Dihydroartemisinin Shows Promising Effects in the Treatment of Experimental Autoimmune Encephalomyelitis and Maintains Inammatory Homeostasis by Targeting AXL in Microglia

Background: During EAE progression, the endogenous mechanisms mediating nervous autoimmune inammation balance, as represented by AXL, were proved to be pathologically disturbed, immune balance and axon repair. Therapeutically, by activating AXL signaling, the inammatory rebalance from promotion to resolution has attracted increasing attention and showed advantages in autoimmune disease treatment. Previous studies implied that DHA had potential effects in treating autoimmune diseases. However, the detailed mechanisms in inammation regulation, especially in CNS, remain unclear. Methods: C57BL/6 mice were immunized with MOG 35-55 and treated daily with DHA. Then clinical scores, pathology, and ethology features of EAE were assessed through histological staining (H&E, LFB staining), TEM and gait analysis. Moreover, DHA-responsive cells and genes were screened by 10x Genomics. The immunological responses to DHA were measured by ow cytometry and uorescence microscope in BV2 cells. The concentrations and bio-activities of chemokines were respectively evaluated through ELISA and trans-well assay. Results: After DHA treatment, the clinical scores and body weight were signicantly improved. Histologically, mice showed slighter spinal cord lesion, less inammatory cuffs. By using gait analysis, DHA obviously improved physical coordination. 10x Genomics demonstrated that DHA selectively upregulated AXL expression in microglia. Immunologically, by enhancing AXL signaling, the phagocytic and chemotactic potential of microglia and the Treg differentiation followed by upregulating PDL1 were signicantly inuenced by DHA. Conversely, specic blocking of AXL by SGI7079 was sucient to reverse above-mentioned functions. Molecularly, DHA specically rebalanced the overactivated inammation through STAT1:SOCS3: AXL: IFNAR pathway. The present study highlighted the central role of AXL signaling in DHA mediated inammatory normalized ACTB


Background
Featuring the unresolved in ammation in central nervous system (CNS) and persistent demyelination, multiple sclerosis(MS) is the most prevalent and, to date, incurable chronic autoimmune disease, which in icts more than 2 million people worldwide [1]. Sharing the similar pathological features and progression pattern with MS, EAE is the most widely-used mouse model for the research and development of anti-MS drugs [2]. Therapeutically, as represented by methylprednisolone, the exogenous suppression of in ammation was chosen as the limited strategy and widely applied for MS treatments, leaving severe side-effects [3].
Hopefully, leading by the in ammation transition and balancing theory, candidates capable of mediating in ammation resolution have attracted increasing attention in recent autoimmune disease treatment. Microglia, resident CNS immune cells of monocyte lineage, participate in both exacerbating and resolving limited in uence on T cells. [26]. However, the detailed mechanisms of DHA in in ammation resolution, especially in MS, remain poorly understood.
In our study, in EAE, we investigated whether AXL acts as a therapeutic target in switching of in ammation transition to protect CNS against in ammatory injury. Pharmacologically, by enhancing AXL signaling through STAT1:SOCS3 pathway in microglia, we detected how DHA regulates CNS in ammation from promotion to inhibition.

Animal experiment and Reagents
Female C57BL/6 mice were obtained from institute of basic theory of Chinese medicine, China academy of Chinese medicine, and were used at 8 to 10 weeks of age. All mice were housed under speci c pathogen-free conditions. All experiments were performed according to the institutional ethical guidelines 10 or 20 mg.kg -1 ) and Methylprednisolone (P zer; 1mg.kg -1 ) were suspended in 0.5% (w/v) carboxymethylcellulose solution and orally administered via gavage once daily. Clinical assessment of EAE was performed daily and mice were scored for disease according to the following criteria: 0, no overt signs of disease; 1, limp tail or hind limb weakness but not both; 2, limp tail and hind limb weakness; 3, partial hind limb paralysis; 4, complete hind limb paralysis; 5, moribund state or dead.

Cell Culture
All cells were maintained in culture medium supplemented with 10% fetal bovine serum (Gibco, USA) at 37 °C in a humidi ed incubator (5% CO2) (SANYO, JAPAN), with Jurkat T cells in RPMI-1640 medium and PC12 cells and BV 2 cells in DMEM medium.

