ADAM17-deficiency on microglia but not on macrophages promotes phagocytosis and functional recovery after spinal cord injury

A disintegrin and metalloproteinase 17 (ADAM17) is the major sheddase involved in the cleavage of a plethora of cytokines, cytokine receptors and growth factors, thereby playing a substantial role in inflammatory and regenerative processes after central nervous system trauma. By making use of a hypomorphic ADAM17 knockin mouse model as well as pharmacological ADAM10/ADAM17 inhibitors, we showed that ADAM17-deficiency or inhibition significantly increases clearance of apoptotic cells, promotes axon growth and improves functional recovery after spinal cord injury (SCI) in mice. Microglia-specific ADAM17-knockout (ADAM17flox+/+-Cx3Cr1 Cre+/-) mice also showed improved functional recovery similar to hypomorphic ADAM17 mice. In contrast, endothelial-specific (ADAM17flox+/+-Cdh5Pacs Cre+/-) and macrophage-specific (ADAM17flox+/+-LysM Cre+/-) ADAM17-knockout mice or bone marrow chimera with transplanted ADAM17-deficient macrophages, displayed no functional improvement compared to wild type mice. These data indicate that ADAM17 expression on microglia cells (and not on macrophages or endothelial cells) plays a detrimental role in inflammation and functional recovery after SCI.


Introduction
Spinal cord injury (SCI) is characterized by a primary insult triggering a cascade of partially detrimental inflammatory processes, which aggravate initial tissue damage and impair neuronal regeneration (Beattie et al., 2002;Dooley et al., 2016;Gyoneva and Ransohoff, 2015). Microglia, the resident immune cells of the central nervous system (CNS), and infiltrating monocyte-derived macrophages are major players in this neuroinflammatory response (David et al., 2015). Microglia cells respond and enter quickly after injury, whereas monocyte-derived macrophages start entering the injured area 3 to 4 days post-injury (Orr and Gensel, 2018). Being highly plastic cells, microglia and macrophages modulate the inflammatory response by secreting pro-and anti-inflammatory cytokines and by clearing noxious factors such as apoptotic/necrotic cells (David et al., 2015;Fu et al., 2016;Graeber, 2010;Horn et al., 2008). A simplistic way to distinguish subsets of microglia/macrophages is to divide them into classically and alternatively activated cells. Alternatively activated microglia/macrophages are considered to be less inflammatory due to reduced secretion of pro-inflammatory cytokines and reduced production of nitric oxide compared to classically activated cells (David et al., 2015).
The modulation of detrimental inflammatory processes is a major goal in SCI therapy. One key player in the modulation of inflammatory processes is 'A Disintegrin And Metalloproteinase 17′ (ADAM17). ADAM17 is a membrane-bound sheddase involved in the proteolytic cleavage of not only TNFα, but hitherto more than 80 different known substrates including cytokines (e.g. TNFα), growth factors (e.g. TGFα), and multiple receptors (e.g. CSFR1) (Zunke and Rose-John, 2017). Thereby, this major sheddase affects a plethora of physiological processes during development, inflammation and tissue homeostasis as well as tissue regeneration. In neuroinflammatory pathologies, ADAM17 is playing an essential role due to the shedding of prohttps://doi.org/10.1016/j.bbi.2019.02.032 Received 11 December 2018; Received in revised form 12 February 2019; Accepted 27 February 2019 inflammatory mediators and the release of phagocytic receptors impairing the phagocytic capacity of phagocytes. Thereby, ADAM17 is involved in the progression of CNS pathologies as for example multiple sclerosis and traumatic brain injury (Weinger et al., 2009;Zhang et al., 2016). By using rather broad acting pharmacological inhibitors, it has been shown that ADAM17 plays an important role in the survival of glial cells and the regenerative processes after SCI (Rego et al., 2014;Vidal et al., 2013;Wei et al., 2015).
We demonstrated that the systemic inhibition or deficiency of ADAM17 not only improves functional recovery, but also modulates immune cell infiltration and increases regeneration of serotonergic fibers after SCI. More specifically, we identified that microglia-specific ADAM17-deficiency improves functional recovery after SCI at a level similar to that observed following systemic ADAM17 inhibition or deficiency. Furthermore, we showed that in both hypomorphic ADAM17 and ADAM17flox +/+ -Cx3Cr1 Cre +/− transgenic mice, the phagocytic receptor CD36 was significantly upregulated and correlated with more interactions between microglia/macrophages and apoptotic cells, suggesting a favorable inflammatory response upon ADAM17-deficiency. These results suggest that the inhibition of ADAM17 may be a novel therapeutic target to promote functional and histological recovery after SCI.
Hypomorphic ADAM17 mice (ADAM17 ex/ex ) and wild type littermates (ADAM17 wt/wt ) were used for in vitro and in vivo studies (Chalaris et al., 2010). ADAM17 ex/ex mice express low levels of ADAM17 in all cells due to the introduction of an additional exon between exon 11 and 12 of the ADAM17 gene. The new exon starts with an in-frame stop codon but is only used inefficiently due to altered splice/donor acceptor sites resulting in < 5% ADAM17 expression as compared to wild type mice (Chalaris et al., 2010). All in vivo experiments with ADAM17 ex/ex and ADAM17 wt/wt were performed with 10-to 12-week old animals. To further investigate the effect of ADAM17 on myeloid cells, 10-week old ADAM17flox +/+ -LysM Cre −/− (ADAM17 MФWT) and ADAM17flox +/ + -LysM Cre +/− (ADAM17 MФKO) mice were used. Chimeric mice were generated by whole body irradiation with 8 Gy of C57BL/6 females (Envigo, the Netherlands). Femurs and tibias were removed from female ADAM17 ex/ex or female ADAM17 wt/wt mice. Bone marrow was flushed with sterile PBS (phosphate buffered saline, pH: 7.5). Mice were reconstituted with 10 7 cells into the tail vein of recipients within 4 h post-irradiation. Bone marrow was allowed to engraft for at least 8 weeks before the mice were used for experiments. During recovery, mice were treated with Neomycin (FSA Chemicals NV, Belgium) and Polymyxini B sulfas (Fagron, the Netherlands) added to the drinking water.
In all experiments age-and sex-matched mice were used. Mice were housed in groups under standardized conditions with a 12-hour light-dark cycle. The mice had ad libitum access to food and water. All strains were kept in identical housing conditions. The experiments were performed according to the guidelines of EU Directive 2010/63/EU on the protection of animals used for scientific purposes. All experiments were approved by the local ethical committee at Hasselt University.

