Abstract
Microglia are activated after spinal cord injury (SCI), but their phagocytic mechanisms and link to neuroprotection remain incompletely characterized. Docosahexaenoic acid (DHA) has been shown to have significant neuroprotective effects after hemisection and compression SCI and can directly affect microglia in these injury models. In rodent contusion SCI, we demonstrate that DHA (500 nmol/kg) administered acutely post-injury confers neuroprotection and enhances locomotor recovery, and also exerts a complex modulation of the microglial response to injury. In rodents, at 7 days after SCI, the level of phagocytosed myelin within Iba1-positive or P2Y12-positive cells was significantly lower after DHA treatment, and this occurred in parallel with an increase in intracellular miR-124 expression. Furthermore, intraspinal administration of a miR-124 inhibitor significantly reduced the DHA-induced decrease in myelin phagocytosis in mice at 7 days post-SCI. In rat spinal primary microglia cultures, DHA reduced the phagocytic response to myelin, which was associated with an increase in miR-124, but not miR-155. A similar response was observed in a microglia cell line (BV2) treated with DHA, and the effect was blocked by a miR-124 inhibitor. Furthermore, the phagocytic response of BV2 cells to stressed neurones was also reduced in the presence of DHA. In peripheral monocyte-derived macrophages, the expression of the M1, but not the M0 or M2 phenotype, was reduced by DHA, but the phagocytic activation was not altered. These findings show that DHA induces neuroprotection in contusion injury. Furthermore, the improved outcome is via a miR-124-dependent reduction in the phagocytic response of microglia.
Introduction
Spinal cord injury (SCI) is a significant cause of disability worldwide, affecting ~130 000 people annually (1). Despite extensive research, there is still no effective treatment, and the various cascades of secondary pathology post-injury are poorly understood. SCI is associated with a strong inflammatory response, and microglia, resident central nervous system (CNS) immune cells become activated immediately after injury (2,3). Activated microglia take on an amoeboid-like morphology that facilitates their migration to the site of the injury, where they release a range of cytotoxic mediators (4,5). Although activated microglia have many similarities with monocyte-derived macrophages, they also have differences at the genetic and phenotypic levels (6). Furthermore, microglia have been shown to induce a form of primary phagocytosis, termed phagoptosis, of stressed but viable neuronal cells (7–9). However, microglia are also known to act as sources of anti-inflammatory mediators and neurotrophic factors, aiding CNS repair (10,11). Microglia initiate both innate and adaptive immune cell responses post-SCI via pro-inflammatory cytokine release and major histocompatibility complex II antigen expression (5,12). Therefore, microglia exhibit a wide range of functions, which can be neurotoxic or neuroprotective.
A robust and protracted macrophage response accompanies microglial activation after SCI. Whilst acute depletion or inhibition of CNS macrophages is neuroprotective (13,14), augmentation of this immune cell response can also, paradoxically, lead to CNS repair (15,16). These divergent effects of activated microglia/macrophages might be explained by the induction of functionally separate phenotypes, such as the classically activated (M1) and alternatively activated (M2) macrophage responses (3,6). M1 macrophages produce high levels of pro-inflammatory cytokines and oxidative metabolites that are essential for host defence, but can also cause significant bystander damage (17,18). On the other hand, the presence of interleukin (IL) IL-4 or IL-13 promotes the differentiation of M2 macrophages, which are responsible for several neuroprotective and reparative functions (19,20). It is now well documented that in the early phases of SCI, activated macrophages may acquire a short-lived M2 polarized phenotype, identified by the cell marker arginase-1 (arg-1), whereas chronic spinal cord inflammation is characterized by a strong and persistent M1 polarized phenotype, identified by the presence of inducible nitric oxide-2 (iNOS-2; 21). Therefore, macrophage activity may exacerbate secondary injury or promote tissue repair post-SCI, depending on which subpopulation predominates, at different times after the initial injury.
The neuroprotective and macrophage/microglia modulatory effects of docosahexaenoic acid (DHA), and other closely related omega-3 polyunsaturated fatty acids such as eicosapentaenoic acid (EPA), have been well documented in compression and hemisection SCI (22–27). Although the exact pharmacological targets of omega-3 PUFAs driving their neuroprotective effects in SCI remain elusive, a range of beneficial effects have been reported after their acute administration post-injury in models of compression and hemisection injury, including decreased axonal and myelin damage, increased neuronal and oligodendrocyte survival, reduced lipid peroxidation and COX-2 activation and improved locomotor recovery (22,24,25,27–29). However, contusion injury represents a pathophysiologically different type of neurotrauma and may better reflect human SCI than hemisection or compression injuries (30). We have recently shown, using in vivo positron emission tomography imaging in a rat contusion SCI model, that DHA can alter the levels of the translocator protein (TSPO), an inflammatory marker which is up-regulated in glia after injury (31).
In this study, we show that acute DHA treatment after injury promotes sustained locomotor recovery following spinal cord contusion injury in both rats and mice. This was accompanied by increased microglia/macrophage levels, but a reduced myelin phagocytosis, in parallel with an increase in microglial microRNA (miR)-124 expression. In vitro, DHA treatment reduces myelin phagocytosis in primary adult rat spinal microglia cultures and in the murine microglial cell line BV-2. This phagocytic response appears to involve miR-124. The reduction in phagocytosis induced by DHA occurs in parallel with the neuroprotective effects, thus providing a mechanism that may be responsible for the beneficial effects of this fatty acid in contusion SCI.
Results
DHA improves open field locomotor scores in rodent spinal cord contusion injury
To assess the degree of motor functional loss and subsequent recovery in DHA-treated animals, the Basso–Beattie–Bresnahan (BBB) and Basso Mouse Scale (BMS) open field locomotor rating scores were used to quantify recovery of hind limb function following contusion SCI in rats and mice, respectively. Using the scoring parameters mentioned in the methods section, both rats and mice showed complete paralysis of both hind limbs immediately following contusion SCI (Fig. 1A and B). The injury paradigm used for rat SCI induced paralysis for only 1 day post-injury (BBB score of 0.6 for vehicle versus 1.2 for DHA at 1 day post-injury), indicating a moderate spinal contusion (Fig. 1A). In the DHA-treated mice, a total paralysis (BMS score of 0) was observed for 3 days, whereas vehicle-treated mice were paralysed for 5 days post-injury, indicating a severe spinal contusion (Fig. 1B). Although previous dose studies from our laboratory have shown that 250 nmol/kg DHA exerts significant neuroprotective effects in lateral hemisection and compression SCI in rats (22,25,32), 500 nmol/kg compared to 250 nmol/kg DHA exerted a more effective neuroprotection in compression-injured mice (24). Furthermore, preliminary studies in contusion injury in rats did not show significant differences in neuroprotection between 250 and 500 nmol/kg (data not shown). Therefore, to maintain similarity in the DHA dose between the rat and mouse contusion injury model, the higher DHA dose was chosen in this study, for both species. Intravenous administration of DHA (500 nmol/kg) 30 min post-contusion SCI did not induce significantly different BBB scores in DHA- versus vehicle-treated rats on days 0–4 post-contusion. However, significant improvement in locomotor recovery was observed from days 5 and 11 onwards, in rat and mouse models, respectively (Fig. 1A and B). The enhanced locomotor recovery in DHA-treated rats and mice remained significant for up to 28 and 35 days post-contusion, respectively (Fig. 1A and B). These data indicate that a single bolus injection of DHA (500 nmol/kg) can improve locomotor function during the early phase of recovery (7 days), with benefits sustained into the later phases of recovery (28 and 35 days), after moderate and severe contusion SCI in rodents.
DHA improves open field locomotor scores in rodent spinal cord contusion injury. (A) DHA-treated Sprague-Dawley rats showed significantly improved locomotor recovery when compared to vehicle-treated animals on days 5, 6, 7, 14, 21 and 28 post-contusion SCI when using the BBB locomotor rating scale. (B) DHA-treated C57BL/6 mice also showed significantly improved locomotor recovery when compared to vehicle-treated animals on days 11–35 post-contusion SCI when using the BMS locomotor rating scale. The i.v. dose of DHA was 500 nmol/kg. Results are displayed as the mean ± SEM. * and ** represent P < 0.05 and P < 0.01 versus vehicle-treated animals, respectively.
