Address correspondence and reprint requests to Seiji Okada, Department of Advanced Medical Initiatives, Graduate School of Medical Sciences, Kyushu University, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan. E-mail: email@example.com
Acute inflammation is a prominent feature of central nervous system (CNS) insult and is detrimental to the CNS tissue. Although this reaction spontaneously diminishes within a short period of time, the mechanism underlying this inflammatory resolution remains largely unknown. In this study, we demonstrated that an initial infiltration of Ly6C+Ly6G− immature monocyte fraction exhibited the same characteristics as myeloid-derived suppressor cells (MDSCs), and played a critical role in the resolution of acute inflammation and in the subsequent tissue repair by using mice spinal cord injury (SCI) model. Complete depletion of Ly6C+Ly6G− fraction prior to injury by anti-Gr-1 antibody (clone: RB6-8C5) treatment significantly exacerbated tissue edema, vessel permeability, and hemorrhage, causing impaired neurological outcomes. Functional recovery was barely impaired when infiltration was allowed for the initial 24 h after injury, suggesting that MDSC infiltration at an early phase is critical to improve the neurological outcome. Moreover, intraspinal transplantation of ex vivo-generated MDSCs at sites of SCI significantly reduced inflammation and promoted tissue regeneration, resulting in better functional recovery. Our findings reveal the crucial role of an Ly6C+Ly6G− fraction as MDSCs in regulating inflammation and tissue repair after SCI, and also suggests an MDSC-based strategy that can be applied to acute inflammatory diseases.
Acute inflammation triggered by traumatic spinal cord injury (SCI) is a highly complex response developed by a wide variety of chemical mediators and cellular components. Immediately after SCI, resident microglial cells are activated and induce CD11b+Gr-1+ myeloid-derived cell infiltration into the lesion area by expressing several cytokines and chemokines (Kumamaru et al. 2011). CD11b+Gr-1+ myeloid-derived cell infiltration is one of the most remarkable features of acute inflammation, and has been widely recognized as having detrimental effects by producing proinflammatory cytokines, reactive oxygen species, and proteases after SCI (Carlson et al. 1998; Beattie et al. 2000). Many studies have reported that attenuating the inflammatory reaction by suppressing the effects of proinflammatory components results in beneficial effects after SCI (Gris et al. 2004; Okada et al. 2004).
Therefore, inflammatory reaction after SCI has been well documented in human and experimental models of SCI, and all of these studies have reported that this acute inflammation is attenuated within a short period of time (Sroga et al. 2003; Fleming et al. 2006). In fact, the expression levels of pro-inflammatory cytokines such as TNFα, IL-6, and IL-1β dramatically increase within several hours and then spontaneously and rapidly return to near normal levels (Kumamaru et al. 2011). Although the acute inflammation is dramatically attenuated, the strategies to enhance factors involved in the resolution of the acute inflammation have not been investigated.
In the field of SCI, myeloid-derived infiltrating macrophages reportedly acquire a M1 or M2 phenotype in the lesion area (Kigerl et al. 2009; Pineau et al. 2010). Contrary to classically activated M1 macrophages which create an inflammatory environment by expressing IL-1β, IL-6, and TNFα, M2 macrophages release anti-inflammatory cytokines and growth factors. Therefore, M2 macrophages may potentially be involved in the resolution of inflammatory reactions. However, a recent study by Kigerl et al. suggested that M2 macrophages appeared transiently 3 days after SCI (Kigerl et al. 2009), indicating that M2 macrophages do not participate in the attenuation of acute inflammation. It is possible that another specific subset of the infiltrating myeloid-derived cells contribute to the resolution of this intense inflammatory reaction after SCI.
Recent accumulating evidence has demonstrated that CD11b+Gr-1+ immature myeloid cell population, namely myeloid-derived suppressor cells (MDSCs), exerts immunosuppressive effects by modulating macrophage activation toward an immunosuppressive phenotype through arginase 1, inducible nitric oxide synthase (iNOS), and IL-10 (Sinha et al. 2007; Bunt et al. 2009). In this study, we demonstrated that infiltrating Ly6C+Ly6G− myeloid cells functioned as MDSCs and played a critical role in the attenuation of acute inflammation after traumatic SCI. In addition, transplantation of MDSCs at lesion areas significantly attenuated acute inflammation and promoted tissue repair by creating a permissive environment for differentiation into M2 macrophages, which improved neurological outcomes after SCI. Our findings clarified the role of MDSCs after traumatic SCI, and suggested a potential MDSC-based therapeutic strategy for the acute phase of central nervous system (CNS) injury.
Materials and methods
Animals and anesthesia
Adult 8- to 10-week-old female C57BL/6J mice were used in this study. All mice were housed in a temperature- and humidity-controlled environment on a 12 h light–dark cycle. Mice were anesthetized with nembutal (intraperitoneal administration of 70 mg/kg). All surgical procedures and experimental manipulations were approved by the Committee of Ethics on Animal Experiments in the Faculty of Medicine, Kyushu University. Experiments were conducted under the control of the Guidelines for Animal Experimentation.