Experimental spinal cord injury
A T-cut spinal cord hemisection was performed as previously described (Boato et al., 2010;Nelissen et al., 2014;Vidal et al., 2013). Briefly, mice were anesthetized with 3% isoflurane (IsofFlo, Abbot Animal Health, Belgium) and a partial laminectomy was performed at thoracic level 8 (T8). A complete transection of the dorsomedial and ventral corticospinal tract was induced bilateral dorsal T-cut hemisection using iridectomy scissors. Muscles were sutured and the back skin was closed with wound clips (Autoclip, Clay-Adams Co. Inc., US). A glucose solution (20%) was administered i.p. to compensate blood loss during surgery. As post-operative pain treatment, buprenorphine (0.1 mg/kg bodyweight Temgesic, Val d'Hony Verdifarm, Belgium) was administrated subcutaneously close to the lesion site. Mice were placed in a recovery chamber (33°C) post-surgery. Investigators remained blinded to the treatment groups for the duration of the study. Bladders of the mice were emptied daily manually until micturition reflex returned spontaneously.

Assessment of functional recovery following spinal cord injury
Functional recovery after SCI was assessed using the Basso Mouse Scale (BMS) test (Basso et al., 2006). The BMS is a 10-point locomotor rating scale (9 = normal locomotion; 0 = complete hind limb paralysis). Mice were scored by one to two blinded investigators in an open field. During the first 8 days, mice were scored daily and afterwards only every second day. The given scores are based on hind limb movements made in an open field during a 4-min interval. The analysis was done using the mean of the left and right hind limb scores for each animal.

Injection of apoptotic Neuro-2a cells into the spinal cord
The Neuro-2a cell line, a murine neuroblast cell line purchased from American Type Culture Collect (CCL-131; France), was cultured in DMEM (41966-029, Gibco) medium supplemented with 10% fetal calf serum (FCS; Biochrom AG, Germany) and 1% penicillin/streptomycin (referred to as DMEM complete). Apoptosis was induced by UV exposure at 254 nm for 15 min followed by 1 h incubation (Mazaheri et al., 2017). Cells were labelled with PKH26, a red fluorescent cell linker (PKH26MINI-1KT, Sigma-Aldrich) according to the manufacturer's instruction.
Mice were subjected to a laminectomy at T8, but no T-cut hemisection was performed. Instead a motorized stereotaxic injector pump (Stoelting, Ireland) was used as described before . The needle was stereotactically inserted into the spinal cord at 4 positions at a depth of 1 mm and 1 × 10 4 cells in 1 µL of DMEM medium were injected over a time period of 1 min. The needle was subsequently kept in place for an additional minute to allow pressure equilibration and prevent backflow of the injected cells.

Microglia isolation from adult ADAM17flox +/+ -Cx3Cr1 Cre mice
Microglia cells from tamoxifen-treated ADAM17flox +/+ -Cx3Cr1 Cre +/− and control mice were isolated as previously described (Nikodemova and Watters, 2012). Briefly, brains were isolated after transcardial perfusion. All steps were performed on ice unless stated otherwise. Brain stem and choroid plexus were removed from brains. After enzymatic digested for 1 h with collagenase from Clostridium histolyticum (175 U/mL; C2674, Sigma-Aldrich), tissue was mechanically dissociated. After enzymatic digestion with DNase for 15 min at room temperature (RT) and centrifugation, the homogenate was resuspended in 30% isotonic Percoll (17-0891-02, GE Healthcare) and centrifuged at 700 g for 10 min. After an additional centrifugation step, the pellet was again incubated with DNase (0.05 mg/mL) in PBS. After 15 min incubation, cells were stained with CD11b MicroBeads (130-093-636, Miltenyl Biotec) according to the manufacturer's instructions. CD11b + cells were separated in a magnetic field using MS columns (Miltenyl Biotec). The collected CD11b + cells were either lysed with RNA lysis buffer or in RIPA lysis buffer.

