Single‐Cell Sequencing Reveals the Optimal Time Window for Anti‐Inflammatory Treatment in Spinal Cord Injury

Though the occurrence of neuroinflammation after spinal cord injury (SCI) is essential for antigen clearance and tissue repair, excessive inflammation results in cell death and axon dieback. The effect of anti‐inflammatory medicine used in clinical treatment remains debatable owing to the inappropriate therapeutic schedule that does not align with the biological process of immune reaction. A better understanding of the immunity process is critical to promote effective anti‐inflammatory therapeutics. However, cellular heterogeneity, which results in complex cellular functions, is a major challenge. This study performs single‐cell RNA sequencing by profiling the tissue proximity to the injury site at different time points after SCI. Depending on the analysis of single‐cell data and histochemistry observation, an appropriate time window for anti‐inflammatory medicine treatment is proposed. This work also verifies the mechanism of typical anti‐inflammatory medicine methylprednisolone sodium succinate (MPSS), which is found attributable to the activation inhibition of cells with pro‐inflammatory phenotype through the downregulation of pathways such as TNF, IL2, and MIF. These pathways can also be provided as targets for anti‐inflammatory treatment. Collectively, this work provides a therapeutic schedule of 1–3 dpi (days post injury) to argue against classical early pulse therapy and provides some pathways for target therapy in the future.


Introduction
Severe spinal cord injury (SCI) leads to devastating irreversible neurological disabilities.A profound neuroinflammatory response is unleashed to remove the tissue debris and prepare a spinal cord repair after SCI.[3] In recent clinical treatments, anti-inflammatory drugs such as riluzole (RLZ) and methylprednisolone sodium succinate (MPSS) have been proposed as neuroprotective strategies at the early stage after SCI with a function of remodeling the toxic environment in SCI. [4]However, the clinical use remains debatable owing to the controversial results according to different injury types and administration time points.For example, in a previous study, although the neuroprotective function of MPSS was reported and regarded as the "standard of care" for acute SCI at times, different MPSS treatments existing in 24-or 48-h protocols present significant differences and the functional gain remains questionable. [5]urthermore, the use of RLZ is not fully effective in preventing early neuronal loss (3 h) after SCI and might confer benefits at a later time. [6]Altogether, a reasonable time window for clinical trials should be determined depending on the biological process of SCI and the mechanism of the medicine.
Inflammatory response activated after injury protects from pathogens and promotes tissue recovery to maintain the homeostasis of cells.However, excessive inflammation can cause a pathological mark in the immune system, resulting in the dysregulation of cells and chronic diseases. [7,8]The dual role of inflammation response also reflects on the varied function of immune cells, such as microglia, macrophages, and astrocytes.Typically, activated microglia and macrophages are shown to play a dual role in both pro-and anti-inflammation.In a study, the activation of microglia secreted pro-inflammatory cytokines, resulting in further damage, and diminished microglia activation markedly reduced pathological tissue size. [9]However, depletion of microglia at the early phase (1 day) exacerbated the secretion of inflammatory mediators and caused neuronal death. [10]acrophages are divided into two major subtypes with pro-(M1) or anti-(M2) functions in immunity reaction. [11]Upon injury, monocyte-derived macrophages adopt a pro-inflammatory phenotype that secretes cytokines such as tumor necrosis factor- (TNF-), interferon- (IFN-), and interleukin-1 (IL-1).On the other hand, anti-inflammatory molecules such as transforming growth factor- (TGF-) were secreted by a more proregenerative polarized phenotype in the following immunity process. [12]ikewise, reactive astrocytes express inflammation factors such as IL-1, which cause further cell damage. [13]However, they also take part in scar formation and microenvironment remodeling at a sub-acute phase and play a vital role in neuroprotection. [14]ased on recent research, inflammation takes on a combination of positive and negative roles depending on the various functions of immune cells at different injury phases.We hypothesized that an appropriate time window for anti-inflammatory treatment should focus on inhibiting excessive inflammation and avoiding the interruption of the neuroprotection function of the immune system.According to this hypothesis, the debated effect of recent treatment by MPSS or RLZ may be attributed to the mismatching of administration junction to excessive inflammation time point.
In this study, we aimed to reveal the process of inflammation in SCI and find an optimization time for anti-inflammatory treatment.Single-cell sequencing makes it possible to distinguish the subpopulation of cells and reveals the functions at different differentiation phases by gene expression level.We used single-cell data by profiling the tissue proximity to the injury site, where we observed different cell type compositions and chromatin profiles based on the different days post-injury (dpi) and further explored the relationship between cells.Furthermore, MPSS was used as a typical medicine to investigate the appropriate time point and mechanism of anti-inflammatory treatment in SCI.Through this investigation, we discovered that administration at 3 dpi reduced pathological tissue area and relieved excessive inflammation.Notably, we found that the inflammatory function of MPSS acted by the inhibition of pro-inflammatory cell number, including neutrophils, T cells, microglia, and the subpopulation of macrophages through pathways such as interleukin-2 (IL2), macrophage migration inhibitory factor (MIF), and TNF but showed no influence in astrocyte scar formation.In light of these results, we proposed that 1-3 dpi should be an optimal time window for anti-inflammatory treatment, and IL2, MIF, and TNF pathways are attractive therapeutic targets for excessive neuroinflammation.