Histopathology
To assess the degree of CNS in ammation and demyelination, C57BL/6 mice treated with vehicle or DHA were anesthetized by pentobarbital sodium and perfused by intracardiac injection of PBS containing 4% paraformaldehyde. Para n embedded 5 mm sections of spinal cord were stained with H&E and Luxol Fast Blue (LFB) and then examined by light microscopy. Lumbar spinal cords were sliced into 1mm sections. Tin sections were cut, stained with uranyl acetate and lead citrate and then analyzed under transmission electron microscope (Olympus, JAPAN) . Brie y, in ammation was scored as follows: 0, none; 1, a few in ammatory cells; 2, organization of perivascular in ltrates; 3, increasing severity of perivascular cu ng with extension into adjacent tissue. Demyelination was scored as follows: 0, none; 1, rare foci; 2, a few areas of demyelination; 3, large (con uent) areas of demyelination.

Gait analysis
Gait assessment was carried out as described previously [27]. on the 24th day when the EAE mice were at the peak period by the automated computer-assisted method (Xin Hai Hua Yi Instrument Co., Beijing China). Data were collected and analyzed with Gait Analysis Lab software version 5.0. The equipment was located under natural light in a silent room. In brief, the system consists of an elevated 1.2 m-long glass plate which is illuminated with a uorescent light coming from the side and the uorescent light is internally re ected in the glass allowing the paws to re ect light as they come into contact with the glass oor. A ceiling on top of the walkway creates a red background to produce the contour of the animal. A high-speed camera (100 frames) underneath the glass plate captures the images which are subsequently analyzed by the connected computer program. The video acquisition system is sealed with a PVC sheet to ensure a uniform dark environment to insure controlled lighting in the experiment. Prior to the rst testing day, the animals were trained to traverse a glass walkway toward their home cage. On subsequent training days, three complete runs across the walkway were recorded by a camera positioned below. If an animal failed to complete a run within 5 s, walked backwards, or reared during the run, the animal was given an additional re-run. Pixels below a light intensity of 20 units on a 0-255 arbitrary scale were ltered out. Prints can be inspected individually and in combinations, and timing diagrams for paw placements are available.

Isolation of mononuclear cell in ltrated in spinal cord
Spinal cord in ltrated mononuclear cells were isolated as described previously. The mice were anaesthetized with pentobarbital sodium and perfused with 20 ml of cold PBS. The spinal cords were extruded by ushing the vertebral canal with PBS and rinsed in PBS. Tissues were forced through 70 mM nylon cell strainers (BD Falcon), and then the spinal cord cell suspensions were incubated with collagenase IV (2 mg.ml -1 ) and DNaseI (100μg.ml -1 ) at 37℃ for 30 min, and passed again through 70 mM nylon cell strainers to yield single-cell suspensions. CNS mononuclear cells were centrifuged (500g) at room temperature for 20 min over discontinuous 30%/ 70% Percoll gradient (GE Healthcare). Then, lysed by TRIzol Reagent (Invitrogen) for qRT-PCR analyze and loading buffer for western blot analyze.

Phagocytosis assay
Phagocytosis assays was assessed as described previously [28], PC12 cells were induced to an apoptotic state by serum deprivation for 48 hours after being cultured. BV2 cells were plated at a cell density of 2 × 10 5 cells. After induction of apoptosis, PC12 cells were harvested, stained with carboxy uorescein diacetate succinimide ester (CFSE) and placed in contact with treated BV2 cells at a 3/1 ratio for 2 hours so phagocytosis would occur. After coculturing, the supernatant was removed, and BV2 cells were harvested and stained with anti-F4/80-PE conjugated anti-mouse antibody diluted in PBS (1/200) for 30 minutes at 4℃. After the incubation period, the cells taken to an ow cytometer and a total of 10 000 events were acquired. Double-stained events were considered as phagocytosing BV2 cells, and percentage of such events was used to quantify phagocytosis. Then used the uorescence microscope for qualitative phagocytosis analysis.
Sample preparation for sequencing The brain and spinal cords were extruded by ushing the vertebral canal with PBS and rinsed in PBS.
Tissues were forced through 40 mM nylon cell strainers (BD Falcon), and then the spinal cord cell suspensions were incubated with collagenase (1 mg.ml -1 ) DNaseI (100μg.ml -1 ) at 37℃ for 60 min, and passed again through 40 mM nylon cell strainers to yield single-cell suspensions. To ensure the cell viability more than 90% by using the trypan blue.
10x Genomics single cell sequence The single cell sequencing was assessed as described previously [29] and implemented by oebiotech. Cells were processed using the 10x Genomics Chromium Controller and the Chromium Single Cell 5′ Library & Gel Bead Kit (PN 1000006) following the standard manufacturer's protocols (https://tinyurl.com/y96l7lns). Three technical replicates were run in parallel for each sample. In brief, between 14,000 and 21,000 live cells were loaded onto the Chromium controller in an effort to recover between 10,000 and 15,000 cells for library preparation and sequencing. Gel beads were prepared according to standard manufacturer's protocols. Oil partitions of single-cell + oligo coated gel beads (GEMs) were captured and reverse transcription was performed, resulting in cDNA tagged with a cell barcode and unique molecular index. Next, GEMs were broken and cDNA was ampli ed and quanti ed using an Agilent Bioanalyzer High Sensitivity chip (Agilent Technologies).