DHA induces neuroprotection in rodent spinal cord contusion injury
The administration of DHA led to significantly improved neuronal survival on days 7 and 28 post-contusion injury in rats, as demonstrated by the number of neurons immunostained for the neuronal marker NeuN, around the injury epicentre (Fig. 2A–J). In particular, significant neuroprotection in DHA-treated compared to the vehicle-treated rats was observed in the regions rostral and caudal to the epicentre at day 7 post-injury (Fig. 2E–G) and regions caudal to the epicentre at day 28 post-injury (Fig. 2H–J). Furthermore, significant neuronal survival was observed in DHA-treated mice, in the region rostral and caudal to the injury epicentre of the intermediate plane at 35 days post-contusion, when compared to the vehicle-treated group (Fig. 2K–S). These results show that DHA can protect against neuronal loss after a moderate or severe contusion of the cord.
DHA prevents NeuN immunopositive cell loss in rat and mice after contusion injury. (A–D) The sections rostral and caudal to the injury site of the contused rat spinal cord were immunostained for NeuN and analysed. Insets are enlarged images of the corresponding dashed boxes. (E–G) DHA-treated rats showed significantly increased NeuN immunolabelled cells when compared to vehicle-treated animals at 2–6 mm rostral to the injury site at different ventral and intermediate horizontal planes, at 7 days post-injury (dpi). (H–J) DHA-treated rats showed significantly increased NeuN immunolabelled cells when compared to vehicle-treated animals at 2–4 mm caudal to the injury site, at 28 dpi. (K–P) The sections rostral and caudal to the injury site of the contused mouse spinal cord were immunostained for NeuN and analysed. Insets are enlarged images of the corresponding dashed boxes. (Q) No difference in the number of NeuN immunolabelled cells was observed between DHA-treated mice and vehicle-treated mice in the ventral horn at 35 dpi. (R) DHA-treated mice showed significantly more NeuN immunolabelled cells in both caudal and rostral regions when compared to vehicle-treated animals at 35 dpi. (S) DHA-treated mice have significantly more neurons 3 mm rostral to the injury site when both ventral horn and intermediate cord data were combined. Results are displayed as the mean ± SEM. * represents P < 0.05 versus vehicle-treated animals. Scale bars = 100 μm. Ventral and intermediate indicate the ventral and intermediate horizontal planes. Insets are enlarged images of the corresponding dashed boxes.
DHA modulates microglia/macrophage activity in rodent spinal cord contusion injury
After SCI, there is an increase in activated microglia and the macrophage activation at the injury site (21,22). In the present study, at 7 days post-contusion, DHA-treated rats displayed a small but significant increase in ED1 (a marker of activated microglia and monocyte-derived macrophages) expression at 4 mm caudal to the epicentre when compared to control animals (Fig. 3A–C). Immunostaining for Iba1, a marker that labels activated and non-activated microglia and macrophages, was also carried out. At 7 days post-contusion there was also a significant increase in Iba1-positive cells at the epicentre, and in regions rostral and caudal to the epicentre, in the DHA-treated group compared to vehicle-treated rats (Fig. 3D–F). Interestingly, there was also an overall increase in the number of Iba1-positive cells in the white matter of mice at 35 days post severe spinal contusion injury after DHA treatment, compared to the vehicle-treated group (Fig. 3H–K, L). However, in the grey matter, the number of Iba1-positive cells was significantly reduced in the DHA-treated versus the vehicle-treated group (Fig. 3M). However, at 28 days post moderate spinal contusion in rat, there was no significant difference in Iba1 immunostaining between the two groups (Fig. 3G). Furthermore, there was no significant difference when the grey and white matter data were combined, in the mouse contusion study (Fig. 3N). These results demonstrate that DHA can modulate microglia/macrophages in both the white and grey matter caudal and rostral to the injury site after moderate–severe contusion.
DHA induces regional and temporal differences in microglia/macrophage cell populations after spinal cord contusion injury in rats and mice. (A and B) The microglia/macrophages present rostral and caudal to the injury site of the contused rat spinal cord were immunostained for ED1 and analysed. (C) DHA-treated animals showed significantly increased ED1 immunolabelled cells when compared to vehicle-treated animals at 4 mm in the rostral and caudal directions during the early phase (7 dpi; P < 0.05). (D and E) The microglia/macrophages present rostral and caudal to the injury site of the contused rat spinal cord were immunostained for Iba1 and analysed. Insets are enlarged images of the corresponding dashed boxes. (F) DHA-treated animals showed significantly increased Iba1 immunolabelled cells when compared to vehicle-treated animals at the lesion epicentre and both 2–4 mm in the rostral and caudal directions during the early phase (7 dpi). (G) No significant difference in Iba1 immunolabelling was observed between DHA- and vehicle-treated animals during the late phase (28 dpi). (H–K) The microglia/macrophages present in grey and white matter rostral and caudal to the injury site of the contused mouse spinal cord were immunostained for Iba1 and analysed. Insets are enlarged images of the corresponding dashed boxes. Arrows indicate Iba1 immunolablled cells. (L) DHA-treated animals showed significantly less grey matter Iba1 immunolabelling when compared to vehicle-treated animals at the injury epicentre at 35 dpi. (F) DHA-treated animals showed significantly greater white matter Iba1 immunolabelling when compared to vehicle-treated animals, but not in any specific rostro-caudal region at 35 dpi. (G) No significant difference was observed when both grey and white matter data were combined. Results are displayed as the mean ± SEM. * represents P < 0.05 versus vehicle-treated animals. Scale bars = 100 μm.
DHA enhances both M1 and M2 microglia/macrophage phenotypes in spinal cord contusion injury
In order to further characterize the role of DHA in the modulation of activated microglia/macrophages after contusion of the cord, an additional analysis of microglia/macrophages using the CD16/32 and arginase-1 antibodies was carried out in mice and rats, at early or later stages of the SCI, to identify the two major subtypes of activated microglia/macrophage phenotype, namely M1 and M2.
As immunostaining with CD16/32 and arginase-1 antibodies on rat contused spinal cord sections was unsuccessful, these antibodies were instead used for western blotting on freshly dissected tissue obtained from separate groups of animals. To fully appreciate the acute effects of DHA along the entire rostro-caudal spinal cord axis, the expression of CD16/32 and arginase-1 was analysed in sequential 3 mm rostro-caudal segments of contused rat spinal cord at 3 days post-injury (Fig. 4A). Both markers were detected in the contused tissue, and a si gnificant increase in both CD 16/32 and arginase-1 was observed at the injury epicentre (in comparison to βIII tubulin control protein) after DHA treatment (Fig. 4B–F).
DHA enhances M1 and M2 microglia/macrophage expression after spinal cord contusion injury in rats and mice. (A) Schematic diagram showing the rostral and caudal regions examined in the western blot. (B) The expression levels of CD16/32 and (C) arginase-1 were up-regulated after DHA compared to vehicle treatment at 3 dpi in rats. (D) βIII tubulin was used as the protein loading control. (E and F) Data expressed as mean ± SEM from four animals per group showed significant differences in CD16/32 and arginase-1 at the injury epicentre. (G–J) The M1 microglia/macrophages immunostained for CD16/32 (arrows) showed that DHA-treated animals had a significant increase in CD16/32 immunolabelled cells when compared to vehicle-treated animals at the injury epicentre in the grey matter and 1 mm caudal in the white matter at 35 days post-contusion. (K–N) The M2 microglia/macrophages immunostained for arginase-1 (arrows) demonstrated that DHA-treated animals have a significant increase in arginase-1 immunolabelled cells when compared to vehicle-treated animals at the injury epicentre in the grey and white matter and 1 mm caudal in the grey matter at 35 days post-contusion. Results are displayed as the mean ± SEM. * and ** represent P < 0.05 and P < 0.01 versus vehicle-treated animals, respectively. Scale bars = 100 μm. Insets are enlarged images of the corresponding dashed boxes. Nuclei labelled with DAPI are represented in blue.