Single-Cell Analysis and Histochemistry Reveals Cell Types at Different Days Post Spinal Cord Injury
The experiment process of SCI and sample harvest was shown in Figure 1A.Microglia, macrophage, and astrocyte was marked by ionized calcium binding adapter molecule 1(Iba1), Cd68, and glial fibrillary acidic protein (GFAP), then observed by confocal microscopy.From results of immunofluorescence, microglia marked by Iba1 kept increasing during 1-7 dpi along with a morphologic change from homeostatic to activated (Figure 1B,B1-B4).The number of Cd68-positive macrophages reached the peak at 3 dpi and then reduced from 3 to 7 dpi (Figure 1B,B5-B8).We performed single-cell sequencing on the first, third, and seventh day post-SCI (at 1, 3, and 7 dpi) to compare the changes in the gene expression profile.Our analysis revealed the presence of multiple cell clusters that could be fused into 10 main groups of cells.We mapped the cell clusters coming from distinct time points on UMAP plots (Figure 1C).T cells, neutrophils, macrophages, microglia, dendritic cells, fibroblasts, astrocytes, oligodendrocytes, endothelial cells, and mural cells were identified (Figures 1C and Figure S1A, Supporting Information).Neurons were excluded by our dissociation protocol.The highest differentially expressed genes (DEGs) of each cell type, which were identified using the CellMarker database, are presented in Figure S1B (Supporting Information).The cell number ratio of endothelial cells, microglia, and astrocytes decreased dramatically at 1 dpi, which was caused by cell death after SCI (Figure 1D).In contrast, inflammation-related cells, such as macrophages, T cells, neutrophils, and dendritic cells increased at 1 dpi (Figure 1D).Neutrophils, which are short-lived and provide an essential defense in acute inflammation, and dendritic cells, which present antigens to and induce the activation and differentiation of naive T lymphocytes, reached the peak at 1 dpi.Macrophages were not observed in the intact tissues and kept increasing from 1 to 3 dpi since they originated from a blood monocyte.Microglia, a type of small macrophage-like glial cells in the central nervous system (CNS), were found both in the intact and injured tissues and kept increasing from 1 to 7 dpi (Figure 1C,D).We further verified some proteins that respond to SCI at distinct time points by performing Western blotting.Proteins that promote the immune reaction, such as IL2, Ccl2, and TNF- were highly expressed at 3 dpi, while anti-inflammatory markers IL33 and TGF- barely existed at 3 dpi (Figure 1E,F).From these results, we found that T cells, neutrophils, and dendritic cells primarily contributed in 1-3 dpi, while microglia and macrophage were the main cell types that led to the following 3-7 dpi immune reaction.Furthermore, 3 dpi was the most intense period of the immune response.To verify the cell functions on different days post-SCI, GO enrichment analysis was performed and analyzed in regard to three aspects: biological process, cellular component, and molecular function.We found that immune reaction-related leukocyte regulation, cell binding, and cytokine synthesis existed during 1-7 dpi, while at 3 and 7 dpi, functions were inclined to the regulation of cell death (Figure S2, Supporting Information).

Distinct Macrophage and Microglia Subpopulation and Differentiation Trajectory Assessed by Degs and Pseudo-Time Analysis
Macrophages are a crucial cell type during the inflammation process, with diverse cell morphology and functions in response to the microenvironment remodeling process.To further understand the functions of macrophages at different dpi, we explored the subpopulations and differentiation trajectory during the sub-acute phase of SCI.As a result, the subpopulation of macrophage was divided by DEGs into seven clusters, namely, M-interferon, M-anti-inflammation, M-MHC, Mpro-inflammation, M-complement, M-ROS, and M-endocytosis (Figure 2A).At 1 dpi, macrophage migration to the spinal cord was rapidly induced (including M1-like or M2-like).Mitochondrial and phosphorylation genes' high-expression subpopulation, M-ROS, and interferon secretion subpopulation, M-interferon (Ifi205, Ifitm3, Ifit1, Ifi206, and Ifitm6), mainly existed at 1 dpi (Figure S3A, Supporting Information) and returned to baseline by 3 dpi.These macrophages were possibly involved in the regulation of activation, differentiation, transcription, and survival of immune cells. [12][17] M-anti-inflammation (Flor2, Cbr2, Mgl2, C4b, Igfbp4, Mrc1, and Cd163) [18,19] existed at 1 and 7 dpi but were rare at 3 dpi.Besides, M-pro-inflammatory express M1 markers (Ccl6, Clec4n, Inhba, and Cxcl3) [20][21][22] and M-MHC could activate T cell and generate MHC during inflammation [23] was distinguished by MHC-related genes such as H2-DMA, H2-DMb1, H2-Eb1, and H2-Aa present from1 to 7 dpi.However. the ratio of MHC subtype decreased at 3 and 7 dpi (Figure 2B).At 3 dpi, subpopulation M-endocytosis with the function of cell endocytosis (Ctsb, Ctsd, and Pasp) [24][25][26] and cell attachment (Spp1 and Gpnmb) [27] maximally increased.M-complement subpopulation, which expresses genes such as C1qa, C1qb, C1qc, Trem2, and Ypel3, [28,29] generated from 3 dpi and mainly existed at 7 dpi.It was the only subpopulation that kept a high level by 7 dpi, when other subpopulations returned to the pre-injury level.
