Oligodendroglial fibroblast growth factor receptor 1 gene targeting protects mice from experimental autoimmune encephalomyelitis through ERK/AKT phosphorylation

Abstract

Fibroblast growth factors (FGFs) exert diverse biological effects by binding and activation of specific fibroblast growth factor receptors (FGFRs). FGFs and FGFRs have been implicated in demyelinating pathologies including multiple sclerosis. In vitro activation of the FGF2/FGFR1 pathway results in downregulation of myelin proteins. FGF1, 2 and 9 have been shown to be involved in the pathology of multiple sclerosis. Recent studies on the function of oligodendroglial FGFR1 in a model of toxic demyelination showed that deletion of FGFR1 led to increased remyelination and preservation of axonal density and an increased number of mature oligodendrocytes. In the present study the in vivo function of oligodendroglial FGFR1 was characterized using an oligodendrocyte‐specific genetic approach in the most frequently used model of multiple sclerosis the MOG_35‐55‐induced EAE. Oligodendroglial FGFR1 deficient mice (referred to as Fgfr1^ind−/−) showed a significantly ameliorated disease course in MOG_35‐55‐induced EAE. Less myelin and axonal loss, and reduced lymphocyte and macrophage/microglia infiltration were found in Fgfr1^ind−/− mice. The reduction in disease severity in Fgfr1^ind−/− mice was accompanied by ERK/AKT phosphorylation, and increased expression of BDNF and TrkB. Reduced proinflammatory cytokine and chemokine expression was seen in Fgfr1^ind−/− mice compared with control mice. Considering that FGFR inhibitors are used in cancer trials, the oligodendroglial FGFR1 pathway may provide a new target for therapy in multiple sclerosis.

Introduction

Fibroblast growth factors (FGFs) are secreted glycoproteins which exert diverse biological effects by binding and activation of specific fibroblast growth factor receptors (FGFRs) 25, 40. FGF2 is the prototypic ligand of the FGF family with complex paracrine and autocrine modes of action mediated by four receptor tyrosine kinases FGFR1‐4 25, 33, 40. In vitro and in vivo data suggest that FGF play a role in demyelinating pathologies 41, 50, 57. In humans, the most common demyelinating disease is multiple sclerosis (MS). In MS increased expression of FGF1 in remyelinated lesions 50, increased glial expression of FGF9 in active demyelinated lesions 41 and a high number of FGF2+ macrophages/activated microglia in active lesions were described 11. Furthermore, high levels of FGF2 protein expression were found during relapse in serum 27 and cerebrospinal fluid (CSF) 59. In a model of toxic demyelination FGF2 k.o. mice showed enhanced oligodendrocyte repopulation of demyelinated lesions 3, enhanced remyelination 2 and reduced axonal damage 64. In rats application of FGF2 into the CSF over 3 days caused disruption of mature oligodendrocytes, loss of myelin and axon degeneration 8. In contrast to these findings suggesting a negative role for FGF2, in MOG_35–55‐induced experimental autoimmune encephalomyelitis (EAE), the most widely used animal model of MS, FGF2 ablation leads to a more severe disease course, increased lymphocyte and macrophage infiltration and decreased remyelination 57. This finding is supported by a FGF2 gene therapy study in EAE resulting in less disease severity, reduced lymphocyte and macrophage infiltration and an increased number of myelin‐forming oligodendrocytes 58. In vitro activation of the FGF2/FGFR1 pathway in oligodendrocyte lineage cells resulted in downregulation of myelin proteins 13 and altered patterns of FGF receptor expression 4. These data support an important role of FGF2 in MS and demyelinating models including EAE.

The function of the corresponding receptor FGFR1 in MS and EAE is not well‐defined. FGFR1 is present on oligodendrocytes, astrocytes and neurons in humans 22 and rodents 56, where it has been shown to be expressed in both oligodendrocyte progenitor cells (OPC) and differentiated oligodendrocytes. In patients with MS FGFR1 is upregulated in OPC in active lesions and around chronic–active and chronic–inactive lesions 11 and in remyelinated lesions 50. In rats FGFR1 is expressed on macrophages/activated microglia in the acute phase of EAE 43. FGFR1 deficient mice showed a reduction in myelin sheath thickness and expression of myelin proteins, there was no effect on OPC proliferation or oligodendrocyte differentiation in these k.o. mice 19.

In a model of toxic demyelination, conditional FGFR1 deletion in oligodendrocytes results in increased remyelination and axonal density and an increased number of mature oligodendrocytes 69 suggesting that FGFR1 may be an inhibitor of remyelination in this model. Effects of FGFR1 are mediated through ERK/AKT pathways 19, 29. Activation of the (PI3K)/AKT pathway in mice results in enhanced myelination 15, 53 but has no effect on OPC proliferation or the number of mature oligodendrocytes 15. ERK deficient mice show reduced myelin thickness, but there is no effect on the number or differentiation of oligodendrocytes 32. In MOG_35–55‐EAE ERK1 null knockout mice and AKT3^−/− mice present with a more severe disease course, increased inflammation and demyelination 1, 65. Apart from findings in MS or demyelinating disease models, FGFR1 has been shown to play a key role in carcinogenesis 12, and multikinase or selective FGFR inhibitors and monoclonal FGFR antibodies are currently being tested in preclinical and clinical trials 12.

