Background
Following spinal cord injury (SCI), a wide variety of secondary injury processes are provoked, which significantly worsens patients’ outcomes [
1]. Among these, the most complex and important one is the neuroinflammatory response that contributes to progressive neurodegeneration. Macrophages (Mφs) and microglia are one of the key players, with a predominant role for Mφs, which accumulate specifically at the lesion epicenter [
1,
2]. At the injury site, local factors (e.g., cytokines, myelin debris, and apoptotic cells) determine the phenotypic properties of these cells, rendering them either detrimental or beneficial for disease pathology [
1,
2]. A general pragmatic classification divides them into (i) classically activated M1 Mφs, which secrete pro-inflammatory factors and induce axonal dieback, and (ii) alternatively activated M2 Mφs, which display anti-inflammatory properties and support axonal regrowth [
3,
4]. The M1/M2 dichotomy is an oversimplified model as Mφs are very heterogenic with remarkable functional plasticity [
5]. Nevertheless, the M1/M2 paradigm helped to identify significant functional characteristics of selected Mφ populations.
In particular, M2a Mɸs have been identified as essential players in neuroregeneration by counteracting inflammation and promoting wound healing [
6,
7]. This subset is characterized by upregulation of the signature molecules Arginase-1 (Arg1), found in inflammatory zone 1 (FIZZ1), chitinase-3-like protein 3 (Chi3l3, Ym1), and mannose receptor (CD206) [
8,
9]. This M2 Mφ subpopulation arrives first at the lesion site in the acute phase after injury but is only transiently detectable [
3]. After SCI, the perilesional milieu favors M1 polarization, resulting in a persistently hostile environment for repair [
1,
3]. As this is a major contributor to poor regeneration, reducing the M1 dominance and promoting an M2 presence has become an attractive therapeutic approach for SCI.
In this study, the goal was to modulate perilesional Mφs/microglia and specifically exploit their M2a phenotype to influence motor recovery beneficially. For this aim, we used the canonical T-helper cell type 2 cytokine interleukin-13 (IL-13), which is one of the main inducers of this M2a subset [
9]. Following SCI, IL-13 level dramatically decreases in the spinal cord tissue [
10,
11]. This drop suggests that the application of IL-13 in the acute phase after SCI may have therapeutic potential by altering Mφ/microglia polarization. Previous animal studies in the context of central nervous system (CNS) trauma revealed that IL-13 administration, either directly or via cell therapy, exerted anti-inflammatory effects and thereby provided neuroprotection [
12‐
14]. Our prior study demonstrated that cell-based delivery of IL-13 via mesenchymal stem cells (MSCs) enhanced functional recovery after SCI. This improvement was associated with an increased perilesional presence of Arg1
+ Mφs/microglia. Thus, it was tempting to speculate that accumulated M2a Mφs were critical for functional repair [
15]. Therefore, in this study, we used Mφs to continuously deliver IL-13 as (i) Mφs have a high migratory capacity towards inflammatory sites after trauma and (ii) IL-13 can introduce additional anti-inflammatory M2a Mφs at the lesion site [
16].
We show for the first time that the use of genetically modified Mφs as carriers of IL-13 is a potent strategy to improve histopathological and functional recovery after SCI. Using IL-4Rα-deficient mice, we demonstrate that IL-13 signaling is essential to reduce neuronal death and suppress axonal dieback by promoting an anti-inflammatory phenotypic shift of perilesional Mφs/microglia. Overall, these findings underline the therapeutic potential of IL-13 and Mφ/microglia modulation in regeneration after CNS trauma.
Methods
Animals
In vitro experiments were done using female 8–12-week old wild type (WT) BALB/c (Envigo) or C57BL/6J mice (Janvier Labs). CX3CR1
+/GFP mice were kindly provided by prof. Brône (Uhasselt). In vivo experiments were performed with either female 10–14-week old WT C57BL/6j mice (Janvier Labs) or IL-4Rα control and knockout (KO) BALB/cJ mice (#000651, #003514, The Jackson Laboratory) [
17]. As male mice have a substantially higher dropout rate after SCI due to wounding and higher risk of bladder or other infections, only female mice were used in this study. Mice were housed at the conventional animal facility of Hasselt University under stable conditions (temperature-controlled room, 12 h light/dark cycle, food, and water ad libitum). Experiments and sample size calculations were approved by the local ethical committee and were performed according to the guidelines of Directive 2010/63/EU on the protection of animals used for scientific purposes.
