Fast direct neuronal signaling via the IL-4 receptor as therapeutic target in neuroinflammation

See allHide authors and affiliations

Science Translational Medicine  28 Feb 2018:
Vol. 10, Issue 430, eaao2304
DOI: 10.1126/scitranslmed.aao2304

IL-4 empowers axons

Multiple sclerosis (MS) is a neuroinflammatory disorder, and current therapies focus on altering immune activity to reduce symptoms. Vogelaar and colleagues tested the ability of intrathecally applied IL-4, a cytokine typically associated with T helper type 2 responses, to treat established disease in several experimental autoimmune encephalomyelitis (EAE) models. IL-4 treatment led to reduced clinical scores, improved locomotor activity, and diminished axon damage. Somewhat surprisingly, the beneficial effects of IL-4 did not depend on T cell modulation in the chronic disease phase. The receptor for IL-4 was observed in postmortem brain histology of several MS patients, and they demonstrated that IL-4 could act directly on neurons in vitro. They also showed benefits of intranasal IL-4 administration in one of the EAE models, which could be a promising avenue to pursue in the clinic.


Ongoing axonal degeneration is thought to underlie disability in chronic neuroinflammation, such as multiple sclerosis (MS), especially during its progressive phase. Upon inflammatory attack, axons undergo pathological swelling, which can be reversible. Because we had evidence for beneficial effects of T helper 2 lymphocytes in experimental neurotrauma and discovered interleukin-4 receptor (IL-4R) expressed on axons in MS lesions, we aimed at unraveling the effects of IL-4 on neuroinflammatory axon injury. We demonstrate that intrathecal IL-4 treatment during the chronic phase of several experimental autoimmune encephalomyelitis models reversed disease progression without affecting inflammation. Amelioration of disability was abrogated upon neuronal deletion of IL-4R. We discovered direct neuronal signaling via the IRS1-PI3K-PKC pathway underlying cytoskeletal remodeling and axonal repair. Nasal IL-4 application, suitable for clinical translation, was equally effective in improving clinical outcome. Targeting neuronal IL-4 signaling may offer new therapeutic strategies to halt disability progression in MS and possibly also neurodegenerative conditions.


Axons may be injured secondary to demyelination but can also be attacked directly by inflammatory processes in chronic neuroinflammation, such as in human multiple sclerosis (MS). Ongoing and accumulating axon pathology is considered the main feature underlying disability, especially during the progressive disease phase (1, 2). Upon an inflammatory attack, axonal swelling and persisting up-regulation of calcium eventually culminate in beading and degeneration (3, 4). These processes of inflammatory neuronal injury can at least partly be reversed. To date, the axon compartment has not been sufficiently targeted by MS treatment strategies, which typically focus on lymphocyte depletion, preventing lymphocyte invasion or suppressing local inflammation processes (1, 5, 6). Therefore, it is imperative that we gain a full understanding of the pathophysiological processes underlying disease progression and rethink our approaches to treating MS. We hypothesize that the restoration of axonal morphology and function could lead to improved clinical outcome.

T lymphocytes and their cytokines not only do harm but may also display homeostasis-restoring functions in the central nervous system (CNS) (7). In our recent work, we reported that interleukin-4 (IL-4)–producing T helper 2 (TH2) cells exert beneficial effects on neurons upon traumatic CNS injury (8). Both an immune regulatory function of systemically applied IL-4 in experimental autoimmune encephalomyelitis (EAE), the murine model of MS (911), and a destructive role in an asthma model (12, 13) have been reported. Here, we unraveled a function of IL-4/IL-4 receptor (IL-4R) signaling directly in neurons. During EAE, the endogenous IL-4 levels are in the picogram per milliliter range (14). After application of 1 μg of IL-4 to the CNS by intrathecal or, more clinically relevant, intranasal administration during the chronic phase of different EAE models, we found consistent amelioration of clinical signs and axonal morphology without changing inflammation. IL-4 effects were abolished in neuronal IL-4R knockout (KO) mice and were mediated through direct neuronal IL-4R signaling. These findings, combined with our observation that human axons express IL-4R, open up an unconventional concept of immune-CNS interaction.


Intrathecal IL-4 treatment reduces severity, improves functional and structural recovery in chronic EAE, and requires neuronal IL-4R expression

Inspired by our previous finding of a beneficial role of TH2 cells in neurotrauma (8), we investigated whether IL-4R is expressed in human postmortem brain tissue, and we found expression on axons in the CNS of non-MS individuals and in MS patients (Fig. 1, A to D). The signal for IL-4R was mainly observed on the axonal membrane (arrows), surrounding neurofilaments marked with the SMI-31 antibody. Not all axons displayed an IL-4R signal (Fig. 1C), indicating specificity for subtypes of axons. Strikingly, axon swellings in MS displayed particularly strong expression of IL-4R (Fig. 1D, arrowhead).

Fig. 1 IL-4Rα on axons of human postmortem brains.