To characterize the developmental insights of the macrophage post-SCI, we performed pseudo-time reconstruction using the Monocle package.We confirmed cells in the left branch as the origin of macrophage differentiation.One minor bifurcation, one large bifurcation, and three paths were observed in the analysis.Cells in M-pro-inflammation could be the first activated population during the pseudo-temporal ordering, sequentially followed by M-MHC, M-interferon, and M-ROS (Figure 2C).Then, M-endocytosis acted in the main way for macrophages to clear antigen-positive cells and cell debris appearing in path 2 (Figure 2C).M-complement was expressed at low levels at the starting site but had higher expression levels in path 3 to the end (Figure 2C).Therefore, our results imply that during the sub-acute phase, macrophages first started from an activation and polarization program by 1 dpi, then went through foreign materials and phagocytized debris, and finally turned to be an immune regulation process by complement.
To better understand the biological basis of the differences in microglia, we sought to determine the microglial subpopulations of individual clusters based on their DEGs.Interestingly, four clusters (G-immune cells, G-homeostatic, G-bifunctional, and Gneurotoxic) respectively existed on different days post-SCI and were identified and presented by UMAP (Figure 2D,E).For example, we found that one cluster named G-homeostatic exhibited high expression of homeostatic genes (Tmem119, P2ry12, and Siglech) [15] and constituted nearly 100% of microglia in the intact tissue.We also found a G-activated subpopulation with high expression of pro-inflammation effect (mt.Atp6, mt.Nd113, and Ccl12) [17,30] at 1 dpi and G-neurotoxic with the expression of neurotoxic molecules secretion, endocytosis, and axon injury gene (Ctsd16, Tmsb4x, and Rpl10a) [24,31,32] at 3 dpi, corresponding to cluster on the UMAP plot (Figure 2D and Figure S3B, Supporting Information).At 7 dpi, a G-bifunctional cluster was found en-riched with the expression of genes implicated in thymocyte proliferation and pro-inflammation (IL1b and IL1a) and cell death inhibition (Bcl2a1b and Bcl2a1d), consistent with a role of regulation including both neuron protection and chronic inflammation.Pseudo-time analysis was performed to define the differentiation orientation of microglia, and according to the result, the bottom branch was defined as the beginning (Figure 2F).Differentiation started with the activation of microglia, then polarized into three phenotypes, namely, G-immune (path1), which participated in pro-inflammatory action and G-bifunctional and G-neurotoxic (path2) with the regulation of cell death and inflammation by cytokine secretion (Figure 2F).Our analysis indicated that microglia at 1 and 3 dpi was mainly consistent with proinflammation and axon injury.However, at 7 dpi, microglia presented regulatory functions such as the inhibition of cell death, axon protection, and chronic inflammation.

Cell Communication Analysis Reveals that Signaling Events Anticipated in Different Cell Types Respond to Sci
Cellular communication can group signaling pathways by defining similarity measures and performing diverse learning from both functional and topological perspectives.Application of this analysis revealed five patterns of outgoing and incoming information (Figure 3A).This result suggests that the majority of incoming and outgoing signals from macrophages are characterized by mode 1, representing multiple pathways, including but not limited to TGF-, galectin, mitogen-activated protein kinase (TWEAK), IL2, and CCL.TGF- is known to be involved in the anti-inflammatory function of macrophages, and in contrast, IL2 promotes macrophage polarization to the M1 phase, a proinflammatory phenotype. [15]In addition, the TWEAK pathway is a member of the TNF group and is involved in apoptosis, fibrosis, and pro-inflammation. [33]The galectin pathway has always played multiple roles, such as in T cell regulation, tissue repair, and oxidative inactivation. [34]icroglial release signals were characterized by three patterns (Figure 3B), representing growth factor pathways (IGF), proinflammatory and apoptosis-related pathways (TNF and PROS), and macrophage colony-stimulating factor (CSF).These results suggest that macrophages and microglia simultaneously activate multiple signaling modalities and pathways, including pro-and anti-inflammatory pathways, apoptosis, and growth factor secretion, among others.At the same time, we analyzed the top 12 highly involved cellular pathways and found that the immunerelated cellular pathways CCL, IL2, and IL1 are mainly composed of macrophages, dendritic cells, T cells, microglia, and monocytes (Figure 3B).SPP1, which is also known for osteopontin (OPN), allowed neuron cells to respond to growth factors such as ciliary neurotrophic factor (CNTF) and brain-derived   neurotrophic factor (BDNF) [35] and built a cell chat network among all cell types.We found that the classic pro-inflammatory pathway, TNF, was mainly mediated by microglia, and the anti-inflammatory pathway, TGF- connected macrophages, microglia, and dendritic cells.Other pathways such as CSF, MIF, TWEAK, and VISFATIN pathways were important linkages between macrophages, microglia, fibroblasts, endothelial cells, and astrocytes.These results also indicated that macrophages and microglia play an important role in the inflammation process during SCI.