Macrophage culture
Bone-marrow-derived Mɸs were obtained from female mice. In brief, primary bone marrow cells were isolated by flushing the femur and tibia with 1× phosphate-buffered saline (PBS, Lonza). Cells were cultured in RPMI 1640 medium (Lonza) supplemented with 10% heat-inactivated fetal calf serum (FCS, Gibco), 1% penicillin/streptomycin (100 units/ml penicillin, 100 μg/ml streptomycin, Invitrogen), and 15% L929 conditioned medium at 37 °C and 5% CO2. Every 2–3 days, partial or complete medium change was done. After 7 days, naive M0 Mɸs were obtained. To induce M1 or M2 polarization, Mɸs were stimulated with 200 ng/ml lipopolysaccharide (LPS, Merck) for 24 h or 33.3 ng/ml recombinant IL-13 (rIL-13, PeproTech) for 48 h, respectively. Conditioned medium (CM) was collected after stimulation.
Lentiviral transduction
To generate IL-13 overexpressing and Arg1 overexpressing Mɸs, naive Mɸs were transduced with pIRES-mIL-13-puro or pIRES-Arg1-puro lentiviral particles provided by the Laboratory for Viral Vector Technology & Gene Therapy (Leuven Viral Vector Core of the KU Leuven) according to previously optimized protocol with minor adaptations [
15]. Mɸs were seeded in a 24-well plate at 1.5 × 10
5 cells/well in standard medium. Positive selection of the transduced cells was made using puromycin (2.25 μg/ml, InvivoGen) for 72 h. CM was collected after puromycin selection.
qPCR
Cells were lysed using QIAzol (Qiagen), and RNA was extracted using the RNeasy mini kit (Qiagen), according to the manufacturer’s instructions. Conversion to cDNA was done with qScript cDNA supermix (Quanta Biosciences). Quantitative PCR was performed on a StepOnePlus detection system (Applied Biosciences) under universal cycle conditions. Primer sequences can be found in the Additional information (Additional file
1: Table S1). The comparative Ct method was used to determine relative gene expression. Data were normalized to the most stable housekeeping genes (CYCA, GADPH, HMBS, HPRT, YHWAZ) determined by GeNorm.
Western blot
Cells were lysed using RIPA buffer supplemented with protease and phosphatase inhibitors (Sigma-Aldrich). The protein concentration was determined by the Pierce™ BCA assay kit (ThermoFisher Scientific), according to the manufacturer's guidelines. Samples (4–10 µg) were separated on 7.5% or 12% (dependent on size target) sodium dodecyl sulfate–polyacrylamide gels by electrophoresis (1 h at 200 V) and transferred to a polyvinylidene fluoride membrane (1.5 h at 350 mA, Merck). After blocking using 5% non-fat dry milk in 1× tris-buffered saline supplemented with 0.05% Tween 20 (TBS-T), blots were incubated with primary antibodies overnight at 4 °C. Used antibodies included: mouse anti-Arg1 (1:1000, sc-271430, Santa Cruz Biotechnology), mouse anti-inducible nitric oxide synthase (iNOS, 1:500, N6657, Sigma-Aldrich), and mouse anti-β-actin (1:2000, sc-47778, Santa Cruz). After washing, secondary antibodies (anti-mouse HRP, 1:2000, Dako) were applied for 1 h at RT (room temperature). β-Actin was used as a loading control. Visualization was done using the Pierce™ ECL plus Western blotting substrate (ThermoFisher Scientific). Images were taken with an Amersham Imager 680 (GE Healthcare Bio-Sciences) and analyzed using the ImageQuant TL software. Pictures displayed in the figures were digitally enhanced to improve visibility.
ELISA
IL-4 and IL-13 secretion in the supernatants of cell cultures was measured using ELISA kits (ThermoFisher Scientific), according to the manufacturer’s instructions.
Griess assay
Nitrite secretion in the supernatants of cell cultures was measured using the Griess reagent system kit (Promega), according to the manufacturer’s instruction.
