Acute injury in the peripheral nervous system triggers an alternative macrophage response

Injury to the peripheral nervous system (PNS) induces a well-orchestrated cellular process that leads to the complete disintegration of the nerve segment distal to the lesion site, termed Wallerian degeneration (WD) [1]. As axons are disconnected from their cell bodies, they are rapidly fragmented by an intrinsic active process of self-destruction [2]. Due to the loss of axonal contact, the myelinating Schwann cells (SC) dedifferentiate into an immature phenotype, start proliferating, and help in the degeneration of myelin. Wallerian degeneration typically triggers a strong neuroinflammatory response in which the SCs are believed to play an important role. Being in close contact with the nerves, SCs are among the first to respond to nerve damage. They induce the production of pro-inflammatory cytokines such as TNF, IL-1α, and IL-1β within hours after nerve injury [3‐5]. Subsequently, these cytokines induce the expression of additional immune mediators such as IL-6, GM-CSF, and IL-10 in both Schwann cells and fibroblasts [3, 4, 6, 7]. The production of MCP-1 and MIP-1α, which reaches a maximum at 1 day after injury, promotes the recruitment of macrophages to the damaged nerves [4, 8]. Moreover, mast cells accumulate in the endoneurium of injured nerves [9] and release mediators that contribute to the recruitment of macrophages and neutrophils [9, 10]. Infiltration of blood-borne monocytes, which spread over the entire nerve, starts from 2 to 3 days after injury and macrophage accumulation peaks at 7 days post injury [11‐14]. These infiltrated immune cells take over the cytokine production and are responsible for the rapid clearance of myelin debris. Before the infiltration of hematogenous macrophages, local macrophages proliferate and undergo morphological changes consistent with immunophenotypic signs of activation [15]. These resident macrophages, together with SCs, readily contribute to myelin phagocytosis [12, 16]. Two to three weeks after injury, the inflammatory response is turned off and macrophages are rapidly eliminated [12, 13, 17]. Generally, WD is believed to induce a strong pro-inflammatory response, as reflected by the induction of cytokines such as TNF and IFNγ [18‐20], and the reported elevation of iNOS [21]. Still, one would expect that neurodegeneration might trigger a more dampened immune response, which is typically associated with sterile inflammation. To address this question we used a model of WD (axotomy of the sciatic nerve) to analyze which type of immune response is being induced. Our results revealed that WD leads to the expression of several negative regulators of the innate immune system. In addition, a predominant M2-like macrophage response could be observed, reflecting the presence of an immunosuppressive milieu necessary to initiate wound repair and restore tissue homeostasis.

Wallerian degeneration (WD) induces an immune response that is considered to be predominantly pro-inflammatory by expressing several pro-inflammatory molecules such as TNF, IFNγ, and iNOS [19‐21]. To confirm the pro-inflammatory environment, we isolated total RNA of the distal segment of four sciatic nerves isolated at different time points upon axotomy. We analyzed three independent experiments and measured the expression of several cytokine and chemokine transcripts using RT-qPCR. In line with literature data, the inflammatory mediators IL-1β, Cox2, MCP-1, and MIP-1α were strongly up-regulated, with maximum expression levels at 24 h after axotomy (Figure 1a) [4]. Strikingly, the expression levels of the inflammatory genes dropped at later time points after axotomy, with most pro-inflammatory genes returning to the basal condition at 48 h. We determined whether this transient immune response was accompanied with the induction of several negative regulators of the immune system and found that IL-1RA displayed a high induction (Figure 1b). Moreover, MyD88small (MyD88s) and IκBα, both negatively regulating NF-κB activation [26], were up-regulated already 4 h after injury. In contrast, two other negative regulators of the NF-κB pathway, A20 and SIGIRR, were not induced. Suppressor of cytokine signaling 1 (SOCS1) was only weakly induced after axotomy at these early time points (Figure 1b). Functions of the immune mediators and negative regulators are shown in Table 2.