RNA extraction and quantitative PCR
Mice underwent SCI surgery as described above. Mice received an overdose with Dolethal (Vetiquinol NV, Belgium) and were transcardially perfused with Ringer solution including heparin as previously described (Nelissen et al., 2014).Standardized areas of the spinal cord tissue (5 mm rostral and 5 mm caudal to the lesion center) were collected and snap frozen in liquid nitrogen. RNA was isolated using the PARIS kit (AM1921, Life Technologies, Belgium) according to the manufacturer's protocol with minor modifications as described before . RNA was extracted from total cell lysates using the RNeasy Mini Kit (74104, Qiagen, the Netherlands) according to the manufacturer's instructions. After reverse transcription to cDNA using qScript cDNA Supermix (95048-100, Quanta Biosciences, US), a quantitative PCR was conducted on a StepOnePlus detection system (Applied Biosciences, Belgium) using standardized cycling conditions (20 s at 95°C, 40 cycles of 3 s at 95°C and 30 s at 60°C). Primer sequences are listed in Table 1.

Isolation and preparation of protein samples
Protein concentration was measured in spinal cord tissue following SCI at 1, 3, 7, 14 and 28 days post-injury as described above. Mice received an overdose with Dolethal and were transcardially perfused with Ringer solution including heparin as previously described (Nelissen et al., 2014). Protein was isolated using the PARIS kit (Life Technologies) as described before .  Sommer, et al. Brain, Behavior, and Immunity 80 (2019)

Cytometric bead array -cytokine/chemokine protein levels
Cytokine/chemokine protein levels were determined locally in the injured spinal cord at 3 and 7 dpi. The protein levels of pro-inflammatory cytokines were quantitatively determined via flow cytometry analysis using the mouse Cytometric Bead Array (CBA; 560485, BD Biosciences) Th1/Th2/Th17 cytokine kit according to the manufacturer's instructions and as previously described .

Protein analysis
Western blot analysis was performed against different protein targets using protein samples isolated from spinal cords or cell lysates.

Quantitative image analysis
For measurement of lesion size and demyelinated area, 4-7 sections per animal containing the lesion center as well as consecutive rostral and caudal areas were analyzed. Lesion size (GFAP immunofluorescence), demyelinated area (MBP immunofluorescence), astrogliosis (GFAP expression) and microglia/macrophage activation (Iba-1 expression) were analyzed as previously described Vangansewinkel et al., 2016). All images were acquired at once and microscope settings remained the same for all the images. To quantify (alternatively) activated microglia/macrophages at the lesion site, the number of positive cells was counted on sections stained against MHCII and Arginase-1. CD4 + T cells were identified through a double labelling against CD4 and Iba-1 to exclude possible CD4 + microglia cells. The number of CD4 + T cells was counted throughout the entire spinal cord section as previously described . The mean number of TMEM119 + cells counted rostral and caudal to lesion was calculated per animal. To quantify CD36 expression on microglia/macrophages, a double labelling against Iba-1 and CD36 was performed. The mean number of CD36/Iba-1 + -cells counted rostral and caudal to lesion site was calculated per animal. Quantification of microglia/macrophage interactions with apoptotic cells was performed doing a double labelling for Iba-1 and cleaved caspase 3 (a well-known marker for apoptosis). An interaction was defined as the formation of a contact of microglial/macrophage processes or the formation of a microglia/ macrophage pouch with an apoptotic cell (Abiega et al., 2016). The mean number of interactions between Iba-1 + and cleaved caspase 3 + cells determined rostral and caudal to the lesion was calculated per animal. Quantification of microglia/macrophage and axon interactions was performed by counting the number of contacts between neurofilament + dystrophic axon bulbs and Iba-1 + microglia/macrophages. Dystrophic axonal bulbs were identified based on their globular and bulbus morphology extending from an axon fiber. A contact was determined when a cell-cell interaction was observed between a dystrophic axonal bulb and an Iba-1 + cell which contained a DAPI + Fig. 2. ADAM17-deficiency results in a better functional recovery, modified immune cell infiltration and a greater number of serotonergic fibers after SCI. (A) ADAM17 gene expression in spinal cord tissue from ADAM17 ex/ex and ADAM17 wt/wt mice. n = 3/group (B) Representative immunoblot showing ADAM17 expression in spinal cord homogenates of ADAM17 wt/wt and ADAM17 ex/ex mice. (C) Quantification of TNFα levels in spinal cord homogenate from ADAM17 ex/ex and ADAM17 wt/wt mice at 3 and 7 dpi. Uninjured mice were used as control. n = 3-9/group, *p < 0.05; two-way ANOVA with a Bonferroni post-hoc test (D) Schematic representation of the experimental set-up. (E) Functional recovery after SCI from ADAM17 ex/ex and ADAM17 wt/wt mice. n = 28-30/group from 4 independent experiments. ***p < 0.001 vs ADAM17 wt/wt mice; two-way ANOVA with a Bonferroni post-hoc test. (F-T) Histological analysis on spinal cord sections at 28 dpi. Quantification of (F) demyelinated area, (G) lesion size, (H) astrogliosis, (I) microglia/macrophages, (J) number of CD4 + T cells, (K) Arginase-1 + cells and (L) MHCII + cells. (M,N) Representative images from MHCII + cells in ADAM17 wt/wt and ADAM17 ex/ex mice. Scale bars represent 100 µm. (O-Q) Quantification and representative images of TMEM119 + cells at 28 dpi Scale bars represent 100 µm. (R-T) Quantification and representative images of 5-HT + fibers in ADAM17 wt/wt and ADAM17 ex/ex mice at 28 dpi. Scale bars represent 100 µm. n = 11-20/group; *p < 0.05; **p < 0.01 vs ADAM17 wt/wt ; Mann Whitney U test. Data are shown as mean ± SEM. D. Sommer, et al. Brain, Behavior, and Immunity 80 (2019) 129-145 nucleus. Analysis was performed in two standardized areas rostral and caudal from the lesion epicenter and the mean number of contacts in these two areas was calculated per animal as previously described . For analysis of serotonergic fibers, sections were stained for 5-HT as described previously (Boato et al., 2010). The total fiber length was assessed caudal and rostral to the lesion using ImageJ open source software. The mean ratio of 5-HT + fibers caudal and rostral to the lesion was calculated per animal.

