Macrophages play a leading role in determining the direction of astrocytic migration in spinal cord injury via ADP-P2Y1R axis

After spinal cord injury (SCI), inflammatory cells such as macrophages infiltrate the injured area, and astrocytes migrate, forming a glial scar around macrophages. The glial scar inhibits axonal regeneration, resulting in significant permanent disability. However, the mechanism by which glial scar-forming astrocytes migrate to the injury site has not been clarified. Here we show that migrating macrophages attract reactive astrocytes toward the center of the lesion after SCI. Chimeric mice with bone marrow lacking IRF8, which controls macrophage centripetal migration after SCI, showed widely scattered macrophages in injured spinal cord with the formation of a huge glial scar around the macrophages. To determine whether astrocytes or macrophages play a leading role in determining the directions of migration, we generated chimeric mice with reactive astrocyte-specific Socs3−/− mice, which showed enhanced astrocyte migration, and bone marrow from IRF8−/− mice. In this mouse model, macrophages were widely scattered, and a huge glial scar was formed around the macrophages as in wild-type mice that were transplanted with IRF8−/ bone marrow. In addition, we revealed that macrophage-secreted ATP-derived ADP attracts astrocytes via the P2Y1 receptor. Our findings revealed a mechanism in which migrating macrophages attracted astrocytes and affected the pathophysiology and outcome after SCI.


Introduction
Although spinal cord injury (SCI) causes severe disability, effective treatments have not been established 1 . When the blood-spinal cord barrier is disrupted by mechanical injury, in ammatory cells, such as neutrophils and macrophages in ltrate into the spinal cord 2 , and these cells secrete in ammatory cytokines, such as interleukin (IL)-1α, IL-1β, IL-6, and tumor necrosis factor (TNF)-α, causing secondary damage 2 . After that, extracellular matrix deposition and astrocyte migration occur, and scar tissue is formed. The extracellular matrix and in ammatory cells are located at the center of the injury, and astrocytes migrate toward them to form a glial scar that serves as a physical barrier. Since this barrier prevents axonal regeneration, elucidation of the mechanism of glial scar formation is critical for developing new treatments for SCI.
We have previously reported that Stat3 regulates astrocyte migration 3 . However, the factors that attract astrocytes to the injury site remain unknown, although astrocyte migration is one of the most important processes in glial scar formation. In this study, we attempted to identify factors that attract astrocytes to the injury center by focusing on macrophages inside the glial scar. The proportion of macrophages at the injury center varies over time. Especially after the subacute stage, when the glial scar is formed, the in ltration of macrophages is particularly pronounced (7 days post-injury) 4 , suggesting that macrophages may also be involved in the formation of the glial scar. We hypothesized that macrophages might also be involved in the appearance of glial scars. We reported that in IRF8 −/− mice, in which macrophages are widely scattered after SCI due to impaired migration, larger glial scars were formed in comparison to wild-type mice 5 . In the central nervous system (CNS), astrocytes form close intercellular communication with other cell types, such as neurons and microglia, in which purine receptors play a signi cant role, for example, in regulating homeostasis, with consequences for synaptic transmission, and higher-order cognitive processes 6,7 . In addition, macrophages have been shown to attract other macrophages by secreting ATP, the ligand for the purine receptor 8 . Therefore, we hypothesized that macrophages also induce astrocytes via purine receptors. In this study, we demonstrated the effect of macrophages on astrocyte migration after SCI, which is critical for glial scar formation.