Real-time PCR Assay
Total RNA was isolated using Trizol reagent (Invitrogen), reverse transcribed, and polymerase chain reaction ampli ed using speci c primers. three-Step Real-time PCR was performed with SYBR Green PCR Reagents (CWbio, China) was used with a total of 100 ng of RNA per reaction with PDL1, AXL, GAS6, RANTES, and ACTB (sequence was shown in supplementary data 1) primers according to the manufacturer's instructions. All results were normalized to ACTB as an internal control. The reactions were aliquoted in triplicate in an optical 96 well plate; RNase-free water was used as a blank. Reactions were run in the LightCycler 480 , USA real-time PCR machine (Roche) with the following cycles: 95°C for 10 min to activate the DNA polymerase, followed by 45 cycles at 53°C for 15 seconds, and 72°C for 1 min. The real-time quanti cation was monitored directly by the StepOnePlus software and the comparative thresholds were identi ed for each gene with each RNA sample and calculated at the end of Protein extraction and Western Blot analysis BV2 cells were lysed using lysis buffer containing protease inhibitor phenylmethylsulfonyl uoride (PMSF). Protein concentrations were measured using a bicinchoninic acid assay kit. Total sample was separated by 10 % sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then blotted to polyvinylidene uoride (PVDF) membranes (Merck Millipore, IPVH00010). Membranes were blocked for 2 h at room temperature with 5 % bovine serum albumin (BSA). Next, incubate with primary antibodies overnight at 4°C. Horseradish peroxidase-conjugated antibodies against mouse and rabbit were used as secondary antibodies. After extensive washing, blots were developed with an enhanced chemiluminescent plus assay kit (Thermo Scienti c), developed on X-ray lm, and analyzed by Image J software (National Institutes of Health, Bethesda, MD, USA).

Flow cytometry
For cell surface staining, cells rst blocked with 3% FBS for 15 min in room temperature and then were stained with antibodies anti-PDL1-FITC (Proteintec, USA) for 1h 4 ℃ to analyze whether DHA in uence the PDL1 expression on microglia and macrophages. For intracellular staining, cells were rst used Triton-100 for 15 min in room temperature, and then used anti-FOXP3-FITC (Proteintec, USA) for 1 h 4 ℃ to analyze whether DHA promote T cell differentiation into Treg cells. Finally, all cells were sorted on the BECKMAN Flow Cytometer.

Trans-well assay
A trans-well (8μm pore size; Falcon, USA) assay was used to further analyze chemotaxis ability of BV2 cells according to the manufacturer's instructions. BV2 cells at the density of 5 × 10 4 /mL were placed in the upper chambers in 0.3ml 5% serum media, and the lower chambers were added with 0.6 mL LPS 50 ng.mL -1 , 1.0 μM DHA+LPS 50 ng.mL -1 , 4.0μM DHA+ LPS 5050 ng.mL -1 , or 10% serum DMEM as negative control. Cells were allowed to transmigrate into the lower chamber following incubation for 12 h at 37°C; nonmigrating cells on the upper chamber surface were removed. The membranes were xed with 4% paraformaldehyde in PBS for 30 min and stained with crystal violet in PBS for 15 min. The representative images (magni cation, ×200) were randomly taken by an inverted microscope. Each experimental group was repeated three times.