Spinal cord injury
Mice were anesthetized and subjected to a severe (70 kdyn) contusion injury at the 10th thoracic level using an Infinite Horizons Impactor (Precision Systems Instrumentation, Lexington, KY, USA). After injury, the overlying muscles were sutured, and the skin was closed with wound clips. During recovery from anesthesia, the animals were placed in a temperature-controlled chamber until thermoregulation was re-established. Sham-operated controls were subjected to laminectomy only.
Mice were intraperitoneally injected with either 100 μg of anti-Gr-1 (clone RB6-8C5) or 100 μg of anti-Ly6G (clone 1A8) antibody beginning 1 day prior to SCI until 5 days after SCI and repeated at an interval of 1 day. Control mice were given equivalent amount of purified rat immunoglobulin G (IgG). The antibodies were purchased from BioXCell (West Lebanon, NH, USA). The efficiency of antibody administration was confirmed by FACS.
Animals were re-anesthetized and transcardially perfused with normal saline followed by 4% paraformaldehyde in 0.1 mol/L phosphate-buffered saline (PBS). The spinal cord was removed and immersed in the same fixative at 4°C for 24 h. A spinal segment centered over the lesion epicenter was transferred into 10% sucrose in PBS for 24 h and 30% sucrose in PBS for 24 h and embedded in an optimal cutting temperature compound. The embedded tissue was immediately frozen in liquid nitrogen and stored at −20°C until needed. Frozen sections of the spinal cord were cut on a cryostat in the sagittal plane at 14 μm, and mounted onto glass slides. Spinal cord sections were permeabilized with 0.01% Triton X-100 and 10% normal goat serum in PBS, pH 7.4, for 60 min. Primary antibodies included anti-CD68 (1 : 200, Serotec, Oxford, UK), anti-Ly76 (1 : 200, Biolegend, San Diego, CA, USA), anti-Iba1 (1 : 200, Wako, Osaka, Japan), anti-heme oxygenase-1 (HO-1) (1 : 200 Assay Designs, Inc., Ann Arbor, MI, USA), anti-NeuN (1 : 200, Chemicon, Temecula, CA, USA), anti-GFAP (1 : 200, Dako, Carpinteria, CA, USA), anti-PECAM (1 : 200 Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-tissue factor (1 : 100, American Diagnostica Inc., Stamford, CT, USA), anti-BrdU (1 : 200, Abcam, Cambridge, UK), and anti-laminin (1 : 200 Sigma, MO, USA) applied to the sections at 4°C overnight. The sections were then incubated with secondary antibodies. Alexa Fluor 488-conjugated goat anti-rat IgG (1 : 200, Invitrogen, Carlsbad, CA, USA), Alexa Fluor 488-conjugated goat anti-mouse IgG (1 : 200, Invitrogen) or Alexa Fluor 568-conjugated goat anti-rabbit IgG (1 : 200, Invitrogen) were used as secondary antibodies. Negative control experiments included a normal spinal cord section with primary as well as secondary antibodies and injured spinal cord sections incubated with secondary antibody alone or with non-specific rat/mouse IgG and secondary antibody. To evaluate proliferating cells at the lesion areas, mice were injected with bromodeoxyuridine (BrdU, 100 μg/g body weight; Sigma) daily from the day of injury until 3 days after SCI for histopathological analysis. For BrdU detection, sections were pre-treated with 2 N HCl for 15 min at 37°C. Immunofluorescent images were analyzed using an Olympus BX51 microscope (Olympus, Tokyo, Japan) or Zeiss Axiovert 200M microscope with a LSM 510 Meta confocal system (Carl Zeiss, Oberkochen, Germany).
Analysis of locomotor function
Motor function of the hindlimbs after SCI was evaluated with a locomotor open field rating scale [the Basso-Beattie-Bresnahan (BBB) Scale; Basso et al. 1996]. Each mouse was assessed 1 day and 7 days post-operatively and weekly thereafter for 6 weeks. A team of three independent examiners evaluated each animal for 4 min and assigned an operationally defined score (0–21) for each hindlimb. Every test was performed in a double-blinded fashion.
Quantification of intralesional hemorrhage
For quantification of intralesional hemorrhage after SCI, we performed a spectrophotometric assay using Drabkin's reagent (Sigma-Aldrich, St Louis, MO, USA). Spinal cord tissue specimens (6 mm long) were obtained from freshly killed controls or spinal cord-injured animals, and each spinal cord was treated individually as follows. Drabkin's reagent (500 μL) was added to each spinal cord; this was then homogenized (20 400 g, 60 s) and allowed to stand for 15 min at 20°C. Drabkin's reagent oxidizes hemoglobin to cyanmethemoglobin which has maximum absorption at 540 nm. The homogenate was then centrifuged at 15 000 rpm for 10 min. The cyanmethemoglobin-containing supernatant was collected, and a quantitative spectrophotometric hemoglobin assay was performed. To prepare a cyanmethemoglobin standard absorbance curve, normal spinal cord tissue (6 mm long) was dissected from uninjured mice after blood removal by cardiac puncture and homogenized with Drabkin's reagent containing incremental aliquots of blood.