(A) Interleukin-4 receptor α (IL-4Rα; green, left) in the human isocortex of an individual with no history of neurological disease. (B) Overview of IL-4Rα expression on SMI-31+ axons (red, middle) in the human isocortex of a multiple sclerosis (MS) patient at the gray matter (GM)–white matter (WM) border. (C) Higher magnification of (B) of IL-4Rα staining on membranes (arrows) of some SMI-31+ axons (asterisk). (D) IL-4Rα immunoreactivity on swollen axons (arrowhead) at the site of lesion. Scale bars, 50 μm (B) and 5 μm (A, C, and D).

Subsequently, we applied IL-4 intrathecally in mouse models of chronic EAE that mimic aspects of MS. Here, we found a marked amelioration of disease progression. When injecting IL-4 directly into the cerebrospinal fluid every other day for 14 days, starting from day 5 (d5) after the first disease peak in myelin oligodendrocyte glycoprotein 35–55 (MOG35–55) C57Bl6 EAE, we observed a significant and reproducible reduction in clinical score (P < 0.05; Fig. 2A). Initially, a dose-response curve was performed using 10 ng, 100 ng, and 1 μg of IL-4, with only the 1-μg dosage resulting in a significant difference to phosphate-buffered saline (PBS), although the other concentrations showed a positive trend (fig. S1). Beneficial effects of IL-4 on the severity of disability were also achieved in a secondary progressive (SP) MS EAE model, where the abovementioned treatment regimen was started at a more chronic stage, namely, 14 days after the peak. A rise in clinical score during the chronic phase was reversed by IL-4 (P < 0.05; Fig. 2B). In addition, IL-4 ameliorated disease in male TCR1640 mice that express a transgenic MOG-reactive T cell receptor and develop spontaneous EAE resembling primary progressive (PP) MS (15). Treatments were applied after reaching a clinical score of 3 (complete hindlimb paralysis). PBS-treated mice rapidly progressed to death, whereas animals treated with IL-4 were able to walk throughout the treatment period (Fig. 2C). These data indicate that IL-4 improved the clinical score independently of the EAE model.

Fig. 2 Improvement of clinical score, axon pathology, and locomotor recovery after IL-4 treatment during chronic EAE.

(A) Disease progression in C57Bl6 myelin oligodendrocyte glycoprotein 35–55 (MOG) experimental autoimmune encephalomyelitis (EAE) mice injected intrathecally for 2 weeks with IL-4 (blue, n = 9) or phosphate-buffered saline (PBS; red, n = 9) during the chronic phase (gray bar; representative of three independent experiments). IL-4 treatment in a model of (B) secondary progressive (SP) MS (n = 4 each group) and (C) primary progressive (PP) MS (TCR1640). Individual TCR1640 male mice are shown from the first day of treatment with IL-4 (blue) or PBS (red) onward. (D) Left: IL-4 effects in neuron-specific IL-4R floxed/floxed (fl/fl) calcium/calmodulin-dependent protein kinase IIα (CamKIIα) Cre+ mice [IL-4, dark blue (n = 18); PBS, red (n = 17)] and Cre mice [IL-4, light blue (n = 12)]. Right: Reverse transcriptase polymerase chain reaction for IL-4R with elongation factor 2a (EF2a) as control on lymphocytes and microglia isolated from Cre+ and Cre mice. (E) Representative images of horizontal sections of the corticospinal tract (CST) labeled with yellow fluorescent protein (YFP) and stained for amyloid precursor protein (APP). Scale bar, 100 μm. DAPI, 4′,6-diamidino-2-phenylindole. (F) Quantification of axon swellings corrected for the analyzed area, normalized to the before-treatment (d20) group (n = 3 each group, average of four sections per animal). (G) Quantification of the CST axon density in pixels per area, normalized to the before-treatment group. (H) Illustration of the parameters stride length and base of support (BOS) in the CatWalk output files from a mouse at d0 (healthy; upper panel) and the same mouse at d35 after PBS treatment (middle panel), as well as an IL-4–treated mouse at d35 (lower panel). LH, RH, LF, RF, left and right hind- and forelimbs. (I to J) Quantification of selected CatWalk parameters (I) stride length and (J) BOS (IL-4, n = 8; PBS, n = 7). Statistical analysis was performed using two-way analysis of variance (ANOVA) for repeated measures with Bonferroni correction for clinical score, one-way ANOVA with Tukey’s multiple comparison test for histology, and Mann-Whitney U for CatWalk. *P < 0.05, **P < 0.01, ***P < 0.001.