While many reports already described the induction of cytokines and chemokines in WD, it is less obvious what type of immune response is triggered by injury in the PNS. Therefore, we decided to focus on gene expression profiles for genes associated with M1 vs. M2 macrophages, representative for the two extremes of a merely pro-inflammatory vs. a merely anti-inflammatory/wound healing phenotype [25, 36, 37]. The main functions of these genes are described in Table 3. We first determined when macrophages start to accumulate in our model, by analyzing the presence of three universal markers for macrophages (Iba1, CD11b, and F4/80) using RT-qPCR [38, 39]. In general, it is considered that a first contribution to the immune response in the nerve is mediated by resident cells (such as Schwann cells, resident macrophages, and fibroblasts) because blood-borne monocytes infiltrate the nerve only 2 to 3 days after injury [12]. Macrophages, expressing Iba1, CD11b, and F4/80, start to accumulate in the injured nerves from day 3 onwards as determined by RT-qPCR (Figure 2a) and immunohistochemistry (Figure 3). Coinciding with the accumulation of macrophages, a second peak in the immune response could be observed, as shown by the biphasic induction of IL-6 and IL-1β expression (Figure 2b). As expected, MCP-1, a chemoattractant for macrophages produced by Schwann cells, is expressed right before macrophage accumulation (Figure 2b). In order to determine the phenotype of the macrophages present in the peripheral nerve after injury, we analyzed markers typically associated with M1 vs. M2 macrophages. None of the M1 markers such as iNOS, IL-12p40, and IFNγ were induced after axotomy at any time point investigated (Figure 2c). On the other hand, the M2-associated genes, arginase-1 and Ym1, were clearly induced. The expression of these genes reached a maximum at 1 day after axotomy and returned to basal level at day 7. Another typical marker for M2 macrophages, Trem2, was induced from day 3 onwards and its expression level remained elevated till day 14 after axotomy (the latest time point investigated in this study) (Figure 2d). The expression of Trem2 appeared to be mediated by the accumulating macrophages, as its expression level displayed a similar pattern as the general macrophage markers (Figure 2b and d). Some markers (arginase-1 and Ym1) were also slightly induced in sham-operated animals, however this induction was only minor compared to the induction seen after axotomy (Figure 2e). Altogether, these data suggest that acute peripheral nerve injury favors an M2 macrophage environment. Additional analyses confirmed this hypothesis. We found that receptors known to trigger M2 cells (such as IL-4Rα and IL-13Rα1), and to stimulate macrophage suppressor function (such as IL-10R), were induced in injured peripheral nerves at 7 and 14 days after injury (Figure 4a). The IFNγR1 receptor, which characterizes M1 macrophages, was not enhanced (Figure 4b). Moreover, scavenger receptors (such as Clec7a and Rage), which are typically expressed by M2 macrophages, showed an increased expression level after axotomy at the late time points relative to the uninjured control nerve (Figure 4c).

The M2 gene expression profile is typically triggered by the cytokines IL-4 and/or IL-13 [60]. In order to determine if these cytokines play a role in the induction of the alternative macrophage environment after axotomy, their expression level was investigated at early time points using RT-qPCR. The IL-4 expression was hardly detectable at the mRNA level in our model of acute peripheral nerve injury and did not seem to be induced. The IL-13 expression, however, was induced upon axotomy at the earliest time point (4 h) investigated (Figure 4d). Importantly, also the anti-inflammatory cytokine IL-10 was induced after injury (Figure 4d). The high IL-10 and low IL-12p40 expression levels are representative of a typical M2 activation profile (Figures 2c and 4d).