Spinal cord water content
Spinal cord edema was evaluated by analyzing the water content of the spinal cord. Injured spinal cords were dissected (8 mm long), weighed, and then dried for 72 h at 80°C to determine the dry weight. Percentage of water content in spinal cord tissue was calculated by the following calculations: spinal cord water content (%) = (wet weight−dry weight)/wet weight × 100%.
Evans Blue blood–brain barrier permeability assay
Mice were injected intraperitoneally with 80 μg/g b.w. Evans Blue dye in PBS. Three hours after injection, the mice were anesthetized and perfused with PBS for 5 min. After perfusion, the injured spinal cords were dissected (8 mm long) and homogenized (15 000 rpm, 60 s) in 0.5 mL of 50% trichloroacetic acid. The sample homogenate was diluted 1 : 1 in 100% ethanol and then centrifuged at 15 300 g for 10 min. The supernatant was collected and absorbance (620 nm) was measured.
Blood samples and spinal cord samples were prepared for flow cytometry as previously described (Saiwai et al. 2010). For the blood samples, after red blood cells were removed with hypotonic lysis buffer (17 mmol/L Tris-HCl, pH 7.2; 100 mmol/L NH4Cl), the resulting suspension was pelleted by centrifugation and washed twice in PBS. Spinal cord samples were dissected and mechanically dissociated with collagenase (175 U/mL; Invitrogen Life Technologies, Carlsbad, CA, USA) for 30 min at 37°C. Cells were washed in Dulbecco's modified Eagle's Medium (DMEM) containing 10% fetal bovine serum, and passed through a 40 μm nylon cell strainer (BD Biosciences, San Jose, CA, USA) to isolate tissue debris from the cell suspension. The resulting suspension from the spinal cord and blood samples was centrifuged and the pellet was re-suspended and incubated for 5 min on ice with Fc block, and for 30 min on ice with the fluorescent antibodies. Samples were stained with anti-CD45, -CD11b, -Gr-1, -Ly6G, -Ly6C, -F4/80, -CD115 purchased from Biolegend. Before analysis, propidium iodide was added to determine the cell viability. The samples were analyzed on a FACSAria II flow cytometer (BD Biosciences). The data were analyzed using FACSDiva software (BD Biosciences).
RNA isolation and real-time PCR
Total RNA was isolated from the FACS-sorted Ly6C−Ly6G+, Ly6C+Ly6G−, and Ly6C−Ly6G− fractions using an RNeasy Micro Kit (Qiagen, Hilden, Germany) and from injured spinal cords using an RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. RNA concentration was determined by Nanodrop (Thermo Scientific Inc., Wilmington, DE, USA), and RNA was reverse transcribed to cDNA using a PrimeScript reverse transcriptase (TaKaRa, Shiga, Japan). Quantitative real-time RT-PCR was performed using primers specific for the genes of interest and SYBR Premix Ex Taq II. The mRNA expressions of each gene in FACS-sorted fractions were normalized to the level of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) for each sample as a standard and expressed as an arbitrary unit. The mRNA expressions of spinal cord tissues were normalized with GAPDH levels and then normalized to the sham-operated spinal cord tissue.
Quantification of angiogenesis/vasculogenesis
Quantification of the extent of angiogenesis/vasculogenesis was performed under a fluorescence microscope by calculating the ratio of the laminin-immunopositive area to the total transverse spinal cord area 1 mm rostral and caudal to the injury epicenter. Image analysis was performed using a computer with ImageJ software (http://imagej.nih.gov/ij/).
In vitro MDSC generation
Bone marrow cells were harvested from the femurs and tibias of uninjured green fluorescent protein-transgenic mice [GFP-Tg mice; (Okabe et al. 1997)]. After being passed through a 40 μm nylon cell strainer (BD Biosciences), cells were plated at 1 × 106 cells/mL in DMEM plus 10% fetal calf serum, 50 mM 2-mercaptoethanol, 10 mM N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid buffer, 1 mM sodium pyruvate, 100 U/mL penicillin, 100 mg/mL streptomycin, and amino acid supplements (1.5 mM l-glutamine, l-arginine, and l-asparagine). G-CSF was added at 100 ng/mL and mouse GM-CSF was added at 20 ng/mL, as described previously (Highfill et al. 2010). Cultures were incubated at 37°C in 10% CO2 for 4 days. On day 4 of culture, MDSCs were positively selected with Ly6C by magnetic-activated cell sorting (Miltenyi Biotec, Bergisch Gladbach, Germany).
Immediately after SCI, transplantation of the generated Ly6C+Ly6G− MDSCs derived from GFP-Tg mice was performed using a glass micropipette and a stereotaxic injector (KDS 310; Muromachi-kikai, Tokyo, Japan). MDSC injections were performed bilaterally at four sites of intact parenchyma adjacent to the epicenter and rostral and caudal to the epicenter.
