Long- and short-term intravital imaging reveals differential spatiotemporal recruitment and function of myelomonocytic cells after spinal cord injury

Abstract

After spinal cord injury (SCI), resident and peripheral myelomonocytic cells are recruited to the injury site and play a role in injury progression. These cells are important for clearing cellular debris, and can modulate the retraction and growth of axons in vitro. However, their precise spatiotemporal recruitment dynamics is unknown, and their respective roles after SCI remain heavily debated. Using chronic, quantitative intravital two-photon microscopy of adult mice with SCI, here we show that infiltrating lysozyme M (LysM(+)) and resident CD11c(+) myelomonocytic cells have distinct spatiotemporal recruitment profiles, and exhibit changes in morphology, motility, phagocytic activity and axon interaction patterns over time. This study provides the first in vivo description of the influx of inflammatory and resident myelomonocytic cells into the injured spinal cord and their interactions with cut axons, and underscores the importance of precise timing and targeting of specific cell populations in developing therapies for SCI.

Key points

• Inflammatory cells such as myelomonocytic cells are key players in the progression and recovery from spinal cord injury (SCI).

• However, the precise spatiotemporal distributions and roles of different subpopulations of myelomonocytic cells remain unclear in part because their dynamics have not been examined in vivo.

• Using chronic in vivo two-photon microscopy techniques in adult transgenic mice with SCI, we show that infiltrating and resident myelomonocytic cells have differential spatiotemporal distribution patterns to injury sites.

• We also show that infiltrating myelomonocytic cells are associated with the collapse of certain cut axon terminals thus potentially impeding recovery, whereas resident myelomonocytic cells clear axon debris, which may be important for axon regrowth and recovery.

• These results set a framework to understand the roles of different subpopulations of myelomonocytic cells in SCI, and may be important for the development of therapies that target specific immune cell populations at precise times post-injury.

Introduction

Following spinal cord injury (SCI), peripheral and resident myelomonocytic cells are the main effectors of the inflammatory response, yet their roles in injury progression remain controversial. For instance, macrophages of myelomocytic origin have been shown to play an essential role in reducing inflammation (Shechter et al. 2009) and breaking down the glial scar (Shechter et al. 2011), but also to cause the collapse and dieback of dystrophic axons, thus impeding regeneration (Horn et al. 2008; Busch et al. 2009, 2011). In the latter case, macrophages of both microglial and peripheral origin were shown to cause rapid axon collapse and dieback in vitro (Horn et al. 2008; Busch et al. 2009, 2011), whereas in the former case only infiltrating macrophages were beneficial (Shechter et al. 2009). Knowing their precise spatiotemporal recruitment dynamics and interactions should lead to a better assessment of their functions.

To help unravel the differential roles of infiltrating and resident myelomonocytic cells after SCI, we developed a relevant pre-clinical mouse model suitable for longitudinal non-invasive multicolour imaging at subcellular resolution. We used intravital two-photon microscopy through chronic glass windows implanted over the exposed spinal cords of adult mice with ‘pin-prick’ SCIs (Fenrich et al. 2012). We provide a quantitative assessment of recruitment and redistribution dynamics of infiltrating lysozyme M (LysM(+)) cells and resident CD11c(+) cells over the long term in the same animals, as well as their real-time interactions with axons after SCI. Our findings show that infiltrating LysM(+) cells are rapidly recruited to injury sites and are associated with the collapse of distended axon terminals (DATs) caudal of injury sites. Conversely, resident CD11c(+) cells peaked at the injury sites later, and are more closely associated with clearing axon debris rostral of injury sites.