Next we analyzed the macrophage phenotype at protein level by using western blot and immunohistochemistry. As the balance between arginase-1 and iNOS expression is highly indicative of the macrophage phenotype, these two markers were used in the following experiments [36]. Western blot analysis of protein lysates of the distal segment of the sciatic nerve showed an induction of arginase-1 protein after axotomy. Arginase-1 protein was detectable from day 1 after injury and reached a maximal signal at day 3. Albeit showing a small decrease over time, the arginase-1 protein level remained high until day 14 after axotomy (Figure 5a). iNOS was not detectable at any time point by western blot analysis, confirming our RT-qPCR data (Figure 5b). As a positive control, peritoneal macrophages were stimulated in vitro with either IL-4/IL-13 or LPS/IFNγ to obtain M2 and M1 macrophages, respectively. As expected, the M2 macrophages expressed arginase-1 and the M1 macrophages expressed iNOS protein (Figure 5a, b). Immunohistochemistry of paraffin-embedded sciatic nerves confirmed the temporal expression profile for arginase-1 shown by western blot. Arginase-1 is rapidly expressed throughout the entire injured nerve. The expression level peaked at 3 days post injury and remained high until day 14 (Figure 5c). Double immunofluorescence staining revealed that arginase-1 was present in F4/80 positive cells and not in S100 positive Schwann cells, which identifies macrophages as the main source for arginase-1 (Figure 6). While at earlier time points all cells that expressed F4/80 were found to be positive for arginase-1, at later time points arginase-1 negative macrophages were present as well. Immunohistochemical staining for iNOS confirmed that this protein was not induced after axotomy (Figure 5c). We only observed strong iNOS staining in blood capillaries in particular regions on the nerve that was present independently of the axotomy, showing that the antibody staining was working properly (Figure 5d).

Finally, we determined whether the M2-predominated immune response triggered after nerve injury is typical for the PNS or whether it is specific for neurodegeneration. To this end, we investigated at different time points the expression of M1 and M2 markers in sciatic nerves from mice intravenously injected with TLR-ligands. We used lipopolysaccharide (LPS), a TLR4 ligand known to induce a classical type I immune response, and Pam3Cys, a TLR1/2 ligand. Intravenous injection of LPS as well as Pam3Cys elicited a rapid and strong immune response in the sciatic nerve, as shown by the induction of inflammatory genes such as IL-1β, Cox2, MIP-1α, and MCP-1 (Figure 7a). Interestingly, the pro-inflammatory cytokine IL-12p40 and typical M1 immune mediator iNOS, both representative for a type I immune response, were induced after LPS injection (Figure 7b). Several negative regulators, such as IL-1RA, MyD88s, and SOCS1, which mediate a negative feedback loop, were also induced by LPS injection (Figure 7c). Injection with Pam3Cys, however, clearly induced a mixed immune response as reflected by the expression of the M1 associated cytokine IL-12p40 and the expression of Ym1, which is an M2-associated macrophage marker (Figure 7b). iNOS was not detectable after Pam3Cys injection (Figure 7b) and none of the other M2-associated genes such as arginase-1 and Trem2 were induced (Figure 7b). These data show that a prototypical type I immune response can be observed in the nerve after injection of LPS, while Pam3Cys seems to induce a mixed immune response. Both TLR-mediated responses clearly differed from the immune response induced after acute peripheral nerve injury.

In response to an infection, a strong pro-inflammatory immune response is triggered. The recruited inflammatory cells are activated when they encounter pathogen-associated molecular products (PAMPs) such as LPS. Hereupon, these cells phagocytose infectious agents and produce pro-inflammatory mediators such as iNOS, IL-12, ROS, and RNS to fight off the invading pathogen. These agents, however, can also cause tissue damage. The innate immune system also detects the presence of endogenous molecules; so-called danger associated molecular patterns (DAMPs) that are only exposed in conditions of injury. Under conditions of cellular stress or injury, one might expect a more dampened, strictly controlled immune response as the cost-benefit ratio is higher. As we and others have shown, pro-inflammatory mediators such as IL-1β and Cox2 and chemokines such as MIP-1α and MCP-1 are rapidly induced in WD, a model of sterile inflammation in the nerve [3, 4]. In the present study we show that the expression of these inflammatory genes is strictly controlled as the mRNA levels of all cytokines and chemokines return to basal level at 48 h. Negative regulators of the pro-inflammatory signaling pathways are induced prior to the decline in inflammatory gene expression, thereby limiting the pro-inflammatory immune response and also the excessive damage caused by the immune system.