To clarify the contribution of the neuronal IL-4R to the beneficial effects of IL-4 in chronic EAE, we crossed IL-4Rα floxed/floxed (fl/fl) mice (C57Bl6 background) with the neuron-specific calcium/calmodulin-dependent protein kinase IIα (CamKIIα) Cre line (16). The CamKIIα Cre–driven IL-4Rα KO mice did not display differences in EAE disease induction or peak severity, indicating that the immune system in these mice is functionally normal (Fig. 2D, left panel). Polymerase chain reaction (PCR) analysis of lymphocytes and microglia isolated from the CNS of these mice revealed IL-4Rα expression by these cells, regardless of the Cre genotype (Fig. 2D, right panel). The beneficial effect of IL-4 on the clinical score was abolished in these neuronal IL-4R KO mice (P < 0.05; Fig. 2D), indicating that IL-4 effects were mediated through neuron-specific mechanisms. Using immunohistochemistry, we elaborately characterized the IL-4R expression in these conditional KO mice. IL-4Rα was localized on neurons in the cortex, especially layer V neurons in the motor cortex, and on hippocampal CA1 neurons of Cre controls but was absent in the Cre+ animals, confirming neuronal KO (fig. S2, A and B). IL-4R was detected on astrocytes and microglia (fig. S2, C and D) in both Cre+ and Cre mice, confirming the PCR results. The absence of the beneficial effect of intrathecal IL-4 treatment in these neuronal KO mice suggests that the IL-4 effects in wild-type (WT) mice were mediated by neuronal IL-4R and independent of the immune system.

To investigate axonal pathology, we then used yellow fluorescent protein (YFP)–H mice in which axonal tracts specifically affected by MS (17) express YFP (18). We observed prominent axonal swellings marked by amyloid precursor protein (Fig. 2E), a marker for axon pathology in EAE (19). Quantification of axon swellings in the corticospinal tract (CST) from YFP-H mice was performed at a time point before treatment (d20) and after 2 weeks of intrathecal IL-4 or PBS (d35). IL-4–treated mice showed no difference in the number of axonal swellings compared to d20 in contrast to PBS animals, which accumulated axonal swellings over time (P < 0.001; Fig. 2F). This indicates that IL-4 was able to halt progressive axonal pathology. This was also reflected by CST axon density, which was not different from the density before treatment, whereas the PBS group displayed a significant decrease (PBS to before, P < 0.01; IL-4 to PBS, P < 0.05; Fig. 2G). In the neuron-specific IL-4R transgenic mice, we found an IL-4–induced preservation of axons in the Cre controls in which IL-4R was expressed. Neuronal IL-4R KO (Cre+) mice displayed significant axonal loss when treated with PBS or IL-4 (P < 0.01; fig. S3, A and B). This indicates that the effect of IL-4 on spinal cord axons is dependent on neuronal IL-4R expression. Performing immunohistochemical analysis of the spinal cord, IL-4R was expressed in numerous axons both in the dorsal and ventral white matter in Cre (fig. S4, A and B) but not in Cre+ (fig. S4, C and D). In and around lesions, we found IL-4R expression still at d36 of EAE (fig. S4, E and F). IL-4 reduced demyelination in Cre mice; this amelioration was abolished in Cre+ IL-4R KO mice (P < 0.01; fig. S3, C and D), suggesting that this reduction was secondary to the neuroprotection. We observed no differences in lymphocyte infiltration (fig. S3, E and F). At d36, the morphology of the microglia in the dorsal CST region was slightly more complex in IL-4–treated mice (increased shape factor: 7.8 ± 0.6 for PBS Cre+, 9.3 ± 0.5 for IL-4 Cre, P < 0.001 compared to PBS, and 8.7 ± 0.4 for IL-4 Cre+, P < 0.05 compared to PBS; fig. S3, G and H). Because the microglial morphology was only marginally different and increased in both Cre and Cre+ mice, this indicates that the functional and structural IL-4 treatment effects were independent of a slight effect of IL-4 on microglia. These data show that the recovery of the clinical score, the increased axon density, and the reduced demyelination all depended on the neuronal expression of the IL-4R.

Because the clinical score is a rather coarse measure of the clinical status, we then quantified locomotor parameters using CatWalk-automated gait analysis (20) as a functional correlate to axonal pathology. Mice were habituated to the walkway so that they learned to perform the test even while sick, with reliable walking patterns produced up to a clinical score of 2. All IL-4–treated mice were able to walk during the complete testing period; however, one PBS-treated mouse was unable to walk at d35 and was therefore excluded from the analysis. Because of the disease, the step size (stride length) decreased (P < 0.001), and the distance between the hindpaws [base of support (BOS)] increased (P < 0.001; Fig. 2, H to J, and movies S1 to S3). In PBS-treated mice, the reduction in stride length persisted over time (P < 0.001; Fig. 2I), and the BOS increased even further (P < 0.001 compared to d0, P < 0.01 compared to d20; Fig. 2J). The pathological CatWalk parameters corresponded to the increased axonal pathology observed in the histological analysis (Fig. 2, E to G). In contrast to PBS-treated mice, IL-4 mice displayed no further deterioration of the BOS compared to d20 (P = 0.45 compared to d20; IL-4, P < 0.05 reduction compared to PBS; Fig. 2J). Even more striking, the stride length completely recovered to healthy levels because of IL-4 treatment (P = 0.07 compared to d0; Fig. 2I). More parameters, such as walking speed and print area, showed a similar recovery to normal values in IL-4–treated mice, whereas PBS mice deteriorated further (fig. S5 and Table 1).

Table 1 Effects of IL-4 in comparison to PBS controls on locomotor parameters measured with CatWalk.

Effects of treatments on CatWalk parameters at d10 (preclinical), d20 (clinical), and d35 [end point of treatment with IL-4 (n = 8) or PBS (n = 7)].