Methods

Animals, surgery and glass window implantation

Thy1-cyan fluorescent protein (CFP)-23 (Feng et al. 2000), LysM-green fluorescent protein (GFP; Faust et al. 2000) and CD11c-enhanced yellow fluorescent protein (EYFP; Lindquist et al. 2004) mice were backcrossed to C57/Bl6 mice and then crossbred to create Thy1-CFP//LysM-GFP//CD11c-EYFP triple transgenic mice with multiple fluorescent cell populations. All surgical procedures were approved by the National Animal Studies Committee of France (authorization no. 13,300), as well as approved and authorized by the National Committee for Ethic in Animal Experimentation (Section N°14; project 86-04122012).

The window implantation protocol has previously been described in detail (Fenrich et al. 2012). Briefly, mice were deeply anaesthetized with ketamine/xylazine (120 mg kg⁻¹; 12 mg kg⁻¹), and supplemented hourly (40 mg kg⁻¹; 4 mg kg⁻¹). Following a dorsal midline incision over T12 to L2, the muscles between the spinous and transverse processes were resected using a scalpel. The animals were suspended from a spinal-fork stereotaxic apparatus (Harvard Apparatus). The dorsal musculature was further resected to expose the vertebrae, and the tips of modified staples were inserted along the edges of the T12 and L2 and glued into place with cyanoacrylate. A layer of dental cement was applied to form a rigid ring around the vertebrae, and a modified paperclip was implanted to serve as a holding point for surgery and imaging. To produce unilateral pin-prick SCI as described previously (Dray et al. 2009; Fenrich et al. 2012), a 26G needle was slowly inserted along the dorsoventral axis at the border between the gracile and cuneate fasciculi until half of the bevel was below the surface of the spinal cord (∼500 μm), and immediately withdrawn. Injuries were made roughly in the middle of the exposed spinal cord rostrocaudally. A line of liquid Kwik-Sil (World Precision Instruments) was applied to the dura mater surface along the midline of the spinal cord, and the glass window was immediately glued and cemented over the spinal cord. Post-operative analgesia was obtained by administration of dexamethazone (0.2 mg kg⁻¹) and rimadyl (5 mg kg⁻¹) s.c. immediately following surgery and every other day for the first 10 days. Animals were imaged immediately following window implantation. Animals were killed by either cervical dislocation under deep anaesthesia or anaesthetic overdose after experiment completion.

Histology

Mice were perfused with normal saline followed by 4% paraformaldehyde for ∼10 min. Spinal cords were dissected and post-fixed overnight at 4°C. Tissue sections (25 μm) at the injury sites were cut in the transverse plane using a cryostat. Sections were incubated in Rb-anti-Iba1 primary antibodies (Sobioda; 1:200), followed by incubation in DyLight650-conjugated Dk-anti-Rb secondary antibodies (Abcam; 1:100). Images were acquired with a Zeiss LSM780 microscope, and 20× or 40× water objective lenses. Fluorophores were excited using 488 nm and 633 nm lasers.

Intravital imaging

The imaging protocol has previously been described in detail (Fenrich et al. 2012). Briefly, for each imaging session mice were lightly anaesthetized with ∼1.75% isoflurane (Baxter) (v/v) in air for ∼2 min, followed by ketamine/xylazine (100 mg kg⁻¹; 10 mg kg⁻¹). For long sessions (>1 h), animals were supplemented with ∼0.4–1.0% isoflurane (v/v) in air from ∼45 min after the start of the session until completion. In all animals either 120 μl of Rhodamine B isothiocyanate-Dextran 70 kDa (Sigma; 20 mg ml⁻¹ in phosphate-buffered saline (PBS)) or 30 μl of QDot-655 (Qtracker 655 non-targeted quantum dots; Invitrogen; 50 μl ml⁻¹ in PBS) was injected i.v. before each imaging session. Throughout imaging the animals were freely breathing and the microscope chamber was warmed to ∼32°C. Following each imaging session the animals were returned to their cage with a piece of tissue for nesting and kept warm until they recovered from anaesthesia.