Furthermore, to investigate whether an increase in CD16/32 and arginase-1 expression is also detected in another model of SCI, we studied a lateral hemisection model in mice using immunohistochemistry. In the lateral hemisection mouse model at 7 day post-injury, there was limited CD16/32 (M1 phenotype) expression level, but high arginase-1 (M2 phenotype) expression level that was confined to the ipsilateral side of the injury (Supplementary Material, Fig. S1). In comparison to the vehicle group, at 35 days post-contusion in mice, markers for both phenotypes were detected. Treatment with DHA significantly increased the number of CD16/32 immunopositive cells at the injury epicentre in the grey matter (Fig. 4G–I) and immediately caudal to the injury epicentre in the white matter (Fig. 4J). Similarly, in comparison to the vehicle group at 35 days post-contusion in mice, treatment with DHA significantly increased the number of arginase-1 immunopositive cells at the injury epicentre and immediately caudal to the epicentre in the grey matter (Fig. 4K–M), as well as at the injury epicentre within white matter (Fig. 4N). These results suggest that DHA is able to enhance the M1 and M2 phenotypes at the acute and chronic stages of spinal cord contusion injury, in the two species.
DHA reduces the M1 phenotype in peripheral primary murine bone marrow-derived macrophages, but not the phagocytic activity
To characterize the effect of DHA on microglia and monocyte-derived macrophages originating in the periphery, we first tested primary murine bone marrow-derived macrophages, which were used to investigate the effect of DHA on the peripherally originating macrophage population. The concentration of DHA used was similar to that previously used in other primary cultures studies (25). DHA at 1 and 3 μm did not significantly alter the number of non-activated (M0) or M2 macrophages (Fig. 5A–I). However, DHA at 3 μm significantly reduced the number of M1 macrophages (Fig. 5A–H). These results suggest that DHA can significantly reduce the M1 phenotype, but not the M0 and M2 phenotypes of peripheral monocyte-derived macrophages. In an additional experiment, the effect of DHA on the phagocytic activation of M1 or M2 macrophages was studied. No significant change was observed between the DHA-treated group at 3 micro, compared to the vehicle control group (Fig. 5J–R). Therefore, the observed modulatory effects of DHA on M1/M2 markers in the spinal cord after injury are unlikely to involve peripherally originating macrophages.
DHA reduces selectively the number of murine M1 bone marrow-derived macrophages (BMDMs), but does not alter their phagocytic activity. (A–F) The BMDMs can either be in a non-polarized phenotype (A and B) or be selectively polarized into an M1 phenotype using LPS and IFN-γ (C and D) or M2 phenotype using IL-4 (E and F). (G and I) For the non-polarized (M0) and M2 phenotypes, DHA (1–3 μM) did not significantly alter the number of cells when compared to the control treatment. (H) However, for the M1 phenotype, DHA (3 μm) significantly reduced the number of cells when compared to the control treatment. (J–O) BMDM immunostained for CD16/32 (green) can phagocytose DiI-labelled myelin after 24 h co-incubation. There was no significant change in DiI-labelled myelin levels within BMDM treated with 3 μm DHA compared to vehicle treatment at M0 (P), M1 (Q) and M2 (R) phenotypes. Two wells per treatment and four independent experiments were carried out. Results are displayed as the mean ± SEM. * represents P < 0.05. Scale bars = 100 μm.
DHA increases miR-124 in microglia and reduces phagocytosis of myelin in vivo
Since microglia have phagocytic activity in the CNS (5,33), the significant increase in activated microglia by DHA could result in increased phagocytosis and improved debris clearance. To test this hypothesis, the levels of myelin co-localized within microglia/macrophages were analysed using immunohistochemistry. At the injury epicentre level, there was a high level of myelin present within Iba1-positive cells of vehicle- and DHA-treated rats, at 7 days post-contusion (Fig. 6A–F). However, to our surprise, the level of myelin present within Iba1-positive cells at 4 and 6 mm rostral to the injury epicentre in DHA-treated rats was significantly reduced compared to vehicle-treated rats (Fig. 6G–M). To further understand how DHA may alter microglial function, we studied a particular microRNA, miR-124, which has been shown to have an important association with microglia/macrophage activation. Thus, miR-124 has been previously shown to correlate inversely with microglia/macrophage activation, as well as inducing the M2 neuroprotective phenotype (34,35). Using the fluorescent in situ hybridization (FISH) technique, we show that there was a significant increase in miR-124 expression in the Iba1-positive microglia/macrophages of DHA-treated compared to vehicle-treated SCI animals at 7 days post-injury in contused rats (Fig. 7A–G).
DHA reduces microglial phagocytosis of myelin in the white matter in rat spinal cord 7d after contusion. (A–L) Microglia/macrophage immunostained with Iba1 (middle colum) can phagocytose myelin (first colum) at 7d after spinal cord contusion. (A-F) Vehicle-or DHA-treated animals showed high levels of phagocytosis at the epicentre level. (G–M) Vehicle-treated animals showed a significantly higher level of phagocytosis of myelin compared to DHA-treated animals at 7d after spinal contusion. Arrows illustrate high phagocytosis of myelin and arrowheads illustrate low phagocytosis of myelin. Results are displayed as the mean ± SEM. * and *** represent p < 0.05 and p < 0.001 vs vehicle-treated animals, respectively. Scale bars = 50 mm.
DHA increases miR-124 expression in microglia in white matter in rat spinal cord 7d after contusion. (A–F) Microglia/macrophage immunostained with Iba1 (top row) have a lower miR-124 expression (middle row) in the vehicle-treated group compared to DHA-treated animals. (G) A significant increase in miR-124 expression was observed at the injury epicentre and rostral regions along the spinal cord. Arrows indicate microglia with miR-124 expression. Results are displayed as the mean ± SEM. *, ** and *** represent p < 0.05, p < 0.01 and p < 0.001, vs. vehicle-treated animals, respectively. Scale bars = 100 mm.
To investigate how these changes are associated with neurological outcome, we studied the phagocytosis of myelin in mouse contusion SCI. Recently, the P2RY12 gene (purinergic receptor P2Y12) has been found uniquely in microglia and not monocyte-derived macrophages (6,36,37). Therefore, the co-expression of P2Y12 and myelin-binding protein (MBP) as a myelin marker was analysed in microglial cells in vehicle- and DHA-treated mice, at 7 days post-contusion, and the neurological outcome was assessed in parallel (Fig. 8A–K). In the rostral and caudal regions, there was a significant reduction in the MBP staining within P2RY12-positive cells, after administration of DHA (Fig. 8B–K). These observations in both contused rats and mice suggest that the phagocytic property of microglia is reduced by DHA. Furthermore, we also showed a significant increase in miR-124 expression in P2Y12 positive microglia of DHA-treated compared to vehicle-treated SCI mice at 7 days post-injury (Fig. 9). Importantly, intraspinal administration of a miR-124 inhibitor significantly reduced the miR-124 levels (Fig. 9) and prevented the reduced phagocytic response of microglia, and this was parallel with the compromised neurological recovery (Fig. 8). This suggests that the reduced phagocytic activity of microglia is associated with an increase in miR-124 expression within microglia. Interfering with this modulation of miR-124 activity appears to have a major impact on the beneficial effects of DHA.