GraphPad Prism 5.0 software (GraphPad Software Inc., La Jolla, CA, USA) was used to analyze the data. Two-way repeated-measures anova (Fig. 3a) and paired t-tests (Figs 6c and 7c) were used to compare BBB scale data. Other statistical evaluations of comparisons of two samples were performed using the Mann–Whitney U-test. The Kruskal–Wallis test followed by Bonferroni's correction of the Mann–Whitney U-test was used for multiple comparisons. Results were considered statistically significant at p < 0.05. All data are expressed as the mean ± SEM.
Characterization of each subset of infiltrating myeloid-derived inflammatory cells after SCI
A large number of CD11b+Gr-1+ inflammatory cells infiltrated the lesion area after SCI leading to secondary damage of neural tissue. Previous studies have reported that suppression of the inflammatory reaction by inhibition of adhesion molecules or chemoattractants without a specific cellular target led to improved neurological outcomes. Although Gr-1 surface antigen is a common epitope on Ly6C and Ly6G which express monocytic and granulocytic subsets, respectively, the detailed role of each subset at lesion areas remains elusive. We first evaluated the temporal change in the infiltration of Ly6C+Ly6G−, Ly6C−Ly6G−, and Ly6C−Ly6G+ cell subsets among CD45+CD11b+ fraction by flow cytometry from 4 h until 7 days after SCI. Flow cytometry analysis demonstrated that the infiltrating Ly6C−Ly6G+ and Ly6C+Ly6G− fractions showed a similar pattern of change, peaking at 12 h after injury, whereas Ly6C−Ly6G− fraction increased gradually with time (Fig. 1a).
To investigate the physiological roles of these myeloid-derived inflammatory cell subsets after SCI, we used FACS to isolate each subset based on their expression of Ly6C and Ly6G cell surface antigens (Fig. 1b). We confirmed that the flow cytometry-sorted Ly6C−Ly6G+ fraction expressed significantly higher levels of CXCR1 and CXCR2, Ly6C+Ly6G− fraction expressed a higher level of CCR2, and Ly6C−Ly6G− fraction expressed higher levels of CX3CR1 (Fig. 1c), which indicated that each subset was regulated by different chemokines. The infiltrating Ly6C−Ly6G+ fraction showed proinflammatory properties with higher expression of IL-1β and TNFα. In contrast, we confirmed that Ly6C+Ly6G− fraction up-regulated both iNOS and arginase 1 expression. This expression pattern is a typical feature of MDSCs which are immunosuppressive immature myeloid cells. In addition, Ly6C+Ly6G− fraction had higher expression of anti-inflammatory mediators such as IL-10, TGFβ, and VEGF, which is also consistent with the typical features of MDSCs (Fig. 1c) (Gabrilovich and Nagaraj 2009).
Anti-Gr-1 antibody (Ab) administration depletes both Ly6C−Ly6G+ cells and Ly6C+Ly6G− cells, whereas anti-Ly6G Ab specifically depletes Ly6C−Ly6G+ cells from the peripheral blood
To determine the specific roles of each fraction in lesion areas after SCI, we developed myeloid cell depletion models using neutralizing Abs. We confirmed that administration of 100 μg of anti-Gr-1 Ab completely depleted both Ly6C−Ly6G+ and Ly6C+Ly6G− cells, and administration of 100 μg of anti-Ly6G Ab specifically depleted Ly6C−Ly6G+ cells from circulating blood (Fig. 2a). As shown in Fig. 1a as well as in previous reports (Saiwai et al. 2010; Kumamaru et al. 2011), the number of infiltrating Ly6C−Ly6G+ and Ly6C+Ly6G− myeloid cells at the lesion area peaked at 12–24 h after SCI, but almost disappeared 7 days after injury. Therefore, we hypothesized that an initial 1-week depletion after injury was sufficient to eliminate the effect of these fractions (Fig. 2b).
Using CD68 immunohistological analysis, we also confirmed the apparent reduction in the degree of CD68+ macrophage infiltration in anti-Gr-1 Ab-treated mice compared with the findings in control and anti-Ly6G Ab-treated mice (Fig. 2c). Because the administration of 100 μg of anti-Gr-1 Ab completely depleted Ly6C+Ly6G− cells from the peripheral blood, most of the CD68+ cells detected in anti-Gr-1 Ab-treated mice were presumed to be resident microglial cells.
Depletion of infiltrating Ly6C+Ly6G− MDSCs impairs functional recovery and exacerbates tissue damage
To investigate the involvement of each fraction in the pathophysiology of SCI, we first evaluated open field locomotor scores for anti-Gr-1 Ab-, anti-Ly6G Ab-, and control IgG-treated mice on day 1, day 7, and on a weekly basis thereafter using the modified BBB scale for mouse hindlimb motor function (Joshi and Fehlings 2002; Okada et al. 2005, 2006). Interestingly, the scale revealed a significantly low functional recovery after SCI in anti-Gr-1 Ab-treated mice (n = 14) compared with control (n = 14) and Ly6G Ab-treated (n = 15) mice at each time point (Fig. 3a).