Isolation and stimulation of microglia cells for in vitro studies
Primary microglia cultures were prepared from postnatal P0-P3 pups derived from ADAM17 wt/wt and ADAM17 ex/ex mice as previously described (Tamashiro et al., 2012). The tissue was dissociated in DMEM complete. Cell suspension was seeded per T75 culture flask. Cells were maintained in humidified incubators at 37°C with 5% CO 2 in air atmosphere. After reaching 80% confluency, DMEM complete containing 1/3 LCM was added. Microglia cells were isolated 6-7 days later using the shake-off method (200 rpm, 2 h, 37°C). Microglia were seeded in 24-well plates at a density of 2 × 10 5 cells per well in 2/3 DMEM complete and 1/3 LCM 24 h prior to stimulation. Cells were treated with a respective stimulus (0.2 µg/mL of LPS; 0.1 µg/mL of IFNγ; 0.033 µg/mL of IL-4) for 24 h or 48 h. After 24 h or 48 h, conditioned medium was collected and samples were collected for gene or protein analysis, respectively.

Nitrite formation
The nitrite concentration in conditioned medium was measured using the Griess reagent system kit (G2930 Promega).

In vitro phagocytosis assays
BMDM or microglia isolated from ADAM17 ex/ex mice and wild type littermates were seeded in a 24-well plate at a density of 2 × 10 5 cells per well in RPMI complete supplemented with 15% LCM or DMEM complete. After 24 h, cells were treated with a respective stimulus (0.2 µg/mL of LPS; 0.1 µg/mL of IFNγ; 0.033 µg/mL of IL-4). After overnight incubation DiI-labelled apoptotic Neuro-2a cells (ratio 4:1) (Wagner et al., 2011) were added to the cells. After 90 min, the cells were washed with PBS and collected in FACS buffer (1 × PBS supplemented with 2% FCS and 0.1% sodium azide). The amount of phagocytosed DiI-labelled apoptotic cells was determined using the FACS Fortessa (BD Biosciences).

Statistical analysis
All statistical analyses were performed using GraphPad Prism 5.01 software (GraphPad Software Inc., US). Functional recovery in vivo and histological analyses of astrogliosis and Iba-1 intensity were statistically analyzed using two-way ANOVA for repeated measurements following Bonferroni post-hoc test. Differences between three groups were evaluated using a Kruskal Wallis test following Dunn's multiple comparison test. Differences between two groups were analyzed using the nonparametric Mann Whitney U test. For in vivo and in situ studies, n values refer to the number of mice. A p value of < 0.05 was considered to be statistically significant. Data in graphs are presented as mean ± standard error of the mean (SEM) unless otherwise stated.