The objective of this study was to characterize the role of oligodendroglial FGFR1 in MOG_35–55‐induced EAE. Based on recent findings on the function of oligodendroglial FGFR1 in toxic demyelination we hypothesized that deletion of FGFR1 would result in a less severe disease course, reduced demyelination and axonal injury, and decreased inflammation. B6.Cg‐Tg(PLP1‐cre/ERT)3Pop Fgfr1^tm5.1Sor mice were administered tamoxifen to induce conditional Fgfr1 deletion in oligodendrocyte lineage cells (referred to as Fgfr1^ind−/− mice). In agreement with the hypothesis a milder disease course, reduced myelin degeneration, increased axonal density, and decreased lymphocyte and macrophage infiltration were found in Fgfr1^ind−/− mice. The underlying mechanisms include increased ERK and AKT phosphorylation, an increase in BDNF and TrkB expression, and decreased expression of LINGO‐1. These data suggest that oligodendroglial FGFR1 is a modulator of demyelination and axonal degeneration in MOG_35–55‐induced EAE.

Materials and Methods

Generation of Fgfr1 conditional knock‐out mice

B6;129S4‐Fgfr1^tm5.1Sor/J, B6.Cg‐Tg(Plp1‐cre/ERT)3Pop/J and C57BL/6 mice were purchased from The Jackson Laboratories (Bar Harbor, ME, USA). B6;129S4‐Fgfr1^tm5.1Sor/J mice were crossbred with B6.Cg‐Tg(Plp1‐cre/ERT)3Pop/J mice and backcrossed to a C57BL/6J background. Genomic DNA was isolated (DirectPCR‐Tail, Peqlab, Erlangen, Germany) to identify the deletion of FGFR1 floxed allele and PLP transgene, mice were genotyped by PCR with the following primers: FGFR1 forward primer 5′‐GGACTGGGATAGCAAGTCTCTA‐3′ and reverse primer 5′‐GTGGATCTCTGTGAGCCTGAG‐3′; PLP transgene forward 5′‐GCGGTCTGGCAGTAAAAACTATC‐3′ and reverse primer 5′‐GTGAAACAGCATTGCTGTCACTT‐3′; CRE internal positive control primer forward 5′‐CTAGGCCACAGAATTGAAAGATCT‐3′ and reverse primer 5′‐GTAGGTGGAAATTCTAGCATCATCC‐3′. The FGFR1 primers in our genotyping assay flank the upstream loxP site in intron 3. Genotyping protocols were followed as provided by The Jackson Laboratories. PCR products were detected by agarose gel electrophoresis (Peqlab, Erlangen, Germany). For induction of Cre recombinase 4‐week‐old B6.Cg‐Tg(PLP1‐cre/ERT)3Pop Fgfr1^tm5.1Sor female mice were treated with 1 mg of tamoxifen (Sigma‐Aldrich, Steinheim, Germany) in 100 µL sunflower oil/ethanol i.p. over five consecutive days (referred to as Fgfr1^ind−/−) (Figure 1A). B6.Cg‐Tg(PLP1‐cre/ERT)3Pop Fgfr1^tm5.1Sor littermate control mice received a sunflower oil/ethanol mixture (referred to as controls) (Figure 1A). Animal experiments were approved by the local state authorities of Hesse, Giessen, Germany (GI 20/23‐Nr. 31/2008).

Figure 1.

Tamoxifen‐induced, conditional deletion of Fgfr1 in oligodendrocytes followed by EAE induction with the MOG_35‐55 peptide (A). About 100 µL of Tamoxifen (Fgfr1^{ind −/−}) or vehicle (sunflower oil/ethanol; control) were injected on five consecutive days in 4‐week‐old female B6.Cg‐Tg(PLP1‐cre/ERT)3Pop Fgfr1^tm5.1Sor mice. EAE was induced at 8–12 weeks of age. Clinical symptoms were monitored until day 62 p.i. Samples were collected on day 18–20 p.i. (acute phase) and on day 62 p.i. (chronic phase). B, Fgfr1^ind−/− mice showed a delayed onset of disease. The maximum EAE score was at day 14 p.i. in control and day 16 p.i. in Fgfr1^ind−/− mice. A milder disease course was found in Fgfr1^ind−/− mice from day 33 p.i. to the end of the chronic phase (with exception of days 47 and 48). In the chronic phase Fgfr1^ind−/− mice presented with a mild paraparesis whereas controls still had a severe paraparesis. n = 22 for each group, data are presented as mean ± SEM. C, Fgfr1^ind−/− mice showed improvement in EAE scores compared with control mice. Mean improvement per animal was calculated by subtracting the EAE score at day 62 from the peak EAE score. Fgfr1 expression was less in Fgfr1^ind−/− mice compared with control mice in the acute (D) and chronic phases (E) of EAE.

Induction and clinical evaluation of MOG_35‐55‐induced EAE

EAE was induced in 8‐ to 12‐week‐old female Fgfr1^ind−/− mice and control mice (Figure 1A) by subcutaneous injection of 300 µg myelin oligodendrocyte glycoprotein peptide (MOG_35‐55; Charité Hospital, Berlin, Germany) emulsified in complete Freund′s adjuvant (Sigma, Steinheim, Germany) containing 10 mg Mycobacterium tuberculosis (Difco, Michigan, USA). Intraperitoneal injections of 300 ng pertussis toxin (Calbiochem, Darmstadt, Germany) were given at the time of immunization and 48 hours later. The clinical disease course was monitored until day 62 in a blinded manner. Mice were sacrificed at days 18–20 p.i. for the acute phase and at day 62 p.i. for the chronic phase (Figure 1A). Three independent experiments were performed. EAE disease course was evaluated using a scale that ranged from 0 to 5: 0 = normal, 0.5 = distal tail weakness, 1 = complete tail weakness, 1.5 = mild hind limb weakness, 2 = ascending hind limb weakness, 2.5 = severe hind limb weakness, 3 = hind limb paralysis, 3.5 = hind limb paralysis and moderate forelimb weakness, 4 = hind limb paralysis and severe forelimb weakness, 4.5 = tetraplegia and incontinence, to 5 = moribund/death.