Figure 4 depicts the results of sc-RNA sequencing at distinct time points after SCI and describes an immune process during the sub-acute phase of SCI.The bone marrow origin immune cells were able to enter the injured tissue owing to blood-spinal cord barrier disruption.The immune process varies along a continuum of differentiation, proliferation, cell adhesion, cell activation, endocytosis, and neurotoxic.Macrophages and microglia with specific phenotypes and functions were polarized rapidly at 1 dpi.High-released ROS and interferon further promote necrosis and polarization.Through G-immune, M-MHC, M-pro-inflammation, and immune cells including dendritic cells, T cells, and neutrophils, macrophages provide innate immunity to remove antigen-positive cells.Furthermore, M-anti-inflammation subpopulations exhibit tissue repair properties and secrete immunosuppressive cytokines to upregulate ECM components and growth factors.The endocytosis function of macrophage was enhanced at 3 dpi and further facilitated debris removal, bacterial removal, and sterilization.Microglia at 3 dpi facilitated neurotoxin cytokine secretion and the transition of neurons toward the necrosis process.During 3-7 dpi, the proliferation of microglia occurred maximally, which is a unique proliferation event in SCI and induces scar-forming astrocytes.Microglia continued proliferation at 7 dpi and assisted scar formation and axonal dieback reduction for wound healing.Complement release began at 3 dpi in macrophages and was kept at a high level at 7 dpi.Complement is a crucial cytokine in immune regulation that can improve endocytosis and cytolysis and induce astrocyte activation, but it can also inhibit axon regeneration.Briefly, cell activation and polarization occurred before 1 dpi, followed by antigen and debris clearance at 1-3 dpi and immune regulation and scar formation at 3-7 dpi.

Optimization of the Time Window for Excessive Immunity Reaction Inhibition Mediated by MPSS
Since we noticed the process of immune reaction during 1-7 dpi, we then optimized the time window to mediate neuroinflammation for preventing excessive immunity reaction by assessment of the injury site area and immunohistochemical staining observation.MPSS or PBS solution was injected into the injury site at a series of time points, namely, 1, 3, and 7 dpi, and evaluated by immunofluorescence staining at 14 dpi.As a result, we found that animals injected at 1 and 3 dpi had a higher gray value of GFAP compared with animals injected at 7 dpi (Figure 5B,C).Although injury site and cavity formation were observed at all time points after injection, the injury area at 3 dpi injection was significantly lower than other time points after injections (Figure 5B,C), suggesting that injection early point may cause further tissue dam-age by interrupting the beneficial immunity reaction, preventing cell polarization and antigen clearance in the initial phase.In addition, higher Iba1-positive cells were found in animals injected at 7 dpi (Figure 5B,C), which suggested that excessive activation of microglia might occur before 7 dpi, causing serious damage to neurons and spared axons.These results indicated that 3 dpi could be an optimal time window for MPSS to remodel the dysregulated inflammation and keep more spare tissue.

The Mechanism For MPSS Treatment In Preventing Excessive Neuroinflammation
Above evidence has uncovered that the optimal treatment window during SCI was 3 dpi.To ascertain whether MPSS could alleviate excessive inflammatory response and protect spare tissue, we performed in situ injection with MPSS or PBS at 3 dpi.Immunohistochemical image and quantification data showed that MPSS injection resulted in a one-third injury site area the size of PBS injection, and Iba1-positive microglia number was nearly half of that in PBS injection (Figure S4B,B3,B4, Supporting Information).However, the GFAP-positive astrocyte scar at the margin of the injury site largely existed in both PBS and MPSS injection (Figures S4B,B1,B2, Supporting Information), indicating that MPSS could not prevent the activation of scar-forming astrocytes.