Spinal cord injury model
A T-cut spinal cord hemisection injury was performed as previously described [
15,
18]. Briefly, mice were anesthetized with 2–3% isoflurane (IsofFlo, Abbot Animal Health) and underwent a partial laminectomy at thoracic level 8. Iridectomy scissors were used to transect left and right dorsal funiculi, the dorsal horns, and the ventral funiculus, resulting in a T-cut hemisection injury. The back muscles were sutured, and the skin was closed with wound clips (Autoclip, Clay-Adams Co. Inc.). Post-operative treatment included blood-loss compensation by glucose and pain relief by buprenorphine (0.1 mg/kg bodyweight, Temgesic). Mice were placed in a recovery chamber (33 °C) until they regained consciousness. Bladders were voided daily until animals were able to urinate independently.
Cell transplantation
For in vivo transplantation, Mφs were isolated from the same mouse strain as the recipient mice in the in vivo experiment, being either C57BL/6J or BALB/c. Mφs were cultured using Nunc™ Dishes with UpCell™ Surface (ThermoFisher Scientific) to avoid enzymatic harvesting. Immediately after SCI induction, mice were randomized and received an intraspinal injection of vehicle or (GFP
+) Mφs. As previously described, a motorized stereotaxic injector pump (Stoelting) with a 10 µl Hamilton syringe (34-gauge needle) was positioned 1–3 mm rostral to the lesion [
19]. The needle was stereotactically inserted into the spinal cord at a depth of 1 mm. Either vehicle (PBS or RMPI) or cells (1 × 10
4 in 2.5 µl) were injected over a 4 min period, after which the needle was kept in place for again 4 min to allow pressure equilibration and prevent backflow. Investigators were blinded to the treatment groups during the complete experiment.
Locomotion testing
Functional recovery was assessed using the standardized Basso Mouse Scale (BMS) for locomotion (0 = complete hind limb paralysis, 9 = normal motor function) [
20]. Scores are based on hind limb movements in an open field during a 4 min interval. At the day of operation, animals were checked for locomotion, and no abnormalities were observed (BMS score = 9). The evaluation was done by an investigator blinded to the experimental groups and was performed daily from 1 until 7 days post-injury (dpi), followed by a scoring every 2nd or 3rd day. For the analysis, the mean BMS score of the left and right hind limb was used for each animal. Mice who did not display any increase in BMS score were excluded.
Immunohistochemical analysis
At 8 or 28 dpi, mice received an overdose of dolethal (Vetiquinol B.V.) and were transcardially perfused using ringer solution supplemented with heparin, followed by perfusion with 4% paraformaldehyde (PFA) in PBS. Longitudinal cryosections of 10 µm were obtained. Immunofluorescent stainings were done as previously described [
15,
18].
To determine the presence of GFP+ cells, sections were only counterstained with DAPI (1:1000, Sigma-Aldrich) for 10 min. To investigate lesion size, demyelinated area, astrogliosis, and immune cell infiltration, sections were blocked using 10% protein block (Dako) in PBS containing 0.5% Triton-X-100 for 1 h at RT. Primary antibodies included mouse anti-glial fibrillary acidic protein (GFAP, 1:500, G3893, Sigma-Aldrich), rat anti-myelin basic protein (MBP, 1:250, MAB386, Merck), rabbit anti-Iba-1 (1:350, 019-19741, Wako), goat anti-ionized calcium binding adaptor molecule 1 (Iba-1, 1:250, NB100-1028, Novus Biologicals), rabbit anti-neurofilament (NF, 1:100, MA5-14981, ThermoFisher Scientific), mouse anti-NeuN (1:1000, MAB377, Merck), rabbit anti-cleaved caspase 3 (1:100, 9661, Cell Signaling) were diluted in PBS with 1% protein block and 0.5% Triton-X-100 and were incubated overnight at 4 °C. To investigate Arg1+ and major histocompatibility complex class ll (MHCII)+ cells, an extra permeabilization step using 0.1% Triton-X-100 in TBS for 30 min was performed before blocking the sections for 1 h with 10% protein block in TBS. Primary antibodies, goat anti-Arg1 (1:50, sc-18354, Santa Cruz) and rat anti-MHCII (1:200, sc-59322, Santa Cruz), were dissolved in TBS containing 10% non-fat dry milk powder and were incubated overnight at 4 °C. Following washing, secondary antibody incubation was done for 1 h at RT. The following secondary antibodies were used: goat anti-rat Alexa fluor 488 (1:250, A11006, ThermoFisher Scientific), goat anti-mouse Alexa fluor 568 (1:250, A11004, ThermoFisher Scientific), goat anti-mouse GaM IgG1 fluor 488 (1:400, A21121, Invitrogen), donkey anti-rabbit IgG 555 (1:250, A31572, Invitrogen), goat anti-rabbit Alexa fluor 488 (1:250, A11008, ThermoFisher Scientific), donkey anti-goat Alexa fluor 488 (1:400, A11055, ThermoFisher Scientific) and donkey anti-goat Alexa fluor 555 (A21432, ThermoFisher Scientific). The specificity of secondary antibodies was tested by omitting the primary antibody. Counterstaining with DAPI was performed for 10 min. Pictures were taken using a Nikon Eclipse 80i fluorescence microscope and analyzed with NIS-Elements Viewer 4.0 software or using a LEICA DM4000 B LED microscope and LAS X software (Leica Microsystems).
