Effects of FGFR Tyrosine Kinase Inhibition in OLN-93 Oligodendrocytes

Fibroblast growth factor (FGF) signaling is involved in the pathogenesis of multiple sclerosis (MS). Data from neuropathology studies suggest that FGF signaling contributes to the failure of remyelination in MS. In MOG35–55-induced EAE, oligodendrocyte-specific deletion of FGFR1 and FGFR2 resulted in a less severe disease course, reduced inflammation, myelin and axon degeneration and changed FGF/FGFR and BDNF/TrkB signaling. Since signaling cascades in oligodendrocytes could not be investigated in the EAE studies, we here aimed to characterize FGFR-dependent oligodendrocyte-specific signaling in vitro. FGFR inhibition was achieved by application of the multi-kinase-inhibitor dovitinib and the FGFR1/2/3-inhibitor AZD4547. Both substances are potent inhibitors of FGF signaling; they are effective in experimental tumor models and patients with malignancies. Effects of FGFR inhibition in oligodendrocytes were studied by immunofluorescence microscopy, protein and gene analyses. Application of the tyrosine kinase inhibitors reduced FGFR1, phosphorylated ERK and Akt expression, and it enhanced BDNF and TrkB expression. Furthermore, the myelin proteins CNPase and PLP were upregulated by FGFR inhibition. In summary, inhibition of FGFR signaling in oligodendrocytes can be achieved by application of tyrosine kinase inhibitors. Decreased phosphorylation of ERK and Akt is associated with an upregulation of BDNF/TrkB signaling, which may be responsible for the increased production of myelin proteins. Furthermore, these data suggest that application of FGFR inhibitors may have the potential to promote remyelination in the CNS.


Introduction
Demyelination of the central nervous system (CNS) causes degeneration of axons [1] associated with permanent disability [2][3][4]. The physiological response to demyelination is robust and leads to effective remyelination [5]. However, remyelination often fails in inflammatory demyelinating diseases such as MS [2,6,7]. The pivotal impediment for remyelination in MS is most likely not the recruitment of oligodendrocyte progenitor cells (OPCs) to lesions, but OPCs' inability to differentiate into mature myelin-producing oligodendrocytes (OLs) [8,9]. Growth factors have been associated with the differentiation of oligodendrocytes in demyelinating pathologies. These include brain derived neurotrophic factor (BDNF), nerve growth factor (NGF), insulin-like growth factor (IGF), platelet derived growth factor (PDGF), ciliary neurotrophic factor (CNTF) and FGF [10,11]. Molecular

Experimental Treatment of Cells
Four groups were deployed and treated with either dovitinib (Selleck Chemicals, Houston, TX, USA), AZD4547 (Selleck Chemicals, Houston, TX, USA), FGF2 (R&D Systems, Minneapolis, MN, USA) or utilized without treatment as a control. AZD4547 and dovitinib were dissolved in dimethyl sulfoxide (DMSO) (Carl Roth GmbH, Karlsruhe, Germany) and stored at −20 • C until use. Compounds were applied at a concentration of 1 µM (AZD4547, dovitinib) or 25 ng/mL (FGF2). DMSO controls have been included in all experiments. The inhibitors were applied at an effective concentration as previously established [18,26]. All treatments were added to DMEM growth medium and setups were incubated for 24 h, after which appropriate amounts of cells were taken for further analysis.

Proliferation Measurement
To assess cell proliferation under different treatment conditions several assays were utilized. Cells were grown in DMEM culture medium supplemented with 10% FBS and the respective compounds at humidified conditions. After 24 h of incubation manual counting was performed using a Neubauer improved chamber (Karl Hecht "Assistant", Altnau, TG, Switzerland) and Trypan blue dye (Carl Roth, Karlsruhe, Germany) as described by others [27]. For photometric evaluation of proliferation, cells were seeded at 2 × 10 4 cells/well in a 96 well plate (Greiner Bio-One, Frickenhausen, Germany) before treatment. After 24 h of treatment, a WST-1 assay (Roche Applied Science, Mannheim, Germany) was carried out as described by manufacturer. Absorption was measured by ELISA-Reader (Multiscan EX, Thermo Fisher Scientific, Langenselbold, Germany) at a wavelength of 405 nm using a reference wavelength of 492 nm.