To further understand the mechanism of MPSS in inflammation inhibition, we sought to examine the difference in gene expression between MPSS and PBS injection through single-cell sequencing.We were able to recognize similar clusters that were identified in the previous analysis depicted in Figure 1.In terms of the cell number ratio, we found that the MPSS treatment resulted in a decrease in microglia, T cells, and neutrophils compared with the PBS group (Figure 6A).In contrast to these immune cells, oligodendrocytes and astrocytes in the MPSS group increased, reflecting the anti-inflammatory and protective function of MPSS after SCI.This result is also consistent with the immunohistochemical observation, smaller injury site area, and fewer Iba1-positive microglia in MPSS-treated animals than in PBS control animals (Figure 5B).However, for macrophage, we found almost no difference in the cell ratio number between MPSS and CTRL groups.Therefore, a further analysis was performed to verify the subpopulation and gene expression changes in macrophages between MPSS and PBS groups.According to the analysis, the macrophage subpopulation (cluster 0) with high expression of immune response regulation and interferonsecreting genes (Ifitm1, Ifitm2, Fth1, and Ftl1) [36,37] was inhibited by MPSS, while the subpopulation (cluster 2) high expressing complement-associated genes (C1qa, C1qb, and C1qc) increased in the MPSS treatment.(Figures S5A,C, Supporting Information).In contrast, we found no dramatic changes in microglia subpopulations by MPSS treatment (Figures S5B,D, Supporting Information).We then screened out four important inflammation genes that largely changed in MPSS treatment.Gene IL2 promoted inflammation, and Tnfaip2 decreased, while anti-inflammatory gene IL33 increased in MPSS injection animals, which was also confirmed by protein expression level in Western blotting (Figure 6C,D).Moreover, the expression of proinflammation genes IL2 and Tnfaip2 started at 1 dpi and reached  the peak at 3 dpi.This result also explained the reason why the optimal time point after injection should be 3 dpi (Figure S6, Supporting Information).
Cell communication between cells was also investigated, and the top 20 pathways were similar in the MPSS and CTRL groups except for the absence of TNF and calcitonin receptor (CALCR) in the latter group (Figure 6E).TNF is a pro-inflammatory pathway regulated between microglia and other cells, as depicted in Figure 3.The contributions of ligands and receptors also present a strong connection between MIF and Cd44 and Cd74 in PBS condition compared with the MPSS condition (Figure 6F).This binding of MIF and CD74 induced the formation of a complex with CD44 and was found to be essential for initiation of a signaling cascade required for immune cell maintenance. [38]And the binding of angptl4 to receptors may contribute in tubule formation of endothelial cells and reducing vascular leakage. [39]ccording to these results, we found that the MPSS treatment could inhibit excessive inflammation and protect spared tissue by reducing active microglia number and regulating macrophage subpopulation.Furthermore, the anti-inflammatory function of MPSS could attribute to the downregulation of immune pathways TNF, MIF, and IL2.These pathways were also screened as important pathways in cell communication between immunity cells, especially T cells, microglia, and macrophages in SCI tissues (Figure 3A).

Discussion
Based on our findings, we recommend a window for antiinflammatory treatment between 1 and 3 dpi for the following reasons: i) An anti-inflammation treatment before or at 1 dpi may interfere the process of antigen clearance, polarization, and anti-inflammation.Although, antigen clearance process maintained up to 7 dpi, the ratio of M-MHC subtype decreased after 1 dpi.Therefore, the interference with antigen clearance by antiinflammation treatment at 3 dpi should be weaker than at 1 dpi.ii) Cytokine expression, such as IL and TNF was highest in 3 dpi and may decrease in following 1-2 days.However, it existed during the immune reaction, [40,41] thus the optimized window for treatment should be between 1-3 dpi.iii) Endocytosis subtype macrophage and neurotoxic subtype of microglia which mediate cell necrosis [42,43] mainly existed at 3 dpi.
Microglia play critical roles in the inflammatory response, and their role in SCI is heavily debated. [44,45]Our results suggest that they assume many distinctive subpopulations over time, and these subpopulation cells can exert either beneficial or detrimental effects depending on their states. [46]As we observed an advanced protection function with a reduction of activated microglia in MPSS injection at 3 dpi, we speculate that an exces-sive inflammation occurred at 3 dpi and was probably led by neurotoxicity and phagocytosis subpopulation in microglia and macrophages.Early activated microglia exerting in the recruitment of monocyte-derived-macrophage were noted to have a beneficial effect in relieving a degradative macrophage phenotype in pathology tissues. [47]However, microglia with the function of chemokine signal amplifying exacerbated the pathology and worsened the neuron function.Interestingly, the clearance of delayed microglia was reported to improve functional recovery in SCI, [48] which also indicated that early intervention with antiinflammatory treatment may cause detrimental resistance in tissue repair.Another important role for microglia is scar formation, by regulating astrocyte survival, and adhesion. [49]In our result, astrocyte scar formation was not influenced by MPSS treatment.Astrocyte scar was generally regarded as a barrier to axon regrowth, with evidence that axon regrowth failed in the presence of mature astrocytes. [50]However, recent studies have shown that contrary to the widespread view, astrocyte scar formation aids tissue repair, and CNS axon regeneration. [44,51,52]It is difficult to ascertain whether astrocyte scarring is beneficial or not.