Pharmacological inhibition of ADAM10 and ADAM17 improves functional recovery and reduces the number of MHCII + cells after SCI
Due to the sequence homology between ADAM10 and ADAM17 and their shared substrates (Dreymueller and Ludwig, 2017), we investigated whether ADAM10 and/or ADAM17 can modulate functional and histological recovery after SCI. To do so, we followed an indirect strategy using a specific ADAM10 and a combined ADAM10/ADAM17 pharmacological inhibitor (Hundhausen et al., 2003). C57BL/6 mice were subjected to SCI and were treated either with GW280264x (combined ADAM10/ADAM17 inhibition), GI254023x (specific ADAM10 inhibition), or PBS + DMSO (control) (Fig. 1A). Mice treated with GW280264x showed significantly improved functional recovery compared to the control group (Fig. 1B). In contrast, inhibition of ADAM10 alone had no significant effect on functional recovery compared to control treated mice (Fig. 1B).
A trend towards a decreased demyelinated area in GW280264x compared to control mice was observed ( Fig. 1C-E), whereas no difference was observed between GI254023x and control treated groups (Fig. 1F). There was no difference in lesion size, astrogliosis and microglia/macrophages (Fig. 1G-I) in GI254023x, GW280264x and control treated mice. However, there was a significant decrease in the number of MHCII + cells in mice treated with GI254023x and GW280264x compared to controls (Fig. 1J-M). No difference was observed in the number of Arginase-1 + cells or CD4 + T cells in GI254023x or GW280264x compared to control treated mice (Fig. 1N-O). Furthermore, there was no difference in the ratio of 5-HT + fibers between the 3 groups (Fig. 1P). Taken together, these results indicate that the combined pharmacological inhibition of ADAM10/ ADAM17ADAM10/ADAM17 reduces the number of MHCII + cells and improves functional recovery following SCI suggesting that ADAM17 is responsible for these effects.
Quantification of demyelinated area at 28 dpi revealed a significant increase in ADAM17 ex/ex mice compared to wild type littermates (Fig. 2F). There was no difference in lesion size, astrogliosis and microglia/macrophages presence (Fig. 2G-I) in ADAM17 ex/ex mice compared to wild type controls. Quantification of CD4 + T cells throughout the spinal cord revealed a significant increase in ADAM17 ex/ex mice compared to wild type littermates (Fig. 2J). While there was no difference in the number of Arginase-1 + cells (Fig. 2K), a significant reduction in the number of MHCII + cells at the lesion site was determined in ADAM17 ex/ex compared to wild type mice (Fig. 2L-N). Further analysis in vitro revealed that ADAM17-deficiency does not lead to a consistent phenotype switch in bone marrow-derived macrophages and primary microglia (supplementary Fig. S1). The number of TMEM119 + microglia cells was significantly increased in ADAM17 ex/ex mice compared to controls (Fig. 2O-Q). ADAM17 ex/ex mice displayed a significant increase in the ratio of 5-HT + fibers compared to wild type littermates ( Fig. 2R-T). We further investigated whether ADAM17-deficiency affects axonal dieback and counted therefore the number of microglia/macrophage-axon interactions using Iba-1 and neurofilament staining ( Supplementary Fig. S2A). No difference in the number of microglia/macrophage-axon contacts between ADAM17 ex/ex and wild type littermates was observed.

ADAM17-deficiency increases apoptotic receptor CD36 expression levels and Iba-1-apoptotic cell interactions after SCI
The data provided above demonstrate that ADAM17-deficiency leads to an improved functional recovery after SCI. In addition, ADAM17-deficiency may increase the number of microglia cells and alter the inflammatory environment by reducing levels of soluble TNFα and the number of MHCII + cells. As professional phagocytes, microglia and macrophages play an important role in the resolution of inflammation. Since ADAM17 affects the expression levels of several phagocytic receptors, we investigated whether ADAM17-deficiency influences phagocytosis after SCI.
We investigated gene expression levels of selected phagocytic receptors in ADAM17 ex/ex and wild type control mice sacrificed at 1, 3 and 7 dpi. Gene expression analysis of pro-and anti-inflammatory cytokines revealed no clear differences between ADAM17 ex/ex mice and wild type littermates (Suppl. Fig. S3). CD36 expression levels were significantly increased at 7 dpi in ADAM17 ex/ex mice compared to wild type littermates (Suppl. Fig. S4A). CD91, TREM2 and Megf10 gene expression levels were significantly increased at 3 dpi in ADAM17 ex/ex compared to wild type control mice (Suppl. Fig. S4B-D), whereas no significant differences were observed in the expression of CD163, MerTK, CD93 and Megf8 ( Supplementary Fig. S4E-H). Next, we investigated protein expression levels in the receptors which showed a significant change on gene expression after SCI. Western blotting revealed a significant increase in CD36 expression at 7 dpi in ADAM17 ex/ ex compared to wild type mice ( Fig. 3A and B). No significant differences were determined in CD91, TREM2 and MEGF10 levels ( Fig. 3C-E).
Histological analysis revealed no differences in Iba-1 expression and in the number of MHCII + and Arginase-1 + cells between ADAM17 ex/ex mice and wild type littermates (Fig. 3F, J, G-H, K-L). Quantification of CD36/Iba-1 double positive cells revealed an increase in numbers of CD36 + microglia/macrophages in ADAM17 ex/ex mice compared to wild type controls (Fig. 3I, M−O).
To test whether ADAM17-deficiency affects migration and phagocytosis of Iba-1 + cells, we injected apoptotic neurons into the spinal cord of ADAM17 ex/ex mice and wild type littermates. Apoptotic cells play a detrimental role during secondary injury after SCI and are known to release so-called "find-me" signals which can attract professional phagocytes (David et al., 2015;Ravichandran, 2010). Histological analysis revealed a significantly higher Iba-1 intensity at the injection site in ADAM17 ex/ex compared to wild type mice (Fig. 3P-X).
Increased CD36 expression on phagocytes (Fig. 3I, M−O) may indicate increased clearance of apoptotic cells. Therefore, we investigated the interaction of microglia/macrophages with apoptotic cells using Iba-1 and cleaved caspase 3 double labelling. While the number of Iba-1 + microglia/macrophages and cleaved caspase 3 + cells were similar between the groups, the number of microglia/macrophage-apoptotic cell interactions was significantly increased at 3 and 7 dpi in ADAM17 ex/ex mice compared to wild type littermates (Fig. 3Y-AF). Furthermore, in vitro analysis with ADAM17-deficient macrophages and microglia cells revealed a trend towards a better phagocytic capacity of apoptotic cells by ADAM17-deficient BMDM and a significantly higher phagocytic capacity of apoptotic cells by ADAM17-deficient microglia upon IL-4 stimulation (Suppl. Fig. S5) Taken together, these results indicate that ADAM17-deficiency leads to a significant increase in phagocytic receptor expression, particularly CD36, and the interaction of microglia/macrophages with apoptotic cells after SCI.