To investigate lipid load, an oil red O (ORO) staining was performed. Sections were incubated with a 0.33% ORO solution (Sigma-Aldrich) for 15 min, followed by a 5 min incubation with 60% isopropanol. To wash away any unbound particles, the sections were rinsed with distilled water. Pictures were taken using the Leica DM 2000 LED microscope (Leica Microsystems).
Immunohistochemical quantification
Quantification was carried out on the original, unedited pictures. Pictures displayed in the figures were digitally enhanced to improve visibility. Quantification of histopathological parameters was performed as previously described and was done by investigators blinded to experimental groups [
15,
18]. To investigate cell survival, the number of GFP
+ cells at the injection, perilesional, and lesion site were counted. To quantify lesion size (GFAP
− area) and demyelinated area (MBP
− area), 5–7 sections per animal containing lesion center and consecutive rostral and caudal area were analyzed. The proximal and distal end of the lesion was detected by identifying the area, where the integrity of the dorsal side of the spinal cord tissue is disrupted. As no intensity was measured, contrast was altered differently among groups. To determine astrogliosis (GFAP expression) and Mφ/microglia presence (Iba-1 expression or Iba-1
+ cells), an intensity analysis was performed, or the number of positive cells was counted using ImageJ [
15]. To identify alternatively and classically activated phagocytes, Arg1
+ or MHCII
+ cells were counted caudally and rostrally from the lesion site. Quantification of foam cell presence was carried out using ImageJ software according to the method previously described by Deutsch et al. with slight adaptations [
21]. To determine neuronal cell death at the lesion site, the number of cleaved caspase 3
+NeuN
+ cells was counted at both rostral and caudal areas from the lesion epicenter and normalized to the amount of NeuN
+ cells. To quantify the interaction between axons and Mφ/microglia, the number of contacts between NF
+ dystrophic axons and Iba-1
+ cells was counted in the perilesional environment. Of note, dystrophic axons were identified by their globular and bulbus morphology extending from an axon fiber [
22,
23]. A contact was detected when a cell–cell interaction between a dystrophic axonal bulb and an Iba-1
+ cell contained a DAPI nucleus. For analysis, two standardized areas, rostral and caudal, from the lesion epicenter were used. The mean number of contacts in these two areas per animal was quantified.
Neuroblastoma cell culture
The murine Neuro2A cell line (CCL-131™, ATCC®) was cultured in RPMI 1640 medium supplemented with 10% FCS and 1% penicillin/streptomycin. The human SH-SY5Y cell line (Sigma-Aldrich) was cultured in DMEM (Gibco) supplemented with 10% FCS and 1% penicillin/streptomycin. Cells were maintained at 37 °C and 5% CO2. For immunocytochemistry, cells were seeded in a 24-well plate at 3 × 104 cells/well for 48 h. To perform cell viability assay, cells were seeded in a 96-well plate at 5 × 103 cells/well and left untreated for 24 h to attach.
Immunocytochemistry
Cells were fixed with 4% PFA and blocked for 1 h at RT with 10% protein block in PBS. Anti-IL-13Rα1 and anti-IL-4Rα (1:250, ab79277, Abcam, and 1:50, NBP1-00884, Novus Biologicals) antibodies were diluted in PBS with 1% protein block and were incubated overnight at 4 °C. Following washing, secondary antibody (donkey anti-rabbit IgG 555, 1:500) incubation was done for 1 h at RT. The specificity of the secondary antibody was tested by omitting the primary antibody. Counterstaining with DAPI was performed for 10 min. Pictures were taken using a LEICA DM4000 B LED microscope and LAS X software (Leica Microsystems).