DHA reduces microglial phagocytosis of myelin via miR-124 predominantly in the white matter of mouse spinal cord 7 d after contusion. (A) Mice receiving DHA and intraspinal negative control injections (blue line) showed a significantly improved locomotor activity when compared to animals receiving vehicle and the negative control (red line), DHA with the miR-124 inhibitor (blue-dashed line) and vehicle with the miR-124 inhibitor (red-dashed line) when using the BMS locomotor rating scale, (B–K) DHA-treated mice with intraspinal negative control injections showed significantly fewer co-localizations of MBP positive myelin (red, white arrows) within microglia immunostained with P2Y12 (green) compared to the other treatment groups, in particular, in the rostral (F) and caudal (H), but not the epicentre (G) white matter regions at or near contusion site. No significant change was observed in the rostral (I) and caudal (K) and majority of the epicentre (J) grey matter regions at or near contusion site. Results are displayed as the mean ± SEM. † represent P < 0.05, **, ^^ and †† represent P < 0.01, and ††† and ^^^ represent P < 0.001. *, †, ^ symbols represent vehicle with negative control, vehicle with miR-124 inhibitor and DHA with miR-124 inhibitor compared against DHA + negative control, respectively. Scale bars = 100 μm.
DHA increases microglial miR-124 predominantly in the white matter of mouse spinal cord 7d after contusion. (A–C) Contused mice with vehicle and intraspinal negative control injections showed limited miR-124 (arrow heads, middle row) expression within microglia immunostained with P2Y12 (arrows, top row). (D–F) However, animals receiving DHA treatment and intraspinal negative control injections exhibited a significant increase in miR-124 expression (arrows, middle row) within the P2Y12+ microglial cells (arrow, top row). (G–L) In mice treated with the intraspinal miR-124 inhibitor, limited miR-124 was detected in both the vehicle- and DHA-treated mice. In particular, the rostral (M), epicentre (N) and caudal (O) white matter regions at or near the contusion site showed DHA induced miR-124, which was reduced with the miR-124 inhibitor. (P–R) In the grey matter regions at or near the contusion site, some specific cord regions also showed significant differences between the groups. Results are displayed as the mean ± SEM. *, **, *** represent p < 0.05, p < 0.01 and p < 0.001, respectively. Scale bars = 100 mm.
DHA acts via miR-124 to reduce microglial phagocytosis of myelin in cultures
To further provide evidence that DHA can reduce the phagocytic response of microglia, DHA was added to primary adult spinal microglia cultures which had been exposed to myelin stained with the fluorescence dye DiI, to observe phagocytic effects. In the presence of DHA (1 and 3 μm), there was a general reduction in DiI-labelled myelin detected within adult primary microglia, with a significant reduction observed at 1 μm DHA at 24 h after incubation, as detected using fluorescent microscopy (Fig. 10A–G). To determine if DHA directly influences microglia via miR-124, the microglial miR-124 expression was analysed using quantitative reverse transcription polymerase chain reaction (RT-PCR). MiR-124 expression was significantly increased in the DHA-treated primary adult microglia cultures, specifically at 1 μm, when compared to the control group (Fig. 10H). To test whether the effect seen is specific to miR-124, another microRNA that modulates the microglia immune response, miR-155, was also investigated (38). Interestingly, no significant difference was observed in miR-155 expression at any of the DHA concentrations studied (Fig. 10I). To further test the effect of DHA on the phagocytic ability of microglia, the commonly used BV2 microglial cells were also studied. DHA at 3 μm caused a significant reduction in the amount of DiI-myelin present within the BV2 cells (Fig. 10J–K and M). In addition, inhibition of miR-124 using a miR-124 inhibitor (3 pmol/ml) was able to significantly reverse the ability of DHA to suppress myelin phagocytosis by BV2 cells (Fig. 10L and M). These results further suggest that the phagocytic property of microglia is reduced by DHA, and this is dependent on changes in miR-124.
DHA reduces microglial phagocytosis of myelin debris via miR-124 in primary adult microglial and microglial BV2 cell cultures. (A–F) Primary microglia from adult rat spinal cord immunostained with Iba1 (green) can phagocytose myelin labelled with DiI (red) after 24 h co-incubation. (G) DHA-treated primary microglia exhibited a significant decrease in DiI-labelled myelin when compared to vehicle-treated microglia at 1 μm DHA. (H and I) This significant reduction in phagocytosis correlated with the upregulation of miR-124, but not miR-155. Four independent experiments were carried out. (J–L) BV2 microglial cells immunostained with IB4 (green) and Hoechst (blue) can phagocytose myelin labelled with DiI (red) after 24 h co-incubation. (M) DHA-treated BV2 microglial cells exhibited a significant decrease in DiI-labelled myelin when compared to vehicle-treated microglia at 3 μm DHA. (L–M) In the presence of the miR-124 inhibitor, the reduced microglial phagocytosis was abolished. Results are displayed as the mean ± SEM. * and *** represent P < 0.05 and P < 0.001. Scale bars = 100 μm.
DHA reduces phagoptosis of UV-stressed neurones by microglia
In recent studies, activated microglia have been shown to phagocytose stressed but viable neurons, alongside lesion debris and dead cells (9). Therefore, DHA may confer neuroprotection by preventing phagoptosis (the phagocytosis of stressed but viable cells), thus providing time for neurons to self-repair or recover. To test this hypothesis, UV-stressed PC12 neurons were incubated with DHA-treated BV2 microglial cells. PC12 neurons were exposed to 200 mJ/cm2 and then left for 16 h, which caused a significant increase in apoptosis as detected by labelling for phosphatidylserine (using Annexin V; Supplementary Material, Fig. S2A–C). UV-stressed PC12 cells were then labelled with TMRE (red) and Hoechst (blue), a live mitochondria marker and nucleus stain, respectively. The use of TMRE and Hoechst enabled the visualization of live stressed PC12 cells (red and blue). Of note, when live stressed (apoptotic) PC12 cells come into contact with BV2 microglia, the Hoechst (blue) staining remains, whereas the red colour from TMRE is no longer present. In addition, the BV2 microglia were labelled with CMFDA (green) when co-cultured with neurons.
Using live imaging microscopy, we show that BV2 microglia induce a TMRE loss (mean ± standard error of the mean; SEM) of 63.7 ± 2.9% in UV-stressed PC12 cells (Fig. 11A and C). However, when BV2 microglia were pre-treated with DHA, there was a marked and significant reduction in the number of cells with TMRE loss, of 32.8 ± 11.5 and 28.5 ± 9.2, at the 1 or 3 μm DHA concentrations, respectively (Fig. 11A–C). These results suggest that DHA reduces the phagoptotic effect of microglia, thus providing one of the potential mechanisms enhancing the survival of viable stressed neurons.
DHA reduces microglial BV2 cell phagoptosis of UV-stressed neuronal PC12 cells. (A) In the absence of DHA, UV-stressed neuronal PC12 cells at the start0 of the experiment (0 min post co-incubation) express live mitochondria (stained red with TMRE) and nuclei (stained blue with Hoechst). Within 18 min of cell–cell contact with a microglial BV2 cell (stained green with CMFDA), the expression of TMRE is diminished and thereafter ‘engulfed’ by the microglial BV2 cell by 180 min post co-incubation. (B) In the presence of DHA (3 μm), even after prolonged cell–cell contact between a UV-stressed neuronal PC12 cell and a microglial BV2 cell, the neuronal PC12 cells remained alive and not engulfed. (C) A significant reduction in the percentage of UV-stressed neuronal PC12 cells was observed with microglial BV2 cells in the presence of DHA (1–3 μm). Results are displayed as the mean ± SEM. * represent P < 0.05. Scale bars = 50 μm.
Discussion
The present study demonstrates that DHA promotes significant functional improvement as well as neuroprotection after spinal cord trauma in two rodent contusion injury models—and this consolidates the neuroprotective profile shown for this fatty acid in hemisection and compression models, thus enhancing the clinical translation potential. The neuroprotective benefits of DHA are associated with an increase in activated microglia/macrophages at the injury epicentre, both for the M1 and M2 phenotype. In a model of peripheral monocyte-derived macrophages, DHA promoted only a reduction in the M1 phenotype of bone marrow-derived macrophages and had no significant effect on either the M0 or M2 phenotypes. In vivo, DHA promoted an increase in microglial miR-124 expression, with reduced phagocytosis of myelin at 7 days post-contusion. In vitro, DHA promoted a reduction in the phagocytosis of myelin by two types of microglial cell cultures. Furthermore, DHA prevented the microglial phagoptosis of stressed but viable neurons. The reduction in the myelin phagocytic response in vitro was linked to a DHA-induced increase in miR-124, and not miR-155.