Macroscopic findings relating to the injured spinal cords indicated that the lesion areas in anti-Gr-1 Ab-treated mice were considerably larger than those in the other groups of mice (Fig. 3b), which was confirmed by quantitative measurement of hemorrhage in the lesion areas using Drabkin's reagent (n = 5 per group, Fig. 3c).
These results indicate that infiltrating Ly6C+Ly6G− MDSCs are important for coagulation to occur at lesion areas after SCI. After tissue damage because of traumatic injury, the extrinsic coagulation cascade is activated by tissue factor, which is a known trigger of this reaction (Daubie et al. 2007). We confirmed that infiltrating macrophages express tissue factor at the lesion areas by immunostaining injured spinal cord sections (Fig. 3d).
In addition to hemorrhage, edema formation was observed within minutes and lasted up to several days after SCI (Sharma 2005). The injured spinal cord in anti-Gr-1 Ab-treated mice showed significant exacerbation of edema and increased the permeability of the blood–brain barrier, measured by a water content test (n = 6–8 per group, Fig. 3e) and Evans Blue Assay, respectively (n = 5 per group, Fig. 3f).
To clarify the effect of the infiltrating Ly6C+Ly6G− subset on the microenvironment of injured spinal cords, we evaluated the expression of several genes (IL-6, IL-1 beta, TNF alpha, IL-10, TGF beta1, and VEGF) in injured spinal cords in these three groups at 12 h after SCI. The expression of IL-6 in anti-Gr-1 Ab-treated mice was significantly higher than that in anti-Ly6G Ab-treated and control mice. In addition, IL-10 expression in anti-Gr-1 Ab-treated mice was significantly reduced compared with that in the other two groups of mice. Interestingly, IL-1 beta expression was significantly reduced in anti-Gr-1 Ab-treated mice compared with that in anti-Ly6G Ab-treated and control mice. There were no significant differences in the expression levels of TNF alpha, TGF beta1, and VEGF among the control, anti-Gr-1 Ab-treated, and anti-Ly6G Ab-treated mice (Fig. 3g).
As a result of this exacerbation of blood–brain barrier disruption and enhanced inflammatory reaction at the lesion area, anti-Gr-1 Ab-treated mice had a significantly larger GFAP-negative necrotic tissue area surrounded by reactive astrocytes than the other groups (Fig. 3h).
Infiltrating Ly6C+Ly6G− MDSCs accelerate the removal of hematomas and have anti-inflammatory properties after SCI
Under pathological conditions, one of the important functions of infiltrating myeloid-derived inflammatory cells is phagocytosis to digest microorganisms and tissue debris. As shown in Fig. 4a, we observed infiltrating macrophages phagocytose interstitial erythrocytes to eliminate hematomas from the lesion areas. To assess the extent of hematoma clearance, we performed Iba-1/Ly76 double immunostaining for macrophages and erythrocytes on injured spinal cord sections. Although most of the residual hematoma was eliminated in control and anti-Ly6G Ab-treated mice 7 days after SCI, a massive amount of residual erythrocytes was still observed in anti-Gr-1 Ab-treated mice (Fig. 4b). Double immunostaining for erythrocytes and laminin revealed that the resident hematoma at the lesion area inhibited laminin from replacing the remaining/existing hematomas (Fig. 4c). Laminin plays an important role as a component of vascular basal membranes and scaffolds for the extracellular matrix; therefore, hematoma elimination by erythrophagocytosis is an essential process for the subsequent regenerative response (Sorokin 2010).
In addition, hemolyzed erythrocytes from the residual hematoma release free heme which has toxic effects because of oxidation. The stress-responsive heme oxygenase-1 (HO-1) catabolizes free heme to protect spinal cord tissues from further oxidative damage. In our SCI model, we confirmed that the dominant cell source of HO-1 production was the infiltrating CD68+ macrophages (Fig. 4d), and HO-1 expression to a lesser extent was observed in anti-Gr-1 Ab-treated mice (Fig. 4e). In addition, we observed that the peak of HO-1 expression was at 4 days after SCI (n = 5 mice per time point, Fig. 4f). As alternatively activated M2 macrophages were reported to be the main source of HO-1, M2 macrophage activity was presumed to be strongly up-regulated 4 days after SCI. This result was consistent with a report that M2 macrophages only appeared transiently between 3 to 7 days after SCI (Kigerl et al. 2009).
Ly6C+Ly6G− MDSC infiltration plays a crucial role in angiogenesis/vasculogenesis after SCI
After a period of acute intensive inflammation, the lesion area gradually undergoes tissue repair. As an indicator of tissue repair response, we assessed the extent of angiogenesis/vasculogenesis at the lesion area in three groups of mice. We confirmed that angiogenesis/vasculogenesis had already begun 4 days after SCI by observing PECAM/BrdU double immunopositive proliferating vascular endothelial cells at the lesion areas (Fig. 5a).