Cytotoxicity Assessment
Possible cytotoxic effects of applied compounds were studied by measuring lactate dehydrogenase (LDH) in the supernatant of incubated cells (96 well plate, 0.5 × 10 4 cells/well) using a prefabricated kit (Cytotoxicity Detection Kit (LDH), Roche Diagnostics, Mannheim, Germany). Cell plates were centrifuged for 10 min at 250× g and supernatant was used to measure absorbance at 492 nm (reference wavelength: 620 nm) to calculate cytotoxicity as described by manufacturer.

Western Blotting and Antibodies
For protein extraction, cells were incubated in T-75 flasks for 24 h as described above. Media was then removed, and cells were washed with PBS. After centrifugation, buffer was removed, cells were lysed and protein concentration was quantified in a nanophotometer (Implen GmbH, Munich, Germany) according to the manufacturer's instructions. After normalization, proteins were separated by 10% SDS-PAGE gel electrophoresis. Separated protein was transblotted to a nitrocellulose membrane (GE Healthcare, Amersham™ Protran™, Buckinghamshire, UK) and after blocking with 5% BSA (Merck, Darmstadt, Germany) primary antibodies were applied in an appropriate dilution in tris buffered saline with tween 20 (TBST) with 5% BSA and incubated overnight. The antibodies used in this study can be found in Supplementary (Supplementary Table S1). SuperSignal™ West Pico chemiluminescent substrate (Thermo Fisher Scientific, Waltham, MA, USA) was used to visualize the protein bands imaged in an ECL Chemocam Imager (Intas-Science, Göttingen, Germany) and quantified using ImageJ software (NIH, Stapleton, NY, USA).

RNA Analysis
For RNA extraction, cells were incubated in T-75 flasks for 24 h as described above. Media was then removed, and cells were washed with PBS. After centrifugation, buffer was removed and the cellular RNA extracted with peqGOLD Total RNA Kit as described by the manufacturer (VWR, PEQLAB, Darmstadt, Germany). Total amount of RNA was quantified in a nanophotometer and 1 µg of RNA was used for subsequent cDNA synthesis with QuantiTect Reverse Transcription Kit (Thermo Fisher Scientific, QIAGEN, Hilden, Germany). All samples were diluted in 200 µL RNAse free H 2 O from which 1 µL was used for comparative quantitative real time PCR (qRT-PCR) MicroAmp ®® Fast Reaction Tubes (Applied Biosystems, Darmstadt, Germany). A total 7 µL of H 2 O, 1 µL of forward and reverse primer (FP and RP, respectively) and 10 µL of SYBR ®® Green qPCR Supermix (Bio-Rad, Hercules, CA, USA) was added. Primers for FGFR1, BDNF, TrkB, PLP, CNPase, SEMA3A, TGFβ and GAPDH as internal control were purchased from Eurofins Genomics (Ebersberg, Germany). The primers used in this study can be found in Supplementary  Table S2. PCR was performed in StepOne ®® Real-Time PCR system (Applied Biosystems, Darmstadt, Germany) using StepOne™ Software v2.3 (Applied Biosystems, Darmstadt, Germany) in 40 repetitive cycles: Denaturation at 95 • C for 15 s followed by annealing at 60 • C for 1 min. The three biological replicates performed contained negative controls without template cDNA and were executed in technical duplicates. Quantification of the results was carried out using the 2 −∆∆Ct method [28].

Immunofluorescence
To address the morphological alterations induced by FGFR inhibition, immunofluorescence was performed. OLN-93 cells were seeded on sterilized coverslips (R. Langenbrinck, Emmendingen, Germany) in 24-well-plates (GreinerBioOne, Frickenhausen, Germany) at a final density of 6 × 10 4 cells/well and treated according to their assigned group as described in section experimental treatment of cells. Cells were washed with PBS, fixed using 4% paraformaldehyde and incubated in PBS containing 0.5% Triton X-100 (Sigma-Aldrich, Steinheim, Germany) for permeabilization. After blocking in PBS with 5% BSA, cells were incubated overnight at 4 • C with primary antibodies, washed and stained with respective fluorescent secondary antibodies (as listed in Supplementary Table S1). Negative controls omitting the secondary antibody were included for each treatment. Cells were counter-stained with DAPI (Carl Roth, Karlsruhe, Germany) and subsequently mounted onto microscope slides using Fluorescence Mounting Medium (DAKO, Agilent, CA, USA). Images were taken with Axioplan 2 Fluorescence Microscope (Carl Zeiss, Jena, Germany) at a magnification of 50× and 400× using ZEN software for microscope (Zen 2.3, ZEISS, Jena, Germany) for image acquisition and processing. For counting of positive cells, 10 areas on every slide were randomly selected and a total of 30 cells was counted in each spot at 50× magnification regarding the intensity and location of signal. The total number of cells was determined by the DAPI counterstain.