View this table:

IL-4 applied during chronic EAE does not affect the immune system

Because lymphocyte infiltration was not affected by IL-4 treatment (fig. S3, E and F), we then investigated immune cells in more detail by fluorescence-activated cell sorting (FACS) analysis of MOG35–55 C57Bl/6 EAE mice at d37, where the difference between IL-4– and PBS-treated groups was greatest. We observed no differences in numbers and subtypes of CD4+ lymphocytes or the percentage of CD11b+ major histocompatibility complex class II+ microglia and macrophages (Fig. 3). This suggests that IL-4 ameliorated the clinical score despite ongoing inflammation. To check whether our IL-4 regimen was able to change the immune system under other conditions, we injected IL-4 intrathecally at disease onset (clinical score of 0.5 to 1) and performed FACS analysis after the 2-week treatment period. This early IL-4 treatment indeed led to less severe disease, accompanied by an increase in Gata3+ CD4+ TH2 cells (P < 0.05; fig. S6), which is in accordance with previously described IL-4 effects in early EAE (10, 21, 22). This indicates that IL-4, when applied early in EAE, reduced inflammation, whereas IL-4 treatment during the chronic phase ameliorated progressive disease via nonimmune mechanisms.

Fig. 3 Absence of effects of IL-4 on immune cells.

(A) Gating strategy for the immune cells isolated from the central nervous system (CNS) of C57Bl6 MOG mice treated with PBS or IL-4. FACS, fluorescence-activated cell sorting; SSC, side scatter; FSC, forward scatter; MHCII, major histocompatibility complex class II; TNFα, tumor necrosis factor α; IFN-γ, interferon-γ; GM-CSF, granulocyte-macrophage colony-stimulating factor. Lymphocyte subtypes and CD11b+MHCII+ cells in the (B) CNS and (C) spleens at end point (d37). Statistical analysis was performed using unpaired t test (n = 3 per group, confirmed in two independent experiments).

IL-4 effects on neurons are due to fast direct neuron-specific signaling

To dissect IL-4 activity on neurons, we moved to in vitro models for neuroprotection and outgrowth. We performed a neuron viability assay by adding N-methyl-D-aspartate to dissociated cortical neurons and were able to show that many PBS-treated neurons (80%) incorporated propidium iodide in comparison to IL-4–treated neurons (P < 0.001; Fig. 4, A and B), indicating that the neurons were protected by IL-4 against damage due to excitotoxicity. We then created an explant culture model of mouse motor cortex layer V, thus specifically modeling the CST axons we analyzed in vivo (Fig. 4C). Explants displaying comparable axon outgrowth after 24 hours in culture were quantified regarding basal axon length. Axon growth at 48 and 72 hours was measured after incubation with IL-4 (50 ng/ml) or equal volumes of PBS. IL-4 significantly increased axon outgrowth in WT cortex (48 hours, P < 0.001; 72 hours, P < 0.01; Fig. 4D) but not in explants from CamKIIα Cre–driven IL-4Rα KO mice (Fig. 4E). To investigate whether the CST also displayed increased axonal outgrowth in vivo, we injected the anterograde tracer rhodamine-conjugated dextran into the motor cortex 7 days before EAE induction. IL-4 treatment was performed as always from 5 days after the peak (P < 0.05; Fig. 4F). Animals were sacrificed after five injections (10 days), and spinal cord tissue was processed for confocal imaging. Because sprouting may occur at any position along the CST, we quantified only those images in which actual sprouts were observed. IL-4–treated animals displayed increased sprouting as compared to PBS (P < 0.001; Fig. 4, G and H). The tracing density was not different between the groups. We then tested whether IL-4–treated axons could overcome inhibitory molecules by cultivating the cortical layer V explants on plates coated with Nogo-A (23, 24). Nogo-A is a known axon growth inhibitor (25, 26), and Lingo-1, which is part of the Nogo receptor, is known to inhibit oligodendrocyte differentiation and myelination (27). Control explants treated with PBS hardly grew on Nogo-A (5 μg/ml), whereas IL-4–treated explants displayed significant axonal growth at 24 hours (P < 0.01; fig. S7, A and B) and exhibited extensive growth at 48 hours (fig. S7C). These data indicate that IL-4 induces regenerative plasticity of spinal cord axons.

Fig. 4 Beneficial effects of IL-4 on neurons and axons.

(A) Neuron viability assay. Incubation of dissociated cortical neurons with 1 μM N-methyl-D-aspartate and concomitant treatment with PBS or IL-4 (50 ng/ml) followed by incubation with propidium iodide (PI). Immunocytochemistry showing neurons marked with tubulin-beta-III (Tubb3; red), PI (green), and DAPI (blue). Scale bar, 10 μm. (B) Quantification of PI+ neurons (n = 4 per group, representative of two independent experiments). (C) Explants of layer V of the motor cortex after two to three days in vitro. The 10 longest axons are marked with black dots. Scale bar, 100 μm. (D) Quantification of effects of IL-4 treatment at 48 and 72 hours (IL-4, n = 5; PBS, n = 8, pooled from three experiments). WT, wild-type. (E) Quantification of axonal outgrowth in cortical explant cultures of the neuron-specific IL-4R knockout mice (n = 3 each group). (F) Injection of anterograde tracer 7 days before EAE induction and treatment with IL-4 during the chronic phase (gray bar) of C57Bl6 MOG EAE showing the typical disease course and treatment response (PBS, n = 8; IL-4, n = 9). (G) Representative images of the traced dorsal CST (upper left, dashed border) displaying sprouting into the gray matter (arrows). Scale bar, 50 μm. (H) Quantification of CST sprouting and tracing density (n = 4 animals, three sections per animal). Statistical analysis was performed using unpaired t test for cell culture and histology and two-way ANOVA for repeated measures with Bonferroni correction for clinical score. *P < 0.05, **P < 0.01, ***P < 0.001.