Brain slice isolation and culture
Hippocampal entorhinal cortex slices were prepared from 8 days old pups. In short, brains were isolated in dissection medium (MEM, 2 mM
l-glutamine, and distilled water, Gibco) and using a tissue chopper (McILWAIN, H. Saur), 350 μm thick slices were made. These were cultured on inserts (0.4 μm porous, Merck) in incubation medium containing 50% MEM, 25% HBSS, 25% heat-inactivated horse serum, 4 mM
l-glutamine, 4 ng/ml insulin, 800 ng/ml ascorbic acid, 2.4 mg/ml glucose and 1% penicillin/streptomycin, Gibco) [
24]. In addition, varying concentrations of rIL-13 (0, 5, 50, and 500 ng/ml) were added to the medium for 72 h.
Neurospheroid culture and staining
Neurospheroids were generated by seeding human-induced pluripotent stem cell-derived neural stem cells (hiPSC-NSCs, 1.6 × 104 cells/well), stably expressing luciferase in an ultra-low attachment 96-well plate. Half medium changes with cNEM (1:1 advanced DMEM/F12: neurobasal medium, 1× neural induction supplement, Gibco) were performed every other day. Cultures were kept under constant orbital shaking (88 rpm).
After 4 weeks, neurospheroids were fixed with 4% PFA, and 10 μm thick slices were made for staining. The slices were permeabilized using 0.1% Triton-X-100 in TBS for 30 min before blocking the sections for 1 h with 20% goat serum in TBS. The primary antibody (mouse anti-Tuj1, 1:250, MAB1195, R&D Systems) was diluted in 10% milk in TBS and was applied overnight at 4 °C. Following washing, secondary antibody (goat anti-mouse IgG AF555, 1:200, A21425, Invitrogen) incubation was done for 1 h at RT. Counterstaining with DAPI was performed for 10 min. Pictures were taken using a BX51 fluorescence microscope equipped with an Olympus DP71 digital camera (Olympus).
Cell viability assays
In the neuroblastoma monoculture, cell death was induced via incubation with different SNAP concentrations (S-nitroso-N-acetylpenicillamine, 100–800 μM, Abcam). Cells were either pre-treated for 24 h or co-treated with various rIL-13 concentrations (5, 33.3, 50, and 500 ng/ml). After 48 h or 72 h, 0.5 mg/ml MTT (3[4,5-Dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide, Sigma-Aldrich) was added to the standard medium for 4 h, after which the cells were lysed and the formazan crystals dissolved in a mixture of DMSO and glycine (0.1 M). Absorbance was measured at a wavelength of 570 nm.
In the brain slices culture, cell death was induced via incubation with 100 μM NMDA (N-methyl-d-aspartic acid, Sigma-Aldrich) for 4 h. After washing, PI (propidium iodide, 5 µg/ml, Invitrogen) was added for 10 min. Pictures were taken using a Nikon Eclipse 80i fluorescence microscope. The PI uptake (fluorescence intensity) from the cells of the cornu ammonis region (CA1–CA3) of the hippocampus was analyzed using ImageJ software.
Neurospheroids (4 weeks) were subjected to control or oxygen–glucose deprivation (OGD) conditions for 6 h. OGD conditions were achieved by incubating the neurospheres in glucose-free NSC medium (DMEM/F12 w/o l-glutamin, w/o HEPES, w/o glucose [VWR], 1× B27 [Gibco], rhEGF and rhFGF-2 [20 ng/ml, Immunotools]) in a humidified Bactron IV anaerobic chamber (Shel Lab, USA) containing 0% O2, 5% CO2, 95% N2. After OGD, glucose (4.5 mg/ml, Gibco) was administered, and spheroids were brought back to atmospheric O2 level (21%). After 90 min of reoxygenation, human rIL-13 (10 ng/ml, Immunotools) was added to the medium for 48 h. Luminescence of the neurospheroids was determined 96 h following reoxygenation by incubating with 150 µg/ml Beetle luciferin (Promega) for 48 h.