DHA improves neurological outcome and protects neurones after contusion SCI
A number of studies have examined the neuroprotective effects of omega-3 fatty acids following SCI induced by hemisection or compression (22,24,25,27,29,39). So far, several favourable effects of DHA administration (behavioural and histological) post-injury have been described, including decreased axonal and myelin damage, increased neuronal and oligodendrocyte survival, decreased macrophage/microglia cell recruitment and reduced lipid peroxidation and COX-2 activation (22,24,25,27,29,39,40). The results of the current study show that DHA treatment 30 min post-injury significantly improves locomotor function in rodents, both in the early (7 days) and late phase (35 days) of spinal cord contusion injury. As transection or hemisection injuries are rarely seen in human spinal cord trauma, contusion injuries are a more relevant model for investigating the potential of various new treatments for human SCI (30,41,42). We have previously demonstrated that DHA can improve neurological outcomes in contusion-injured rats in the first week after injury (31), and we further demonstrate in the present study that significant locomotor improvement continues to 28 and 35 days post-injury in rats and mice, respectively; therefore the beneficial effect is robust and persistent. In accordance with improved motor recovery, increased neuronal survival was also observed at 7 and 28 days post-contusion in DHA-treated rats. Furthermore, DHA-treated mice also showed significantly improved neuronal survival when compared to vehicle-treated animals at 35 days post-contusion. Thus, the demonstration of significant locomotor recovery alongside improved neuronal survival after contusion injury supports a considerable neuroprotective role for DHA, acutely administered post-SCI.
However, the observations reported so far in SCI have not differentiated the impact of this fatty acid on different microglia/macrophage cell populations and phenotypes, and this was a major focus of the present study. First, we showed that DHA treatment significantly increased Iba1 expression levels during the early phase (7 days) of contusion SCI in rats. Overall, Iba1 immunopositive microglia/macrophage cell counts were either unaffected or decreased by DHA treatment during the late phase of contusion SCI in rats and mice (28 and 35 days). As activated microglia/macrophages have been shown to acquire an alternative M2 polarized phenotype (identified by CD206+/CD204+/arg-1+) during the early phases of SCI (3,21), by enhancing the M2 response DHA could provide additional neuroprotective support. Later on, as microglia/macrophages are said to drift towards a predominantly M1 polarized phenotype (identified by CD16+/CD32+/CD86+/iNOS-1+) during the post-acute and chronic phases of SCI (3,21), DHA treatment may also potentially dampen harmful cell phenotypes over time.
Pre-stimulated macrophages implanted into damaged rodent spinal cords greatly improve functional recovery after injury (15,43). Taking this complexity into account, DHA treatment has the potential to modulate microglia/macrophage activity in both the early and late phases of SCI, in a manner which may simultaneously enhance beneficial functions and limit deleterious effects of microglia/macrophages. Importantly, we did not find evidence to support the claim that DHA is able to boost only the M2 microglia/macrophage populations after SCI. In fact, our data show an increase in the M1 (CD16+/CD32+) as well as the M2 (Arg-1+) population, via western blot at 3 days and tissue cell counts at 35 days, in mice (Fig. 4), strongly suggesting that DHA can augment both M1 and M2 cell phenotypes post-contusion SCI. Although these results did not match our initial hypothesis, we believe two plausible interpretations exist. Firstly, increased M1 populations during the late phase of SCI could represent an earlier DHA-induced expansion of M2 phenotypes, which have subsequently acquired the M1 profile over time. Secondly, the enhancement of M1 microglia/macrophages by DHA treatment may not necessarily represent per se the acquisition of a neurotoxic phenotype. For example, DHA treatment of lipopolysaccharide-activated microglia has been shown to inhibit pro-inflammatory mediator release (i.e. NO, iNOS, TNF-α and IL-6) without inducing an alternative anti-inflammatory M2 phenotype (i.e. Arg-1 and IL-10; 44). Our results imply that DHA is capable of inducing region-specific M1 responses within both grey and white matter, which might have separate neuroprotective roles. Furthermore, despite its utility as a working model to study macrophage heterogeneity in vivo, the M1/M2 classification of activated microglia/macrophages is an over-simplification, with at least one additional macrophage activation state, as well as several M2 sub-classifications believed to exist (3,21,45,46). For this reason, it can be assumed that more than one M1 subtype may be present after injury, which, contrary to current views, may serve important neuroprotective functions.
DHA reduces microglial phagocytosis of myelin debris
It is thought that phagocytosis of myelin debris is crucial for lesion clearance and subsequent axonal repair (47). Recently, Chen et al. (48) showed that DHA and EPA can enhance myelin phagocytosis in vitro. To our surprise, in our study DHA did not enhance phagocytosis of myelin debris in primary adult microglia cultures; reduced DiI-labelled myelin clearance was observed at 1 μm DHA after 24 h of incubation (Fig. 8). Furthermore, our in vivo data also confirm that DHA treatment causes a reduced phagocytosis of myelin at 7 days after injury (Fig. 6). The differences in in vitro results may reflect the fact that Chen et al. used primary microglia from neonatal brain and a high DHA concentration (20–80 μm). In contrast with their work, we used primary microglia from the spinal cord of adult animals (which is directly relevant to the microglia population studied in the post-contusion tissue), and in addition a microglia cell line (BV2 cells), and a DHA concentration that was between 1–3 μm, and thus more relevant physiologically. In fact, myelin phagocytosis has been shown to induce the production of many pro-inflammatory mediators (i.e. ROS, TNF-α. IL-6 and IL-1), causing significant bystander damage if left unregulated (49–52). As phagocytosis of myelin has been shown to be neurotoxic in nature and elicits a M1 macrophage response (53), DHA may serve to dampen excessive myelin removal, protecting against further bystander damage. Interestingly, a pro-regenerative treatment induced by lentiviral vector-driven expression of chondroitinase ABC in contusion SCI has been shown to lead to an amplification of a phagocytic microglia/macrophage phenotype at 3 days post-injury (72), whereas at 2 weeks after injury the opposite effect was seen. Furthermore, in the same study, a significant amplification of the increase in CD206-expressing microglia triggered by injury was also seen at 2 weeks after treatment.
When co-cultured with neurons, microglial cells were able to selectively phagocytose and remove dying cells, whilst leaving healthy neurons intact (54). We speculate that the phagocytosis of dying cells by microglia (and possibly also by macrophages) may explain the increased Iba1 immunopositive expression levels observed at 7 days post-contusion injury. Later on, increases in white matter Iba1 expression levels and CD16/32 and arginase-1 immunopositive cell counts (epicentre to 1 mm caudal) at 35 days suggest that DHA may support remyelination in these areas and reduce excessive and unnecessary myelin phagocytosis. Our interpretation is supported by studies on demyelination/remyelination processes (55) which did not find evidence supporting the idea that M1 microglial phenotypes drive demyelination whilst M2 populations enhance remyelination. In fact, the microglia phenotype present during demyelination persists over time post-injury, suggesting a gradual and more subtle switch in microglia functions that support remyelination (55). Furthermore, both M1- and M2-phenotype markers were shown to be up-regulated during periods of demyelination and remyelination (55). Microglial cells were also shown to support myelination by providing important trophic support to oligodendrocyte progenitor cells (55). Similarly, concurrent microglial iNOS expression showed moderately protective effects against demyelination (55). Therefore, DHA treatment may be able to prevent microglia/macrophage cell death after injury, thus enabling the acquisition of more subtle neuroprotective roles of these cells over time.