Next, we aimed to explore the neuronal IL-4R signaling pathway. Immunoprecipitation with an IL-4Rα antibody was performed on homogenates of dissociated cortical neurons in culture and revealed coprecipitation of SH2 domain (Src homology 2 domain)–containing adaptor protein (Shc) and insulin receptor substrate 1 (IRS1; Fig. 5A). Phosphatidylinositol-3 kinase (PI3K) was recruited to the IL-4R–IRS1 complex in IL-4–treated neurons (P < 0.001; Fig. 5, A and B). The phosphorylation of IRS1 was significantly increased after only 10 min of treatment of the cortical neurons with IL-4 in comparison to PBS control treatment (P < 0.01; Fig. 5, C and D). One of the downstream molecules of the IRS1-PI3K pathway is protein kinase C (PKC), which is known to phosphorylate growth-associated protein–43 (GAP-3), a regeneration-associated protein (28, 29). Here, we show an up-regulation of phosphorylation of the CST marker PKCγ (P < 0.05), accompanied by an increase in GAP-43 phosphorylation (P < 0.01) after 10 min of IL-4 incubation (Fig. 4, C and D).

Fig. 5 Direct signaling of IL-4 in neurons.

(A) Immunoprecipitation (IP) using IL-4Rα antibody and detecting antibodies for SH2 domain (Src homology 2 domain)–containing adaptor protein (Shc), insulin receptor substrate 1 (IRS1), and phosphatidylinositol-3 kinase (PI3K), the latter after treatment with PBS or IL-4. Input (in): Protein lysates of dissociated cortical neurons, washing steps (w1 to w3). Output (out): Sample after IP elution. (B) Quantification of PI3K recruitment after incubation with IL-4 (n = 3). (C) Phosphorylation assays on dissociated cortical neurons treated for 10 min with IL-4 (50 ng/ml) or PBS. Western blots for phospho- and total IRS1, protein kinase C γ (PKCγ), and growth-associated protein–43 (GAP-43). (D) Ratios of phosphorylated protein through total protein for IRS1 (n = 3), PKCγ (n = 9 to 11), and GAP-43 (n = 6 to 7). (E) Quantification of cortical axon growth with IL-4 treatment in the presence of PKC inhibitor bisindolylmaleimide I (BisI; n = 3 to 5 each time point per group). (F) Phosphorylation of signaling molecules in response to IL-4 treatment of cortical neurons from the IL-4R fl/fl CamKIIα Cre mice. (G) Representative images of immunohistochemistry for pIRS1 (green) on the CST and dorsal columns (DC) in IL-4– or PBS-treated EAE mice. Scale bar, 50 μm. (H) Quantification of pIRS+ axon profiles corrected for the analyzed area (n = 5). Statistical analysis was performed using unpaired t test for phosphorylation assays and one-way ANOVA with Tukey’s multiple comparison test for growth assay. *P < 0.05, **P < 0.01, ***P < 0.001.

Incubation of cortex explants with the PKC inhibitor bisindolylmaleimide I (BisI) completely abolished the effects of IL-4 (P < 0.01), whereas BisI alone did not influence basic axon outgrowth (Fig. 5E). Even more striking, cortical neurons isolated from the neuron-specific IL-4R KO mice displayed reduced phosphorylation of all signaling molecules and GAP-43 (P < 0.05; Fig. 5F). This was a partial reduction because these signaling molecules can be activated by other pathways stimulated by components of the medium. Moreover, only excitatory neurons are CamKIIα Cre+ (30); therefore, not all neurons in the culture are IL-4–deficient. Finally, we performed immunohistochemistry on EAE animals treated with IL-4 or PBS and found a clear increase in the number of axons containing pIRS1 (P < 0.01; Fig. 5, G and H). This indicates that IL-4 activates the above-identified IL-4R–IRS1 signaling pathway in axons in vivo.

IL-4Rα chain can interact with two different co-receptor chains, the common γ chain or the IL-13Rα chain, resulting in type I or type II IL-4R, respectively (31). Here, we show that dissociated cortical neurons and their axons expressed IL-4Rα as well as both co-chains and therefore all receptor types (fig. S8A). Because IL-13 is capable of binding to IL-4R type II [but with a 100-fold lower affinity (31, 32)], we tested IL-13 in the outgrowth assay but found no increased axonal outgrowth (fig. S8B). In addition, IL-13 had no significant effect on the phosphorylation of PKCγ and GAP-43 (fig. S8C). This indicates that the signaling and outgrowth effects observed above were IL-4–specific.

