. 1 B and C). The Arg1-expressing cells were mainly distributed at the periphery of the infarction core at 7 dps when the expression level was at its highest (Fig. 1D). Arg1 expression was not observed in peri-infarct neurons visualized by microtubule-associated protein 2 (MAP2) (Fig. 1E). The expression of Arg1 was restricted within the glial fibrillary acidic protein (GFAP)-positive astrocytic scar boundary (Fig. 1E). The majority of Arg1-expressing cells were colocalized with Ionized calcium-binding adaptor molecule-1 (Iba-1)-positive macrophages or microglia (Fig. 1E), indicating that Iba-1-positive myeloid cells were the main source of Arg1 expression after an ischemic stroke. When a photothrombotic infarction was induced in the CX3CR1-GFP mice where GFP is expressed in CX3CR1-positive resident microglial cells (29), only 20% of Arg1-positive cells were colocalized with GFP (Fig. 1 F and H). Lysozyme M is expressed exclusively in hematogenous macrophages and neutrophils but not in microglia (30). To determine whether macrophages infiltrating from blood express Arg1 (30), we employed an R26-stop-enhanced yellow fluorescent protein (EYFP) reporter line crossed with LysM-cre mice. We observed that approximately 90% of Arg1-positive cells colocalized with YFP-positive cells (Fig. 1 G and H). When the cellular mapping analysis was done at 3 dps, the percentage of Arg1-positive cells colocalized with either CX3CR1 or LysM was similar to that at 7 dps, although there was a trend toward an increase in CX3CR1 and a decrease in LysM expression between the two time points (SI Appendix, Fig. S1 A and B). Since LsyM is also expressed in almost half of the neutrophils (30, 31), we also investigated whether Ly6G-positive neutrophils express Arg1. Ly6G-positive neutrophils were observed within the infarcted tissue at 1 and 3 dps when the neutrophils actively infiltrate the brain parenchyma (32). However, Arg1 was not expressed in the Ly6G-positive neutrophils at either time point (SI Appendix, Fig. S2 A–C). Collectively, immunolabeling studies suggest that LysM+ macrophages are a principal source of Arg1 after ischemic stroke.

Fig. 1.

Characterization of Arg1 expression after photothrombotic stroke in mice (A) Eriochrome cyanine R staining of brain tissue at 7 dps, showing cortical ischemic region. (Scale bar, 1.0 mm.) (B) Immunohistochemistry of Arg1 at the lesion core at 1, 3, 7, and 14 dps. (Scale bar, 50 μm.) (C) Quantification of the number of Arg1 positive cells. *, **, and *** indicate P < 0.05, P < 0.01, and P < 0.001, respectively, by one-way ANOVA followed by post hoc Bonferroni’s multiple comparison test. N = 3 animals per time point. (D) A representative image shows localization of Arg1 positive signals in the periphery of lesion core at 7 dps. Lesion border is indicated in the yellow dotted line. The corpus callosum and cortex are indicated in white dotted lines. (Scale bar, 500 μm.) cc: corpus callosum. (E) Representative immunohistochemistry images labeled with Arg1 and cell-type-specific markers including MAP2, GFAP, and Iba-1 at 7 dps, respectively. The yellow dotted line indicates lesion border. The right-most magnified images were obtained from the white solid squares in Iba-1 and Arg1 colabeled image. (Scale bar, 50 μm.) (F and G) Representative immunohistochemistry images of Arg1 in Cx3cr1-GFP (F) or LysM-cre::R26-stop-YFP (G) animals at 7 dps. Scale bar, 500 μm for low magnification images and 50 μm for high magnification images. White solid squares indicate regions that are magnified on the right side. (H) Quantification of the percent Arg1 expressing Cx3cr1+ or LysM+ cells out of the total Arg1+ cells at 7 dps. *, **, and *** indicate P < 0.05, P < 0.01, and P < 0.001, respectively, by an unpaired t test. N = 3 animals per line. Each data point in all the graphs represents a single animal.

Deletion of Arg1 in Macrophages Promotes Recovery of Motor Function Following Photothrombotic Stroke.

To investigate the functional role of Arg1, we generated Arg1 conditional knockout (cKO) mice by crossing Arg1^flox/^flox and LysM-cre mice lines, aiming to selectively delete Arg1 in LysM-expressing macrophages. In this model, the Cre recombinase, expressed under the control of the LysM promoter, targets the two loxP sequences inserted upstream of exon 7 and downstream of exon 8, respectively, of the Arg1 genomic sequence. It then removes the genomic sequence between the two loxP sites, resulting in the deletion of exons 7 and 8 (SI Appendix, Fig. S3A), which encompass the substrate binding region of Arg1 (amino acids 126 to 277). Induction of Arg1 messenger RNA (mRNA) in the brain tissue after stroke was almost completely abolished at 7 dps in Arg1 cKO mice (SI Appendix, Fig. S3 B and C). We confirmed that Arg1 protein levels were substantially reduced following stroke in Arg1 cKO mice (Fig. 2A). The number of Arg1-positive cells was markedly decreased in Arg1 cKO mice, although a small proportion of Arg1-expressing cells were still detected in immunohistochemistry within the infarcted region (Fig. 2 B and C).

Fig. 2.

Motor behavior recovery in Arg1 cKO animals after photothrombotic stroke (A) Western blot of Arg1 protein in the brain tissue at 7 dps. (B) Immunohistochemistry of Arg1 in Arg1 cKO animals or flox (Arg1 flox/flox) control animals at 7 dps. (Scale bar, 200 μm.) Infarction: yellow dotted line, cortex: white dotted line. (C) Quantitative analysis of the number of Arg1-expressing cells per mm². *** indicates P < 0.001 followed by an unpaired t test. (N = 3 animals per group). Each data point in the graph represents a single animal. (D–G) Quantitative analysis of the pellet test, cylinder test, ladder walk test, and modified neurological severity scoring (mNSS), respectively. Repeat measures two-way ANOVA repeated measure was performed to match the time difference. Bonferroni’s multiple comparison was conducted to assess group differences at each time point. (N = 14 animals per group). # and ## indicate P < 0.05 and P < 0.01, respectively, by two-way ANOVA column factor (overall group difference). * indicates P < 0.05 by post hoc Bonferroni’s multiple comparisons.