ADAM17-deficiency on myeloid cells does not lead to a better functional recovery and does not affect immune cell infiltration after SCI
Overall, the results provided so far suggest that ADAM17 plays a detrimental role in functional recovery after SCI. Since immune cell infiltration and activation were altered in ADAM17-deficient mice after SCI (Fig. 2) and given that ADAM17 is a ubiquitously expressed protease, we further investigated the effect of ADAM17-deficiency on specific cell types. Due to the effect of ADAM17 on barrier function, we investigated the effect of ADAM17-deficiency on endothelial cells. Our results strongly indicate that endothelial-specific ADAM17-deficiency does not affect functional recovery and immune cell infiltration Functional recovery after SCI of C57BL/6 ex/ex and C57BL/6 wt/wt mice. n = 11-13/group. Graph bars represent SEM. (N-P) Representative images and quantification of lesion size. Scale bars represent 500 µm. Quantification of (Q) demyelinated area, (R) astrogliosis, (S) microglia/macrophages, (T) MHCII + cells, (U) Arginase-1 + cells and (V) CD4 + T cells in at 28 dpi in C57BL/6 ex/ex and C57BL/6 wt/wt mice. n = 8-13/group, *p < 0.05 vs C57BL/6 wt/wt; Mann Whitney U test. Data are shown as mean ± SEM. D. Sommer, et al. Brain, Behavior, and Immunity 80 (2019) 129-145 following SCI (Suppl. Fig. S6).
In order to investigate whether ADAM17-deficiency specifically on myeloid cells leads to a better functional recovery after SCI, we crossed mice carrying a floxed ADAM17 allele (ADAM17flox +/+ ) with LysM Cre transgenic mice expressing Cre-recombinase under the control of the LysM-promoter. This promoter is mainly expressed in myeloid cells, and thus, we compared ADAM17flox +/+ -LysM Cre +/− (ADAM17 MФKO) with ADAM17flox +/+ -LysM Cre −/− (ADAM17 MФWT) mice (Clausen et al., 1999;Horiuchi et al., 2007). To confirm the knockdown efficiency in this mouse line, we determined ADAM17 mRNA expression (Fig. 4A) and protein levels (Fig. 4B) in BMDM isolated from ADAM17 MФWT and ADAM17 MФKO mice. Functional recovery after SCI was assessed (Fig. 4C). No difference in functional recovery in ADAM17 MФKO compared to ADAM17 MФWT mice was observed (Fig. 4D). Histological analysis also revealed no differences in demyelinated area, lesion size, astrogliosis and the number of microglia/ macrophages, MHCII + , Arginase-1 + or CD4 + T cells in ADAM17 MФKO compared to ADAM17 MФWT mice (Fig. 4E-K). Although the LysM Cre system is widely used to study myeloid-specific deletions, it is well-known to be highly variable in its deletion efficiency (Murray and Wynn, 2011). Thus, to provide further evidence that ADAM17-deficiency on myeloid cells is not responsible for the beneficial effects of ADAM17 reduction in ADAM17 ex/ex mice, we transplanted ADAM17 ex/ ex or ADAM17 wt/wt bone marrow into irradiated C57BL/6 mice. Functional recovery of recipients of ADAM17 ex/ex (C57BL/6 ex/ex) and ADAM17 wt/wt (C57BL/6 wt/wt) bone marrow was assessed (Fig. 4L). There was no difference observed in functional recovery between the two groups (Fig. 4M). Quantification of the lesion size revealed a significant decrease in C57BL/6 ex/ex compared to C57BL/6 wt/wt mice ( Fig. 4N-P). Further histological analysis revealed no differences in demyelinated area, astrogliosis, microglia/macrophages and in the number of MHCII + , Arginase-1 + or CD4 + T cells (Fig. 4Q-V). Together, these results indicate that ADAM17-deficiency on myeloid cells does not significantly affect functional recovery and immune cell infiltration after SCI.