Statistical Analysis
All experiments were conducted at least in 3-6 replicates. Data were analyzed using Sigmaplot 14 (Systat, San Jose, CA, USA). Depending on distribution of data, differences between groups were analyzed either by using ANOVA or Kruskal-Wallis ANOVA followed by a Tukey post hoc test or by Dunn's Method. p < 0.05 was considered statistically significant (*/ #= p < 0.05, **/ ##= p < 0.005, ***/ ###= p < 0.001). Values are shown as mean ± standard error of the mean (SEM).

Inhibition of FGFR Is Cytotoxic to OLN-93 Cells
As these compounds inhibit growth factor receptors involved in cell survival, we investigated possible cytotoxic effects on OLN-93 cells by means of a LDH assay. Both substances were deployed at a concentration of 1 µM throughout the experiment. LDH levels in the supernatant of the treated cells revealed a significant cytotoxic effect of AZD4547 (p < 0.001), while dovitinib did not exert cytotoxicity in comparison to control ( Figure 1B).

FGFR1 Expression Is Reduced by FGFR Inhibition
In the present study we investigated specific effects of pharmacological FGFR inhibition in oligodendrocytes by immunostaining and comparative qRT-PCR, after previously characterizing the effects of genetic FGFR silencing in OLs in the EAE model [16,17]. Application of AZD4547 and dovitinib resulted in a significant reduction of FGFR1 + cells compared to control (p < 0.001) ( Figure 1C,D). FGF2 did not affect FGFR1 expression on cellular or mRNA levels ( Figure 1C-E). While dovitinib treatment decreased the overall number of FGFR1 + cells, it did not affect FGFR1 mRNA expression. Cells treated with AZD4547 expressed less FGFR1 mRNA (p = 0.009) ( Figure 1E).

FGFR Inhibition Decreased Phosphorylation of FGFR Downstream Signaling via Akt and ERK
After studying the proliferative and cytotoxic effects of AZD4547 and dovitinib, we investigated whether these substances might be involved in FGFR signaling. For this purpose, we conducted Western blot and immunofluorescence staining for pERK and pAkt to examine the relative activation of these pathways following FGFR inhibition. The number of pAkt + cells was significantly reduced by AZD4547 and dovitinib (p < 0.001) (Figure 2A,B). We observed a trend towards lower abundance for dovitinib compared to control (p = 0.131) ( Figure 2C,D). Fewer pERK + cells were detectable by immunofluorescence after treatment with AZD4547 or dovitinib (p < 0.001) ( Figure 3A,B). pERK protein levels were unchanged ( Figure 3C,D). . Quantification of fluorescence exposed a reduction in pAkt + cell count for both inhibitors to less than 60% (B). Representative Western blot images of pAkt analysis (C). pAkt levels as quantified by Western blot (D). Data are presented as mean ± SEM. *** p < 0.001 vs. control; # p < 0.05, ### p < 0.001 vs. FGF2. . The number of pERK + cells was reduced by more than 40% in all groups treated with FGFR inhibitors (B). Representative images of pERK Western blot (C). pERK protein levels were reduced by dovitinib (D). Data are presented as mean ± SEM. *** p < 0.001 vs. control; # p < 0.05, ### p < 0.001 vs. FGF2.