The animals were subjected to a battery of behavior tests to assess the recovery of motor functions. In the pellet retrieval test, Arg1 cKO animals tended to achieve a higher rate of successful pellet retrieval than control Arg1 floxed animals, although the genotype effect on the successful retrieval rate was not statistically significant according to the two-way repeated measures ANOVA (F_{(1, 26)} = 1.309, P = 0.263) (Fig. 2D). In the cylinder test, mice without stroke use both forelimbs equally to touch the cylinder wall. After stroke, they started to preferentially use the nonimpaired forelimb, and the ratio between impaired and nonimpaired limbs became approximately 0.5 at 7 dps. Arg1 cKO mice began to exhibit more frequent use of the impaired forelimbs starting from the 2 wk-time point, resulting in a higher ratio than Arg1 flox animals (Fig. 2E). Two-way repeated measures ANOVA revealed a significant difference between the control and cKO group over time (F_{(1, 26)} = 5.520, P = 0.028). In the ladder rung test, control Arg1 floxed animals made frequent mistakes at 7 dps, and the number of errors declined thereafter. Arg1 cKO animals exhibited much faster recovery in this test, showing lower percentage errors from the 2-wk time point (Fig. 2F), and the genotype effect on the difference in the error rate was also statistically significant (F_{(1, 26)} = 10.35, P = 0.004). Semiquantitative assessment of neurological functions using modified neurological severity score did not reveal a difference between the two groups (Fig. 2G). When the motor functions were assessed in male and female animals separately, Arg1 cKO mice exhibited a similar pattern of improvement in recovery regardless of the sex difference (SI Appendix, Fig. S4), indicating that sex did not influence functional recovery by Arg1 deletion.

Arg1 cKO Reduces Poststroke Fibrosis and Promotes Peri-Infarct Remyelination.

To determine the pathological outcomes that may underlie the enhanced functional recovery in Arg1 cKO mice, we first examined whether Arg1 deletion altered the activation of macrophage and microglial cells. The intensity of IBA-1 immunoreactivity representing both macrophages and microglial cells was comparable within the stroke lesion core at 7 dps (1 wk) between flox and cKO mice (SI Appendix, Fig. S5 A and B) IBA-1-positive cells in the peri-infarct cortex, which highly likely represent microglial cells, exhibited similar immunoreactivities (SI Appendix, Fig. S5 A and C). The IBA-1 intensity tended to decrease by 4 wk poststroke at both lesion core and peri-infarct region, but there was no significant difference between the two groups (SI Appendix, Fig. S5 D–F). In addition, morphological features of microglial cells, including the number and the length of cytoplasmic branches and the ramification index, which was used to quantify the extent of microglial activation (33), were not different between the two groups at either 1 or 4 wk poststroke (SI Appendix, Fig. S5 G–N). We also compared the astroglial reactivity surrounding the infarct core at 7 dps. The intensity of GFAP immunoreactivities surrounding the infarct core was not significantly different between flox and Arg1 cKO mice at 7 dps (SI Appendix, Fig. S5 O and P). At 4 wk poststroke, GFAP immunoreactivities were markedly increased compared to those at 7 dps, indicating consolidation of the glial scars by this time point (SI Appendix, Fig. S5 P and S). Nonetheless, there was no significant influence of Arg1 cKO on the glial scar formation (SI Appendix, Fig. S5 R and S). Importantly, the volume of infarcted tissue was also not affected by Arg1 deletion in macrophages at both 1- and 4-wk time points (SI Appendix, Fig. S5 Q and T).

Arginase activity in macrophages is linked to increased fibrosis in peripheral organs (34, 35). Recent studies revealed that ischemic stroke induces fibrotic scarring, which may exacerbate impairment of functional recovery and inhibit the formation of poststroke axonal connections (36, 37). To assess the extent of poststroke fibrosis, cortical tissues were stained using antibodies against fibronectin, an extracellular matrix (ECM) protein comprising fibrotic scars. We observed significantly reduced fibronectin expression in Arg1 cKO animals compared to Arg1 flox control animals following stroke (Fig. 3 A and B). Western blot analysis also confirmed the reduction of the fibronectin level in cortical lysates from Arg1 cKO mice (SI Appendix, Fig. S6 A and B). Chondroitin sulfate proteoglycans (CSPGs) are produced by astrocytic glia and other components of fibrotic scars such as fibroblasts and macrophages (38, 39). CSPGs deposited in injured CNS can contribute to the formation of fibrotic microenvironment (40, 41). We found that the photothrombotic stroke led to a marked accumulation of CSPGs (measured by CS-56 immunoreactivity) at the border of the infarcted tissue as well as the lesion core (Fig. 3C). The CSPG deposition was sharply attenuated in Arg1 cKO animals (Fig. 3 C and D), indicating that Arg1 in macrophages may be implicated in the development of poststroke fibrotic microenvironment.

Fig. 3.

Arg1 cKO reduces poststroke fibrosis and promotes peri-infarct myelination (A and B) Representative images of coronal brain sections subjected to immunohistochemical staining of fibronectin (A) and quantification graphs comparing the intensity of fibronectin immunoreactivity (B). (C and D) Representative images of coronal brain sections subjected to immunohistochemical staining of CS-56 (CSPGs) (C) and quantification graphs comparing the intensity of CS-56 immunoreactivity (D). The tissue sections were obtained from Arg1 cKO or flox control animals killed 4 wk after a photothrombotic stroke. Dashed rectangles indicate regions that are magnified on the right side. Dashed lines indicate the outer edges of the cerebral cortex. (Scale bars, 100 μm.) ** and *** indicate P < 0.01 and P < 0.001, respectively, by an unpaired t test. (N = 9 animals per group). (E and F) Representative images of coronal brain sections subjected to immunohistochemical staining of neurofilament M (NFM) (E) and quantification graphs comparing the density of NFM immunoreactivity (F). (G and H) Representative images of coronal brain sections stained for SMI-32 (Sternberger Monoclonals Incorporated-32, nonphosphorylated neurofilament) (G) and quantification graphs comparing the density of SMI-32 immunoreactivity (H). Images were also obtained in the peri-infarct cortex at the same 4-wk time point. (Scale bars, 100 μm.) (I) Double immunofluorescence staining of myelin basic protein (MBP) and fibronectin in brain sections obtained at 4 wk poststroke. (Scale bars, 100 μm.) Yellow dashed lines indicate the infarction borders. White dashed lines spaced at 50 μm intervals indicate boundaries for quantification based on distance from the border. (J) A quantification graph of MBP immunoreactivity at the peri-infarct cortex. * indicates P < 0.05 by an unpaired t test. (N = 9 animals per group). (K) A quantification graph of MBP immunoreactivity based on distance from the infarct border. ** indicates P < 0.01 by an unpaired t test. (L) Representative images of APC-CC1 immunohistochemistry in coronal brain sections obtained at 4 wk poststroke. (Scale bars, 100 μm.) Yellow dashed lines indicate the infarction borders. White dashed lines spaced at 50 μm intervals indicate boundaries for quantification based on distance from the border (M) A quantification graph of APC-CC1-positive cell number at the peri-infarct cortex. (N = 9 animals per group). (N) A quantification graph of APC-CC1-positive cell number based on distance from the infarct border. Each data point in all the graphs represents a single animal.