Different from microglia, the influence of MPSS in macrophages reflects on the regulation of subpopulations instead of the total cell number, pro-inflammation subpopulation was inhibited in MPSS treatment but an increase in C1q high-expressing subpopulation was observed.C1q acts as an opsonin in the immune system, driving autocatalytic complement cascade activation.C1q was reported to anticipate various cellular functions, including cellular adhesion, clearance of cell debris, neurodegenerative pathogenesis, chemotaxis, and cellular differentiation. [53,54]However, the view on the complement function in SCI injury was not consistent.In CNS injury, complement plays a role in improving neurite outgrowth by interacting with myelin-associated glycoprotein, resulting in reduced activation of growth-inhibitory signaling in neuron. [53]Also, C1q was essential for cell phagocytosis, and deficiency of C1q led to a failure in apoptotic cell debris clearance. [55]However, the view that complement activation was detrimental to secondary injury was also proved, with evidence that the deficiency or inhibition of C1q could improve histological and functional locomotor recovery at the early stage of secondary injury. [56]This possibility of complement functions in SCI injury is worth investigating further.
According to our pathway analysis, downregulated genes belonging to TNF, IL2, and MIF pathways expressed in MPSS treatment were regarded as a mechanism of MPSS antiinflammatory treatment.Commonly, these pathways comprise elements of pro-inflammation. [57]Typically, the TNF pathway was well known for its pro-inflammatory function.Functional activity of immune cells such as dendritic cells, monocytes, www.advanced-bio.com and macrophages can be markedly increased by signaling through the TNF family.In addition, improvements in inflammation and cell apoptosis are always consistent with the upregulation of TNF in acute SCI. [57]IL2 was identified as a cytokine capable of driving the expansion of the activated T cell population and manipulating the differentiation of T cells. [58]MIF, mainly secreted by monocytes/macrophages, is described as a kind of pro-inflammatory cytokine involved in inflammation-associated pathophysiology.It forms the CD44 complex by linking to the CD74 receptor, which was reported to upregulate lymphocyte survival. [59,60][63] Studies have shown that the suppression of the NF-B pathway can promote the recovery of SCI. [64,65]ased on the understanding of pathways, we demonstrated that the function of MPSS in excessive inflammation inhibition was mainly attributed to the reduction in cell activation, such as microglia, T cells, and macrophages.Therefore, we supposed that the subpopulations G-neurotoxin and Mendocytosis, which maximally existed at 3 dpi, should be the critical cell types that induce excessive inflammation.
In summary, our results argue against the usage of antiinflammatory medicine at an early stage (several hours) of SCI.To the best of our knowledge, this is the first work to propose an appropriate time window for anti-inflammatory treatment depending on cell heterogeneity changes through the immunity process.The mechanism of typical anti-inflammation medicine MPSS found that it attributes to the activation inhibition of cells with pro-inflammation phenotype through the downregulation of pathways such as TNF, IL2, and MIF.These pathways also could be provided as targets for anti-inflammation treatment and for target therapy in the future.
Animals and Surgical Procedure: Female Sprague-Dawley rats (≈250 g) were obtained from the Zhejiang Academy of Medical Sciences.All experiments were approved by the Zhejiang University Animal Experimentation Committee and were in complete compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (ZJU20220394).After the animals were anesthetized with pentobarbital sodium (0.5 mL/100 g, injected intraperitoneally), a dorsal laminectomy was performed at the 10th thoracic vertebral level (T10-T11) to expose the dorsal surface of the spinal cord.To perform a crush injury animal model, iridectomy scissors were used to transect ventrally and laterally.After the injury, the muscles and skin were sutured.Bladder care was provided twice daily until spontaneous voiding resumed.Samples for analysis of different time points were collected at 1, 3, and 7 dpi at the injury site.For drug injection, MPSS was dissolved in PBS solution at a concentration of 10 mg mL −1 .The solution was loaded in a micropipette and injected at a rate of 200 mL min −1 at five different sites (1 μL per site) and two different depths, 0.6 and 1.2 mm at 3 dpi.For control animals, we performed an injection with PBS.Animals were sacrificed for analysis two weeks post-injury.
Immunohistochemistry and Western Blotting: To obtain tissue section for immunohistochemistry, 4% paraformaldehyde in 0.1 m phosphate buffer (pH 7.2-7.4)was used in animal perfusion.Then spinal cords were dissected and fixed in 4% paraformaldehyde for one night at 4 °C.The fixed spinal cords were immersed in 15% and 30% sucrose solutions for dehydration before embedding.Sagittal sections of spinal cord tissue (1-cm long), including the lesion site, were sectioned using a cryostat (Leica CM1950, Germany) and thaw-mounted onto Super Frost Plus slides (Fisher Scientific, USA).Next, sections were blocked with 5% donkey serum and 0.3% Triton X-100 in PBS buffer, followed by incubation of primary antibodies.After one night of incubation at 4 °C, sections were washed with PBS thrice and incubated with the appropriate secondary antibodies for 3 h at room temperature.Sections were finally observed using a confocal laser scanning microscope (A1Ti, Nikon, Japan).Quantification was determined in terms of the gray value of cells per section stained by different antibodies.We defined a baseline at the epicenter of the injury site, and an area with 1 mm rostral and caudal to the baseline was used in quantification.Three randomly selected sections in one animal and three animals per condition were used in this analysis.To assess the inflammatory protein levels in the injury site, a segment of 2 mm rostral and caudal to the injury site was dissected, and total protein was extracted by homogenate in a lysis buffer and centrifuged at 12 000 g, 4 °C for 10 min.Forty microgram sample protein was loaded onto 10% polyacrylamide gel and went through a separation process by SDS-PAGE.Proteins were then transferred to polyvinylidene fluoride membranes and blocked in 5% skimmed milk in PBST for 1 h at room temperature.Primary antibodies were added to membranes overnight at 4 °C.Secondary antibodies were used in membrane incubation after washing thrice with PBST.Finally, protein bands on membranes were observed using an ECL kit and quantified by the Image Lab Software provided by Bio-Rad.