Microglia-specific ADAM17-deficiency improves functional recovery and reduces the number of MHCII + cells after SCI
Despite the fact that tissue analysis in ADAM17 ex/ex mice at 28 dpi revealed no significant difference in Iba-1 intensity, a significant increase in numbers of TMEM119 + microglia compared to wild type littermates was observed (Fig. 2O). Thus, we investigated whether microglia-specific ADAM17-deficiency results in an improved functional recovery after SCI. To assess the effect of microglial-specific ADAM17-deficiency in vivo, ADAM17flox +/+ mice were crossed with Cx3Cr1 CreERT2 transgenic mice expressing Cre-recombinase under a tamoxifen-inducible Cx3Cr1 promoter, a promoter expressed in monocytes, monocyte-derived macrophages, tissue-specific macrophages and microglia Yona et al., 2013). This model allows microglia-specific targeting due to the longevity and self-renewal of microglia cells compared to monocytes and monocyte-derived macrophages (Goldmann et al., 2013).

Discussion
In this study, we demonstrate for the first time that the absence of ADAM17 specifically on microglia, but not macrophages or endothelial cells, promotes phagocytosis and improves histological and functional recovery after SCI in mice. We have used 4 different genetically modified mouse lines as well as bone marrow chimera to demonstrate in vivo that microglial ADAM17 plays a detrimental role in the inflammatory and regenerative processes after SCI.
Only a few studies investigated so far the role of ADAM17 in neuroinflammatory pathologies such as traumatic brain injury and multiple sclerosis (Geribaldi-Doldán et al., 2018;Katakowski et al., 2007;Plumb et al., 2006;Zunke and Rose-John, 2017). The role of ADAM17 in SCI is still elusive and until now only studied using rather broad acting pharmacological inhibitors (Vidal et al., 2013;Wei et al., 2015). ADAM10 and ADAM17 are closely related sheddases within the ADAM family. Both sheddases play an essential role in a number of physiological and regenerative processes (Saftig and Reiss, 2011;Zunke and Rose-John, 2017). Interestingly, our results using specific and selective pharmacological inhibitors for ADAM10 and ADAM17 demonstrate that both enzymes affect the pro-inflammatory environment after CNS injury, whereas only ADAM17 plays an essential role in functional recovery after SCI. We further confirmed these data by using the hypomorphic ADAM17 knockin mouse model with a reduction of ADAM17 protein levels in all tissues (Chalaris et al., 2010). In the present study, ADAM17 ex/ex mice showed a better regeneration of serotonergic fibers and a better functional recovery after SCI. As expected, the demyelinated area was increased in ADAM17 ex/ex mice due to the important role of ADAM17 in the maturation of oligodendrocyte precursor cells (Palazuelos et al., 2015).
One important function of ADAM17 is the shedding of pro-inflammatory mediators such as TNFα, IL-6R or IL-1R from the plasma membrane. We show that upon ADAM17-deficiency, the pro-inflammatory response is attenuated after SCI, as the number of MHCII + cells (interpreted as classically activated microglia/ macrophages) is reduced. Whereas the number of TMEM119 + microglia and CD4 + T cells is increased upon ADAM17-deficiency. The specific subtype of the CD4 + T cells is unclear. Specific T cell phenotyping after CNS trauma is highly challenging and is prone to artifacts due to the relatively low number of T cells present in the CNS (Beck et al., 2010;Hendrix et al., 2013). Corroborated by others, ADAM17-deficiency or inhibition appears to reduce the number of classically activated microglia or macrophages in the CNS, whereas no changes in the number of alternatively activated macrophages have been observed Madrigal et al., 2002). In contrast to the study by Vidal et al., no differences in lesion size or Iba-1 expression were determined between ADAM17 ex/ex and wild type mice (Vidal et al., 2013). Furthermore, given that we did not observe differences in Iba-1 expression and lesion size with our specific and selective pharmacological ADAM10/ADAM17 inhibitors, we assume that this discrepancy may be a result of off-target effects of the pharmacological inhibitor.
The clearance of apoptotic cells is essential to resolve inflammation and to prevent further tissue damage (Neumann et al., 2009). Among the different phagocytic receptors, only CD36 was found up-regulated in ADAM17 ex/ex mice following SCI. Together with the increased interaction between professional phagocytes and apoptotic cells in ADAM17 ex/ex mice, our results are consistent with previous studies showing that ADAM17-mediated shedding of phagocytic receptors is impairing the resolution of inflammation by affecting the phagocytic behavior of professional phagocytes (Driscoll et al., 2013;Thorp et al., 2011).
The improved functional recovery observed in ADAM17 ex/ex compared to wild type mice began in the sub-acute phase after SCI. Based on this, it is tempting to speculate that ADAM17-deficiency affects processes taking place during the acute and sub-acute period after SCI. This is further corroborated by the fact that high amounts of soluble TNFα are almost immediately released upon CNS injury leading to a cascade of cellular dysfunctions and excessive inflammatory processes (Pineau and Lacroix, 2007;Takahashi et al., 2003).