Histopathology and immunohistochemistry

Fgfr1^ind−/− mice and control mice were anesthetized and transcardially perfused with 4% paraformaldehyde in the acute and chronic phases. Spinal cords were removed and embedded in paraffin blocks. A minimum of six spinal cord cross sections was examined per animal. Spinal cord sections were stained for inflammatory infiltrates (hematoxylin and eosin), myelin loss (Luxol fast blue/periodic acid‐Schiff) and axonal degeneration (Bielschowsky silver impregnation). The extent of myelin loss, encompassing primary demyelination and myelin loss caused by axonal damage, was calculated by relating the area of myelin loss to the total white matter area in LFB/PAS‐stained sections. Axonal densities were evaluated in Bielschowsky silver impregnated sections as described earlier 68. Counting was performed in the white matter in spinal cord sections comprising of cervical, thoracic and lumbar spinal cord at 1000× magnification. Axons were counted in minimum of 5 white matter lesions with microscopic fields using a 25‐point ocular grid eyepiece from Olympus (Hamburg, Germany). The number of points crossing axons was counted as a fraction of the total number of points of the ocular morphometric grid. Axonal density within the lesion was compared with the normal appearing white matter. Axonal density in lesions is given as the percentage of axons compared with the normal appearing white matter. Importantly, MOG_35–55‐induced EAE in B6 mice is characterized by substantial axonal pathology 16, 37. To differentiate primary demyelination from myelin loss caused by axonal damage and loss, a side‐by‐side assessment of myelin histochemistry (LFB) and axonal preservation (Bielschowsky silver impregnation) was performed for all lesions. For immunohistochemistry, spinal cord sections were deparaffinized, hydrated and antigen‐retrieval was performed by boiling the sections in citrate buffer (10 mM, pH 6). Endogenous peroxidase was blocked for 10 min with 3% hydrogen peroxide. Sections were then incubated with 10% FCS for 1 hour and with primary antibodies for macrophages/activated microglia (Mac 3, clone M3/84, 1:200, Pharmingen, San Diego, CA, USA), activated B cells (B220, clone RA3‐6B2, 1:200, Pharmingen, San Diego, CA, USA), T cells (CD3, clone CD3‐12, 1:150, Serotec, Oxford, UK), Olig2(+) and Nogo‐A(+) oligodendrocyte populations (Olig 2, 1:300, IBL, Gunma, Japan; Nogo‐A(+), 1:50, Santa Cruz Biotechnology, CA, USA) overnight. Afterward, biotinylated secondary antibodies [goat anti‐rat (Mac 3, B220) and goat anti‐rabbit (CD3, Olig2, Nogo‐A)] were applied and the signal was detected by incubation with an avidin‐biotin complex. Microscopic images were captured using a light microscope (Olympus BX51, Hamburg, Germany) and acquired by a digital camera (Olympus DP71, Olympus America Inc., Center Valley PA, USA). For analysis of myelin loss, total white matter areas and areas of myelin loss were determined by manually tracing the regions by using the Image J software (ImageJ 1.47d, National Institute of Health, USA). The percentage of myelin loss per mouse was obtained by dividing the total area of myelin loss by the total area of white matter. The density of axons was counted with an ocular 25‐point grid at 1000× magnification (oil immersion). Mac3(+), B220(+) and CD3(+) cells were counted at 400× magnification with an ocular morphometric grid. Cellular immunohistological quantifications were performed using an ocular morphometric grid. T cells, B cells and macrophages/activated microglia were done in a minimum of six inflammatory lesions per mouse from the entire spinal cord comprising cervical, thoracic and lumbar spinal cord. Analyses were not restricted to the subpial areas. The average numbers of positive cells were normalized to an area of 1 mm².

ERK/AKT and BDNF/TrkB receptor protein quantification

In the acute and chronic phase spinal cords were separately homogenized in lysis buffer with Tissue ruptor (Qiagen Instruments, Hombrechtikon, Switzerland). The amount of protein was quantified (Pierce^® BCA Protein Assay Kit, Thermo Scientific, IL, USA) and normalized. Equal amounts of protein (60 µg) were loaded and separated by 10% SDS‐PAGE, transferred (Trans Blot, Semi dry Transfer cell, BioRad) to a nitrocellulose membrane (GE Healthcare, Amersham^TM Hybond ECL, Buckinghamshire, UK) and blocked with 5% BSA for 1 hour. The membranes were incubated overnight at 4ºC with the primary antibodies against pERK and pAKT (1:1000; Cell Signaling, MA, USA), BDNF and TrkB receptor (1:500; Santa Cruz Biotechnology, CA, USA), Membranes were then incubated for 1 hour with goat anti‐rabbit secondary antibody (1:1000, Santa Cruz Biotechnology, CA, USA) and developed using SuperSignal West Pico Chemiluminescent Substrate (Thermo, Pierce Biotechnology, Rockford, IL, USA) using a Fusion Fx7 chemiluminescence system (Peqlab, Erlangen, Germany). GAPDH (Santa Cruz Biotechnology, CA, USA) was used as a loading control; proteins were analyzed by use of the Fusion BioID software (Peqlab, Erlangen, Germany).