Statistics
Data analysis was performed using GraphPad Prism version 7 (GraphPad Software). D’Agostino and Pearson omnibus test was used to assess normality. The BMS results were analyzed using a two-way ANOVA for repeated measurements with a Bonferroni post hoc test. GFAP and Iba-1 intensity were analyzed using a two-way ANOVA with a Bonferroni post hoc test. All other data were analyzed using either a one-way ANOVA with a Bonferroni post hoc test or with a non-parametric Kruskal Wallis test with a Dunn’s multiple comparison test. Data are presented as mean ± standard error of the mean (SEM). Differences with P values < 0.05 were considered significant (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001).
Discussion
It is generally accepted that neuroinflammation is a significant driver of secondary injury-related damage in SCI. During the last years, the potential of IL-13 as a treatment for neuroinflammation has been shown by us and others in different animal models of neurodegeneration [
9,
12‐
15,
19,
32]. Here, we show for the first time that IL-13 delivery by genetically modified Mφs improves functional repair after SCI and that these beneficial effects are mediated by IL-13 and not the M2a phenotype of the IL-13 Mφs. Mφs were selected as carriers above MSCs due to their strong homing ability towards inflammatory sites and because MSC-mediated actions after SCI were attributed to the induction of M2a Mφs [
15,
33]. In this study, we demonstrate that transplantation of IL-13 Mφs after SCI is an efficient approach for eliciting anti-inflammatory effects and thereby create a neuroprotective environment less prone to neuronal death and axonal retraction. Our study supports IL-13 administration as an effective immunomodulatory therapy and underlines the importance of reprogramming local Mφs/microglia to provide neuroprotection and improve functional outcome after SCI.
IL-13 secretion by Mφs has not been previously reported under physiological conditions, only in the context of pulmonary fibrosis [
34]. Through genetic transfer, we induced IL-13 secretion by Mφs. Phenotyping of these IL-13 Mφs prior to in vivo administration showed that IL-13 Mφs secreted IL-13 at similar concentrations to that used for ex vivo polarization of M2a Mφs [
12,
13]. In addition, CCR2 and CCR5 upregulation by IL-13 Mφs suggests an increased homing potential towards the lesion site, as their ligands, CCL2, and CCL3 are upregulated during the 1st days after SCI [
16,
26]. Whether the upregulation of these factors in the IL-13 Mφ group also contributes to neuroprotection remains to be elucidated. The effects of the CCL2–CCR2 pathway after SCI are controversial. While CCR2 deficiency improves outcome, CCL2 contributes to M2 Mφ polarization [
26,
35]. Interestingly, CCL2 treatment counteracted NMDA-induced cell death of motor neurons [
36,
37]. The CCR5 axis has been described to play a pro-inflammatory role after SCI, as in the absence of CCR5 or its ligand, CCL3, recovery was improved [
38,
39]. Therefore, it was rather surprising that CCR5 was upregulated by IL-13 Mφs. However, the role of CCR5 in SCI seems to be more complex as recombinant CCL5 treatment caused upregulation of M2 Mφ and axonal regeneration markers [
26]. Overall, CCL2 and CCL5 have been described to be able to provide anti-inflammatory actions [
26,
35]. Yet, since recombinant IL-13 alone has a similar neuroprotective effect in the brain slice culture, it is feasible that the CCR2 and CCR5 upregulation in IL-13 Mφs plays a minor role in protecting the spinal cord after lesion. The polarization of the IL-13 Mφs was indicated by the upregulation of the primary M2a markers Arg1, FIZZ1, and Ym1. Furthermore, their anti-inflammatory phenotype was more pronounced than M2 Mφs, potentially due to continuous autocrine signaling provided by IL-13.