Fibrotic scar acts as a strong barrier that impedes axonal growth following CNS injury (37, 42, 43). When we visualized neurofilament-positive axons, there were very few axonal structures within the infarcted tissue in both control and cKO groups. Moreover, the density of the neurofilament-positive axons in the peri-infarct cortex did not differ (Fig. 3 E and F). We also examined the nonphosphorylated neurofilament immunoreactivity [Sternberger Monoclonals Incorporated-32 (SMI-32)], which represents damaged axons (44). The SMI-32 immunoreactivity did not change significantly in the peri-infarct area by Arg1 deletion in macrophages (Fig. 3 G and H). A recent study has reported that fibroblasts may inhibit the differentiation of oligodendrocyte precursor cells (45). Therefore, it is conceivable that reduction of the poststroke fibrosis by Arg1 deletion in LysM-positive macrophages may influence the extent of the peri-infarct myelination. Indeed, we observed a decrease in the intensity of myelin basic protein (MBP) immunoreactive signals in the peri-infarct region adjacent to the infarction border (Fig. 3I). The MBP density tended to be restored in the regions distant from the infarct boundary. Arg1 cKO animals showed significantly higher myelin density than the flox control group (Fig. 3 I and J). The difference between the two groups was noticeable only in the regions within a couple of 100 μm from the border (Fig. 3 I and K), suggesting that the fibrotic microenvironment may have a negative impact on the myelination of oligodendrocyte lineage cells that are positioned close to the fibrotic scar. When we measured the number of mature oligodendrocytes labeled by APC-CC1 antibodies, there was no significant difference between the flox and cKO groups (Fig. 3 L–N). Moreover, we did not find any difference in the number of early myelinating oligodendrocytes marked by BCAS-1 immunoreactivity (46) (SI Appendix, Fig. S7 A–C), or in the count of olig2+ entire oligodendrocyte lineage cells, including oligodendrocyte progenitor cells, between the control and cKO animals (SI Appendix, Fig. S7 D–F). Collectively, these findings suggest that Arg1 cKO in infiltrating macrophages reduces poststroke fibrosis accompanied by enhanced restoration of myelination in the peri-infarct cortex.

Influence of Arg1 cKO in Macrophages on Synaptic Structures and Microglial Synaptic Elimination in the Peri-Infarct Cortex.

Synaptic structures undergo dynamic remodeling in the peri-infarct cortex (47, 48), and the synaptic plasticity in this region plays a crucial role in poststroke functional recovery (49–52). Therefore, we reasoned that the notable improvement in motor functional recovery in Arg1 cKO animals may be linked to a differential regulation of synaptic connections in the peri-infarct region. The excitatory synaptic density was assessed by immunolabeling the excitatory presynaptic marker vGLUT2 and the postsynaptic marker PSD95 (Fig. 4 A–E). Our analysis showed a significant reduction in the number of both excitatory pre- and postsynaptic structures to a similar extent within the peri-infarct region at the 4-wk time point, when compared to animals that received a sham operation. Importantly, the loss of excitatory synapses was prevented by depleting Arg1 in infiltrating macrophages (Fig. 4 B and F–H). When we accurately identified the synaptic contacts with overlapping signals between the pre- and postsynaptic markers using superresolution imaging, the loss of synaptic contacts was more pronounced than either the pre- or postsynaptic marker alone (Fig. 4 E and H). The reduction of the synaptic contacts was also alleviated by Arg1 cKO in LysM-positive macrophages. Together, these results suggest that Arg1-expressing macrophages infiltrating into the infarcted region may play a role in synaptic loss in the peri-infarct cortex, which could contribute to poststroke functional deficits.

Fig. 4.

Influence of Arg1 cKO in macrophages on synaptic structures and microglial synaptic elimination in the peri-infarct cortex (A–D) Representative images of excitatory pre- (vGLUT2) and postsynaptic (PSD95) markers in the peri-infarct region 4 wk following a photothrombotic stroke. (Scale bar, 5 μm.) (E) Visualization of physical contacts between pre- and postsynaptic structures using superresolution microscopic imaging. Colocalized (white) puncta were analyzed as synapses. (F–H) Quantification graphs comparing the number of pre- (F), postsynaptic puncta (G), and synaptic contacts (H). N = 28 animals; flox + sham = 5, flox + stroke = 9, cKO + sham = 5, cKO + stroke = 9). *, **, and *** indicate P < 0.05, P < 0.01, and P < 0.001, respectively, by one-way ANOVA followed by post hoc Tukeys’s multiple comparisons. (I) An experimental scheme for analyzing in vivo microglial synaptic elimination using AAV9-Synaptophysin-mCherry-eGFP virus. Photothrombotic stroke was induced in the caudal forelimb area (CFA) 2 wk after AAV9-SYP-mCherry-eGFP injection into the rostral forelimb area (RFA). Images were captured in the tissue adjacent to the infarction site to observe the dynamics of synaptic connections from the RFA to the peri-infarct region of CFA. (J) mCherry alone synaptic puncta detected in Iba-1 positive microglia. (K) 3D reconstructed Iba-1 and mCherry alone puncta. mCherry alone puncta inside the 3D rendered Iba-1 cellular membrane were converted to 3D spots using IMARIS software. (Scale bar, 10 μm.) (L) Violin plot illustrating the number of mCherry alone synapses per microglia. (N = 135 microglia from five animals for the flox group and 138 cells from five animals for the cKO group). *** indicates P < 0.001, by an unpaired t test. (M) The average number of mCherry alone synapses in Iba-1 positive cells. N = 5 for the flox group and five for the cKO group (16 to 32 cells per subject). * indicates P < 0.05 by an unpaired t test. Each data point in all the graphs represents a single animal.