Although PAMPs and DAMPs are recognized by the same set of receptors, such as TLRs, they can induce a different set of genes [61]. Bacterial compounds trigger a microbicidal environment and classically activated M1 macrophages; while endogenous molecules seem to activate an inflammatory response associated with genes that mediate tissue repair [61, 62]. Since most studies so far focused specifically on the induction of pro-inflammatory mediators, WD in the PNS has always been associated with the induction of a strong pro-inflammatory immune response. We found, however, by analyzing genes associated with M1 and M2 macrophages, that acute peripheral nerve injury rather induces an M2-like macrophage environment. None of the typical pro-inflammatory markers of the M1 subtype of macrophages such as iNOS, IFNγ, and IL-12p40 could be detected, while M2 markers such as arginase-1, Ym1, and Trem2 were highly up-regulated. Intriguingly, other M2 markers like Fizz1 and Cdh1 were not induced. Van den Bossche et al. showed that some M2 markers like Cdh1 are strongly down-regulated by the presence of pro-inflammatory cytokines [63]. This could be the case here as well. The stimulation of the alternative macrophage environment in the nerve appeared to be controlled at the level of IL-13. This cytokine was readily detectable from 4 h after the onset of neurodegeneration, and prior to the expression of arginase-1 and Ym1. IL-13, which is together with IL-4 a central master switch in the M2 phenotype, is typically expressed by macrophages, basophils, mast cells, or activated T cells [64]. Since we detect accumulation of macrophages only from days 2 to 3 onwards, it is less clear at the moment which cells are responsible for the early onset expression of IL-13, arginase-1, or Ym1. In the peripheral nerve resident macrophages, mast cells or SCs could be engaged in the expression of IL-13, while neutrophils could contribute to the expression of arginase-1 and Ym1. Neutrophils are recruited to the damaged nerves at day 1 after injury [65], and are suggested to contribute to the expression of tissue repair genes [66]. Our results demonstrate that damage to the nerve establishes a rapid immunosuppressive reaction within the nerve, and this from very early time points on, which appears to be in contrast with another recent report [67]. Shechter et al. described that axotomy of the optic nerve creates a pro-inflammatory environment in the nerve that was later turned into an anti-inflammatory one by infiltrating macrophages [67]. Macrophages have been shown before to play a beneficial role in WD in the PNS, as depleting them impaired functional recovery [68]. By phagocytosing debris, macrophages contribute to regeneration by removing inhibitory myelin debris and paving the way for neurite outgrowth. Pre-existing auto-antibodies have been shown to play an important role in clearance of myelin debris by promoting a macrophage influx and stimulating their phagocytic activity [69]. In addition, macrophages produce neurotrophic factors, thereby supporting regeneration [70]. The protective role of macrophages in WD might also be explained by their phenotype. The M2 macrophages were shown to be neuroprotective in vitro by stimulating neurite outgrowth, while M1 macrophages were neurotoxic to neuronal cell cultures [71]. Moreover, potent inducers of a systemic Th2 switch, such as glatiramer acetate and statins, support the neuroprotection and/or nerve regeneration [72‐74]. The Th2-inducing adjuvants, such as IFA and Alum, promote axon regeneration better than the Th1-inducing adjuvant CFA [75, 76]. Also Th2 cells support neuronal survival in vitro to a greater extent than Th1 cells [77].