Reverse transcription‐PCR

In the acute and chronic phase spinal cords were removed, snap frozen in liquid nitrogen and homogenized with Tissue ruptor (Qiagen Instruments, Hombrechtikon, Switzerland). Total RNA was isolated (NucleoSpin RNA II, Macherey‐Nagel, Düren, Germany). cDNA was synthesized from 1 µg of total RNA using the QuantiTect^® Reverse Transcription Kit (Qiagen GmbH, Hilden, Germany). Quantitative PCR was performed for FGFR1, proinflammatory cytokines, CX3CL1, CX3CR1, FGF2, LINGO‐1, SEMA3A and TGFβ using the iTaq^TM Universal SYBR^® Green qPCR Master Mix (Bio‐Rad, CA, USA) by 40 cycles at an annealing temperature of 60ºC using the StepOne^® Real‐Time PCR system (Applied Biosystems, Darmstadt, Germany). In our mice model exon 4 of the Fgfr1 gene is flanked with lox sites. Upon tamoxifen injection Cre mediated deletion of the Frs2/3‐binding site on Fgfr1 occurs in oligodendrocytes and the primers will pair on exon 4. Quantification of target genes was performed using the following primers: FGFR1, forward 5′‐CAGATGCACTCCCATCCTCG‐3′ and reverse 5′‐GGGAGCTACAGGGTTTGGTT‐3′; TNFα, forward 5′‐CGGTCCCCAAAGGGATGAGAAGT‐3′ and reverse 5′‐ACGACGTGGGCTACAGGCTT‐3′; IL1β forward 5′‐TACCTGTGGCCTTGGGCCTCAA‐3′ and reverse 5′‐GCTTGGGATCCACACTCTCCAGCT‐3′; IL6 forward 5′‐CTCTGCAAGAGACTTCCA‐3′ and reverse 5′‐AGTCTCCTCTCCGGACTT‐3′; FGF2 forward 5′‐GGC TGC TGG CTT CTA AGT GT‐3′ and reverse 5′‐ACT GGA GTA TTT CCG TGA CCG‐3′; LINGO‐1 forward 5′‐TCATCAGGTGAGCGAGAGGA‐3′ and reverse 5′‐CAGTACCAGCAGGAGGATGG‐3′; TGFβ forward 5′‐CTCCTGCTGCTTTCTCCCTC‐3′ and reverse 5′‐GTGGGGTCTCCCAAGGAAAG‐3′; SEMA3A forward 5′‐GGATGGGTCCTCATGCTCAC‐3′ and reverse 5′‐TGGTGCTGCAAGTCAGAGCAG‐3′; GAPDH was used as the internal control gene, and the comparative ΔΔCT method was used to evaluate gene expression.

Statistics

All analyses were performed blinded to the genotype. All mice and samples were included into the analyses. Positively labeled cells in lesions of randomly selected spinal cord sections were counted. Each category analyzed included five or more spinal cord sections per mouse and three mice per condition. Specific numbers of animals per sample are noted in the figure legends. Statistical analysis for differences between the genotypes in the clinical course and histological data was performed using the Mann–Whitney U test. A T‐test was used to analyze reverse transcription‐PCR and protein data. Statistical significance was set at P ≤ 0.05. Values are expressed as mean ± standard error of mean. * indicates P ≤ 0.05, ** indicates P < 0.01, *** indicates P < 0.001. Statistical analysis was performed using Systat 12 software (Systat, San Jose, CA, USA). Graphs were prepared using the Sigmaplot 10 software (Systat, San Jose, CA, USA).

Results

Fgfr1^ind−/− mice show a milder EAE disease course

Mice with oligodendroglial Fgfr1 deletion (Fgfr1^ind−/−) were analyzed for their susceptibility to MOG_35‐55 peptide EAE induction (Figure 1B). EAE was induced in 22 eight‐ to twelve‐week‐old female Fgfr1^ind−/− mice and compared with 22 age‐ and sex‐matched littermate controls. The onset of clinical symptoms was on day 9.7 ± 0.3 in controls and on day 10.9 ± 0.2 in Fgfr1^ind−/− mice (P < 0.001). From day 33 p.i. on (with exception of days 47 and 48) Fgfr1^ind−/− mice showed a milder disease course than controls (P ≤ 0.05). In this disease phase, Fgfr1^ind−/− mice presented with a mild paraparesis whereas controls still suffered from a severe paraparesis. There was a marked improvement in function in Fgfr1^ind−/− mice (P = 0.01, Figure 1C). Disease incidence in both genotypes was 100%. No effect of gene deletion on mortality was observed. There were no differences in body weight between Fgfr1^ind−/− and control mice. Fgfr1 mRNA expression was less in Fgfr1^ind−/− mice in the acute (P = 0.013, Figure 1D) and chronic (P = 0.017, Figure 1E) phase of EAE.

Reduced myelin loss and increased axonal density in Fgfr1^ind−/− mice

The differences in the clinical course were most evident in the chronic phase of EAE. As the next step, Fgfr1^ind−/− mice were assessed for their susceptibility to myelin loss and axonal density in spinal cord white matter. Myelin loss was evaluated by Luxol fast blue staining and axonal density by Bielschowsky staining. In the acute phase of EAE no difference in the extent of myelin loss (Figure 2D–F) was observed between Fgfr1^ind−/− mice and controls. At this time point, the axonal density was higher in Fgfr1^ind−/− mice than in controls (P = 0.05) (Figure 2G–I). In the chronic phase of EAE (Figure 2M–R) Fgfr1^ind−/− mice showed less myelin loss (P = 0.046) and a higher axonal density (P = 0.05).