DHA enhances microglial miRNA-124 expression
To understand how DHA may directly influence microglial cells, the effect of DHA on miR-124 was analysed using FISH and quantitative RT-PCR techniques. Our results indicate that DHA is able to significantly increase microglial miR-124 expression in the injured spinal cord and primary adult microglial culture (Figs 7 and 9). In uninjured spinal cord, miR-124 in microglia was not detectable (data not shown). As miR-124 has been shown to correlate inversely with microglia/macrophage activation, as well as inducing M2 phenotypic changes after CNS trauma (34,35), our data suggest that DHA is capable of modulating microglial activity after SCI via the enhancement of miR-124 expression. A very important observation in this study is that the early interference with the activity of miR-124 compromises motor improvement after DHA, suggesting that the microglia modulation is an important aspect of the neuroprotective effect of the fatty acid.
Long-chain omega-3 fatty acids such as DHA have a multitude of molecular targets that could drive the effect on miRNA and on microglia activity (56,57), such as the peroxisome proliferator-activated receptors and retinoid receptors, or TREK-1 channels—targets which are all possible candidates for a modulation of microglia (58,59), and this aspect remains to be characterized in future studies.
Conclusion
The results of the present study show that DHA can act directly on microglia and reduce their phagocytic activity. DHA induces neuroprotection after moderate or severe contusion SCI in rodents and a markedly improved neurological outcome. Interestingly, DHA seems to be able to enhance early grey matter microglia/macrophage (Iba1+) cell populations (7 days) when compared to control animals, which then remained unchanged or decreased in the late phase (28 and 35 days) after SCI. Therefore, DHA is able to augment early microglia/macrophage responses after severe contusion. DHA is shown to enhance grey and white matter M1 (CD16/32+) responses in mouse contused cords at 35 days. Increased M1 populations may represent an earlier DHA-induced expansion of M2 phenotypes that have subsequently acquired M1 reactivity over time, or the acquisition of a non-reactive M1 phenotype, enabling improved locomotor recovery. Furthermore, increased white matter CD16/32+ microglia/macrophage populations over time may enhance remyelination and support oligodendrocyte cells in this process. DHA treatment reduces myelin phagocytosis in vitro (a process linked to phagoptosis) and enhances miR-124 expression (an important microRNA associated with the acquisition of M2 phenotypes), in vivo and in vitro. Therefore, our data demonstrate that DHA is associated with a complex modulation of the microglia/macrophage response, which occurs in parallel with a significant neuroprotective effect after contusion SCI in rodents.
Materials and Methods
In vivo study
Spinal cord contusion injury
All animal surgery was carried out under the UK Home Office Scientific Procedures Act (1986) and EU directive 2010/63/EU. The ARRIVE guidelines were followed for this study. Female Sprague-Dawley rats (230–250 g) were anaesthetized intraperitoneally with a mixture of medetomidine (Domitor, 0.5 mg/kg) and ketamine (Ketaset, 60 mg/kg), after which they received a contusion injury at spinal thoracic level T10/11 level using the PinPoint controlled cortical impactor (velocity, 1.5 m/s; dwell time, 100 ms; penetration depth, 1.7 mm; tip diameter, 2 mm; Hatteras Instruments, USA). Adult female C57BL6 mice (age 12 weeks old) were anesthetized intraperitoneally with a mixture of medetomidine (0.5 mg/kg) and ketamine (75 mg/kg) before receiving a spinal thoracic level T10/11 contusion injury using the PinPoint controlled cortical impactor (velocity, 0.7 m/s; dwell time, 85 ms; penetration depth, 0.8 mm; tip diameter, 1.5 mm). All animals underwent daily weight monitoring, twice daily buprenorphine treatment and manual bladder expression until it was no longer necessary.
DHA treatment
SCI surgery was performed in two groups of animals receiving either an intravenous (i.v.) injection of vehicle (saline containing 0.2% ethanol) or DHA (free fatty acid, 500 nmol/kg; Sigma-Aldrich, Gillingham, U.K. D2534) 30 min post-injury. All i.v. injections were administered via the tail vein in a volume of 1 ml/kg. Animal surgeons were blinded to treatment type. Sprague-Dawley rats were sacrificed at 7 and 28 days after injury, whereas C57BL6 mice were sacrificed at 35 days after injury.
Intraspinal miR-124 inhibitor and intravenous DHA injection
In a separate study in mice with contusion SCI, animals received 0.5 μl intraspinal injections in both the caudal and rostral region of the contused spinal cord using a glass micro glass needle via a microinfusion pump, as previously described (27,60). Each mouse received either 25 pmol of miR-124 miRCURY LNA miRNA inhibitor or negative control A, which acts as a miRCURY LNA miRNA inhibitor control (Exiqon, Vedbaek, Denmark Y104101112, Y100199006) mixed with Lipofectamine 2000 (Invitrogen, 2 μl in 50 μl phosphate buffer saline; PBS) as the transfection reagent. Thirty minutes post-contusion injury, DHA or vehicle was intravenously administered as described above.
Behavioural analysis
The BBB open field locomotor rating score (0–21) and BMS (0–9) were used to rate hind limb movement following contusion (61,62). BBB and BMS scoring was conducted in an open field (4 min sessions) on days 17, 14 and 28 post-injury (Sprague-Dawley rats), as well as every day for the first week, then every other day until 35 days post-injury (C57BL/6 mice). All tests were performed and analysed by individuals who were blinded to the treatment type.
Immunohistochemistry
Animals were perfused with saline followed by 4% paraformaldehyde, for histology. Spinal cord segments containing the lesion site including the rostral and caudal regions were dissected out and embedded in OCT compound. Serial horizontal cryostat sections (rat, 20 μm; mice, 15 μm) were collected across SuperFrost Plus microscopes slides and processed for immunohistochemistry. Each slide contained a selection of dorsal, intermediate and ventral cord sections. Sections underwent 3 × 5 min 10 mM PBS washes before incubation in 0.3% hydrogen peroxide in PBS for 10 min to remove endogenous peroxidase. For all washes, sections were gently agitated on a shaker. For immunohistochemical labelling, sections were blocked with 10% normal goat serum (containing 10 mM PBS, Triton X-100 and sodium azide) for 1 h before incubation with the primary antibody. Antibodies used were as follows: ionized calcium binding adaptor molecule 1 (Iba1, 1:1000, Wako, Neuss, Germany 019-19741), which labels activated and non-activated microglia/macrophages; P2Y12 (1:200, Cambridge, Cambridge, U.K. Bioscience, ANA55043A), which labels microglia; CD16/32 (1:500, BD Pharmingen, Wokingham, U.K. 553141), a specific marker of M1 activated microglia/macrophages; Arginase-1 (1:250, Santa Cruz, Dallas, Texas sc-18354), a specific marker of M2 activated microglia/macrophages, NeuN (1:1000, Chemicon, Watford, U.K. MAB377), for neuronal cells; myelin basic protein (MBP, 1:500, Abcam, Cambridge, U.K. ab62631) for the most abundant protein component of myelin. Sections were incubated with primary antibody overnight in a humidified chamber at room temperature. The following day, sections underwent 3 × 5 min PBS washes before incubation with secondary Alexa Fluor antibodies (Invitrogen®, Waltham, Massachusetts). Tissue sections were incubated in a humidified chamber for 2 h at 4°C. After application of secondary antibodies, sections underwent a further 3 × 5 min PBS washes before mounting with Vectashield and 4ʹ,6-diamidino-2-phenylindole.
FISH and immunohistochemistry
The FISH procedure was carried out using the 5′-biotin labelled rno-miR-124 microRNA in situ probe (miRCURY LNA™ detection probe, Cat. No. 614870–370, Exiqon) according to manufacturer’s instructions similarly to previously published (63). Briefly, horizontal rat spinal cord sections containing the injury site were washed 3 × 5 min DEPC-treated 10 mM PBS then dehydrated through a graded series of alcohol and air-dried. Sections were incubated overnight with 40 nM miR-124 microRNA probe diluted with microRNA in situ hybridization buffer (Cat No. 90004, Exiqon) at 59°C. RNase A-treated spinal cord sections were used as a negative FISH control. Next day, slides were washed in a series of pre-heated DEPC-saline-sodium citrate (SSC) solutions at 59°C for 5 min each. After further washes in 0.2× SSC and 1× PBS at room temperature for 5 min each, slides were incubated with 0.3% hydrogen peroxide for 30 min. Thereafter, the biotin signal on the miR-124 microRNA was amplified using the Avidin-Biotin complex (Vectastain ABC Elite Kit, Vector Labs, Peterborough, U.K.) and biotinyl tyramide (1:75, NEN Life Science Products, Boston, Massachusetts) and visualized with ExtrAvidin®-FITC (1:400, Sigma, Sigma-Aldrich, Gillingham, U.K.). After the FISH procedure, the sections were incubated with Iba1 antibody overnight followed by secondary Alexa Fluor 594 for 2 h and coverslipped with Vectashield and DAPI as mentioned above.