Recent studies suggest that microglial cells regulate the number of synapses in peri-infarct regions by phagocytosing synaptic structures (53, 54). We explored the possibility that microglial synaptic elimination in the peri-infarct cortex was affected by the deletion of Arg1 in macrophages within the infarcted tissue. To assess microglial synaptic elimination in vivo, we used AAV9-synaptophysin-mCherry-eGFP viruses to label neuronal synaptophysin. Given that the pKa of the red fluorescence proteins is lower than that of GFP, the mCherry signal is preserved in acidic environments such as in the phagocytic lysosome, whereas the GFP signal is rapidly dissipated (55). After delivery of AAV into the rostral forelimb area (RFA), we induced ischemic stroke in the caudal forelimb area (CFA) to examine synaptic connections originating from RFA to the adjacent ischemic region (Fig. 4I). mCherry alone signals observed within Iba-1-positive cells were determined as engulfed synapses by microglial cells (Fig. 4J). We quantified the number of the mCherry signals exclusively present in Iba-1-positive microglia using 3D rendered microglia and synapses (Fig. 4K). The violin plot showed that microglial cells from Arg1 cKO animals contained a significantly lower number of mCherry alone puncta than control floxed animals (Fig. 4L). The average number of mCherry alone puncta found within Iba-1 positive microglial cells was 7.7 ± 1.0 in control animals, and the average number decreased to 5.2 ± 0.4 in Arg1 cKO group (Fig. 4M), revealing a significant reduction in microglial synaptic elimination by deletion of Arg1 in macrophages.

Arg1 cKO in Macrophages Alters Microglial Inflammatory Gene Signatures in the Peri-Infarct Cortex.

To understand potential mechanisms underlying changes in microglial activity by Arg1 cKO in infiltrating macrophages, we conducted cytokine PCR array to profile expressions of inflammation signature genes from FACS-sorted microglia at 7 dps (Fig. 5A). Microglia in the peri-infarct cortex was sorted based on CD11b immunoreactivity and low expression of monocyte marker CD45 (SI Appendix, Fig. S8). A total of 53 cytokine genes were detected in CD45^low microglial cells isolated from the peri-infarct cortical tissue, as listed in the heatmap plot (Fig. 5B). Gene expression levels were compared between cKO and flox control animals under stroke conditions. The fold change values from the sham condition were subtracted from those observed in the stroke condition to account for the baseline gene expression resulting from the deletion of Arg1. All detected cytokine genes were plotted on a volcano plot to visualize the significant cytokine differentially expressed genes (DEGs) in Arg1 cKO compared to flox control animals following a photothrombotic stroke (Fig. 5C). Among the genes listed on the volcano plot, 10 were significantly downregulated, and only four genes were upregulated. Down DEGs included proinflammatory cytokines such as Interleukin (IL)-6, IL-6 family oncostatin M (Osm), and transforming growth factor (TGF-beta) with TGF-beta receptor ligands, including GDF9 and GDF15, while TGF-beta receptor antagonists such as BMP1 and BMP7 showed significant upregulation. K-means clustering classified the significant DEGs into four clusters (Fig. 5D). Cluster 1 included proinflammatory cytokines such as IL-6, Osm, IL-1b, and IL-16. TGF-beta related genes, Tgfb1, Tgfbr1, GDF1, and GDF15 composed of cluster 2. Cluster 3 consisted of BMP1, BMP7, and Fgf10, and TNF superfamily 12 and 13b (Tnfsf12 and Tnfsf13b), implicated in proinflammatory NFkB activation, were included in cluster 4. Interactions among the cytokines in each cluster were quantified using the STRING protein interaction analysis, which revealed IL-6 and TGF-beta as the top two cytokines with the highest interactions (Fig. 5E). Gene ontology (GO) analysis showed that the cytokine DEGs were enriched in the proinflammatory pathway (regulation of interleukin-6 production, inflammatory response, interleukin-1-mediated signaling pathway) and the TGF-beta signaling (TGF-beta receptor signaling pathway, Small Mothers Against Decapentaplegic protein phosphorylation) (Fig. 5F). Collectively, the gene expression study and bioinformatic analysis indicated that deleting Arg1 in macrophages infiltrating into the infarcted region reduced proinflammatory activity and TGF-beta signaling in CD11b⁺/CD45^low microglial cells in the peri-infarct cortex.

Fig. 5.

Arg1 cKO in macrophages alters microglial inflammatory gene signatures in the peri-infarct cortex (A) A schematic diagram of the experimental design. A chunk of the cortical tissue, including the infarcted and spared peri-infarct cortex, was obtained at 7 dps, and the infarct core was carefully dissected and removed. The remaining peri-infarct cortical tissue was dissociated into single cells using the Adult brain dissociation kit, and the dissociated cells were subjected to FACS to isolate CD11b+/CD45^low microglial cells from macrophages. (B) A heatmap plot comparing all detected cytokine expressions between flox and cKO groups in the sham or stroke conditions. Log₂(fold change) of the gene expression value of each cKO subject compared to the average expression value of the flox group was color-coded. Asterisks indicate the statistical significance of comparing cKO and flox groups in the stroke condition normalized by subtracting the difference between cKO and flox groups in the sham conditions. * and ** indicate P < 0.05 and P < 0.01, respectively, by an unpaired t test. (C) A volcano plot of DEGs by Arg1 cKO in the stroke condition. Significantly regulated genes were determined based on log₂(fold change) > 0.5 and P < 0.05 criteria. Down DEGs were shown in blue, and up DEGs in red. The fold change value comparing cKO and flox control animals in the stroke condition was normalized by subtracting the difference between cKO and flox groups in the sham condition. (D) K-means clustering of the significant cytokine DEGs. (E) The degree of interactions within each cluster was calculated using the web-based STRING analysis. Protein interaction value was obtained by dividing the number of node degrees by the total number of nodes. (F) A bubble plot of the enriched GO processes (P < 0.05 = FDR < 95.0). The bubble size indicates the number of annotated genes in the GO process. (G) Representative images of semi-PCR products of microglial phagocytic markers (CD68, CR3, CD36, GAL-3, and TREM2). mRNA samples were obtained from FACS-isolated microglia out of the peri-infarct cortex 7 d poststroke. (H–L) Quantitative comparison of mRNA expressions by real-time PCR. CD68 and CR3 showed significantly reduced expression in Arg1 cKO animals compared to flox animals after stroke. *, **, and*** indicate P < 0.05, P < 0.01, and P < 0.001, respectively, by one-way ANOVA followed by post hoc Tukey’s multiple comparison. Each data point in all the graphs represents a single animal.