In this phase, there were significant morphological differences in laminin-positive vessels at the lesion area between control, anti-Ly6G Ab-treated, and anti-Gr-1 Ab-treated mice. The vessel diameters were larger and the vessel walls were thicker in control and Ly6G Ab-treated mice compared with anti-Gr-1 Ab-treated mice (Fig. 5b). To quantitatively evaluate the vessel distribution density after SCI, we calculated the ratio of laminin-positive area to total transverse spinal cord area at 1 mm rostral and caudal to the injury epicenter. The ratio was significantly reduced in anti-Gr-1 Ab-treated mice (Fig. 5c and d). Barrette et al. (2008) also demonstrated in a peripheral nerve injury model that CD11b+ myeloid cells are essential for the formation/stabilization of blood vessels in the injured nervous system. Therefore, these results suggest that Ly6C+Ly6G− MDSC infiltration was critical in promoting angiogenesis/vasculogenesis after SCI.
Ly6C+Ly6G− MDSC infiltration within 24 h after SCI was a critical factor in promoting functional recovery
We explored the possibility that augmentation of the effects of Ly6C+Ly6G− MDSCs at the lesion area after SCI could be beneficial to the tissue repair process. To determine the most effective phase to amplify the effect of Ly6C+Ly6G− MDSCs, we compared the functional recovery between mice with SCI in an anti-Gr-1 Ab ‘pre-treatment’ group and an anti-Gr-1 Ab ‘delayed treatment’ group, as indicated in Fig. 6a. Mice in the ‘pre-treatment’ group were administered anti-Gr-1 Ab daily from 1 day prior to injury to 1 day after SCI. We began anti-Gr-1 Ab treatment 24 h after SCI followed by daily administration until 3 days after SCI in the ‘delayed treatment’ group to allow Ly6C+Ly6G− MDSC infiltration within the initial 24 h after injury. Histopathological analysis showed that CD68+ monocyte/macrophage infiltration was suppressed in the ‘pre-treatment’ group 4 days after SCI, which was a similar result to that of a group of mice that underwent treatment with anti-Gr-1 Ab for 1 week (Fig. 6b). In contrast, CD68+ cell infiltration in mice in the ‘delayed treatment’ group was barely inhibited. The functional recovery in the ‘pre-treatment’ group was significantly exacerbated compared with that in the ‘delayed treatment’ group (n = 10 per group, Fig. 6c), suggesting that the infiltration of Ly6C+Ly6G− MDSCs within 24 h after SCI was critical to improve neurological outcome. From this result, we concluded that immediate augmentation of Ly6C+Ly6G− MDSCs would be effective in promoting tissue repair.
In vitro-derived Ly6C+Ly6G− MDSC transplantation effectively promotes the tissue repair process and functional recovery after SCI
Highfill et al. have shown that Ly6C+Ly6G− MDSCs could be efficiently induced by in vitro culture of bone marrow cells with granulocyte colony-stimulating factor (G-CSF) and granulocyte–macrophage colony-stimulating factor (GM-CSF) (Highfill et al. 2010). We also confirmed that the percentage of Ly6C+Ly6G− MDSCs with a high expression of CD115 and F4/80 significantly increased in media supplemented with G-CSF and GM-CSF compared with media alone or media supplemented with G-CSF (Fig. 7a). We isolated Ly6C+Ly6G− MDSCs by positive selection using magnetic beads from GFP-Tg bone marrow cells cultured with G-CSF and GM-CSF, and transplanted them (5 × 105 cells/lesion area) into the epicenter after contusion SCI.
In the histological sections 4 days after transplantation, we confirmed that some of the transplanted Ly6C+Ly6G− MDSCs were BrdU positive indicating their survival and proliferation in the lesion area (Fig. 7b). The MDSC transplantation (Tp) group (n = 13) showed significantly better functional recovery at each timepoint after SCI compared with the control group (n = 12) in terms of the open field motor score (Fig. 7c). To clarify the mechanisms of this functional improvement by Ly6C+Ly6G− MDSC Tp, we compared mRNA expression levels in the lesion areas between the MDSC Tp and control groups. Injured spinal cords were harvested at 12 h when the inflammatory reaction was most intense, and at 7 days when the repair process was already initiated after SCI. We confirmed that the expression levels of arginase 1 and iNOS were significantly enhanced in the MDSC Tp group compared with the control group, both at 12 h and 7 day after SCI (Fig. 7d), indicating that in vitro-derived MDSCs continuously exerted their immunosuppressive effects on the lesion area for at least a week. In addition, we found that the expression levels of anti-inflammatory cytokines (IL-10, TGFβ) and anti-oxidant HO-1 were significantly enhanced in the MDSC Tp group (Fig. 7d). We also evaluated the expression levels of VEGF, IGF, and HGF 12 h and 7 day after SCI because these growth factors have been shown to exert angiogenic, neuroprotective, anti-inflammatory, and anti-apoptotic effects after SCI (Hung et al. 2007; Kitamura et al. 2007). The expression of these growth factors was elevated in the MDSC Tp group after SCI (Fig. 7d). Taken together, these results suggest that treatment with in vitro-generated MDSCs exert pleiotropic effects and result in an improved neurological outcome.