Single-Cell RNA Sequencing (SC-RNA-seq) Analysis: Animals (five animals for each condition) were anesthetized deeply with pentobarbital sodium, and from the center of the SCI site, a 5-mm section of the spinal cord was dissected.For extraction of RNA in cells, the tissue was cut and chopped into small pieces, then incubated in 2 mL of collagenase (1 mg mL −1 in PBS) for 30 min at 37 °C and centrifuged at 300 g for 5 min at 4 °C.Pelleted cells were collected in 2 mL of 0.25% trypsin-EDTA solution and incubated at 37 °C for 5 min.Then, digestion by trypsin was stopped by FBS (10%), and the pellet collected via centrifugation was dispersed in PBS with 0.5% BSA.The dispersed cell pellet was strained through a 40-μm cell strainer twice, ensuring that no large cell cluster existed.To avoid the influence of blood cells, 10 mL of erythrocyte lysate was added to the cell for 10 min and the cell suspension was collected by centrifuging.Finally, single-cell suspension was resuspended in PBS for counting.Libraries were prepared according to 10× Genomics Chromium Single Cell 3′ Reagent Kit v3 (10× Genomics) and Single Cell 3' Gel Bead Kit v2 instructions, and an Agilent 2100 Expert High Sensitivity DNA Assay was used to assess the quantity of cDNA.Finally, cDNA libraries were sequenced using an Illumina HiSeqTM 4000 system, and the outputs were mapped to the mouse GrCm38 genome using Cell Ranger v.3.0.2.All raw data and proceeded files were provide in GEO (GSE222082).
Data Normalization and Dimensional Reduction: Initial processing of peripheral cells was performed using the Seurat R Package (v4.0.1).Individual cells were filtered for the total number of genes expressed and the percentage of mitochondrial reads.This filtering was set to retain cells with 200-2500 genes and percent mitochondrial reads less than 5%.Individual cells were then normalized using log normalization with a scale factor of 10 000.After processing, clustering was performed using the Seurat package.To identify the shared and unique cell types in different samples, an integration workflow from the Seurat package was used.In brief, canonical correlation analysis was first performed between two samples to find anchors by FindIntegrationAnchors() function.Then, depending on the consistency of anchors between cells, a gene expression matrix could be generated.To verify the largest variation in gene expression, dimensional reduction was performed.The top 10 calculated principal components were selected using a jackstraw analysis to quantify p-value distributions.The FindCluster() function was used to define cell types based on similar gene expressions.For visualization, the UMAP cluster was used to present cell clusters based on examining the standard deviations of the top 10 principal components and running.The differential markers between the clusters were isolated by comparing significantly upregulated genes as defined as adjusted p-values < 0.05.We distinguished and named cell types by DEGs based on the CellMarker database.
Cell-Cell Communication: To infer the interaction between several cell types, we performed a ligands-receptors analysis.The transmission between cells is based on the interaction of ligands and receptors on the cell surface and cytokines produced by cells.Cell Chat is a package that provides a database including ligands, receptors, cofactors, and polymers.Cell-cell communication was performed using Cell chat (v0.5.5) (https: //github.com/sqjin/CellChat).In brief, a Cell Chat DB.mouse database with ligands-receptors were added as a reference list.Clustered cell data from upstream analysis using Seurat was used to construct the cell chat object.The highly expressed gene was screened by identifyOverExpressed-Genes () function, and the corresponding pathway was identified based on these genes.With significant differences in gene expression, the possibility of ligands-receptors relationship pairs between cells was calculated by the computeCommunprob() function.Cells with cell communication possibilities below 20% were removed by a parameter of cut-off.Finally, results were compared with the imported ligands-receptors database, and an alluvial and circle plot was built for visualization.
Pseudo-Time Analysis: Cell trajectory and pseudo-time analysis were performed using the Monocle R package (v2.18.0) and the reverse-graph embedding machine learning algorithm.First, genes at different differentiation phases were selected for ordering.In our case, samples at distinct time points were used in the analysis.Therefore, DEGs at different time points were selected.These genes were ordered by the orderCells() function after principal component analysis.Each cell in a high-dimensional space was represented as a point corresponding to the expression level of ordered genes.Then, Monocle 2 constructed a differentiation tree based on the selected data, and the algorithm moved the cell to the nearest vertex in the tree and updated the position to fit the cell well.Therefore, a new spanning tree was constructed after the iteratively continued process performed until all cells converged.In the whole process, ordering and dimension were both conducted, and once the tree was constructed, Monocle selected a tip as the "root."The geodesic distance from each cell to the root was calculated to automatically assign its branches.Differential gene testing for the pseudo-time analysis was based on the previously identified cell clusters in the Seurat object and a cut-off for significance p-value < 0.01.