Image analysis
Spinal cord sections with NeuN, Iba1, CD16/32, Arg-1 and MBP immunolabelling and miR-124 FISH staining were viewed on a Leica epifluorescence microscope. Images were obtained using a 20× objective lens. In rat spinal cord tissue sections, two adjacent regions of interest (ROIs) were selected for analysis at the contusion epicentre as well as regions rostral and caudal to the injury epicentre (2, 4 and 6 mm rostral and caudal). Images of each ROI were captured at three different depths (dorsal, central and ventral sections of the cord). For rat contused cords, the area occupied by Iba1 immunopositive cells was measured using the MetaMorph image analysis program. Percentage threshold area (area occupied by immunopositive cells) was calculated via the use of intensity thresholding. Co-localization of Iba1+ cells with MBP or miR-124 was analysed using ImageJ. For mice contused cords, immunopositive cell number and size were recorded every 1 mm in the rostro-caudal plane and at three different depths, using the ImageJ software program, National Institutes of Health, U.S.A. Cell counts and cell size were averaged per ROI and then per whole cord to give overall values for the lesion epicentre as well as ROI rostrally and caudally. For the localization of myelin within microglia, a customized macro was used to identify the co-localization of myelin within Iba1+ cells in the white matter from 7 d contused rat spinal cord. The difference in some of the image analysis used between different experiments was due to the studies being carried out simultaneously and independently by different members of the research group. For the identification of myelin within P2Y12+ microglial cells in the spinal cord of 7 d contused mice, a customized ImageJ macro was used. A 500 μm × 500 μm square was placed within the ROI of each side of the spinal cord. The macro enabled all P2Y12+ staining within the ROI of the image and all the MBP staining within the ROI of the same image to be identified. Thereafter, the images were superimposed and only MBP staining present within the P2Y12+ cells was considered for the quantification. The MBP staining value for each ROI was the average MBP staining per P2Y12+ microglial cell.
Western blot analysis
The method used has been previously described (64). Briefly, adult rats with spinal cord contusion injury were either treated with DHA or vehicle at 30 min post-contusion as mentioned above. Three days post-injury, the animals were sacrificed, and the spinal cord was quickly removed onto dry ice and then stored at −80°C until further processing. The spinal cord was dissected into 3 mm thick cross sections, with the centre of the injury site as the epicentre. A further two and three segments were also removed caudal and rostral to the epicentre, respectively. The protein from each cord segment was harvested in 1 ml ice-cold lysis buffer (20 mM HEPES pH 7.4, 100 nM NaCl, 100 mM NaF, 1 mM Na2VO4, 5 mM EDTA, 1% Nonidet P-40 and 1X protease inhibitor cocktail; Roche, Basel, Switzerland). The homogenate was rotated for 2 h at 4°C before centrifugation at 13500g for 15 min at 4°C. The total protein concentration of the supernatant was determined using the Pierce TM bicinchoninic acid protein (BCA) assay kit (ThermoFisher Scientific, Dartford, U.K. Cat. No. 23225).
Fifteen micrograms of total protein were electrophoresed on 12% acrylamide gel before transfer onto Hybond P membranes and incubated overnight at room temperature with rat anti-CD16/32, rabbit anti-arginase-1 or mouse anti-βIII tubulin (control protein). To avoid batch differences, each blot contained an animal from each treatment group.
In vitro study
Bone marrow-derived macrophages
The method used has been previously described (21). Briefly, the tibias and femurs of C57BL/6 mice were removed, and the bone marrow flushed out using RPMI 1640 with 10% foetal bovine serum. Cells were triturated three to five times and red blood cells lysed in lysis buffer (0.15 M NH4Cl, 10 mM KHCO3, 0.1 mM Na2EDTA; pH 7.4). After a single wash in media, cells were cultured in RPMI 1640 containing 1% penicillin/streptomycin, 1% HEPES, 0.001% β-mercaptoethanol, 10% FBS and 20% supernatant from sL929 cells. After 7–10 days, when the bone marrow cells have differentiated into macrophages, the cells were either treated with media alone (termed M0) or differentiated into M1 or M2 macrophages. The differentiation into M1 macrophages required treatment with LPS (100 ng/ml) and IFN-γ (20 ng/ml). The differentiation into M2 macrophages required treatment with IL-4 (20 ng/ml). After 30 min of either control or treatment to differentiate into M1 or M2, DHA (1–3 μm) was incubated with the macrophages. Twenty-four hours later, the cells were fixed with 4% paraformaldehyde and underwent immunostaining for Iba1, CD16/32 and DAPI.
The number of cells and immunostaining intensity was measured using the Thermo Scientific™ ArrayScan™ XTI High Content Analysis Reader. Ten regions of each well with each treatment and condition consisting of two wells were analysed. Four independent experiments were carried out.
Primary adult rat microglia culture
Primary microglia cultures from adult rat spinal cord were carried out as previously described (65). Briefly, adult Wistar rats were deeply anaesthetized with sodium pentobarbital (120 mg/kg, i.p.) then perfused with saline to flush away monocytes/macrophages. The spinal cord was dissected out and placed into ice-cold Hanks Balanced Salt Solution media. The dura mater and spinal roots were removed before cutting longitudinally at 200 μm segments using a McIlwain tissue chopper. After digestion in papain followed by dissociation, the cell suspension was passed through a 70 μm sieve. A total of 5 μl of cell suspension was stained with 5 μl DAPI. Using an epi-fluorescent microscope DAPI immunopositive cells were counted, and 40 000 cells were plated into each well containing a sterile ethanol-cleaned 13 mm diameter cover slip. After 2 h, unattached cells were washed away with DFP media [containing Dulbecco’s modified eagle medium (DMEM), 15% heat-inactivated foetal bovine serum (HI-FBS) and penicillin–streptomycin; 100 IU/ml] and allowed to incubate overnight at 37°C and 5% CO2 and were used the next following day for experimental studies.
For data analysis, Iba1 and DiI immunopositive cells were counted using the ImageJ program. Ten ROIs were selected for analysis per well (65). The average amount of DiI within each cell per DHA treatment concentration was analysed. Four independent experiments in each treatment group, control (DFP media containing 0.0003% ethanol) 1 and 3 μm DHA were studied.
Retroviral immortalized murine (BV2) microglia culture
The BV2 microglia cell line was cultured as previously described (66). Briefly, BV2 cells were cultured in DMEM supplemented with 10% HI-FBS and penicillin–streptomycin under 37°C and 5% CO2. During experimental studies, the HI-FBS was reduced to 5%.
Myelin phagocytosis
Preparation of spinal cord myelin
Myelin was purified from rat or mouse spinal cord tissue by sucrose gradient centrifugation as previously described (67,68). Briefly, the spinal cords were homogenized with a glass–glass homogenizer in 0.32 M sucrose and then layered onto 0.85 M sucrose. After centrifugation at 75000g for 30 min, the crude myelin was transferred into 20 volumes of 0.32 M sucrose and centrifuged 13 000g for 25 min. The white myelin pellet was resuspended in 25 volumes of sterile water and centrifuged at 15000g for 25 min. The pure myelin pellet was dissolved in 2.5 mM Tris (pH 7.0) and the protein concentration was determined using the Pierce BCA assay.