Figure 2.

Inflammation, myelin loss and axonal density in spinal cord sections in acute (A–I) and chronic (J–R) EAE in controls and Fgfr1^ind−/− mice. Representative images of spinal cord sections are shown. The inflammatory index (A–C) and myelin loss (D–F) were not different between Fgfr1^ind−/− mice and controls in the acute phase. Axonal density was increased in Fgfr1^ind−/− mice (G–I). In the chronic phase Fgfr1^ind−/− mice showed a reduced inflammatory index compared with control mice (J–L). The degree of myelin loss was less in Fgfr1^ind−/− mice (M–O). The axonal density was increased in Fgfr1^ind−/− mice (P–R). n = 3. Bar: 200 µm, 20 µm (insert).

Inflammation in Fgfr1^ind−/− mice is characterized by reduced numbers of lymphocytes and macrophages/microglia

To study the underlying cellular mechanisms that drive myelin loss and axonal damage, the degree of inflammation was analyzed in spinal cord white matter lesions on H&E stained sections of Fgfr1^ind−/− mice and littermate controls. In the acute phase the inflammatory index was not significantly different between Fgfr1^ind−/− mice and controls (Figure 2A–C), whereas in the chronic phase the inflammatory index was reduced in Fgfr1^ind−/− mice (P = 0.05; Figure 2J–L). With regard to cellular infiltration the number of CD3(+) T cells was not affected by FGFR1 deletion in the acute phase of EAE (P = 0.275, Figure 3G–I). However, the number of Mac3(+) cells (P = 0.05) and B220(+) B cells (P = 0.05) was reduced in Fgfr1^ind−/− mice compared with controls (Figure 3A–F). In the chronic phase a reduction of Mac3(+) cells (P = 0.05), B220(+) B cells (P = 0.05) and CD3(+) T cells (P = 0.05) was observed in Fgfr1 ^ind‐/‐ mice compared with controls (Figure 3J–R).

Figure 3.

Mac3(+) cells (macrophages/activated microglia), B220(+) B cells and CD3(+) T cells in white matter lesions of spinal cord sections in the acute (A–I) and chronic phases (J–R) of EAE in controls and Fgfr1^ind−/− mice. Representative images of spinal cord sections are presented. The number of Mac3(+) cells per mm² was reduced in Fgfr1^ind−/− mice in both the acute (A–C) and chronic (J–L) phase of EAE. B220(+) B cell numbers per mm² were significantly reduced in Fgfr1^ind−/− mice in both the acute (D–F) and chronic (M–O) phase of EAE. The number of CD3(+) T cells per mm² was not different between Fgfr1^ind−/− mice and controls in the acute phase (G–I), whereas in the chronic phase the number of CD3 (+) T cells was reduced significantly (P–R). n = 3. Bar: 200 µm, 20 µm (insert).

Deletion of Fgfr1 does not affect numbers of oligodendrocyte lineage cells

In the mouse CNS Olig2 36 and Nogo‐A 38 are expressed by oligodendrocyte lineage cells. To answer the question whether deletion of oligodendroglial FGFR1 affects oligodendrocyte lineage cells in EAE the number of Olig2(+) and NogoA(+) oligodendrocytes was studied in spinal cord white matter lesions in the acute and chronic phase. There was no effect of FGFR1 deletion on the number of Olig2(+) oligodendrocytes at both time points (Figure 4A–C for acute phase, Figure 4G–I for chronic phase). A trend toward an increased number of NogoA(+) oligodendrocytes in Fgfr1^ind−/− mice was found both in the acute phase (P = 0.124, Figure 4D–F) and the chronic phase (P = 0.118, Figure 4J–L).

Figure 4.

Number of Olig2(+) and NogoA(+) oligodendrocytes in spinal cord sections in controls and Fgfr1^ind−/− mice. Representative images of spinal cord sections in the acute phase (A, B, D, E) and the chronic phase (G, H, J, K) are shown. There were no differences in the number of Olig2(+) oligodendrocytes in the acute (C) or chronic phase (I). A trend toward an increased number of NogoA(+) oligodendrocytes in Fgfr1^ind−/− mice was seen at the acute phase (F) and the chronic phase (L). n = 3. Bar: 200 µm.

Increased phosphorylation of ERK1/2 and AKT in Fgfr1^ind−/− mice

To study the underlying mechanisms associated with less myelin loss and an increase of axonal density in the Fgfr1^ind−/− mice we examined ERK1/2 and AKT phosphorylation in the spinal cord of the acute and chronic phases of EAE. No differences in ERK1/2 and AKT phosphorylation were seen in the acute phase of EAE (Figure 5A,B). In the chronic phase of EAE, however, ERK1/2 (P = 0.016) and AKT (P = 0.024) phosphorylation was increased in Fgfr1^ind−/− mice compared with controls (Figure 5E,F).

Figure 5.

Protein quantification of ERK1/2 and AKT phosphorylation, and brain‐derived neurotrophic factor (BDNF)/TrkB receptor expression in spinal cord lysates of controls and Fgfr1^ind−/− mice. Representative Western blots are shown for the acute and chronic phases of EAE. In the acute phase no effect of Fgfr1 deletion on ERK1/2 and AKT phosphorylation (A,B) and BDNF/TrkB receptor protein expression (C,D) was seen. In the chronic phase ERK1/2 and AKT phosphorylation (E,F), and BDNF and TrkB receptor protein expression (G,H) were upregulated in Fgfr1^ind−/− mice. n = 3.