In the present study, we show a reduction in TNFα in spinal cord homogenates of ADAM17 ex/ex mice at 3 dpi, which may be correlated with reduced inflammation in the (sub-) acute phase following SCI. As expected, TNFα levels reduced to baseline in spinal cord homogenates at 7 dpi. Thereby, no difference between ADAM17 ex/ex mice and wild type controls was observed (Nelissen et al., 2014). Transmembrane and soluble TNFα have distinct effects. Transmembrane TNFα is preferentially signaling via TNFR2 and is regulating gene expression associated among others with cellular survival and resolution of inflammation. Soluble TNFα is preferentially signaling via TNFR1 and has neurotoxic effects, influences the blood-brain permeability as well as neuronal plasticity and is therefore a mediator of neuronal damage and neuropathologies (Olmos and Lladó, 2014).
ADAM17 is the major sheddase of numerous substrates involved in inflammatory responses such as TNFα, IL-6R and others (Zunke and Rose-John, 2017). It is tempting to speculate that ADAM17 influences, due to the release of its substrates, the activation state of macrophages and microglia. This is further based on studies showing that IL-6 signaling through membrane-bound IL-6R is required for differentiation of macrophages into an alternatively activated state (Guerrero et al., 2012;Mauer et al., 2014). In contrast to our expectations, our in vitro results indicate that the release of ADAM17 substrates from macrophages and microglia cells has no strong effect on their activation state.
Since ADAM17 is expressed on various cell types, we aimed to identify whether ADAM17-deficiency on a specific cell type is affecting inflammation and regeneration after SCI. Therefore, we conditionally inactivated ADAM17.
No difference in functional recovery in myeloid-specific ADAM17-deficient mice was observed. We observed a significant reduction in lesion size in C57BL/6 ex/ex mice compared to C57BL/6 wt/wt mice. However, there is no clear correlation between lesion size and functional recovery (Inman and Steward, 2003). Lesion size is only one of several factors, which can influence the functional recovery after SCI. As shown by Driscoll and colleagues, ADAM17-deficient macrophages display a better resolution of apoptotic cells in vivo (Driscoll et al., 2013). Thereby, it can be speculated that ADAM17-deficiency on macrophages may have led to a better clearance of inflammatory and inhibitory factors, thereby to reduced tissue loss and a reduction in lesion size in vivo. Our data demonstrated that ADAM17 expression on infiltrating myeloid cells does not affect the inflammatory environment or functional recovery after SCI. In ADAM17 ex/ex mice, a higher number of TMEM119 + microglia cells was observed in spinal cord tissue at 28 dpi. Consequently, we hypothesized that ADAM17-deficiency affects microglia cells and further investigated this using an inducible microglia-specific ADAM17knockout mouse model. Microglia-specific ADAM17-deficient mice displayed significantly more fiber regeneration compared to control mice after SCI. Axon regrowth is a prerequisite for functional recovery after SCI. Major obstacles for axonal regrowth following SCI are among others environmental factors such as glial scar, axonal dieback and inhibitory cellular debris (Filous and Schwab, 2018). We did not observe differences in astrogliosis in the hypomorphic ADAM17 mouse model and the microglia-specific ADAM17-deficient mouse model; therefore, it is feasible to speculate that ADAM17-deficiency exerts its beneficial effects not via the modulation of the glial scar. The lengthy retraction of damaged axons from the injury site is known as axonal dieback. Together with the significantly fewer classically activated microglia/macrophages observed at 28 dpi, it was tempting to speculate that ADAM17-deficiency might affect microglia/macrophage-induced axonal dieback Evans et al., 2014). In contrast to our expectations, ADAM17-deficiency did not alter the number of microglia/macrophage-axon contacts following SCI. Myelin and cellular debris are obstacles for remyelination and neurite outgrowth as well as axonal regeneration (Kotter et al., 2006;Kottis et al., 2002). In both, ADAM17 ex/ex and microglia-specific ADAM17 knockout mice, the observed higher expression of CD36 and larger number of microglia/ macrophage interactions with apoptotic cells, suggest an improved clearance of apoptotic cells compared to controls. This in turn may have reduced the levels of inhibitory factors, resulting in improved histopathological and functional outcome. In vitro phagocytosis assays are known to have several shortcomings among other due the missing variety of stimuli influencing the phagocytic capacity of microglia and monocyte-derived macrophages in vivo (Hellwig et al., 2013). Keeping these limitations in mind, our in vitro data support the hypothesis that ADAM17-deficiency on microglia and monocyte-derived macrophages influences their phagocytic capacity.
In summary, in this study we show for the first time that ADAM17deficiency specifically on microglial cells (but not on macrophages or endothelial cells) promotes histological and functional recovery after SCI in mice. Unexpectedly, ADAM17 does not substantially alter the macrophage/microglia phenotype (Cabron et al., 2018). Increased expression of the phagocytic receptor CD36 and a higher number of interactions between microglia/macrophages and apoptotic cells suggest an increase in phagocytic activity. These data combined with higher numbers of spared, sprouting or regenerating serotonergic fibers in ADAM17 ex/ex mice (and a trend in microglia-specific ADAM17-deficient mice), may suggest a greater clearance of apoptotic bodies as well as other noxious factors, thereby reducing inflammation. It can be speculated that a less inflammatory milieu promotes regeneration of serotonergic fibers and consequently functional recovery after SCI. Taken together, these data provide first evidence that microglial ADAM17 plays a key role in post-SCI inflammation and exerts detrimental effects on functional and histological recovery after SCI.