In this study, we demonstrated for the first time that infiltrating Ly6C+Ly6G− myeloid cells significantly contribute to attenuate acute inflammation when total myeloid cell infiltration peaked after SCI. A detailed transcriptional evaluation of each inflammatory cell fraction revealed that this subset showed typical features of MDSCs (Fig. 1c). We evaluated a SCI model with complete depletion of Ly6C+Ly6G− cells and observed their pleiotropic role in tissue repair to further determine the specific roles of Ly6C+Ly6G− MDSCs. These cells were associated with the coagulation process by enhancing the expression of tissue factor, which is the initiator of the coagulation cascade. They also made a significant contribution to the resolution of a hematoma by enhancing erythrophagocytosis in the lesion area. In addition, they promoted the subsequent tissue repair process, including angiogenesis/vasculogenesis, after attenuating intense acute inflammation. Our results indicate that Ly6C+Ly6G− MDSCs are promising for therapeutic application during the acute phase of SCI because they contribute synergistically to inflammatory resolution and the subsequent tissue repair process.
MDSCs are defined as an immature CD11b+Gr-1+ population that exerts an immunosuppressive effect in vitro and in vivo (Sica and Bronte 2007). MDSCs are a heterogeneous population because the Gr-1 antigen was identified on both Ly6G and Ly6C surface markers. MDSCs are subdivided into two major subsets: Ly6C−Ly6G+ granulocytic and Ly6C+Ly6G− monocytic MDSCs. Movahedi et al. reported that Ly6C−Ly6G+ MDSCs and Ly6C+Ly6G− MDSCs are completely different subpopulations, because each subset uses distinct signaling pathways and effector molecules, even though they both commonly exert immunosuppressive effects as their main function (Movahedi et al. 2008). In this study, we demonstrated that the Ly6C+Ly6G− fraction critically contributes to the attenuation of acute inflammation during the acute phase of SCI by functioning as MDSCs, and that most of the infiltrating Ly6C−Ly6G+ fraction consisted of typical inflammatory neutrophils. Interestingly, Ioannou et al. reported that granulocytic Ly6C−Ly6G+ MDSCs are a critical factor for the amelioration of EAE (Ioannou et al. 2012). These differences are apparently attributable to the causative inflammatory reaction of EAE (adaptive immune system) and acute phase of SCI (innate immune system).
Until now, phenotypic characterization of mouse monocytes has been well performed and can be divided into two distinct populations according to Ly6C and CX3CR1 expression. These two populations have been described as the Ly6C+/CX3CR1−/CCR2+ subset and the Ly6C−/CX3CR1+/CCR2− subset (Geissmann et al. 2003; Gordon and Taylor 2005). In this study, we confirmed that the infiltrating Ly6C+ subset at the lesion area showed higher CCR2 expression and lower CX3CR1 expression, indicating an immature monocytic subset (Fig. 1c). From the result of a temporal change in infiltrating cells in Fig. 1a, most infiltrating Ly6C+ MDSCs at the acute phase were supposed to be eliminated from the lesion area before they differentiated into Ly6C− macrophages. However, we also demonstrated that CD68+ macrophage infiltration 4 days after injury was significantly decreased by depleting the Ly6C+ subset, suggesting that part of the Ly6C+ subset would survive and proliferate in the lesion area to contribute to subsequent tissue repair processes.
Some studies have reported the existence of anti-inflammatory and neuroprotective cell fractions among infiltrating leukocytes after SCI such as M2 macrophages (Kigerl et al. 2009; Busch et al. 2011). Shechter et al. demonstrated that the infiltrating monocytic fraction has an anti-inflammatory property in the lesion area by injecting Gr-1+ monocytes intravenously after SCI (Shechter et al. 2009). In this study, we further clarified that the Ly6C+ subset was involved in the resolution of acute inflammation, showing the same characteristics as Ly6C+ MDSCs by directly isolating the subset and evaluating gene expression.
For a further evaluation of the Ly6C+ subset, we used two different kinds of neutralizing Abs: anti-Gr-1 Ab to deplete both the Ly6G+ and Ly6C+ fractions and anti-Ly6G Ab to deplete only the Ly6G+ fraction. Depleting a specific leukocyte fraction using anti-Gr-1 Ab is a widely used technique to evaluate the role of a cell subset. Using the affinity differences of anti-Gr-1 Ab for Ly6G and Ly6C epitopes, Kim et al. demonstrated that low dose (125 μg) administration of anti-Gr-1 Ab depletes neutrophils only, whereas a high dose (400 μg) depletes both monocytes and neutrophils (Kim et al. 2009). In fact, Stirling et al. reported that neutrophil depletion by the anti-Gr-1 Ab modulates following inflammation and impairs the tissue repair process after SCI (Stirling and Yong 2008). However, they confirmed that Ly6C+ monocytes are not depleted at all after administrating this Ab. In our study, complete depletion of Ly6G+ neutrophils by anti-Ly6G Ab treatment did not result in significant impairment of neurological recovery compared with that in the control group, which was different from the results of Stirling et al. Interestingly, in a peripheral nerve injury model, neutrophil depletion by anti-Ly6G Ab treatment ameliorated allodynia (Nadeau et al. 2011). Therefore, further research is necessary to clarify the differences in the effects of anti-Gr-1 Ab and anti-Ly6G Ab when they deplete only neutrophils after SCI.