PI3K converts phosphatidylinositol-4,5-bisphosphate to phosphatidylinositol-3,4,5-trisphosphate, leading to the opening of Ca2+ channels in neurons (33, 34). A rise in Ca2+ is known to activate PKC (28, 35), which fits with our observations above, and disrupt the connection between GAP-43 and calmodulin (CaM). We performed immunoprecipitation of GAP-43 from neurons treated with IL-4 or PBS and found a reduction in CaM binding (P < 0.01; Fig. 6, A and B). PKC, GAP-43, and CaM all play a role in the modification of the actin cytoskeleton via different actin-remodeling proteins (28, 36, 37). Therefore, we made use of phalloidin to mark filamentous actin (F-actin) in dissociated cortical neurons. After 30 min of incubation with IL-4, we observed a marked shift in the F-actin signal. Whereas PBS-treated neurons displayed a patch-like clustering of F-actin in the cell bodies and neurites (Fig. 6C), IL-4 induced strong filamentous patterns in the neurites and a shift toward the neuron’s surface (Fig. 6D). BisI abolished this effect (Fig. 6E), indicating that the IL-4–induced changes in F-actin were indeed downstream of PKC. We then used the Ca2+ chelator BAPTA-AM [1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid acetoxymethyl ester] to confirm that these effects were due to Ca2+. The IL-4–induced shift in F-actin was not observed in cultures treated with IL-4 + BAPTA-AM (Fig. 6F). The length of the actin filaments was quantified and displayed a significant increase in IL-4–treated neurons (P < 0.001; Fig. 6G).

Fig. 6 Cytoskeletal remodeling after IL-4 treatment in vitro.

(A) IP with GAP-43 antibody to detect calmodulin (CaM) binding. (B) Quantification of IL-4–induced release of CaM (n = 3 per treatment). (C to F) Immunocytochemistry for Tubb3 (red, left) and phalloidin (green, middle), a marker for filamentous actin (F-actin), with merged image (right) for dissociated cortical neurons treated with (C) PBS, (D) IL-4 (30 min, 50 ng/ml), (E) IL-4 + BisI, and (F) IL-4 + BAPTA-AM [1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid acetoxymethyl ester]. Patchy clusters of F-actin are marked with asterisks, strong filamentous F-actin signals in neurites are marked with arrows, and a shift toward the surface of the cell body is marked with arrowheads. (G) Quantification of actin filament length (n = 5). Scale bar, 10 μm. Statistical analysis was performed using unpaired t test for CaM release and one-way ANOVA with Tukey’s multiple comparison test for F-actin quantification. **P < 0.01, ***P < 0.001.

These findings strongly indicate that IL-4 acts directly via neuronal IL-4R through IRS1-PI3K-PKC signaling, with GAP-43 and CaM as main regulators of F-actin cytoskeletal modulation (fig. S9). This pathway likely results in reduced axon pathology and increased axon outgrowth and sprouting, thereby leading to the marked functional recovery in EAE mice.

Intranasal application of IL-4 leads to similar functional recovery

To provide a more clinically relevant route of application, we compared lumbar and nasal treatment of IL-4. As before, C57Bl6 MOG animals were treated with 1 μg of IL-4 or PBS from d5 after the disease peak, every other day for 2 weeks. Strikingly, similar to lumbar IL-4, nasal IL-4 markedly improved the clinical score as compared to nasal PBS controls (P < 0.05; Fig. 7A). Analysis of the CST axons revealed that nasal IL-4 was even more effective than lumbar IL-4 in reducing axonal swellings (nasal IL-4 to PBS, P < 0.001; nasal IL-4 to lumbar IL-4, P < 0.05; Fig. 7, B and C). We checked whether this route of application affected systemic and CNS inflammation, and like before, no difference in the numbers or subtypes of CD4+ lymphocytes (fig. S10, A and B) as well as no effect on demyelination and only a minor effect on microglial morphology (shape factor: 5.7 ± 0.2 for PBS, 8.4 ± 0.2 for nasal IL-4, P < 0.001; fig. S10C) were observed. Therefore, we conclude that nasal IL-4 treatment, like lumbar IL-4, ameliorates clinical signs through neuronal mechanisms, independently of inflammation.

Fig. 7 Improvement of clinical score and axon pathology by nasal IL-4 treatment during chronic EAE.

(A) Disease progression in C57Bl6/YFP-H MOG EAE mice treated for 2 weeks with IL-4 via intrathecal injection (blue; n = 5) or intranasally with IL-4 or PBS (light blue and red; n = 12 each) during the chronic phase (gray bar). (B) Representative images of horizontal sections of the YFP-labeled CST counterstained for APP, showing axonal swellings. Scale bar, 50 μm. (C) Quantification of axon swellings corrected for the analyzed area (n = 3 each group, average of three to five sections per animal). Statistical analysis was performed using two-way ANOVA for repeated measures with Bonferroni correction for clinical score and one-way ANOVA with Tukey’s multiple comparison test for quantification of axonal swellings. *P < 0.05, ***P < 0.001.