Macrophages attract astrocytes in SCI
After SCI, macrophages in ltrated into the injury site and migrated toward the epicenter with the migration of reactive astrocytes toward the peri-injury area 5,9 . Initially, scattered macrophages and astrocytes migrate toward the epicenter over time as well, but eventually, they are located in different sites, with macrophages in the center of the lesion and astrocytes at the margin of the injury (Fig. 1a, b). These results suggest that macrophage migration may affect astrocyte migration. To investigate what happens to astrocytes when macrophage migration is impaired, a bone marrow chimera mouse model was created by transplanting bone marrow from IRF8 −/− mice (which have been reported to show impaired macrophage migration) 5 into Nes-Cre-EGFP mice, in which reactive astrocytes expressed EGFP. Using the chimeric mice models, we compared the scar area at 7 and 14 days after SCI and found that the glial scar area in mice with impaired macrophage migration was signi cantly greater than that in the mice with normal macrophage migration at both time points (Fig. 2a, b). These results suggest that the migration of macrophages in uences the migration of reactive astrocytes and subsequent astrocytic scar formation. In addition, we investigated whether the intrinsic nature of astrocytes was altered with impairment of macrophage migration. First, since astrocytes are known to polymerize at the injury margins in scar formation, we evaluated the effect on polymerization via cell density. We found no signi cant difference in cell density at either 7 or 14 days post-injury, indicating that loss of IRF8 in macrophages had no signi cant effect on astrocyte polymerization ( Fig. 2c-h). Next, since activated astrocytes are hypertrophied, we investigated whether astrocyte activity was affected by assessing the area of astrocyte cell bodies. As a result, the area of EGFP + astrocytes was not altered by impaired macrophage migration, suggesting that impaired macrophage migration had no signi cant effect on the cellular activation of reactive astrocytes (Fig. 2c-h). Finally, the proliferative capacity of astrocytes was evaluated by Ki67 staining. No signi cant differences were found in proliferation between astrocytes in these two chimeric mice (Fig. 2c-h). These ndings suggest that the macrophages attract astrocytes extrinsically, without altering their proliferative potential. We previously reported that Slc39a6 is involved in the Stat3-mediated migration mechanism of reactive astrocytes 3 . Therefore, we assessed the expression of Slc39a6 and found no signi cant difference (Fig. 2i). This result suggests that the effect of macrophages on astrocyte migration is not mediated by Stat3 in astrocytes. These ndings indicate that IRF8-mediated macrophage migration affects reactive astrocyte migration without changing the proliferation, activation, or Stat3-mediated migration capacity of reactive astrocytes.
Macrophage migration is the leading factor in SCI.
In a previous study, we reported that the rapid migration of reactive astrocytes leads to the early accumulation of macrophages using Nes-Socs3 −/− -EGFP + mice, in which Socs3 gene in reactive astrocytes was deleted 3 . In contrast, the poor migration of IRF8 −/− macrophages was accompanied by the impaired migration of reactive astrocytes (Fig. 2a, b). These facts suggest the possibility of interaction between in ltrating blood-derivedmacrophages and reactive astrocytes after SCI. To further investigate whether the migration of IRF8 −/− macrophages could affect the migration of reactive astrocytes, we generated two additional chimeric mice: [macrophages; IRF8 +/+ ]/[reactive astrocytes; Nes-Socs3 −/− -EGFP + ] and [macrophages; IRF8 −/− ]/[reactive astrocytes; Nes-Socs3 −/− -EGFP + ] using EGFPnegative IRF8 +/+ or IRF8 −/− bone marrow cells with Nes-Socs3 −/− -EGFP + recipient mice (Fig. 3a). In the former chimeric mice, IRF8 +/+ macrophages showed extremely fast migration with the rapid migration of Nes-Socs3 −/− reactive astrocytes after SCI, as described in our previous study ( Fig. 3b-d) 3 . However, even with Nes-Socs3 −/− -EGFP + reactive astrocytes, the migration disorder of IRF8 −/− macrophages was not rescued in the latter chimeric mice ( Fig. 3b-d). Instead, the Nes-Socs3 −/− -EGFP + reactive astrocytes with IRF8 −/− macrophages migrated more slowly than those with IRF8 +/+ macrophages (Fig. 3b, d). These results mean that while the migration of macrophages and reactive astrocytes interact with one another, the migration of macrophages is more strongly affected by macrophage IRF8 than by the migration of reactive astrocytes. We further examined the interaction between the migration of macrophages and reactive astrocytes on functional recovery after SCI. There was no signi cant deference among WT, IRF8 −/− and these chimeric mice before SCI, or at 1 day post-injury ( Fig. 3e). However, at 14 days postinjury, although the former chimeric mice with WT macrophages and Nes-Socs3 −/− reactive astrocytes showed signi cantly ameliorated functional recovery in comparison to WT mice, the latter chimeric mice with IRF8 −/− macrophages and Nes-Socs3 −/− reactive astrocytes exhibited greater deterioration of the motor functional recovery in comparison to WT and the former chimeric mice (Fig. 3e). These ndings revealed that, for motor recovery after SCI, autonomous migration of macrophages via IRF8 is more important than reactive astrocyte migration. Regarding astrocyte migration and macrophage migration, these ndings clari ed that macrophage migration plays a leading role in the interaction of their migration. In addition, Socs3 is a negative feedback factor for Stat3 10,11 , and it is known that Stat3 regulates reactive astrocyte migration via the Rho A small G protein 12 . In this study, the loss of IRF8 in macrophages also altered the migration of astrocytes with or without the expression of Socs3. This result also supports the hypothesis that the pathway by which macrophages attract astrocytes is not mediated by Stat3 (Fig. 2i).