Deletion of FGFR1 causes changes in BDNF and TrkB receptor expression

To investigate whether less myelin loss and increased axonal density in the spinal cord was accompanied by changes in BDNF synthesis, the effect of oligodendroglial ablation of FGFR1 on the neurotrophin BDNF and its receptor TrkB was studied on protein level in the acute and chronic phase. In the acute phase BDNF and TrkB receptor protein expression was similar between Fgfr1^ind−/− mice and controls (Figure 5C,D). In the chronic phase BDNF (P = 0.010) and TrkB receptor protein (P = 0.003) was increased in Fgfr1^ind−/− mice when compared with controls (Figure 5G,H).

Altered pattern of cytokine and chemokine expression in Fgfr1^ind−/− mice

To characterize the underlying molecular mechanisms of cellular infiltration, proinflammatory cytokine gene expression was analyzed in the spinal cord in the acute and chronic phase (Figure 6). In the acute phase Fgfr1^ind−/− mice showed decreased expression of TNFα (P < 0.001), IL1β (P = 0.019) and IL6 (P < 0.001) compared with controls (Figure 6A–C). In the chronic phase no differences in proinflammatory cytokine expression for TNFα, IL1β, IL6 were observed (Figure 6F–H). Expression of inducible NOS (iNOS) and IL12 was not changed in Fgfr1^ind−/− mice in both phases of EAE. In addition, to define further molecular mechanisms of cellular infiltration, the expression of the chemokine CX3CL1 and its receptor CX3CR1 in the spinal cord were studied in the acute and chronic phase. In the acute phase no effect of FGFR1 deletion on gene expression of CX3CL1 or CXC3R1 was seen (Figure 6D,E). In the chronic phase CX3CL1 (P = 0.020) and CX3CR1 (P = 0.028) expression was decreased in Fgfr1^ind−/− mice compared with controls (Figure 6I,J).

Figure 6.

Gene expression of proinflammatory cytokine and chemokine/receptor levels in the spinal cord of controls and Fgfr1^ind−/− mice. Findings in the acute phase (A–E) and the chronic phase (F–J) are shown. In the acute phase decreased mRNA expression of TNFα, IL1β and IL6 was observed in Fgfr1^ind−/− mice compared with controls (A–C). In the chronic phase TNFα, IL1β, IL6 were not different between Fgfr1^ind−/− mice and controls (F–H). There was no difference in the expression of the chemokine CX3CL1 and its receptor CX3CR1 in Fgfr1^ind−/− mice and controls in the acute phase (D,E) whereas in the chronic phase CX3CL1 and CX3CR1 were reduced in Fgfr1^ind−/− mice (I,J). n = 6.

Ablation of FGFR1 affects remyelination inhibitor expression

To study whether differences in myelin damage in the spinal cord were accompanied by changes in inhibition of remyelination, the effect of FGFR1 on remyelination inhibitors such as FGF2, LINGO‐1, SEMA3A and TGFβ were studied on mRNA level (Figure 7). In the acute phase no differences for FGF2, LINGO‐1, SEMA3A and TGFβ, expression were seen between Fgfr1^ind−/− mice and controls (A‐D). In the chronic phase gene expression of LINGO‐1 was decreased in Fgfr1^ind−/− mice (P = 0.018, Figure 7F), a trend toward reduced expression of FGF2 was seen in Fgfr1^ind−/− mice (P = 0.071, Figure 7E). No differences were found for SEMA3A and TGFβ expression (Figure 7G,H).

Figure 7.

Gene expression of the remyelination inhibitors FGF2, LINGO‐1, SEMA3A and TGFβ in the spinal cord of controls and Fgfr1^ind−/− mice. Findings in the acute phase (A–D) and the chronic phase (E–H) are shown. In the acute phase FGF2, LINGO‐1, SEMA3A and TGFβ expression was not different between Fgfr1^ind−/− mice and controls. In the chronic phase LINGO‐1 was reduced in Fgfr1^ind−/− mice, furthermore a trend toward a reduced expression of FGF2 was seen. n = 6.

Discussion

Recent studies suggest that FGFR1 plays a role in MS and a model of toxic demyelination 69. FGFR1 is upregulated on oligodendrocyte precursor cells within active and around chronic MS brain lesions 11. In EAE it is expressed on effector cells such as macrophages and activated microglia 43. In the present study the in vivo function of FGFR1 was characterized using an oligodendrocyte‐specific genetic approach, which allowed defining the function of this receptor in a cell‐specific and inducible fashion. Impaired signaling via oligodendroglial FGFR1 resulted in a milder disease course. The underlying neuropathology included less myelin loss and axon degeneration, and reduced cellular infiltration. Furthermore, impaired signaling via oligodendroglial FGFR1 was associated with phosphorylation of the ERK/AKT pathway, enhanced expression of the neurotrophin BDNF and decreased expression of LINGO‐1, and downregulation of proinflammatory cytokines and chemokines.

The key clinical finding of our study is that Fgfr1^ind−/− mice showed a milder disease course in the chronic phase of EAE. Furthermore, Fgfr1^ind−/− mice presented with a delay in onset of symptoms. These clinical findings are in contrast to those obtained after systemic deletion of its key ligand FGF2. FGF2^−/− mice are more severely affected in the chronic phase of EAE 57, a finding, which is supported by data from a FGF2 gene therapy study 58. There was no effect of systemic FGF2 deletion in the acute phase of EAE. Intrathecal injection of a HSV multigene vector engineered with the human FGF2 gene at day 20 after EAE induction also resulted in a milder disease course 5 days later to the end of the study 58. Using cuprizone toxic demyelination in FGF2 null k.o. mice and in oligodendroglial FGFR1 conditional knockout mice, evaluation revealed improved motor function, less demyelination and axon degeneration as in our study 48. In human brain tissue from patients with MS, demyelination in active/chronic lesion areas was associated with an increase of FGF2 and FGFR1 11 suggesting a role of the FGF/FGFR pathway in neurorepair and neuroprotection. Differences in findings between studies may be explained by (a) alternative binding of other ligands to FGFR1 or activation of other FGFRs by FGF2, (b) up‐ or downregulation of members of the FGF family during demyelination, (c) different k.o. approaches including systemic deletion of a ligand and conditional k.o. of a receptor in specific cells, (d) differences in proteins used to induce EAE and (e) various demyelinating disease models.