A tunable femtosecond pulsed laser (Mai-Tai, Spectra-Physics) was coupled to a Zeiss two-photon microscope (LSM 7 MP) equipped with a 20× water immersion objective lens (NA = 1.0) and five non-descanned detectors. The laser was tuned to 940 nm to optimize the simultaneous excitation of the fluorophore combination to be examined, and filter sets were designed to optimize the separation of the emission spectra of multiple fluorophores. For each image stack laser intensity was adjusted according to imaging depth in order to maximize image intensity while minimizing saturation throughout the image stack.

Analysis

Images were analysed using ZEN light edition (Zeiss) and ImageJ software. Analysis was performed on raw data, but all presented images are pseudo-coloured and contrast enhanced for clarity. Occasionally, there was a drift in the field of view for an image stack that was realigned using the StackReg plugin on ImageJ (Thévenaz et al. 1998). Quantitative data from adjacent days were binned for calculating averages.

For LysM-GFP and CD11c-EYFP fluorescent intensity measurements we used tiled image stacks with an acquisition area of 850 × 1700 μm, and optical sectioning every 5.0 μm. We identified a single optical section roughly 20 μm below the dorsal surface of the spinal cord from the tiled image stack obtained at 0 days. Using blood vessels and axons as anatomical markers, we extracted the corresponding optical sections at equivalent depth from the tiled images of all the other imaging sessions. A pattern of five regions of interest, each with an area of 200 × 200 μm, were stacked rostrocaudally and positioned so that the middle box was centred on the injury site for the 0 day image stacks. A sixth box of the same size was positioned equidistant from the midline, but contralateral of the injury site. The average GFP or EYFP fluorescent intensities were measured within each of the boxes, and the average fluorescent intensities of the boxes on the injury side were compared with the average fluorescent intensities of the contralateral boxes.

Results

We used Thy1-CFP//LysM-GFP//CD11c-EYFP reporter mice with CFP expression in most dorsal root ganglia axons (Feng et al. 2000), GFP expression in peripheral myelomonocytic cells including neutrophil granulocytes, circulating macrophage precursors and activated infiltrating macrophages (Cross et al. 1988; Faust et al. 2000; Hume, 2011), and EYFP expression in a subset of myelomonocytic cells resident to the spinal cord including a subset of microglia (Bulloch et al. 2008; Hume, 2011). As a first step, we designed parameters to achieve simultaneous four-colour fluorescence and second-harmonic imaging of SCI sites for extended periods of time (Fig. 1A and B; Supporting video 1). In the uninjured spinal cord, LysM(+) cells were restricted to the vascular lumen (strong LysM-GFP expression) and perivascular space (weak LysM-GFP expression). Specifically, perivascular LysM(+) cells were located between the vessel lumen and the dura (i.e. second harmonic generation emitted by collagen in dura), and were therefore peripheral to the neuropil. Conversely, CD11c(+) cells were sparse, had morphologies typical of microglia (Kettenmann et al. 2011) with ramified processes continuously extending to and retracting from surrounding neural tissues, and were located mainly within the spinal cord parenchyma. The number of circulating CD11c(+) cells was quantified using time-series videos of spinal blood vessels before and after SCI (n= >100 h of video). We found that although circulating LysM(+) cells were common, only one CD11c(+) cell was observed throughout all of the imaging sessions, thus further supporting that CD11c(+) cells are mostly resident to the spinal cord rather than infiltrating from the periphery.

Figure 1. CD11c(+) and lysozyme M (LysM(+)) cells have differential recruitment dynamics to spinal cord injury (SCI) sites over time.