Macrophage-astrocyte Interactions Via P2y1 Receptors
To elucidate how macrophages affect the migration of astrocytes, we focused on P2Y1 receptor (P2Y1R) as a representative receptor in the migration pathways, because P2Y1R was reported to be involved in astrocyte velocity in traumatic brain injury 13 and to regulate cell migration via the ADP-P2Y1R-MAP/ERK pathway 14 . Therefore, we hypothesized that ATP-derived ADP secreted by macrophages might attract astrocytes via P2Y1R. First, we performed immunostaining to con rm the expression of P2Y1R in the astrocytes of the injured spinal cord. As a result, P2Y1R was expressed in spinal cord astrocytes after SCI (Fig. 4a). Second, we performed a transwell assay using primary cultured astrocytes. To investigate whether macrophages attract astrocytes, we co-cultured the cells with macrophages (Fig. 4b). Co-culture with macrophages signi cantly increased the number of migrating astrocytes (Fig. 4c, d). Next, we performed three experiments to clarify the pathway by which macrophages attract astrocytes (Fig. 4e). First, we evaluated whether ADP attracts astrocytes. The number of migrating astrocytes was signi cantly increased when a medium with ADP was used, indicating that ADP attracts astrocytes (Fig. 4f, g). Second, to con rm whether the attraction of astrocytes by macrophages was due to the secretion of ADP, we performed co-culturing of macrophages and astrocytes with apyrase, an enzyme that degrades ADP (Fig. 4e). The results showed that the number of migrating astrocytes did not increase in the presence of apyrase, indicating that astrocyte migration is regulated by macrophage-derived ADP ( Fig. 4f, g). Finally, to determine whether P2Y1R is the receptor on astrocytes on which ADP acts, we cocultured astrocytes and macrophages with MRS-2179, an antagonist of P2Y1R. With MRS-2179, the number of migrating astrocytes did not increase, indicating that astrocytes are attracted to macrophages by the ADP-P2Y1R pathway (Fig. 4f, g). Since ATP secreted from macrophages is degraded to ADP, macrophages would attract astrocytes via the ADP-P2Y1R axis, resulting in astrocyte migration toward the macrophages in the epicenter and astroglial scar formation after SCI.

Discussion
In this study, we showed that the macrophages are essential for astrocyte migration using bone marrow chimeric mice. Impairment of macrophage migration also impaired the migration of astrocytes toward the center of injury. As a result, astrocytes became more widely distributed and the scar area increased with the impaired migration of macrophages. Socs3 −/− astrocytes, whose migration would be promoted with the migration of WT macrophages, did not migrate well toward the center of injury with impaired macrophage migration. These results indicate that macrophages play a leading role over astrocytes in the cell migration of macrophages and astrocytes after spinal cord injury. In addition, the co-culture of macrophages and astrocytes also showed that macrophages attract astrocytes. Using antagonists, we further elucidated that ADP derived from ATP secreted extracellularly by macrophages attracted astrocytes via P2Y1R. These ndings indicated that macrophages affect the direction of astrocyte migration and are important clues to clarifying the pathophysiology of spinal cord injury.
In CNS injury, various cells interact with each other to produce dynamic changes. For example, microglia, neurons, and astrocytes have been reported to interact via P2Y receptors 13,15 . However, the in uence of macrophages on astrocytes is not known. In this study, the great in uence of macrophages on astrocyte migration, which occurs during the acute to subacute phase, was clari ed. Since macrophages comprise the majority of in ammatory cells in the spinal cord scar and astrocytes are the major component of the glial scar, clarifying the interaction between macrophages and astrocytes is essential for the elucidation of the pathophysiology of SCI.
In addition, given that the glial scar, once formed in the subacute phase, is maintained into the chronic phase, the macrophage-derived ADP-astrocytic P2Y1R axis may also be involved in the mechanisms of glial scar maintenance 16 . We previously reported that scar forming astrocytes maintain the glial scar by changing the surrounding naïve astrocytes to scar-forming astrocytes 16 . However, it is not known how astrocytes are recruited to the scar. With regard to scar maintenance in the chronic phase, astrocytes may have been recruited to scar tissue via the macrophage-derived ADP-astrocyte P2Y1R axis.
In this study, we showed a mechanism in which macrophages attract astrocytes to the center of injury. Considering that astrocytes are arranged in a row at the outer edge of the scar and are rarely present inside the glial scar, astrocyte migration is not regulated only by the concentration gradient of ADP; some factor inside the scar stops astrocyte migration. However, the mechanism that stops the migration of astrocytes at the glial scar has not been elucidated. For example, PARP1 may be a candidate for the negative regulation of astrocyte migration. Stat3 is known to be essential for astrocyte migration 3 , and PARP1 has been reported to interact directly inhibit Stat3 phosphorylation by causing poly-ADPribosylation of Stat3 17 . The expression of PARP1 was upregulated in the spinal cord after SCI in comparison to the naïve spinal cord, as was previously shown in our RNA-seq study 18 . This result is consistent with the hypothesis that PARP1 negatively regulates astrocyte migration in SCI.
In conclusion, we demonstrated that the impairment of macrophage migration led to widespread astrocyte migration. Macrophages secrete ATP that is degraded to ADP, and astrocytes migrate toward macrophages via the ADP-P2Y1R pathway. Our ndings provide deeper insight into the interaction between astrocytes and macrophages, and suggest a potential therapeutic target for SCI.