In vitro data suggest that the deleterious effects of FGFR1 activation by FGF2 on differentiated oligodendrocytes are associated with a downregulation of myelin proteins 17. Myelin sheaths play an important role in protecting axons from degeneration 30. Our findings suggest that impaired signaling via oligodendroglial FGFR1 results in less myelin loss and in enhanced axonal density. Similar effects of oligodendroglial FGFR1 were also observed in toxic demyelination, where remyelination and axonal integrity is enhanced 69. The degree of myelin loss in Fgfr1^ind−/− mice in our study is in agreement with findings in FGF2 null mice in toxic demyelination, in which enhanced remyelination 2 and reduced axonopathy were found 64. In contrast, in a PLP‐EAE study in FGF2 null mice increased myelin degeneration and axonal damage was seen in the chronic disease phase 57. Oligodendroglial FGFR1 did not affect the number of oligodendrocyte lineage cells in our model. In agreement, in the model of toxic demyelination reduced oligodendroglial FGFR1 expression did not have an effect on OPC numbers 69. We observed increased phosphorylation of ERK1/2 and AKT in the chronic phase of Fgfr1^ind−/− mice. Both ERK1/2 and AKT3 are predominantly expressed in the brain and play a role in brain development and neurodegeneration. In the hypomyelinated CNS of Fgfr1/Fgfr2 double mutant mice ERK1/2‐MAPK activity was reduced, these mice showed reduced myelin protein gene expression and myelin thickness 19. In contrast to these findings, we observed enhanced ERK1/2 activation in the chronic phase of EAE. The difference between the studies may be caused by the additional k.o. of FGFR2 expressed in mature oligodendrocytes. It is known that active ERK1/2 and AKT phosphorylation increases myelination in the CNS 15, 53. ERK1/2 activation is capable of reinitiating myelin growth by preexisting oligodendrocytes, even after active myelination is terminated 31. Growth factors associated with an activation of the ERK1/2 and AKT pathway include FGF and BDNF 66. In vitro inactivation of ERK1/2 in oligodendrocytes results in reduced extension of oligodendrocyte processes 32. ERK1/2 k.o. mice have normal numbers of oligodendrocytes, delayed myelination and a failure to upregulate key myelin proteins 15, 20, 21. In vivo data suggest that activation of AKT causes enhanced myelination 15 and increases myelin thickness 24, 28. Inhibition of the downstream target of AKT mTOR, results in reduced myelin thickness 53. In agreement with these findings on the function of ERK1/2 and AKT, upregulation of ERK1/2 and AKT in our study was associated with reduced myelin degeneration without an alteration in the number of oligodendrocytes.

In the chronic phase of EAE we found increased BDNF and TrkB expression in Fgfr1^ind−/− mice presenting with less axonal degeneration. In vitro FGF2 has been suggested to promote the role of BDNF through upregulation of the respective receptor TrkB expression 7. BDNF is a neurotrophic factor involved in neuronal survival and differentiation as well as axonal growth 10. Interactions of FGFR1 with other growth factor receptors are not well studied. FGF upregulate BDNF and TrkB in retinal ganglion cells by activating the ERK signaling pathway 5, 61, the TNFα/TNFR pathway 13, expression of Nrf2 67, VEGF/VEGFR2 51, 60 and the PDGF receptor 52. In MS immune cells are a major source of BDNF 35. More BDNF(+) cells are found in active demyelinating lesions compared with inactive lesions 62. Furthermore, neurons around active lesions and reactive astrocytes within lesions express high levels of its high affinity receptor gp145trkB 62. In EAE 39, 42 and other lesion models 45 BDNF is expressed by activated immune cells and it is also associated with axonal protection 42. Activation of TrkB by an agonist reduces the severity of EAE, reduces myelin loss, protects against axonal degeneration and attenuates inflammation 44. Activation of TrkB induces the Akt/STAT3 pathway associated with neuronal survival and myelination 44. In agreement, we observed an increase in BDNF/TrKB expression and AKT phosphorylation in Fgfr1^ind−/− mice.

In both acute and chronic phases of EAE, impaired signaling via FGFR1 was associated with a decreased number of macrophages/activated microglia and B220(+) B cells in the spinal cord. Activated macrophages and microglia release a number of soluble factors such as nitric oxide (NO), excitotoxins and proteases, which cause structural damage in the CNS 26. In Fgfr1^ind−/− mice a significant reduction of CD3(+) T cells was observed in the chronic phase of EAE, there was no difference in CD3(+) T cells number in the acute phase of EAE as described for AKT3^−/− mice, where an increase of T cells was detected 65. We found increased AKT phosphorylation in the chronic phase of EAE suggesting that oligodendroglial FGFR1 mediates inflammation through the AKT pathway. In contrast to our findings, in FGF2^−/− mice the number of macrophages and T cells was increased in acute EAE 57. As discussed above, differences between studies may be explained by alternative binding of other ligands to FGFR1, up‐ or downregulation of members of the FGF family or different k.o. approaches. Data on oligodendrocyte‐specific FGFR1 ablation and immune cell infiltration into the CNS are not available from the toxic demyelination model 52. In brain tissue of patients with MS, FGF2 has been detected in inflammatory cells such as microglia and infiltrating T cells and B cells in chronic active lesions 11. There are only few data on the interaction of FGF/FGFR and inflammation. In patients with rheumatoid arthritis FGF1 activated FGFR1 expressing T cells are increased in the peripheral blood 9 suggesting that immune‐growth factor networks contibute to multiple interactions that maintain chronic inflammation.