In autoimmune diseases of the PNS such as Guillian-Barré Syndrome (GBS) and chronic inflammatory demyelinating polyneuropathy (CIDP), a Th1 response is associated with the early stages of the disease. During recovery of GBS and CIDP, a shift towards a Th2 response is observed, suggesting a protective role for Th2 responses in these diseases [78‐80]. Also from animal models it is apparent that type II immune responses are beneficial; as nasal administration of recombinant IL-4 ameliorates ongoing experimental autoimmune neuritis (EAN) and inhibits demyelination [81]. The self-limiting clinical course of GBS might be explained by the induction of IL-4 and IL-10. The role of the immune system in hereditary neuropathies is less well studied. Patients suffering from inherited neuropathies show endoneurial T-cells in their nerve biopsies and some patients even show inflammatory infiltrates [82, 83]. Studies with animal models such as the heterozygote P0 mice, a model of Charcot-Marie-Tooth (CMT) 1B neuropathy, clearly show a functional degenerative role for macrophages and T-cells [84, 85]. Unfortunately, the type of immune response triggered in hereditary neuropathies has not been addressed.

In CNS injury, macrophages have been implicated in both exacerbating as well as ameliorating tissue damage at the injury site. Kigerl et al. showed that spinal cord injury (SCI) initially induces both M1 and M2 macrophages, however, the M1 phenotype predominates the lesion site after 1 week [71]. The presence of both phenotypes might explain the dual effect of macrophages in this model. Moreover, axonal regeneration after SCI is prevented by an inhibitory environment due to myelin inhibitors. Qui et al. showed that elevating cAMP was sufficient to overcome the myelin-mediated inhibition [86]. Subsequent studies showed that arginase-1 and polyamines played a major protective role downstream of cAMP [45]. By using PNS grafts together with acidic fibroblast growth factor in a model of SCI, the recruited macrophages produced large amounts of arginase-1 and were involved in polyamine synthesis [87]. This strategy significantly improved functional recovery [88]. Altogether these data show a central role for type II immune responses, arginase-1 and downstream polyamines in regeneration.

Lessons from graft implantation in CNS injury showed that PNS tissue induces a permissive environment for regeneration. Macrophages contribute to this permissive environment as spinal cord-injured axons failed to regenerate through peripheral nerve grafts in the absence of CD11b⁺ cells [68]. We now hypothesize that PNS injury triggers an inherent protective environment by inducing an M2 phenotype of macrophages and arginase-1 expression. This model can further be used to unravel how the environment is induced and to elucidate which protective program needs to be elicited. Currently, it remains to be shown how the alternative macrophage environment is established; however there might be a role for IL-13, a typical M2 inducer, as this cytokine is up-regulated very early after injury and prior to the M2-associated gene expression.

Finally, the induction of the alternative macrophage environment appeared to be triggered specifically in response to neurodegeneration. Our results show that when challenged with bacterial products such as LPS, a typical pro-inflammatory immune response, as reflected by a strong IL-12p40 and iNOS signal and the absence of Ym1 or arginase-1, could be detected within the PNS. Intriguingly, injection with Pam3Cys, a TLR1/2 ligand, induced a mixed response marked by presence of both IL-12p40 and Ym1 induction. TLR2, the co-receptor for TLR1, has been associated before with the induction of a type II gene expression [89, 90]. In a recent study we showed that specifically TLR1 was highly induced after acute peripheral nerve injury and hypothesized that it might play a role in detecting neuronal injury [22]. The possible involvement of TLR1/2 in the detection of peripheral nerve injury and in the switch towards the type II gene expression is currently under investigation.

In conclusion, we demonstrate that acute peripheral nerve injury induces an inherent protective response with the initiation of several negative feedback loops, limiting excessive tissue damage. Furthermore, we show that an M2-like anti-inflammatory environment is induced, rather than a pro-inflammatory one. Since type II responses have been shown before to be neuroprotective, we believe that instead of inhibiting the immune responses, shifting the macrophage phenotype or type of immune response towards an alternative activation state or type II response would be a better therapeutic strategy to stimulate repair, as this would create a permissive environment for neuronal regeneration.