A, experimental setup for long-term optical imaging, and diagram showing imaging orientation. Left, image of mouse with implanted window. Middle, enlarged view of spinal cord imaged in B. Right, diagram of intravital two-photon spinal cord imaging zone (dashed box) relative to injury site (SCI), DRG soma (DRG), retracting cut axon bulb still attached to soma (Retract.), degenerating cut axon stump (Degen.), and with blood vessels shown in red. Caudal, up; Rostral, down. B, representative two-photon image stack of an uninjured spinal cord showing 5-colour imaging including Thy1-CFP dorsal column axons, LysM-GFP cell in vessel, a ramified CD11c-EYFP microglial cell in white mater, QDot655 in blood vessels and second harmonic generation (SHG) from collagen in dura mater. C, representative two-photon image stack showing LysM-GFP, CD11c-EYFP and blood vessels labelled with QDot655 at 6 days after injury. Same region as shown in the middle panel of A. Caudal is up; Rostral is down. Scale bar: 100 μm. D–H, quantification of average relative fluorescent intensity of LysM-GFP (green lines) and CD11c-EYFP (magenta lines) caudal of the injury site (D, E), at the injury site (F) and rostral of the injury site (G, H) compared with control side over time. Error bars, SEM (n= 4 LysM-GFP//CD11c-EYFP mice; n= 3 LysM-GFP mice; n= 4 CD11c-EYFP mice). Representative two-photon image stack near an SCI site showing LysM(+) cells (I) and CD11c(+) cells (J) with amoeboid morphology and containing CFP(+) vacuoles (arrows) at 4 and 6 days post-injury, respectively. Scale bars: 20 μm. (K) Time-series images of a CD11c(+) cell undergoing cell division at 3 days post-injury. At 00:00 h, the cell is ramified with multiple branches. By 00:40 h the processes have retracted and the cell is spheroid. The cell divides between 00:40 and 00:50 h. The two new cells extend ramified processes and slowly migrate away from one another over the next several hours (09:10 h). Scale bar: 20 μm.

We then tracked distributions, morphofunctional modifications and interactions of LysM(+) and CD11c(+) cells relative to cut dorsal columns axons following injury. LysM(+) cells were rapidly recruited to the injury sites and peaked within 2–5 days. They redistributed mostly caudally towards retracting axon terminals by 6–8 days (Fig. 1C–H). At very early time-points (≤1 day) most LysM(+) cells near the injury site were spheroid or slightly ramified with few dynamic processes emanating from the soma (c.f. Fig. 2E–F and 3A), consistent with LysM-GFP expression in neutrophil granulocytes and macrophage precursors (Faust et al. 2000). However, starting at 1 day, LysM(+) cell morphology evolved progressively. The cells became larger, more amoeboid and almost always contained vacuoles, some of which were CFP(+), consistent with the morphology of activated macrophages (Fig. 1I). Moreover, nearly all LysM(+) cells had no detectable Iba1 immunoreactivity, a marker for microglia, indicating that LysM(+) cells are unlikely derived from the resident microglial population (Supplementary Fig. 1). In contrast, CD11c(+) cells peaked at the injury site between 18 and 22 days, and were preferentially distributed rostrally towards degenerating axon terminals (Fig. 1D–H). At ≤3 days most CD11c(+) cells near the injury site had a ramified morphology, which was less complex than resting microglia and consistent with progressive microglial activation (Kettenmann et al. 2011; Fig. 1K). Moreover, at 3 days we observed a CD11c(+) cell division near the injury site (Fig. 1K; Supporting video 2), which indicates that the expansion of the CD11c(+) population is due, at least in part, to proliferation. The proportion of amoeboid CD11c(+) cells (Fig. 1J) increased throughout injury progression, and they often contained CFP(+) vacuoles, especially at later time-points, consistent with the morphology of activated microglia (Kettenmann et al. 2011). Histological analysis of the lesion sites showed that 81% of CD11c(+) cells had uniform Iba1 immunoreactivity, 16% had Iba1 immunoreactivity that was either weak or restricted to punctate subcellular regions, and 3% had no detectable Iba1 immunoreactivity (Supplementary Fig. 1). Furthermore, analysis of the in vivo images revealed that <5% of CD11c(+) were also LysM(+) at 6 days, suggesting that there may be limited phenotypic overlap between the infiltrating and resident myelomonocytic cell populations. For both populations we observed a decrease in their motility over time concomitant with their evolution towards activated macrophage phenotypes (c.f. Supporting video 3 at 2 days post-injury; and Supporting video 4 at 6 days post-injury).