DiI labelling of myelin
Labelling of spinal cord myelin with fluorescent dye DiI was carried out as previously described (69). Myelin (1 mg/ml) was labelled by incubating with DiI (0.2 mg/ml, Cambridge Bioscience Ltd, BT60010) at 37°C for 30 min. After resuspension with 40 volumes of sterile DPBS, the solution was briefly sonicated for 15 s to decrease particle size (70). After centrifugation at 15000g for 5 min, the DiI-labelled myelin was resuspended in 5% DMEM (1 mg/ml).
Phagocytosis of DiI-myelin
Microglia or bone marrow-derived macrophages were incubated for 30 min with DiI-labelled myelin (50 μg) per well before DHA (1–3 μm) was added. After 24 h, the media were removed and the cells were either frozen on dry ice for RNA extraction or fixed with 4% paraformaldehyde for immunocytochemistry.
Inhibition of DiI-myelin phagocytosis in vitro
To study the role of DHA on miR-124, 3 pmol/ml miRCURY LNA microRNA inhibitor specifically designed for rno-miR-124 (Exiqon, 4104239-001) complexed with the transfection reagent Lipofectamine 2000 (Invitrogen) was added an hour before application of DHA, as previously described (25).
Phagoptosis of UV-stressed neurones
Neuronal (PC12) cell lines were used to assay the effect of DHA on the phagoptosis of microglia BV2 cells similarly to the method previously published (9).
PC12 cells
PC12 cells were cultured in complete medium (containing DMEM, 5% heat-inactivated horse serum, 5% HI-FBS and penicillin–streptomycin) at 37°C and 5% CO2. During the experimental studies, the PC12 cells were transferred onto collagen (2 mg/ml, Roche Diagnostics, 11179179001) covered 9 cm dish HI-FBS and were differentiated into neurons when cultured in differentiation medium [containing DMEM, 1% heat-inactivated horse serum, nerve growth factor (50 ng/ml, Sigma-Aldrich, N0513), and penicillin–streptomycin].
Live cell imaging
On the day of the experiment, BV2 and PC12 cells were incubated with the relevant fluorescent dyes for 20 min at 37°C and 5% CO2 and then thoroughly washed several times with DPBS to remove any free dye. BV2 cells were labelled with 5-chloromethylfluorescein diacetate (0.4 μM, CellTracker™ green CMFDA, C7025, a green fluorescent dye suited for monitoring cellular movement and location). PC12 cells were exposed to 200 mJ/cm2 irradiation using the CX-2000 UV crosslinker to stress the neuronal cells. The following day, the UV-treated PC12 cells were stained with tetramethylrhodamine ethyl ester (25 nM, TMRE, ThermoFisher, Dartford, U.K. T-669, a red fluorescence dye that readily labels active mitochondria) and Hoechst 33342 (10 μg/ml, ThermoFisher, H1399, a blue fluorescence dye that is a cell-permanent nuclear counterstain). Sixteen hours after the UV treatment, the PC12 cells were harvested by trypsinization and added onto the DHA-treated BV2 cells. Time-lapse microscopy was conducted on an inverted microscope (Zeiss Axiovert 200 M) at ×20 magnification in improved MEM (ThermoFisher, 10373-017) without Phenol red, supplemented with 5 % HI-FBS and using the 24 well plates at 37°C, 5% CO2 and in a humidity-controlled chamber. Two regions within the well were randomly selected and images were time-lapse captured every 9 min for a total of 3 h.
Analysis of time-lapse microscopy
The number of contacts made between a BV2 cell (green) and UV-stressed PC12 cells (red with blue nuclei) was recorded. After the contact was made, the number of PC12 cells which lost the red fluorescence as an indicator of no active mitochondria was also recorded.
UV-induced apoptosis
To confirm that neuronal PC12 cells were stressed from the UV irradiation protocol, PC12 cells plated onto collagen-coated glass coverslips and cultured in differentiation medium were exposed to 200 mJ/cm2 irradiation. Sixteen hours after the UV treatment, the PC12 cells were stained using Annexin V Alexa Fluor® 488 (ThermoFisher, V13241) according to the manufacturer’s instruction for imaging microscopy.
Immunocytochemistry
All microglial cultures underwent 3 × 5 min PBS washes before incubation with primary antibodies for 2 h. Primary microglia were stained using rabbit anti-Iba1 (1:500 Wako) and BV2 microglia cell lines were stained with biotinylated Isolectin-B4 (1:200, Vector Labs B-1205) before applying secondary antibodies for 45 min. After performing three further PBS washes, microglia cultures on glass cover slips were mounted onto glass slides with DAPI and Fluorsave mounting media. The lipophilic dye DiI was used to visualize myelin within microglial cultures, which is a red fluorescence dye when viewed with fluorescent microscopy.
Quantitative RT-PCR analysis of miRNA
Total RNA was extracted from primary adult microglia cultures by using the miRVANA™ PARIS™ isolation kit (AM1560, Life, Technologies Dartford, U.K.) similarly to previously published (71). Microglial cells from 2 wells per treatment were pooled together. Total RNA was quantified using the ND1000 spectrophotometer (NanoDrop Technologies, UK). Analysis of miR-124 expression was carried out using the TaqMan miRNA assays (Assay ID: 001182) and performed on Applied Biosystems 7500 real-time PCR system. The internal control used was RNU6B (Assay ID: 001093). Quantitative RT-PCR analysis was performed in triplicates from four independent experiments.
Statistical analysis
All statistical analysis of results was carried out using the GraphPad Prism program. For behavioural data, and staining for NeuN, Iba1, CD16/32, Arg-1, MBP and miR-124, a two-way repeated measures Analysis of variance (ANOVA) was used to compare the overall effect of DHA and vehicle injection. Individual differences between DHA- and vehicle-treated animals were determined using Tukey’s post-hoc tests. For DHA-treated microglial cultures and quantitative RT-PCR analysis of miRNA, a one-way repeated measures ANOVA was used to compare DHA versus control groups. Individual differences between DHA concentrations were determined using Dunnett’s post-hoc tests. Statistical significance was set at P < 0.05. Data points are represented as the mean +/− SEM.
For the in vitro experiments, 10 regions of each well, with each treatment and condition consisting of 2 wells were analysed. Three or four independent experiments were carried out. The number of cells and immunostaining intensity from the BMDM experiment was measured using the Thermo Scientific™ ArrayScan™ XTI High Content Analysis Reader. The statistical analysis used was one-way ANOVA followed by Tukey’s post-hoc test. Statistical significance was set at P < 0.05. Data points are represented as the mean +/− SEM.
Acknowledgements
The authors would like to thank Dr Kristina Kigerl for help with Thermo Scientific™ ArrayScan™ XTI High Content Analysis Reader, Dr Jan Stoetaert and Dr Belén Martín-Martín of the BALM facility for help with time-lapse microscopy and Dr Xuenong Bo for the supply of PC12 cells.
Conflict of Interest statement. None declared.
Funding
US Department of Defence CDMRP/SCIRP award (W81XWH-10-1-1040 to P.K.Y., T.B. and A.M.T.); The Barts and London Charity (to P.K.Y. and A.M.T.); Rod Flower Vacation Scholarship (to A.L.B.); International Spinal Research Trust (to J.H. and P.G.P.); Ray W. Poppleton Endowment (to P.G.P.); Chang Gung Memorial Hospital, Taiwan (CMRPG3A1051–1054 to Z.-H.L.). M.A.B. is funded by the Spanish Ministry of Economy and Competitivity (Programa Ramón y Cajal: RYC-2017-21804).
Author contributions
The authors(s) have made the following declarations about their contributions: conceived and designed the experiments: P.K.Y., P.G.P. and AMT; performed the experiments and analysed the data: P.K.Y., A.L.B., M.A.B., T.H.R.I., T.B., Z.H.L., M.A.K.E.M., F.G., A.D.L., S.U.R.N. and J.H. J.C.E.H. contributed reagents/materials/analysis tools: P.K.Y., P.G.P. and A.M.T.; wrote the paper: P.K.Y., A.L.B., P.G.P., J.C.E.H., J.V.P. and A.M.T.