Since we observed attenuated synaptic eliminations by microglia in Arg1 cKO animals following stroke, we also examined the expression of phagocytosis-related genes Trem2, Gal3, CD36, CD68, and CR3 (56–60) in FACS-isolated CD11b⁺/CD45^low microglial cells (Fig. 5G). Among the phagocytic markers we measured, CD68, CD36, and Gal-3 were significantly up-regulated in microglia following stroke (Fig. 5 G–L). Arg1 cKO following stroke exhibited significantly down-regulated expression of CD68, but not CD36 and Gal-3 (Fig. 5 H, J, and K). CR3 expression was slightly increased in stroke conditions, and the level was significantly reduced by Arg1 cKO (Fig. 5I). Expression of Trem2 was decreased following stroke in Arg1 cKO animals. However, induction of stroke did not influence Trem2 expression in wild-type animals (Fig. 5I).

Arg1 Activation in Macrophages Influences Microglia to Enhance Synaptosome Phagocytosis In Vitro.

These findings suggest that macrophages infiltrating the infarcted tissue may interact with microglial cells in the peri-infarct cortex to regulate microglial activity. Recent studies also reported dynamic crosstalk between peripherally derived macrophages and microglia in CNS inflammation (19, 20). To directly test the interaction between the two cell types, we designed an in vitro experiment where bone marrow–derived macrophages (BMDMs) were cocultured with microglial cells. To induce arginase activity in BMDMs, zymosan, a toll-like receptor 2 agonist, was treated based on a previous study (61), and zymosan administration indeed markedly enhanced the arginase activity in BMDMs (Fig. 6A). The increased arginase activity was effectively abolished by OATD-02, an arginase inhibitor. Cultured BMDMs were treated with zymosan with or without OATD-02 for 24 h and then cocultured with microglia on a cell culture insert, allowing the two cell types to communicate with each other using diffusible molecules (Fig. 6B). Then, the cell culture insert with BMDMs were removed, and microglial cells were subjected to mRNA measurement or in vitro phagocytosis assay. We measured the mRNA expression of the two hub cytokine genes, TGF-beta and IL-6, downregulated in the peri-infarct microglial cells following a photothrombotic stroke, using real-time PCR. Coculturing microglia with zymosan-treated BMDMs did not alter the expression of TGF-beta1 mRNA (Fig. 6C). However, the mRNA levels of IL-6 were significantly upregulated in these cocultures (Fig. 6D). The upregulation of IL-6 mRNA was attenuated when BMDMs in the coculture system were treated with both zymosan and OATD-02 (Fig. 6D), indicating that the induction of IL-6 expression may be mediated by the enhanced arginase activity in BMDMs.

Fig. 6.

Arg1 activation in macrophages influences microglia to enhance synaptosome phagocytosis in vitro (A) Arg1 activity assay in BMDMs following zymosan or OATD-02 treatment. *** indicates P < 0.001, by one-way ANOVA followed by post hoc Tukey’s multiple comparison. (B) A schematic diagram of the experimental design for the in vitro macrophage–microglia interaction assay. BMDMs were cultured with Zymosan (1.25 μg/mL) with or without Arg1 inhibitor, OATD-02 (1 μM). After 24 h, BMDMs were cocultured with primary microglia for 24 h. (C and D) Expression of transforming growth factor-beta1 (C) and interleukin-6 (D) mRNA expression was measured in microglia using quantitative RT-PCR 24 h after coculture. ** indicates P < 0.01, by one-way ANOVA followed by post hoc Tukey’s multiple comparison. (E) Representative images of microglial pHrodo-synaptosome phagocytosis assay. (Scale bar, 100 μm.) (F) Quantitative graph of pHrodo+ cell counts. (G) The area under cover of the graph (F) for statistical analysis. * and *** indicate P < 0.05 and P < 0.001, respectively, by one-way ANOVA followed by post hoc Tukey’s multiple comparison. Each data point represents an independent culture prepared from two wild-type animals. (H) Arg1 activity assay in BMDMs from floxed (wild-type) or Arg1 cKO animals following zymosan treatment. *** indicates P < 0.001, by one-way ANOVA followed by post hoc Tukey’s multiple comparison. (I) Semiquantitative RT-PCR results of Arg1 and Arg1 gene expression in wild-type or Arg1 cKO BMDMs with or without zymosan treatment. (J) Quantitative graph of pHrodo+ cell counts. (K) The area under cover of the graph (J) for statistical analysis. ** and *** indicate P < 0.01 and P < 0.001, respectively, by one-way ANOVA followed by post hoc Tukey’s multiple comparison. Each data point represents an independent culture prepared from two wild-type animals and two Arg1 cKO animals.

To mimic microglial synaptic elimination, in vitro phagocytosis assay was performed using synaptosomes isolated from mouse cortical tissue. These synaptosomes were labeled with pHrodo, a pH-sensitive fluorescent dye that is expected to emit fluorescence when the synaptosomes are engulfed within an endosome at a low pH level (62). Microglial cells cocultured with zymosan-treated BMDMs exhibited a higher number of pHrodo-positive cells compared to those cocultured with phosphate-buffered saline (PBS)-treated BMDMs (Fig. 6E), suggesting that BMDMs treated with zymosan enhanced the synaptic phagocytic activity of cocultured microglial cells. To determine whether zymosan-treated BMDMs influence microglial phagocytosis specifically for synaptosomes, we used latex beads labeled with pHrodo instead of synaptosomes for the phagocytosis assay. While the baseline microglial phagocytosis levels were comparable for both synaptosomes and latex beads, coculturing with zymosan-treated BMDMs did not increase microglial phagocytic activity (SI Appendix, Fig. S9 A–C), suggesting that zymosan-treated BMDMs specifically influence microglial synaptic phagocytosis.