Figure 2. Axon retraction and deterioration occur at early post-injury times and are independent of interactions with lysozyme M (LysM(+)) and CD11c(+) cells.

A, diagram showing the zones included in the time-series imaging analysis. The legend is the same as in Fig. 1A. Caudal, up; Rostral, down. B, time-series images of a cut axon (arrowheads) that shifted position at 2 days after injury. C, time-series images showing an axon (arrowheads) degenerating at <1 day after injury. D, time-series images of a retracting cut axon terminal (arrowheads) starting at <1 day after injury. E, time-series images of a LysM(+) cell (arrow) that formed a close apposition with an injured axon with a tortuous morphology and irregular swellings (arrowhead) starting at <1 day after spinal cord injury (SCI). F, time-series images showing a LysM(+) cell forming a close apposition with an uninjured axon with normal morphology starting at <1 day after injury. Scale bars: 20 μm. For all images, the injury site was down (B–D), or down and left (E, F) of the field of view.

To examine whether LysM(+) and CD11c(+) cells affect axonal dieback, intravital two-photon time-series movies were acquired of both cut and intact axons caudal of injury sites (Fig. 2). We examined ∼115 h of time-series video between 0 and 12 days post-injury (n= 11 mice). Few changes in axon morphology were observed at time-points >6 days. Conversely, within 3 days of injury, many axons underwent rapid morphological changes such as displacement (Fig. 2B; Supporting video 3), degeneration (Fig. 2C; Supporting video 5) and retraction (Fig. 2D; Supporting video 6). These dramatic morphological changes were largely independent of close appositions with LysM(+) or CD11c(+) cells. The majority of interactions that did occur were short lasting, and the result of LysM(+) and CD11c(+) cells dynamically scanning axons. We therefore examined whether cell scanning resulted in more subtle changes in axon morphology. We focused on appositions that varied in duration from several minutes to several hours. In nearly all cases (see exception below) axon morphology remained unchanged as a result of close appositions between LysM(+) or CD11c(+) cells and injured (Fig. 2E; Supporting video 7) or uninjured (Fig. 2F; Supporting video 8) axons.

In our time-series imaging we also observed one other type of rapid morphological change, the collapse of DATs. DATs contained a large non-fluorescent vacuole surrounded by a thin CFP(+) layer (Fig. 3) and were observed most often at 1–3 days, less frequently at 4–6 days, and rarely at later post-injury times. We tracked the dynamics of 26 DATs in three mice between 1 and 6 days post-injury (average of ∼14.8 h per DAT for a total of 384.5 DAT ‘imaging hours’). The formation steps were observed for 16 DATs, of which 11 expanded from a single large vacuole in retracting axons (Fig. 3A; Supporting video 9), and the remaining five expanded from the merger of several small vacuoles in stationary bulbous terminals (Fig. 3B; Supporting video 10). Once expanded, DATs could undergo rapid changes in both shape and size, including full collapse (n= 8; rapid reduction in diameter of >60% for >30 min) and partial collapse (n= 7; rapid reduction in diameter of 30–60%, sometimes followed by recovery). Closer examination of the collapse events revealed that LysM(+) cells were in close apposition to DATs (i.e. there was no discernible gap between a LysM(+) cell and the DAT) throughout all 15 collapse events, and in several cases (n= 4) collapse occurred immediately following a close apposition between the DAT and at least one LysM(+) cell (Fig. 3C; Supporting video 11). Conversely, significantly fewer collapses occurred while in close apposition with a CD11c(+) cell (n= 4 of 15 DAT collapses; P < 0.05, Wilcoxon Signed-Rank Test). To determine whether DATs are preferentially contacted by LysM(+) cells compared with other axon terminals in the same region, the dynamics of contact between 31 cut axon terminals without DAT morphology (non-DAT) and LysM(+) and CD11c(+) cells were examined in the same three mice used for DAT analysis (average of ∼14.5 h per non-DAT for a total of 449.5 non-DAT ‘imaging hours’). We found that unlike DATs, which frequently had sustained contacts with LysM(+) and/or CD11c(+) cells (i.e. close appositions lasting longer than 10 min), only about half of the non-DATs had a sustained contact with LysM(+) or CD11c(+) cells (n= 15 of 31). Moreover, non-DATs were preferentially contacted by CD11c(+) cells (n= 13 non-DATs contacted by 20 CD11c(+) cells) compared with LysM(+) cells (n= 7 non-DATs contacted by seven LysM(+) cells; Fig. 3D; P < 0.05, Wilcoxon Signed-Rank Test). Together, these data suggest that LysM(+) cells preferentially contact DATs, and that these interactions may ‘trigger’ DAT collapse thus modulating the morphology of cut axons. Having recorded continuously the same DATs for up to 20 h with a 10 min sampling rate on different days post-injury, we found that collapsed axon terminals remained quiescent over days without signs of retraction or elongation (n= 4; Fig. 3E), despite strong in vitro evidence showing that activated macrophages mediate the collapse and actively promote the retraction of dystrophic terminals (Horn et al. 2008; Busch et al. 2009).