Animals
All study protocols involving mice were approved by the Committee of Ethics on Animal Experimentation of our institution and were conducted in accordance with the National Institutes of Health guidelines for the care and use of animals. All mice were housed in a temperature and humidity-controlled environment on a 12-hour light-dark cycle, and had ad libitum access to food and water.
All animal studies were approved by the Committee of Ethics on Animal Experimentation of our institute and conducted in accordance with ARRIVE guidelines (https://arriveguidelines.org). Every effort was made to reduce the number of animals used and to minimize their suffering.

Spinal Cord Injury
Mice were anesthetized with pentobarbital (75 mg/kg intraperitoneally) and subjected to a contusion injury (70 kilodynes) at the 10th thoracic level using an In nite Horizons Impactor (Precision Systems Instrumentation, Lexington, KY) 21 . After the injury, the overlying muscles were sutured, and the skin was closed with wound clips. During the period of recovery from anesthesia, the animals were placed in a temperature-controlled chamber until thermoregulation was re-established. The motor function was evaluated using a locomotor open-eld rating scale, the BMS 22 .

Primary Astrocyte Cultures
Puri ed primary astrocyte cultures were prepared from C57BL/6J mice, as described previously 3,23 . In brief, after removal of the meninges, on postnatal day 2, mouse brain tissues were minced and incubated in a rocking water bath at 37°C for 30 min in DMEM (08456-36, Nacalai Tesque Kyoto, Japan) in the presence of 0.25% trypsin (Nacalai Tesque Kyoto, Japan) and 4 mg/ml DNase I (Sigma, Saint Louis, MO). The dissociated cells were triturated with 0.25% FBS and centrifuged at 300 ×g for 3 min. Following dilution with an astrocyte-speci c medium: DMEM containing 10% FBS (Life Technologies, Carlsbad, CA) and 1% penicillin-streptomycin (Nacalai Tesque Kyoto, Japan), the cells were plated on Poly-L-Lysin coated T75 ask. After 7-10 days in a humidi ed CO 2 incubator at 37°C, the T75 ask was set up on an orbital shaker to remove microglia at 180 rpm for 30 min. We added 20 ml of fresh astrocyte culture medium and then shook the ask at 240 rpm for 6 h to remove oligodendrocyte precursor cells. Astrocytes were detached from T75 asks using 0.25% trypsin and were used for experiments.

Quantitative Reverse Transcription Polymerase Chain Reaction (Rtpcr)
Total RNA was isolated from the astrocytes obtained from spinal cord tissue using the RNeasy Mini kit (Qiagen, Venlo, the Netherlands). cDNA was synthesized from the total RNA using PrimeScript Reverse Transcriptase (Takara, Tokyo, Japan) according to the manufacturer's instructions. RT-qPCR was performed using primers speci c to the genes of interest (Table 1) and a SYBR Premix Dimmer-Eraser (RR091A; Takara Bio, Shiga, Japan). Data were normalized to the level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Real-time PCR was conducted using a CFX Connect Real-Time PCR Detection System (Bio-Rad, Hercules, CA).

Transwell Assay
The transwell assay was performed as described previously 25

Declarations Ethics approval
Ethical approval was obtained from the ethical review committee of Kyushu University Graduate School of Medical Sciences (A22-336-0) in accordance with the provisions of the institution's Regulation for Animal Experiments.

Consent for publication
All listed authors consented to the submission of this manuscript, and all data were used with the consent of the individual who generated the data.

Data Availability
The datasets generated and/or analyzed in the current study are available upon request from the corresponding author.