When microglial cells were cocultured with BMDMs treated with both zymosan and OATD-02, the number of pHrodo-positive microglial cells decreased (Fig. 6E and Movie S1). Quantitative analysis revealed that the zymosan + OATD-02 group showed a substantial reduction in pHrodo-positive cells during the early sessions of the assay, with a trend toward recovery in the later sessions (Fig. 6F). Quantification of the area under the curve (AUC) throughout the assay demonstrated a partial, yet statistically significant, reduction in the extent of synaptic phagocytosis by OATD-02 (Fig. 6G). Since OATD-02 inhibits both Arg1 and Arg2 (63, 64), we conducted the in vitro phagocytosis assay using Arg1 knockout BMDMs instead of OATD-02. The Arg1-deficient BMDMs were derived from Arg1 cKO mice using the same LysM-cre line, as the Cre recombinase activity is notably high in BMDMs from LysM-cre mice (65). Measurement of arginase activity revealed that the zymosan-induced increase in arginase activity was significantly lower in Arg1-deficient BMDMs than that in floxed (wild-type) BMDMs (Fig. 6H). However, a substantial level of arginase activity remained in Arg1-deficient BMDMs compared to those treated with OATD-02, likely due to the intact expression of Arg2 in these BMDMs (Fig. 6I). In the synaptosome phagocytosis assay, when microglial cells were cocultured with either floxed or Arg1-deficient BMDMs treated with zymosan, the number of pHrodo-positive cells decreased when cocultured with Arg1-deficient BMDMs, to an extent comparable to that with BMDMs treated with OATD-02 (Fig. 6 J and K). This finding indicates that the arginase activity of Arg1, rather than Arg2, in macrophages specifically influences synaptic phagocytosis in microglial cells.

Discussion

The current study investigated the functional role of Arg1, a marker for anti-inflammatory macrophage phenotype, in the context of ischemic stroke. Arg1 expression has been documented in infiltrating macrophages or monocyte-derived macrophages in various CNS injuries, including ischemic stroke (28, 66, 67). Arg1 expression has been associated with improved functional outcomes (28, 68) and is often considered a marker for macrophages that promote CNS tissue repair or regeneration (67, 69, 70). However, few studies have directly addressed the functional role of Arg1 expression in macrophages within these contexts. A recent study demonstrated that depletion of Arg1-positive microglia/macrophages using mannosylated clodronate liposomes exacerbated poststroke outcomes by promoting proinflammatory responses (71). However, this study examined the functional roles of Arg1-positive myeloid cells, rather than the Arg1 protein itself. Another study reported that knockout of STAT6 led to increased infarction volume and worsened functional outcomes, partly by suppressing Arg1 expression in macrophages and microglial cells. Lentiviral overexpression of Arg1 restored the anti-inflammatory features (72), supporting a beneficial influence of Arg1 on poststroke recovery. Our study directly examines the functional role of Arg1 in macrophages in an animal model of ischemic stroke. Our findings suggest a potentially detrimental role of Arg1 expressed in infiltrating macrophages on poststroke recovery.

Our cellular mapping study has identified LysM-positive infiltrating macrophages as a primary source of Arg1 following photothrombotic stroke. This finding was corroborated by a marked reduction in Arg1-positive cells upon Arg1 deletion using the LysM-cre line. These results align with an earlier study demonstrating exclusive Arg1 expression in LysM-GFP-positive cells, but not in GFP-negative myeloid cells, in the middle cerebral artery occlusion model (73). A recent study also reported that Arg1 is predominantly expressed in monocyte-derived macrophages traced by Cxcr4-cre compared to microglial cells (33). Given that microglial cells also exhibit endogenous LysM promoter activity (74), it is possible that LysM-cre cannot fully differentiate between macrophages and microglial cells. However, since Arg1 expression is mostly found in myeloid cells within the infarcted cortex, where infiltrating macrophages are dominant after a stroke, the impact of LysM-expressing microglial cells in our study results would be minimal. We observed that approximately 20% of Arg1-positive cells colocalized with CX3CR1-GFP-positive cells at 3 and 7 dps. These cells may represent either genuine resident microglia or microglia-like cells derived from infiltrating hematogenous macrophages. In the latter case, Arg1 in these microglia-like cells would be deleted by the LysM-cre line, given their hematogenous macrophage origin (30). It is noteworthy that not all Arg1-positive cells were LysM-positive, and a small fraction of Arg1-expressing cells persisted in Arg1 cKO animals. This may indicate that not all infiltrating macrophages are LysM-positive, as previous studies have reported that a small population of infiltrating macrophages can express CX3CR1 (75). We speculate that the influence of the small amount of residual Arg1, regardless of its origin, is negligible in the context of our study.

Our study demonstrates that deletion of Arg1 in infiltrating macrophages substantially reduced the accumulation of fibrotic ECM both in the infarcted tissue and at the border between the infarction core and peri-infarct cortex. While fibrosis is a key component of wound healing, excessive fibrosis can impede functional restoration following injury in various organs. In CNS, fibrotic scarring has been shown to inhibit axonal regeneration and diminish functional recovery (43, 76). A recent study revealed that photothrombotic stroke led to demyelination in the peri-infarct cortex, which could be improved by regulating inflammatory signaling (77). We also observed decreased myelination in areas of excessive fibrotic ECM accumulation near the infarction. This pattern is reminiscent of the experimental autoimmune encephalomyelitis model, where perivascular fibroblast infiltration in the spinal cord parenchyma correlates with areas of demyelination (45). Although Arg1 cKO enhanced myelination in the peri-infarct cortex close to the infarction border, the number of oligodendrocyte lineage cells at various maturation stages was not changed in Arg1 cKO animals. These findings suggest that fibrotic ECM accumulation, orchestrated by Arg1 in infiltrating macrophages, may specifically inhibit the remyelination process in the peri-infarct cortex close to the infarction border. We found that deleting Arg1 in LysM-positive macrophages reduced TGF-beta signaling, a crucial pathway regulating fibrosis, in microglial cells in the peri-infarct cortex. This decrease in TGF-β activity in microglia near the infarction border may contribute to the reduced fibrotic scarring observed in Arg1 cKO mice.