Figure 3. Distended axon terminal (DAT) collapse occurs during close appositions with lysozyme M (LysM(+)) cells.

A, time-series images of an injured axon (arrowheads) that retracted and formed a DAT from a single vacuole at 1 day. Scale bar: 20 μm. B, time-series images of the formation of a DAT from a non-retracting bulbous terminal by the merger of several small vacuoles (arrowheads) at 1 day. Scale bar: 20 μm. C, time-series images showing the collapse of a DAT (inset) at 3 days. Left, shows the DAT before collapse and a LysM(+) cell (arrowhead) migrating towards the DAT. Middle images of the DAT during collapse immediately following close apposition with the LysM(+) cell (arrowhead). Right image shows the axon terminal several hours after DAT collapse, and the continued interaction between the axon terminal and the LysM(+) cell (arrowhead). The arrow points to a nearby DAT that did not collapse throughout 20 h of imaging. Scale bars: 50 μm; inset: 25 μm. D, pie graphs showing the ratios of DATs (left) and non-DATs (right) that had sustained contacts with LysM(+), CD11c(+), both LysM(+) and CD11c(+) cells, and with neither cell type. E, time-series images of the same region as shown in C at 9 days. Notice that the upper axon terminal is no longer distended (upper inset), and that neither axon terminal had elongated or retracted from the point of DAT collapse. Scale bars: 50 μm; insets: 25 μm.

Discussion

In this study we used Thy1-CFP//LysM-GFP//CD11c-EYFP reporter mice with chronic glass windows implanted over SCI sites to track the spatiotemporal and morphofunctional distributions of LysM(+) and CD11c(+) myelomonocytic cells in the same animals for extended periods of time. Within hours of injury LysM(+) cells rapidly infiltrated lesion sites, were highly motile and were predominantly spheroid with few processes, typical of neutrophil granulocytes. However, within a few days the LysM(+) cell population underwent a redistribution towards retracting axons, had reduced motility and progressed towards amoeboid morphologies consistent with activated macrophages (Faust et al. 2000; Hume, 2011; Mawhinney et al. 2012). These data suggest that different subpopulations of LysM(+) cells are sequentially recruited to the injury and/or that early recruited populations undergo maturation and phenotypic changes. Conversely, CD11c(+) cell morphology transitioned from that of resting microglia with many highly dynamic processes to activated macrophages with fewer, more quiescent, processes (Bulloch et al. 2008; Hume, 2011; Kettenmann et al. 2011).