ELISA for CCL5 detection
Serum and cerebrospinal uid were taken at the day 26 and culture supernatants from BV2 cells were evaluated using ELISA kit (all from proteintech, USA) according to the manufacturer's instructions.

Statistical analysis
Data was analyzed using GraphPad Prism 7 software (GraphPad, La Jolla, CA, USA), and was presented as the mean ± SEM. Signi cant differences in comparing multiple groups, data were analyzed by one-way ANOVA test. All other statistical comparisons were done using nonparametric statistical tests. Differences with p values of less than 0.05 were considered signi cant.

DHA is a promising candidate showing some therapeutic advantages in EAE
Firstly, we identi ed the therapeutic effect of DHA on the development and severity of mice with EAE. The onset of clinical signs began on the 4th day postimmunization. As time went on, in EAE group, all mice had developed clinical symptoms and the mean clinical score was 3.92 ± 0.25 on the 18th day at the peak of the disease, also accompanied by paralysis of all four limb. In contrast, the mean clinical score in the DHA 10 mg.kg − 1 treated group (1.98 ± 0.21) was signi cantly reduced to nearly two-fold. Strikingly, the disease score was even less than one-fold than treatment with MET (2.84 ± 0.19), the most widely used drug for MS in the clinical treatment (Fig. 1A). Besides, DHA treatment obviously prevented the body weight loss in EAE mice (Fig. 1B). These results predicted that DHA had therapeutic potential in preventing EAE.
Histologically, to make further effort to DHA e cacy veri cation, spinal cord and brain tissue sections were prepared and then H&E staining was performed. Results in Fig. 1C and D showed that DHA signi cantly inhibited the pathological in ltration of in ammatory cells. Then, to histologically detect the protective effect on myelin, we used Luxol Fast Blue (LFB) staining, a speci c myelin detection method. In EAE group, result in Fig. 1E showed the intensity of LFB staining in corpus callosum was weakened, most myelin lipids were pale blue, and blue laments were hardly visible at the edge of spinal cord which can be seen as marker of demyelination. By contrast, DHA treatment remarkably prevented the demyelination.
Morphologically, we directly observed the ultramicroscopic structure of pathological changes by using transmission electron microscopic. Result in Fig. 1F showed an "orbital like" compact lamellar structure with no shrinkage, tightly arranged myelin sheath in negative control group. The impaired axons and inner vacuoles were also rarely observed. On the contrary, in EAE group, there were obvious pathological changes in the myelin sheath, the shape of myelin sheath was irregular, the lamella of myelin sheath was obviously loose, disintegrated, fused, or even lost. Consistent with previous results, the damages in myelin, as quanti ed by demyelinating or intact areas of nerve bers, could be signi cantly prevented after DHA 10 mg.kg − 1 treatment, which even showed more protective effects than MET group.
Of note, by using gait analysis, we for the rst time veri ed the e cacy of DHA. Results were shown in Fig. 1G. DHA treatment can obviously reverse the disorder of walk speed, Lh-Rh Pressure, Lh-Rh support time, Lh-Rh Print area, and ameliorate Body Rotation Average. Furthermore, it showed more therapeutic effects than MET in the perspective of walk speed and Lh-Rh Pressure. In conclusion, results from different aspects convinced us the therapeutic signi cance of DHA on EAE mice, which supported its future application in MS.

AXL in microglia was the potential responsive molecule of DHA in EAE
To detect the drug-responsive cell subset in 10x Genomics, we rstly used tSNE (t-Distributed Stochastic Neighbor Embedding) projection. Then, we used Louvain optimization algorithm to cluster and classify the cell groups. The responsive cell-types to DHA can nally be categorized into 14 sub-populations.
Among them, Macrophage and microglia were for the rst time identi ed as the most responsive cells and illustrated by graph-based clustering ( Fig. 2A).
Furthermore, we also geonomy-widely provided the description for the DHA responsive molecular network and screened the speci c marker genes expression in microglia. Result indicated that AXL was speci cally induced by DHA in microglia from EAE mice. Pathologically, gene de ciency of AXL in EAE would cause much more excessive autoimmune in ammatory reaction, consequently inducing myelin sheath damage [30]. This study gave us great inspiration and reliable evidence that AXL may be a potential target for DHA treatment (Fig. 2B).