Immunolabeling of excitatory synapses revealed that photothrombotic stroke significantly reduced the number of excitatory synapses in the peri-infarct area. Notably, the depletion of Arg1 in infiltrating macrophages mitigated this synaptic loss. Microglial cells are recognized for their essential role in regulating synapse formation in the brain through synaptic pruning during developmental stages (78). In neurodegenerative diseases like Alzheimer’s disease, microglial immune responses lead to the elimination of viable synapses, exacerbating synaptic loss (79). Similarly, synaptic elimination by microglial phagocytosis also occurs in the peri-infarct region following stroke (53, 54). It has been suggested that microglial cells may target dysfunctional synapses, of which removal could lead to functional benefits (80, 81). However, inhibition of synaptic phagocytosis improved poststroke functional recovery (53, 54), indicating that microglia-mediated loss of synaptic structures may contribute to neurological deficits following stroke. Therefore, it is highly likely that decreased microglial synaptic elimination by Arg1 cKO in the current study contributed to the improvement of motor recovery. Since fluorescence-tagged synaptophysin was used to visualize the microglial process of synaptic elimination, we could not analyze excitatory and inhibitory synapses separately. Removal of inhibitory synapses may enhance functional recovery by disinhibition of inhibitory neuronal influence on synaptic excitability. Future studies are necessary to specifically examine how the phagocytosis of inhibitory synapses affects functional outcomes following a stroke.

Microglial states regulating synaptic phagocytosis are influenced by various inflammatory cytokines (82). Recent research reported that SPP1-mediated microglial synaptic elimination is regulated by autonomous TGF-beta signaling in an Alzheimer’s disease model (83), implicating TGF-beta signaling in the activation of microglial synaptic elimination. We also found reduced TGF-beta signaling in CD45^low microglial cells from Arg1 cKO animals. Our data also suggest that CD68 and CR3 may be potential downstream effectors regulating microglial synaptic phagocytosis following stroke. Trem2 is known to regulate microglial synapse elimination not only during normal development but also in neurodegenerative conditions (81, 84). Although Trem2 expression was downregulated following stroke in Arg1 cKO animals, its expression level was not influenced by stroke induction in wild-type mice, raising the question of whether Trem2 is critically involved in microglial synaptic elimination under stroke conditions. Our in vitro experiment directly addressed the question of how Arg1 deletion in infiltrating macrophages in the infarcted tissue alters the poststroke inflammatory milieu in the peri-infarct cortex. We observed that zymosan-treated BMDMs significantly enhanced synaptic phagocytosis in cocultured microglial cells. Importantly, this enhanced phagocytic activity was attenuated by cotreatment with the Arginase 1/2 inhibitor OATD-02. Although BMDMs lacking Arg1 showed some remaining arginase activity derived from Arg2, synaptic phagocytosis in microglial cells was substantially suppressed, highlighting the pivotal role of Arg1 over Arg2 in regulating microglial synaptic phagocytosis. These results imply that Arg1 activity in macrophages infiltrating into the infarcted tissue exerts modulatory influence on microglial cells in the peri-infarct tissue. It is worth noting that the behavioral consequences of Arg1 deletion in LysM-positive infiltrating macrophages became evident 2 wk after stroke induction, while peak expression of Arg1 was observed at the 1-wk time point. We speculate that the inconsistency between the timing of Arg1 expression and the impact of its deletion supports the notion that a detrimental influence of Arg1 is exerted indirectly by affecting microglial cells in the peri-infarct cortex. Modulation of microglial phagocytosis by peripherally derived macrophages was also observed in a spinal cord injury model (19). Conversely, microglial cells can regulate the recruitment and neurotoxic effects of monocyte-derived macrophages (21). These observations suggest a bidirectional interaction between macrophages and microglial cells, which may play a crucial role in shaping the postinjury inflammatory milieu (20).

In conclusion, our study elucidates the detrimental role of Arg1 expressed in infiltrating macrophages following ischemic stroke. We demonstrate that Arg1 deletion in LysM-positive macrophages confers beneficial effects on poststroke recovery through two primary mechanisms: regulation of fibrotic scar formation and modulation of microglial synaptic phagocytosis. We propose that Arg1 in macrophages represents a promising therapeutic target for modulating the poststroke inflammatory environment.

Materials and Methods

Experimental Animals.

All animal experiments were performed under the approval of the Institutional Animal Care and Use Committee of Ajou University Medical Center. Arg1 flox (C57BL/6-Arg1tm1Pmu/J, Strain #:008817), Cx3cr1-GFP (B6.129P2(Cg)-Cx3cr1tm1Litt/J, Strain #:005582), Rosa-stop-eYFP (C57BL/6Gt(ROSA)26Sortm1(HBEGF)Awai/J, Strain #:007900), and LysM-cre (B6.129P2-Lyz2tm1(cre)Ifo/J, Strain #:004781) lines were purchased from Jackson laboratory. Animals were maintained in individually ventilated cages with 12 h of light/dark cycle. To create a cKO of Arg1 animals, specifically in cells expressing LysM, Arg1 flox mice were bred with LysM-cre mice, where the Cre recombinase is expressed under the control of LysM promoter. Animals that were homozygous for the Arg1 flox allele and heterozygous for the LysM-cre allele were designated as Arg1 cKO mice. Arg1 flox homozygous littermates from the production of Arg1 cKO animals were used as controls, sharing similar maternal care and housing conditions. To generate mice expressing YFP under the LysM promoter control, we crossed Rosa-stop-eYFP mice with the LysM-cre line. Mice heterozygous for both alleles were used in experiments.

Induction of Photothrombotic Ischemic Stroke.