Thus, the role of neuroinflammation after SCI remains highly controversial as to whether it has a neurotoxic or neuroprotective effect. Our results suggest that the acute phase of inflammation in particular could be more neurotoxic unless MDSCs infiltrate to regulate this reaction.
The properties of MDSCs are well characterized, particularly in relation to cancer (Talmadge 2007; Ostrand-Rosenberg and Sinha 2009). MDSCs exert their immunosuppressive effects by producing iNOS, arginase 1, and the anti-inflammatory cytokine IL-10. In addition, they promote tumor growth, invasion, and metastasis by increasing the number of tumor-associated macrophages (TAMs) and by inducing angiogenesis/vasculogenesis through secretion of interferon-γ, IL-13, and vascular endothelial growth factor (Sica and Bronte 2007).
As with the TAM phenotypes, which are environment dependent, infiltrating macrophage differentiation and polarization toward the M1 or M2 phenotype after SCI are determined by the cytokine milieu of the lesion microenvironment. Although M1 macrophages have pro-inflammatory properties, M2 macrophage phenotypes are characterized by immunoregulation and tissue remodeling. In this study, we confirmed that Ly6C+Ly6G− MDSCs contributed to up-regulation of IL-10 expression in lesion areas, suggesting that MDSCs promote infiltrating monocytes to differentiate into M2c macrophages, which are associated with tissue repair and remodeling (Munder et al. 1999; Martinez-Pomares et al. 2003). In fact, enhancement of their effects by transplanting MDSCs after SCI successfully achieved anti-inflammatory and neurotrophic effects as well as functional improvements.
Among the anti-inflammatory factors from Ly6C+ MDSCs, IL-10 functions as a potent inducer of HO-1 in macrophages (Lee and Chau 2002). HO-1 is a heme-degrading enzyme that protects tissues from free heme toxicity. In addition, it also has a direct effect of attenuating inflammation (Otterbein et al. 2003). We confirmed that transplantation of MDSCs significantly up-regulated HO-1 expression, suggesting that MDSCs developed an environment favorable for tissue repair. In addition, expression of both arginase 1 and iNOS was enhanced in the lesion areas after MDSC transplantation for at least 1 week after SCI. This up-regulation of both arginase 1 and iNOS was a determining factor to define the characteristics of MDSCs, which are different from M2 macrophages.
Cell-based therapies have great potential for treating many CNS disorders. In this context, bone marrow-derived mononuclear cells (BM-MNC) are one of the most attractive and feasible sources, because bone marrow cells can be obtained easily, made to proliferate rapidly, and can be applied to autologous as well as allogeneic transplantation. However, limited improvements in neurological function have been observed in SCI after BM-MNC transplantation. These cells are whole mononuclear cells from bone marrow, which include the proinflammatory cell subsets such as mature monocytes and natural killer cells. In contrast, our results suggest that endogenously infiltrating Ly6C+Ly6G− MDSCs exerted anti-inflammatory and neuroprotective effects, and that the intralesional transplantation of MDSCs generated in vitro by G-CSF and GM-CSF from bone marrow cells effectively improved SCI pathophysiology. In addition, it has been suggested that the therapeutic efficacy of BM-MNC transplantation is enhanced in combination with GM-CSF administration (Yoon et al. 2007). In the field of SCI, even administration of GM-CSF alone shows neuroprotective effects and improved functional recovery (Ha et al. 2005; Bouhy et al. 2006). Given that Ly6C+Ly6G− MDSCs can be induced from bone marrow cells by in vitro culture with GM-CSF, in vivo administration of GM-CSF would lead to an increase in the number of Ly6C+Ly6G− MDSCs not only in peripheral blood but also in the lesion area.
In this study, we transplanted MDSCs through direct intralesional injection. Intralesional application has the advantage of ensuring the delivery of a greater proportion of MDSCs than is possible with intravenous or intrathecal administration. Takahashi et al. concluded that intralesional transplantation was the most effective and feasible method for transplanting neural stem/progenitor cells (NSPCs) into the injured spinal cord (Takahashi et al. 2011). This is because a greater number of intralesionally grafted NSPCs survived at the site of injury as compared with those surviving after intrathecal injection. However, some studies have reported that intravenous or intrathecal injection of BM-MNCs results in beneficial effects (Akiyama et al. 2002; Cizkova et al. 2011). Further studies may be required to determine the most effective method for bone marrow-derived cell transplantation.
In conclusion, our findings suggest that infiltrating Ly6C+Ly6G− cells in the lesion area functioned as MDSCs and contributed to resolve acute inflammation. In addition, our strategy of promoting attenuation of acute inflammation by enhancing endogenous anti-inflammatory factors may provide a potential therapeutic strategy for the acute phase of traumatic SCI.
This study was supported in parts by Grant-in-aid for Scientific Research (B) on Innovative Areas, for Challenging Exploratory Research, Grant-in-Aid for Japan Society for the Promotion of Science Research Fellows, and research foundations from the general insurance association of Japan.