Adaptive immunity plays a role not only in disease but, according to emerging evidence, most likely also in homeostasis of the CNS (7, 38). Together with our previous findings on beneficial effects of TH2 cells in experimental neurotrauma (8), we hypothesized that IL-4 may have beneficial roles in the inflamed CNS beyond its known immune regulatory function (9, 10). Here, we discovered an unconventional communication between the immune and nervous systems: fast—that is, within minutes—direct IL-4R signaling in neurons, resulting in reduced axonal pathology and increased outgrowth capacity, even in the presence of inhibitory Nogo.

In chronic inflammation of the CNS in patients, rather limited success has been achieved in halting disability progression without side effects or in shutting down the immune response completely. Furthermore, the latter is no guarantee of halting progression because local inflammation may persist, CNS intrinsic mechanisms may continue, or complete inflammation shutdown may come too late (3941). Our findings tackle principal questions of repair in the neuronal compartment.

We demonstrate prominent and reproducible amelioration of clinical scores by intrathecal IL-4 treatment in three different EAE models, suggesting that IL-4 acts regardless of the initial cause of the disease. Furthermore, the effects of IL-4 were independent of inflammation. We note that an earlier study on local application of an IL-4–producing virus showed immune regulation through an increase in local regulatory T cells (22). However, the virus was applied during the inflammatory attack, whereas we applied IL-4 exclusively during the chronic phase of the disease.

The IL-4 effect was completely abrogated in the IL-4R fl/fl CamKIIα Cre mice. In these mice, IL-4R expression was abolished on neurons, whereas expression on astrocytes, lymphocytes, and microglia was maintained. A detailed analysis of the known CamKIIα distribution revealed a widespread neuronal expression in the cerebral cortex, hippocampus, amygdala, striatum, thalamus, and hypothalamus as well as in nuclei of the medulla (16, 30, 42, 43). Most of these are excitatory, but expression of CamKIIα in inhibitory neurons in olfactory bulb and cerebellum (Purkinje cells) has also been described (30). In line with known CamKIIα expression in dorsal root ganglion neurons (44), we demonstrated IL-4R expression on motor and sensory tracts of the spinal cord, which was abolished in the neuronal IL-4R KO. Although there are some reports on CamKII expression in activated microglia (45, 46), the CamKIIα Cre IL-4R KO mice displayed a normal EAE disease course, which is in clear contrast to microglia/macrophage LysM Cre IL-4Rα mice that are protected from EAE (47). This speaks against a microglial influence in the IL-4R fl/fl CamKIIα Cre mice. Accordingly, a recent study, which also made use of CamKIIα Cre mice to distinguish between neurons and microglia to study progranulin expression, found neuron-specific depletion of progranulin, with microglia still expressing this gene (48). We conclude that the CamKIIα Cre KO is neuron-specific; therefore, our data indicate that the observed IL-4 effects on clinical scores and axon integrity are mediated by direct neuronal IL-4R–dependent mechanisms. Moreover, our mice deteriorated after withdrawal of IL-4, indicating that the ongoing inflammation reinduced axon damage, also signifying a neuroprotective role of IL-4.

Our data show that IL-4 prevents axon pathology and leads to functional recovery of locomotion. Axonal swelling was recently characterized as the earliest sign of inflammation-triggered damage, independent of demyelination (4). This leads to a blockade of axonal transport and subsequent axonal dysfunction in various neurological diseases (49, 50). We and others showed that axon swellings can either recover to healthy morphology or proceed further to fragmentation (3, 4, 51). This suggests a dynamic process of axonal repair and degeneration occurring in parallel. Because we found IL-4R expression on axon swellings in human MS patient brains, we looked more closely at axons in EAE. In the PBS group, axonal swellings were massively increased over time, and some locomotor parameters, measured by CatWalk, deteriorated. Overall, this indicates an ongoing axonal pathological process in the chronic phase of EAE. Decreased axon density in control PBS mice is in line with these observations. We observed an increase in axon density and a strong suppression of axonal swelling in IL-4–treated mice. In line with this structural repair, IL-4 treatment of EAE mice led to complete recovery of several hindlimb functions.

Compared to PBS mice, IL-4 caused an improvement in the extent of demyelination; however, this effect was abolished in the neuronal IL-4R KO; therefore, we conclude this to be secondary to the axonal effects. A marginal effect of IL-4 on microglial morphology observed in our study is in accordance with a recent study on a trauma model, where acute IL-4 treatment only affected a small proportion of microglia and induced only some M2 markers without reducing M1 phenotypes (52). In addition, in our study, proinflammatory cytokine-producing T cells remained present throughout, which could explain why IL-4 had only a marginal effect on microglial morphology. The abolishment of IL-4 effects in the neuronal IL-4R KO mice did not depend on these microglial changes because these were observed in both Cre and Cre+ mice. Therefore, the IL-4 effect on clinical score, axon density and morphology, and demyelination were deduced to be completely dependent on neuronal IL-4R expression. Strikingly, the nasal application of IL-4 also neither changed T cell cytokine profiles nor affected demyelination, with the same marginal changes in microglial morphology as observed for lumbar IL-4. Thus, both lumbar and nasal IL-4 reduced axonal swelling without changing inflammation.