Statistical Analysis: All statistical analyses were performed using the GraphPad Prism software (GraphPad Software Inc.).Error bars in all figures represent as mean ± standard error of the mean (SEM).One-or twoway ANOVA with Tukey's multiple comparison test was used in statistical analysis, as indicated in figure legends.A p-value of less than 0.05 was considered significant.

Figure 1 .
Figure 1.Transcriptomic and histochemistry identification of major cell types that comprise the spinal cord injury (SCI) site at the sub-acute time point.A) Experiment process of SCI and tissue collection.B) Immunofluorescence images present DAPI-, GFAP-, Iba1-, and Cd68-positive cells (blue, green, red, and magenta, respectively) in the spinal cord tissue at 1, 3, and 7 dpi and intact condition.Scale bars presented as 1 mm and 200 μm in the enlarged image.C) UMAP visualization of all cells from uninjured and 1, 3, and 7 dpi present major cell types in the injury site.D) Cell number of each cell type from uninjured and 1, 3, and 7 dpi.E,F) Western blotting bands and quantification present the expression level of inflammation proteins, including Ccl2, IL2, IL33, TNF-, TGF-, and GAPDH.Statistical analysis was performed using two-way ANOVA with Tukey's multiple comparison test and shown as mean ± SEM, *p-value < 0.05 and **p-value < 0.01, n = 3.

Figure 2 .
Figure 2. Identification of cell subpopulations and differentiation trajectory in macrophage and microglia cluster.A) UMAP visualization of all macrophage subpopulations combined from uninjured and 1, 3, and 7 dpi identified based on differentially expressed genes (DEGs).B) Cell number of each identified subpopulation at different time points.C) Differentiation trajectory of each macrophage population based on the pseudo-time analysis.D) Subpopulation of microglia identified at different time points.E) Cell number of each microglia subpopulation at different time points.F) Cell differentiation trajectory in microglia subpopulations determined via pseudo-time analysis.

Figure 3 .
Figure 3. Cell communication analysis reveals the ligands-receptors relationship and pathways between cells.A) The alluvial plot shows incoming and outcoming cell signal pattern recognition for each cell type to facilitate intercellular communication network analysis.B) Circle plot showing top 12 signaling networks presenting ligands-receptors pathways sending and receiving signals between cells.

Figure 4 .
Figure 4. Conclusion of immune reaction process during the sub-acute phase of spinal cord injury (SCI) inflammation.Inflammation started with the polarization and migration of cells, followed by antigen clearance guided by macrophage subpopulations at 1 dpi.Immunity reaction keeps proceeding at 3 dpi, reflecting on the endocytosis subpopulation of macrophage and neurotoxic microglia.Complement secretion and scar formation was the main event at 7 dpi.

Figure 5 .
Figure 5. Protective effect of methylprednisolone sodium succinate (MPSS) on spared tissue from excessive inflammation at different time point treatments.A) Spinal cord injury (SCI) and drug injection process, B) representative images of spinal cord section stained with DAPI (blue), GFAP (green), Iba1 (red), and CD68 (magenta) of animals two weeks post-injury.White dotted lines indicate the area of the injury site.Scale bars: 1 mm.C) Gray value quantification of GFAP, Iba1, and CD68 (upper three), and injury area quantification (lower one).Statistical analysis was tested by two-way ANOVA with Tukey's multiple comparison test and presented as mean ± SEM, n = 3, *p < 0.05 and **p < 0.01, respectively.C) The enlarged image shows the details of the scar and Iba1-positive microglia in A. Scale bars: 200 μm.

Figure 6 .
Figure 6.Identification of methylprednisolone sodium succinate (MPSS) mechanism in inflammation inhibition by comparing the difference of cell types and gene expression between MPSS and CTRL samples.A) UMAP visualization of all cells from MPSS-and PBS-treated tissues representing major cell types in the injury site.B) The cell number ratio of each cluster in MPSS and CTRL samples.C) Expression of pro-inflammatory (IL2, Tnfaip2) and antiinflammatory (IL33) genes in MPSS-and PBS-treated tissue.D) Expression level of pro-inflammatory (IL2, Tnfaip2) and anti-inflammatory (IL33) proteins examined by Western blotting.Protein bands and quantification are shown on the left and right.Statistical analysis was tested by one-way ANOVA with Tukey's multiple comparison test and shown as mean ± SEM. *p < 0.05 and **p < 0.01, n = 3, respectively.E) Cell communication through pathways and contribution of each ligands-receptors pair both in MPSS and CTRL samples.Incoming and outgoing signaling through pathways identified in five patterns in the CTRL tissue.Data present the top 20 pathways.F) Contribution of each ligands-receptors pair in both MPSS and CTRL samples.Data present the top 10 pairs.