In the CNS T cells are reactivated resulting in the release of proinflammatory cytokines and chemokines 14. In the acute phase of EAE, impaired signaling via FGFR1 was associated with decreased secretion of proinflammatory cytokines such as TNFα, IL1β and IL6. There is evidence for a deleterious role of the CX3CL1/CX3CR1 pathway in MS. CX3CR1 is present on astrocytes and oligodendrocytes in active and silent MS lesions 54. Increased expression of serum CX3CL1 is observed in patients with MS 34. Data from a genetic study suggest that the CX3CR1 I₂₄₉T₂₈₀ haplotype may protect from switch to secondary progressive MS 63. The role of the CX3CL1/CX3CR1 pathway appears to be controversial. We observed that impaired signaling via oligodendroglial FGFR1 is associated with downregulation of the CX3CL1/CX3CR1 pathway and decreased immune cell infiltration in the chronic phase of EAE. This deleterious role of the pathway is supported by a study where treatment with an anti‐CX3CL1 neutralizing antibody resulted in reduced clinical disease severity and less lymphocyte infiltration in EAE 49 and activation of FGFR1 in mammary tumor cells promotes the recruitment of macrophages in a CX3CL1/R1 dependent manner in vitro and in vivo 6, 55. In contrast to these findings, CX3CR1‐deficient mice show a more severe disease course and increased IFN‐γ expression 23. Further, since we observed this positive effect during chronic phase, it is possible that the difference in the microenvironment of the acute and chronic lesions may play a role in governing the outcome of oligodendroglial Fgfr1 loss, such as astrocyte derived FGF2 which persists in the chronic phase.

Apart from findings on FGF2, other FGF have been shown to play a role in demyelinating lesions of patients with MS. In vivo, FGF1 was localized on astrocytes, neurons, oligodendrocytes, as well as microglia and infiltrating T cells and B cells. FGFR1 and FGFR2 were significantly higher expressed in remyelinated lesions compared with control white matter 50. In vitro, FGF1 induced the leukemia inhibitory factor and chemokine CXCL8 in astrocytes, which caused recruitment of oligodendrocytes and promotion of remyelination 50. Furthermore, in vitro downregulation of myelin basic protein, myelin oligodendrocyte glycoprotein and myelin associated glycoprotein after exposure to FGF1 was observed. FGF9 was found in active demyelinating lesions in patients with MS, and was made responsible for tissue damage, lesion formation through activating proinflammatory cytokine and chemokine production in the CNS. In vitro FGF9 inhibited myelination and remyelination 41.

The nogo receptor‐interacting protein (LINGO‐1) is a key inhibitor of oligodendrocyte precursor cell differentiation and myelination in vitro 47 and in vivo 46. Deletion of LINGO‐1 causes a less severe EAE disease course, improved axonal integrity and promoted axon remyelination 46. An anti‐LINGO‐1 antagonist applied before and after onset of symptoms results in less disease activity in acute EAE 46. LINGO‐1 can interact with the TrkB receptor to inhibit phosphorylation induced by BDNF thereby suppressing oligodendrocyte differentiation 18. In our study impaired signaling via oligodendroglial FGFR1 was associated with downregulated LINGO‐1 expression in the chronic phase of EAE suggesting that this inhibitor is controlled by oligodendroglial FGFR1 as well.

The finding that signaling via oligodendroglial FGFR1 modulates clinical disease activity and inflammation gives the opportunity for a targeted molecular therapy. Selective and multikinase FGFR inhibitors are currently beeing tested in patients with genetically selected tumors with promising efficacy in some of these trials 12. Tyrosine kinase inhibitors are active against kinases such as VEGFR or PDGFR. Luzitanib, a VEGFR/FGFR inhibitor has been applied successfully in FGFR1‐amplified breast tumors 12. In addition, monoclonal antibodies such as FP‐1039 binding to the extracellular domain of FGFRs are used in clinical testings 12. Signaling via oligodendroglial FGFR1 also modulates LINGO‐1 expression suggesting that LINGO‐1 inhibition could be a promising approach to decrease the effects of the receptor. A LINGO‐1 antagonist has been used succesfully to promote remyelination in EAE 46 and inhibition of LINGO‐1 is has been tested in a Phase I clinical MS trial (NCT01244139).

In summary, impaired signaling via oligodendroglial FGFR1 has numerous beneficial effects on MOG_35–55‐induced EAE. Ablation of oligodendroglial FGFR1 resulted in a less severe clinical disease course, less myelin loss and axonal degeneration, and reduced cellular inflammation. These changes are accompanied by an increase of ERK/AKT phosphorylation as well as increased expression of BDNF and TrkB. Furthermore, there is an association between oligodendroglial FGFR1 and the expression of LINGO‐1. Considering that FGFR inhibitors are used in cancer trials, the oligodendroglial FGFR1 pathway may provide a new target for therapy in multiple sclerosis.