Distinct from our previous observation in traumatic injury where we found that IL-4 indirectly stimulated axon growth by enhancing neurotrophin signaling (8), we here discovered a direct and fast (within minutes) IL-4R signaling pathway in the neuron. This signaling via IRS1-PI3K-PKC (fig. S9) and the downstream phosphorylation of GAP-43, a well-known regeneration-associated protein (28, 29, 36), leads to cytoskeletal modification and axon growth. This supports the observed axonal repair and functional recovery. Given that axons in MS patients’ brains express IL-4R, our study provides the basis for a treatment strategy addressing neuronal injury in neuroinflammation and possibly neurodegenerative conditions.

Ultimately, we reveal a principle of immune-CNS cross-talk by demonstrating direct IL-4R–mediated signaling in neurons. When IL-4 was applied directly to the CNS compartment, we observed amelioration of chronic disease in different EAE models, from the conventional C57Bl6 model to SP EAE to spontaneous PP EAE. IL-4 was able to reverse severity from complete hindlimb paralysis to near-normal walking by rescuing axon damage independent of the immune system. Before translation to the clinic is possible, some limitations of this study will need to be addressed, namely, the lack of knowledge regarding systemic side effects as well as pharmacodynamics and pharmacokinetics. Furthermore, the amount of recombinant IL-4 needed in human has yet to be assessed. However, the nasal application of IL-4 being as effective as lumbar IL-4 provides a first step toward clinical translation.


Study design

The aim of this study was to investigate the effects of IL-4 on neuroinflammatory disease. To this end, we conducted controlled laboratory experiments using mouse models. The treatment group size for clinical scoring was typically six to nine animals (based on experience). Animals that did not reach a clinical score of 2 at disease peak were excluded from the study. Before treatment, the animals were randomized so that the initial disease curves were similar between the groups. C57BL6 MOG EAEs were repeated five times, the SP MS model was repeated twice, and for the spontaneous MS model (TCR1640), individual animals were analyzed because disease incidence was low. The IL-4R KO experiment was performed once, and the nasal study was performed twice. FACS and histology were repeated in two independent experiments. For histology, three to four mice per treatment group were randomly selected, and two to five sections per mouse were analyzed. Histology and the nasal studies were performed blindly. Primary data are located in table S1.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 5 (GraphPad Software Inc). Clinical scores were analyzed using repeated-measures two-way analysis of variance (ANOVA) with post hoc Bonferroni correction. CatWalk data were analyzed using Mann-Whitney U test. Data obtained from tissue analysis, in vitro assays, and Western blots were subjected to unpaired t test or one-way ANOVA with Tukey’s test for multiple comparison. Data are plotted as means ± SEM. For all other Materials and Methods, see the Supplementary Materials.


Table S1. Primary data.

Materials and Methods

Fig. S1. Dose-response curve.

Fig. S2. IL-4R expression in IL-4R fl/fl CamKIIα Cre mice.

Fig. S3. Histological analysis of the IL-4R fl/fl CamKIIα Cre+ and Cre spinal cord.

Fig. S4. IL-4R expression in spinal cord axons.

Fig. S5. Quantification of locomotor parameters using the CatWalk system.

Fig. S6. Early IL-4 treatment of C57Bl6 MOG EAE.

Fig. S7. Axonal growth on inhibitory Nogo-A.

Fig. S8. IL-4R subtypes and absence of IL-13 effect.

Fig. S9. IL-4R signaling pathway in neurons.

Fig. S10. FACS analysis and histology after nasal IL-4.

Movie S1. Representative CatWalk run for a healthy mouse (pre-induction, d0).

Movie S2. Representative CatWalk run for a PBS-treated mouse at d35.

Movie S3. Representative CatWalk run for an IL-4–treated mouse at d35.

References (5359)


Acknowledgments: We are grateful to C. Oswald, S. Fregin, A. Zymny, and M. Pfeiffer for excellent technical support. We also thank R. Lu for training of the lumbar injections and C. Ernest for proofreading and editing of the manuscript. Funding: This work was supported by the German Research Foundation (DFG; CRC 1080 to J.K. and F.Z. and CRC-TR-128 to F.Z.), Progressive MS Alliance (PA-1604-08492, BRAVEinMS to F.Z.), and NIH/National Institute of Neurological Disorders and Stroke (NS096967 to J.K.). Author contributions: C.F.V., R.N., and F.Z. conceived the study and designed the experiments; S.M. performed the in vivo experiments; S.L. performed the in vitro experiments; S.M., S.L., U.B., K.B., J.B., C.F.V., and F.Z. analyzed the data; A.S. contributed to immunocytochemistry; C.S.R. provided human material; J.V. supervised human experiments; C.F.V., J.K., and F.Z. drafted the manuscript; and S.B. and C.S.R. were involved in editing and discussion. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Data that support the findings of this study are available from the corresponding author upon request.

Stay Connected to Science Translational Medicine

Navigate This Article