As the inflammatory imbalance following SCI impedes regenerative processes, increasing the fraction of the anti-inflammatory Mφ/microglia phenotype has become a major therapeutic strategy. Ma et al. demonstrated that intravenous injection of M2 Mφs improved functional recovery after SCI [
40]. However, Kigerl et al. showed that 3 days after intraspinal injection of GFP
+-M2 Mφs, only 20–40% maintained their polarization status [
28]. This is not surprising given that lesion-associated factors favor the classically activated Mφ phenotype. Based on these results, it is tempting to speculate that preventing an M1 switch of grafted cells is a prerequisite for functional improvement. To overcome this pitfall, we used genetically manipulated Mφs to secrete IL-13. Even under a robust inflammatory stimulus, these cells maintained IL-13 secretion, and their anti-inflammatory marker expression (Arg1, FIZZ1, and Ym1 expression) was more pronounced during inflammation than rIL-13 stimulated M2 Mφs. A similar observation was made by Boehler et al., who showed that IL-10 transduction was more potent than rIL-10 stimulation to maintain an M2 polarization upon inflammation [
41]. Importantly, in our study, only IL-13 Mφs improved functional recovery after SCI, whereas the M2 Mφs did not. It is important to note that this is only a partial and no full recovery. IL-13 Mφ-treated mice reached an average score of 3.4, which is more than 2 points higher than the vehicle group. We consider this difference as biologically and therapeutically relevant, because it means that highly impaired mice which show only slight ankle movement develop a stepping pattern after IL-13 Mφ transplantation [
20]. However, these data have to be interpreted with care, because in this study, only female mice have been used, and there is evidence that sex differences may influence the BMS outcome or the inflammatory response after SCI [
42‐
44]. For example, female mice have a reduced lesion size and a higher Mφ number compared to males [
42,
43]. Surprisingly, IL-13 Mφs did show an upregulation of iNOS mRNA but not of its protein expression after LPS stimulation. The link between IL-13 and iNOS gene expression was already observed in our previous studies [
32,
45]. For example, IL-13 MSC treatment induced upregulation of Arg1 and iNOS mRNA levels in a cuprizone model [
32]. Still, IL-13 delivery was associated with anti-inflammatory effects and suppressed demyelination.
The M1:M2 ratio is a crucial determinant of functional recovery after SCI [
1]. As IL-13 is well-known for inducing alternative activation of phagocytes, we analyzed whether IL-13 Mφ injection caused a polarization shift in vivo. The perilesional environment of IL-13 Mφ-treated mice was converted as shown by an increased anti-inflammatory cell population (Arg1
+ cells) and a decreased pro-inflammatory cell number (MHCII
+ cells) without affecting the global Mφs/microglia cell presence. Notably, the increase in Arg1
+ cells was absent in the mice who received M2 Mφs. This immunomodulation of local Mφs/microglia by IL-13 is in line with previous stroke and TBI reports [
12‐
14]. Consistently, we previously showed in a study using IL-13 MSC that the Arg1-expressing population consisted mainly of invading monocytes/Mφs [
15].
Lesional lipid load was analyzed as an indirect measurement of foamy cells, which are detrimental in SCI [
46]. Mφs phagocytose apoptotic cells and myelin debris to pave the way for regeneration [
47]. However, in SCI, excessive uptake overwhelms the intracellular lipid homeostasis, subsequently transforming them into large foamy, pro-inflammatory cells [
46]. Wang et al. showed that Arg1 expression was reduced 7 dpi, while foam cell formation was observed [
47]. As our results indicate an increase in Arg1
+ cells, it may be expected that foam cell formation is reduced. Indeed, the ORO staining results showed an overall reduction in lipid load at the lesion site following IL-13 Mφ transplantation. Even though the underlying mechanism remains to be resolved, several studies indicate a role of IL-13 in phagocytosis. The importance of IL-13 in stimulating efferocytosis was described in a TBI animal model where repeated intranasal rIL-13 treatment increased the NeuN
+Iba-1
+ area [
13]. In stroke, the STAT6/Arg1 pathway was found to induce efferocytosis by Mφs/microglia [
48]. Mechanistically, Proto et al. revealed that IL-13 secretion activated IL-10 production by Mφs, which enhanced efferocytosis via autocrine signaling [
49]. In addition, in a mouse model of atherosclerosis, IL-13 acted anti-atherogenic by inducing M2 polarization. In vitro, Mφs stimulated with IL-13 increased the uptake of oxLDL by CD36 upregulation and induced the expression of the lipid exporter (ABCA1), preventing foam cell formation, which is a major driver of atherosclerosis [
50]. Based on these data, it is plausible that enhanced phagocytosis and/or processing may explain a reduction in foamy cells in our model.