The in ammatory-resolving effect of DHA is functionally correlated with AXL
To immunologically elucidate DHA mediated resolution of autoimmune in ammation, the AXLconcentrated e cacy analysis was functionally conducted mainly through three aspects. These included the regulation of phagocytic and chemotactic potential in microglia as well as the regulation in coregulatory signaling during antigen-presentation.
Firstly, to identify the prediction from 10x Genomics, we quanti ed the transcription level of AXL in mononuclear cells (MNCS) collected from CNS tissues in EAE mice, whose total number re ects the degree of in ltrated in ammatory cells [21]. As shown in Fig. 3A, by qRT-PCR assay, DHA treatment can up-regulate the expression of AXL compared with the vehicle control group. In contrast, GAS6, the ligand for AXL, showed no differences among each group (Fig. 3B). Consistent with the transcriptional result, the protein expression of AXL was further detected. Results showed that ( Fig. 3C and D), speci cally under in ammatory condition, DHA can up-regulate AXL compared with modeling group.
As indicated by previous report, relying on AXL, microglia can sense and engulf apoptotic cells, clear the myelin debris in CNS. Facilitating by such mechanism, the over-activated in ammation can be rapidly resolved and effectively con ned within a small area. We therefore, functionally evaluated the impact of DHA on phagocytosis of microglia in vitro. In this study, the 5-Fu induced apoptotic PC12 cells (the apoptosis rate: more than 81%, Fig. 3E) were co-cultured with BV2 cells and the phagocytotic intensities were quanti ed by ow cytometry and further visualized by microscopic observation. As shown in Fig. 3F and G, compared with LPS group, the count and percentage of PC12 cells phagocytized by BV2 cells in DHA (1.0/4.0 µM) groups were both signi cantly increased. This result proved that the phagocytic ability of microglia under in ammation condition can be signi cantly enhanced by DHA, forming the important basis for in ammation resolution.
In the next study, from the perspective of "antigen uptake and presentation", we aim to reveal the in uence of DHA on APCs. Inspired by the essential roles of co-regulatory factors during EAE progression, detection of PDL1 and CD80 were chosen as the markers for the immunological properties during antigen presentation. As shown in Fig. 4A and B, in both Raw264.7 cells and peritoneal macrophages, DHA obviously increased the transcription of PDL1 compared with LPS treated group. However, there are no changes in the transcription of CD80 between DHA and LPS treated groups. Consistently, Flow cytometry analysis showed that the enhanced surface expression of PDL1 can be clearly detected in DHA treated macrophages as well as BV2 microglia cell lines (Fig. 4C, D and E). These results provided molecular indications that, by upregulating PDL1 in microglia, DHA showed potential for enhancing the co-inhibitory signaling during antigen presentation and neutralizing the excessive in ammatory signals in the progression of autoimmune diseases.
To further detect whether DHA directly in uences the differentiation of T cells, we established a co-culture model of BV2-Jurkat T cells. Results in Fig. 4F showed that DHA (1.0/4.0 µM) signi cantly elevated the percentage of Treg cells (CD4 + Foxp3+), compared with LPS group. These results showed that DHA suppressed autoimmune in ammation partially through enhancing of PDL1 expression and promoting the differentiation of Treg cells.
According to the indication from the 10x Genomics, CCL5 is one of the most responsive molecules in DHA treatment (Fig. 5A and B). Inspired by the crucial role of CCL5 in the tra cking of encephalitogenic T cells, we rstly analyzed the level of CCL5 by using the qRT-PCR and ELISA assays (Fig. 5C). In line with the geonomy-wide screening, DHA treatment can obviously inhibit the gene expression in MNCS separated from EAE mice. Besides, the high level of CCL5 in the cerebrospinal uid (CSF) of EAE mice can also be signi cantly reversed by DHA treatment (Fig. 5D). which shows the same pattern in the serum (Fig. 5E).
To functionally determine the in uence of DHA on the chemoattraction of microglia, we used LPS to induce the production of CCL5 for 12 hours. Next, the supernatant was collected for the detection of CCL5 concentration by using ELISA assay. The results in Fig. 5F showed that DHA treatment can obviously reduce LPS-stimulated CCL5 secretion. Trans-well assay showed that DHA (1.0/4.0 µM) dramatically reduced the chemotactic ability of BV2 cells in LPS treated model (Fig. 5G).