Our intravital approach also allowed us to examine axon behaviour in relation to myelomonocytic cells. The differential spatiotemporal distributions of LysM(+) and CD11c(+) cells strongly suggest that despite a common macrophagic function these cells have differential abilities to interact with axons and thus repair and/or worsen injury. The progressive activation of CD11c(+) microglia in the first week after SCI is consistent with the timeline of axon degeneration (Fenrich et al. 2012) and a role in clearing axonal debris, thus potentially contributing to a pro-regenerative environment (Neumann et al. 2009). DATs were transient structures with morphological similarities to stalled and retracting cut axon terminals (Ramon & Cajal, 1959; Tom et al. 2004; Fenrich & Rose, 2011). LysM(+) cells were closely associated with the collapse of DATs, which did not sprout beyond their collapse-point at longer post-injury times, suggesting that the long-term consequences of LysM(+) cell interaction with cut axons may be detrimental to axon sprouting.

Beyond close appositions, the contributions of LysM(+) and CD11c(+) cells to axonal fate might be through modification of the spinal cord microenvironment. This could be through paracrine signalling (Hume, 2008), the release of pro-inflammatory cytokines (Stow et al. 2009) or the recruitment of effector cells not labelled in this study (Minagar et al. 2002). For instance, it is well documented that macrophages and dendritic cells are derived from a common precursor (Hume, 2008; Chow et al. 2011), and that macrophages have antigen presenting capabilities (Hume, 2008). Whether LysM(+) and CD11c(+) cells have differential antigen presenting capabilities or time-evolving antigen presenting capabilities could be important in regulating the adaptive immune response and recovery after SCI (Laliberte & Fehlings, 2013). Finally, it is possible that LysM-GFP and CD11c-EYFP markers do not reveal all infiltrating and resident macrophages, and we therefore cannot exclude the presence of other subpopulations of macrophages, suggesting that a more thorough characterization of these cells using tools such as multi-parametric flow cytometry during the course of spinal cord healing is warranted in future studies. In any case, our observations emphasize that strategies to repair the injured spinal cord by modulating macrophages need a precise knowledge of their recruitment and function to be effective.

Our intravital microscopy approach provides a framework for further studies using relevant mouse models to advance our understanding of the role of immune cells after SCI and testing whether removing some subpopulations, such as blocking the recruitment of LysM(+) cells between 1 and 6 days, may be an effective means to control injury progression and axonal regrowth.

Glossary

CFP

cyan fluorescent protein

DAT

distended axon terminal

EYFP

enhanced yellow fluorescent protein

GFP

green fluorescent protein

LysM

lysozyme M

PBS

phosphate-buffered saline

SCI

spinal cord injury

Additional information

Competing interests

The authors declare no competing of financial interest.

Author contributions

K.K.F., G.R. and F.D. conceived and designed the study, interpreted the data, and wrote the paper; K.K.F and F.D. analysed the data; K.K.F., P.W. and F.D. collected the data.

Funding

This work was supported by an institutional grant from Centre National de la Recherche Scientifique, and by grants from the Association de Recherche sur la Sclérose en Plaque (ARSEP) (to G.R. and K.K.F), Institut Thématiques MultiOrganismes (ITMOs) Neurosciences and Immunologie, (to G.R.), Agence Nationale de la Recherche (ANR JCJC PathoVisu3Dyn), Institut de Recherche sur la Moelle Epinière (IRME) (to F.D.). K.K.F. was supported by an ARSEP fellowship. Imaging was performed on the PIcSIL imaging facility of the IBDML.

Author's Present address

Keith K. Fenrich: University of Alberta, Faculty of Rehabilitation Medicine, Center for Neuroscience, 3-88 Corbett Hall, Edmonton, AB, Canada T6G 2G4.

Supplementary material

Supplementary Fig. 1

Supporting video 1

Supporting video 2

Supporting video 3

Supporting video 4

Supporting video 5

Supporting video 6

Supporting video 7

Supporting video 8

Supporting video 9

Supporting video 10

Supporting video 11