Notably, the beneficial immunomodulatory effects of IL-13 Mφs were accompanied by neuroprotection. In animal models of TBI and stroke, IL-13 treatment also facilitated anti-inflammatory effects and reduced neuronal cell loss [
12,
13]. Likewise, we observed a significant decrease in the number of cleaved caspase 3
+ neurons in IL-13 Mφ-treated mice. Furthermore, damaged axons retract from the injury site after SCI, a process called axonal dieback [
46,
51]. The significant contribution of Mφs to this phenomenon has been shown by Horn et al., who demonstrated a substantial reduction in axonal dieback following peripheral Mφ depletion [
4]. In addition, M1 Mφs, in particular, have been reported to induce axonal retraction via destructive physical contacts [
4]. IL-13 Mφ treatment did not affect the number of lesional Iba-1
+ cells compared to vehicle. However, it reduced the number of contacts between dystrophic axons and phagocytes. Since the number of Iba-1
+ cells was not influenced, this suggests that axonal dieback, a limiting factor for regeneration, was reduced [
52]. Therefore, it can be speculated that induction of an anti-inflammatory perilesional environment by IL-13 Mφs creates a permissive milieu for axon regeneration by reducing neuronal cell death and axonal dieback. It cannot be excluded that direct effects on axonal plasticity may contribute to the observed beneficial effects. However, Vogelaar et al. have shown that there were no direct effects of IL-13 on axonal outgrowth in vitro [
53].
To analyze whether the beneficial effects of IL-13 Mφs were either dependent on IL-13 itself or their altered M2a phenotype, we used an IL-4R
α KO mouse model. IL-13 binding to its receptor, IL-13Rα1, will evoke the dimerization with the IL-4Rα for downstream pathway activation [
54]. Deletion of this part will prevent IL-13-induced signaling. Noteworthy, IL-4 also exerts its effects through this receptor complex and IL-4 signaling will be lost as well [
54]. However, following SCI, IL-4 levels are undetectable, challenging a significant role in the pathology [
55]. In addition, we demonstrate that IL-13 Mφs do not secrete detectable levels of IL-4 (below detection range, Additional file
9: Table S2). Vehicle-treated WT and IL-4R
α KO mice did not show differences in functional recovery after SCI. This is in line with Fenn et al., who did not observe reduced recovery after SCI in adult IL-4R
α KO mice [
56]. The functional improvement induced by IL-13 Mφs was abrogated in IL-4R
α KO mice, highlighting the role of proper IL-4R
α signaling. Besides their IL-13 secretion, IL-13 Mφs represent anti-inflammatory Mφs which overexpress Arg1 compared to rIL-13-stimulated M2 Mφs. To investigate whether this polarization was involved in their beneficial effects, Arg1 overexpressing Mφs, which do not secrete IL-13, were also grafted following SCI. However, no significant impact on functional recovery was observed. These results point towards IL-13 signaling as the critical mechanism by which IL-13 Mφs induce immunomodulation, neuroprotection, and functional recovery after SCI. Consistently, IL-13 delivery by MSCs was able to modulate Mφ/microglia polarization during the active inflammation phase in a cuprizone model [
14]. As we were unable to trace the IL-13 Mφs 8 days after SCI, initial IL-13 secretion seems to be crucial for the functional recovery, while persistence of the cells at the lesion site appears not to be required.
Finally, our results indicate that the neuroprotective effect by IL-13 Mφs was not mediated by direct action of IL-13 on neurons. In both murine and human neuroblastoma cell lines, neither pre- nor co-treatment with IL-13 protected against NO-induced cell death. Similarly, no protective effect of IL-13 on the cell viability of OGD-subjected neurospheroids was observed. However, in an murine organotypic brain slice model, IL-13 (5 ng/ml) rescued neurons from NMDA-induced cell death. This suggests that IL-13 indirectly affects the survival of neurons via IL-13R-expressing microglia, which are present in organotypic brain slice culture but not in the neuronal monocultures. Consistent with these findings, Shin et al. reported that the induction of IL-13 in microglia contributed to the survival of neurons upon inflammation [
57]. Accordingly, Miao et al. showed that IL-13 reduced neuronal death in an Mɸ neuronal co-culture by suppressing NO secretion, again highlighting the indirect neuroprotective effect of IL-13 [
13]. Aside from the impact of IL-13 on Mφs/microglia, astrocytes can also be influenced. Stimulation of astrocytes with IL-13 induced brain-derived neurotrophic factor secretion, which reduces neuronal death, stimulates axonal regeneration and improves functional recovery following SCI [
58,
59]. Regardless, we did not observe a significant difference in astrogliosis.