Animals aged 8 to 12 wk were used for the induction of photothrombotic infarction. They were anesthetized by intraperitoneal injection of ketamine (90 mg/kg) and xylazine (10 mg/kg) mixture before surgery. The head was secured in the stereotaxic frame, and an incision was made to expose the skull after a brief sterilization using 70% ethanol. A cold light source, utilizing a 150 W halogen lamp fiberoptics with a diaphragm of 1.5 mm diameter, was placed on either side of the skull corresponding to the CFA at the predetermined coordinates (AP 0.0 mm, ML ± 1.8 mm). For animals subjected to behavioral assessment, 10-wk-old animals underwent photothrombosis following a 2-wk pretraining period. Illumination was directed to the side contralateral to the preferred forelimb determined during the shaping session for the pellet retrieval test. Animals not subjected to the behavioral study underwent photothrombosis induction in the right hemisphere. Rose Bengal (Sigma-Aldrich, 10 mg/mL, 10 μL/g body weight) was administered intraperitoneally, and illumination started 5 min after injection. The cold light was illuminated continuously for exactly 25 min. To prevent hypothermia, animals were maintained on a 37 °C heating pad throughout surgery and illumination. Animals in the sham group received an intraperitoneal injection of PBS vehicle instead of Rose Bengal, along with the same duration of light illumination.

In Vivo Phagocytosis Analysis and 3D Image Reconstruction.

To analyze in vivo synaptic elimination by microglia after stroke, AAV9-hSYN-mCherry-eGFP was injected into the RFA of the premotor cortex (AP -1.5 mm, ML -1.8 mm, DV -1.5 mm) of mice. Two weeks after viral injection, photothrombotic stroke was induced in the CFA. Animals were killed at 7 dps. The brain was sectioned into 30 μm-thickness and labeled with Iba-1 antibody to visualize microglia. At least five regions of interest (ROIs) were imaged approximately 360 μm away from the stroke boundary using a Nikon A1R confocal microscope. The image files were converted to IMARIS 9.0 software formats for 3D reconstructions. The phagocytosed synapsed Iba-1 signals were reconstructed into 3D surfaces, while mCherry or eGFP signals were converted into 3D spots. Rendered 3D spots outside the Iba-1 3D surface were removed. To detect mCherry signals alone within 3d rendered Iba-1, any mCherry signal colocalized with eGFP was masked. Although mCherry and eGFP+ colocalized signals were also detected in Iba-1 cells, the majority of signals were from mCherry alone. eGFP alone spots were rarely detected in Iba-1 positive cells. Noncolocalizing mCherry spots with Iba-1 surface were considered as “phagocytosed synapses.” The number of phagocytosed synapses was analyzed from 136 or 137 Iba-1+ cells from five animals each for Arg1 flox or Arg1 cKO groups, respectively. Individual cell values and averaged group values were plotted separately.

In Vitro Microglia Synaptosome Phagocytosis Assay.

To isolate synaptosome preparations, cortical tissue (~10.0 mg) was dissected from the brain of C57BL6/N wild-type mice. The tissue was immersed in 100 μL of 0.32 M sucrose and homogenized using a tissue homogenizer. The homogenate was centrifuged at 470 g for 2 min, and the supernatant was collected. The supernatant was then centrifuged again at 10,000 g for 10 min. The resulting supernatant was removed, and the pellet was resuspended in 0.32 M sucrose solution. This suspension was transferred into 0.8 M sucrose solution and centrifuged at 9,100 g for 15 min using a swing bucket rotor. After centrifugation, the synaptosome layer between the 0.32 M and 0.8 M sucrose solution was collected. To label the synaptosomes with pHrodo, they were resuspended in sodium bicarbonate buffer (pH 8.4) at a concentration of 3 mg/mL. pHrodo (Invitrogen, P36600, 3 μM) dissolved in DMSO was added to the synaptosomes and incubated for 45 min at room temperature. The synaptosomes were then diluted in DPBS (Gibco, 14190-144) at a 1:10 ratio and centrifuged for 5 min at 2,500 g, twice, to wash away excess pHrodo. The synaptosomes were resuspended in serum-free DMEM at 3 mg/mL concentration and used for the in vitro phagocytosis assay. After 24 h of coculture with BMDMs, pHrodo-labeled synaptosomes were added to the microglia at a final 0.017 mg/mL concentration. The cell culture plate was placed in the JULI stage (NanoEnTek) and live images were taken for 24 h at 80-min intervals. Five ROIs from each well were imaged using a 10× lens. To analyze the number of pHrodo-positive microglia, the JULI STAT software (ver. 2.0.1) was used, which automatically counted the number of pHrodo-positive cells at each time point. For the phagocytosis assay using latex beads, approximately 1 × 10⁹ Amine Latex Beads (ThermoFisher Scientific, A37362, 1.0 μm) were centrifuged at 15,000 g and resuspended in sodium bicarbonate buffer (pH 8.4). pHrodo (3 μM) was added to the latex beads and incubated at room temperature for 45 min. The beads were washed three times and resuspended in DPBS. Approximately 5 × 10⁶ beads were applied to the cells. TGF-beta and IL-6 mRNA expression in cocultured microglia was measured from extracted mRNA. qRT-PCR was performed using the following primer sets: IL-6: Forward: 5′-AGTTGCCTTCTTGGGACTGA-3′, Reverse: 5′-TCCACGATTTCCCAGAGAAC-3′; TGF-beta: Forward: 5′-TGCGCTTGCAG AGATTAAAA-3′, Reverse: 5′-AGCCCTGTATTCCGTCTCCT-3′.

Statistical Analysis.

Statistical analyses were conducted using GraphPad Prism software (Ver.9.0). An unpaired t test was performed to compare two independent groups. One-way ANOVA was used to compare more than two independent groups. Behavioral assessments were analyzed using two-way ANOVA to account for the time difference within the same subject. Türkiye’s post hoc analysis was performed for the statistical hypothesis test.

The remaining materials and methods are provided in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

Live images of microglial pHrodo-synaptosome phagocytosis assay. Each frame of the video represents 80 minutes, and live imaging was conducted for 24 hours immediately following pHrodo-synaptosome treatment. Microglia were incubated for 24 hours with conditioned media collected from a macrophage and microglia co-culture before pHrodo-synaptosome treatment. Scale bar = 50 μm.

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.