The in ammatory-resolving effects of DHA necessitated the AXL signaling in microglia
Molecularly, in order to clarify whether the inhibition of autoimmune in ammation of DHA is required for the presence of AXL. We used SGI7079, the phosphorylation blocker of AXL at Tyr702. Figure 6A demonstrated that challenging with SGI7079 successfully blocked phosphorylation of AXL and the related regulation on STAT1:SOCS3 pathway. In addition, Fig. 6 also implied SGI7079 is su cient to neutralize the in ammation resolving effects of DHA in microglia. The T cell differentiation (Fig. 6B, C), phagocytic potential (Fig. 6D, E) and chemotactic sensitivity (Fig. 6F, G) consistently showed no statistical differences among DHA, LPS and NC treated group (P > 0.05), indicating that AXL would be a potential target of DHA in treating EAE. 5. DHA speci cally rebalanced overactivated in ammation by regulating AXL related STAT1:SOCS3 pathway Large amounts of researchers have found that, speci cally under the pro-in ammatory conditions, AXL interacted with type I interferon receptor (IFNAR) and potentiated the downstream in ammatory-resolving events as represented by the activation of STAT1:SOCS3 pathway. Of great interest, as the molecular switch for in ammation from promotion to resolution, the transcription of AXL itself is inversely facilitated by STAT1, a central in ammatory sensor regulated by various types of pro-in ammatory signals including IFN-β ligation to IFNAR. This forms the endogenous basis for the negative feedback loop of immune responses and consequently maintains the homeostasis in in ammatory microenvironment (graphically shown in below, Fig. 7.).
In light of this, to molecularly clarify the mechanistic revelation of DHA to IFNAR-STAT1 pathway, in ammatory stimulator to IFNAR (LPS or more speci cally, IFN-β) were used and the in uences of DHA on this pathway were then evaluated in BV2 cells. As shown in Fig. 8, in the presence of DHA, the expression of AXL can be dramatically induced by both LPS (Fig. 8A and D) and IFN-β ( Fig. 8B and E), leading to a molecular pattern with higher in ammation-resolution potential, as suggested by SOCS3 upregulation. Moreover, in sharp contrast to the results obtained from the STAT1-avaliable state, the expression of AXL and SOCS3 were kept at the similar levels comparing to modeling group when phosphorylation of STAT1 was blocked by Fludarabine (FLU) (Fig. 8C and F). This result revealed that the enhanced AXL expression and the negative regulation of DHA in in amed microglia were both required for the normal-responsiveness of STAT1 pathway.
Next, recent ndings further proved that the in ammatory resolution effects of AXL shows high speci city to proin ammation conditions, which suggested that the biological activities of AXL were integrated into the IFN-β-IFNAR activation. To test such speci city, in the non-in amed state, the STAT1:SOCS3 pathway was analyzed in the presence of DHA. In addition, the in ammatory resolving functions mediated by AXL were phenotypically measured in DHA treated BV2 cells. As shown in Fig. 9A, without in ammatory stimulation, DHA failed to regulate STAT1-SOCS3 pathway. Functionally, the differentiation induction of Treg cells (Fig. 9B, C), phagocytic (Fig. 9D) and chemotactic in uences of BV2 cells (Fig. 9E), as aboveproved manipulated by DHA, were totally neutralized. These results indicated that DHA functions as a resolution promoter by regulating AXL related STAT1:SOCS3 pathway speci cally in the condition of overactivated in ammation.

Discussion
Previous studies indicated that there may exist close correlation between DHA and T cells differentiation in treating autoimmune in ammation, leaving large research gap for the systematic understanding about its immunological activities, especially the molecular explanation for the regulation of Th17/Treg balance [24]. In our study, inspired by cell-type based recognition and characterization, we for the rst time identi ed macrophage/microglia as the most responsive cells of DHA in autoimmune-in ammation lesion of EAE model. This nding largely expanded the traditional concept holding that the antiin ammatory effects of DHA was limited in T cells regulation.
Recent researches have proved that microglia actively participate in almost all MS processes. It serves as the major immune-active element and pioneer cells in the in amed CNS. Therefore, modulating the function of microglia can be an attractive therapeutic strategy in coping with CNS in ammation and in ammation-induced demyelination [30]. More importantly, supported by geonomy-wide sequencing and bioinformatics analysis, we for the rst time revealed that AXL was proved to be the potential responsive gene functionally executing the pro-resolution effects of DHA, as demonstrated by the elevated phagocytosis, chemotactic inhibition and co-inhibitory signaling enhancement in microglia from EAE model. This result provided indicative evidences for the deeper understanding of DHA in activating the endogenous self-limitation mechanism during the progression of autoimmune in ammation.
Furthermore, it also revealed a theoretical explanation for the DHA-mediated Treg differentiation as reported by previous researchers.
Relying on the regulatory effect in STAT1:SOCS3 signaling and the direct interaction with IFNAR, AXL was identi ed as a switch in mediating in ammatory transition from promotion to resolution and a crucial maintainer for immune homeostasis. Pathologically, the functional defect of AXL and its abnormal regulation in STAT1-transduced signaling cascades form part of the molecular basis attributing to autoimmune disease progression, especially in MS [18]. In our study, by combining genome-wide screening with molecular biological analysis, we obtained convincing evidence in targeting AXL by DHA. In clear contrast to MET, the most commonly-used drugs for MS in clinic, our research demonstrated that DHA may not be a unidirectional and immoderate immunosuppressant. Instead, DHA is a potential in ammation-conditioned rebalancer, featuring "No in ammation, No inhibition" properties. Such hypothesis can be experimentally supported in our study. When the in ammation was pathologically over-activated, DHA may enhance the IFNAR-AXL interaction and promoted in ammation resolution through transducing signals to STAT1:SOCS3. By contrast, as the in ammation tends to be mitigated, the induction of AXL and activation of SOCS3 were weakened due to the dissociation between INFAR-AXL and the deactivation of IFNAR: STAT1 pathway, which do not further induce sustained over-suppression of the immune response (as proved by Fig. 9).
Hopefully, this mechanistic feature initially indicated the advantages of DHA in treating EAE, suggesting that DHA may be a promising candidate in glucocorticoids-alternative therapeutic strategy. By DHA application, it may also be helpful to avoid the side effects of glucocorticoid, including prednisone, methylprednisolone, and betamethasone, causing excessive immune suppression. To further reveal the therapeutic signi cance of DHA, the speci c mechanisms and validation of pharmacodynamic dependence of DHA and AXL would become the key research project in our future study.

Conclusions
We for the rst time highlighted that, by enhancing AXL signaling, DHA e ciently blocked the in ammatory initiation and expansion in EAE as proved by the reduced chemoattraction, the induction of Treg differentiation and the enhanced efferocytosis of microglia, contributing to autoimmune homeostasis. Our study pharmacologically indicated that, speci cally when in ammation was overactivated, DHA may be a promising candidate for the treatment of EAE. We also proved DHA may show more favorable e cacy comparing with glucocorticoids in MS treatment.

Competing interests
The funding agency had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. The authors have no nancial or non-nancial competing interests to declare.     The in ammatory -resolving effects of DHA is functionally correlated with AXL Peritoneal macrophages from C57BL/6 mice were challenged by 3% sodium thioglycolate three days in advance. A) RT-PCR analysis for PDL1 and CD80 transcription in Raw 264.7 cells and peritoneal macrophages which were treated by LPS (50ng.ml-1) and DHA (1.0/4.0μM). B, C, D) Flow cytometry assay for PDL1 expression in Raw 264.7 cells, peritoneal macrophages (PDL1+/F480+) and BV2 cells by LPS (50ng.ml-1) and DHA  The in ammatory -resolving effects of DHA is functionally correlated with AXL A, B) CNV analysis showed that CCL5 were almost expressed in macrophages and microglia (red, cells with detected CNVs; gray, no detected CNVs). MNCS from EAE, MET(1mg.kg-1) and DHA (2, 10, 20 mg.kg-1) mice were harvested by 30/70% percoll gradient separation. C) qRT-PCR assay for CCL5 were analyzed by LightCycler 480 Roche. D, E) CSF and serum were obtained from EAE, MET(1mg.kg-1) and DHA (2, 10, 20 mg.kg-1) mice, ELISA assay for CCL5 analysis were carried out by using the CSF and serum. The