The Expression of Neuroprotective BDNF and Its Receptor TrkB Is Elevated Following Inhibition of FGFR Signaling
We next studied the neurotrophin BDNF and its receptor TrkB, which are known to play an important role in myelination [29,30]. Application of dovitinib resulted in an increase in the number of BDNF + cells (p = 0.005) ( Figure 4A,B) and TrkB + cells (p < 0.001) ( Figure 5A,B). Treatment with AZD4547, however, increased only the number of TrkB + (p = 0.002) ( Figure 5A,B) but not BDNF + cells ( Figure 4A,B). Dovitinib caused an upregulation of BDNF and TrkB proteins (p < 0.001) ( Figure 4D,E and Figure 5D,E). AZD4547 did not affect protein levels of TrkB or BDNF ( Figure 4D,E and Figure 5D,E). In contrast to these findings, mRNA encoding for BDNF (p < 0.001) was increased by AZD4547 ( Figure 4C). Furthermore, NTRK2 mRNA expression was elevated after AZD4547 treatment (p < 0.001). Treatment with dovitinib did not alter mRNA levels of BDNF or NTRK2.

Expression of the Myelin Inhibitory Genes SEMA3A and TGFß Is Reduced after FGFR Inhibition by AZD4547
Semaphorin 3A (SEMA3A) is one of the factors known to suppress (re)myelination by impeding OPC migration [32]. Transforming growth factor β (TGFβ) has ambivalent roles in demyelination. We, therefore, assessed their expression after FGFR inhibition on mRNA levels using comparative qRT-PCR. The analysis revealed an abating effect of AZD4547 on mRNA levels of TGFB1 (vs. control: p < 0.024) as well as SEMA3A (vs. FGF2: p = 0.007; vs. control: p < 0.001) ( Figure 8A,B). Dovitinib did not regulate TGFB1 gene expression, there was a strong trend towards downregulation of SEMA3A (Dovitinib vs. control: p = 0.062) ( Figure 8A,B).

Discussion
There is increasing evidence for a role of FGF/FGFR signaling in the pathogenesis of MS and its disease model EAE. We recently showed that OL-specific deletion of either FGFR1 or FGFR2 resulted in a less severe disease course in MOG 35-55 -induced EAE [16,17]. Furthermore, detrimental effects of FGFR signaling in OLs on inflammation, myelin and axons were found in this disease model, highlighting the importance for the investigation of FGFR signaling in these myelin-producing cells. Therefore, to extend our understanding of FGFR signaling in oligodendrocytes, we studied the effects of pharmacological inhibition of FGFR in OLN-93 cells using the tyrosine kinase inhibitors dovitinib and AZD4547. We propose that the characterization of effects caused by FGFR inhibition in OLs can be a step towards the development of therapeutics abrogating impaired remyelination in MS.
To date, research on the function of FGFRs in oligodendrocytes has been done in knockout mice. The studies revealed that activation of ERK and Akt is central to the function of OPCs and OLs. These signaling proteins take part in pathways involved in proliferation and differentiation programs including process outgrowth and myelination [33][34][35][36][37]. Thus, ERK and Akt are considered to be important cellular actors in the OL cell cycle [38,39] and presumably the most important signaling molecules downstream of FGFRs [40]. Furthermore, FGFR inhibitors applied to FGFR + cancer cells reduce the downstream signal transducers pERK and pAkt [19,[41][42][43]. However, inhibition of the FGFR pathway specifically in OLs has not been explicitly studied yet. In the present in vitro experiments, both inhibitors of FGFRs reduced the proliferation of OLN-93 cells. Accordingly, FGFR inhibitors exert anti-proliferative effects in tumor cells [44,45] and vice versa, FGFs increase proliferation in various cells including OPCs [46][47][48][49]. This anti-proliferative effect of FGFR inhibition in OLN-93 cells was accompanied by a significant reduction of pErk and pAkt expression. There is considerable evidence that ERK and Akt are promoters of myelination [33][34][35]39]. Crosstalk between these two through mTOR signaling [34,36], as well as activation and inhibition of other pathways are probably involved in myelination and differentiation of OLs. In EAE, OL-specific deletion of FGFR1 and FGFR2 resulted in an upregulation of pAkt in the chronic phase of EAE. Since ERK/Akt levels were not changed in the acute phase of EAE, at a time point when beneficial effects of FGFR deletion were already present, activation of other pathways may account for the effects on myelin [16,17]. In this in vitro study, inhibition of FGFRs in OLs resulted in reduced phosphorylation of both direct downstream targets ERK and Akt. Despite the reduction of key signal proteins, an increase in myelin expression was observed, suggesting that in OLN-93 cells alternative supplementary or separate promyelinating pathways exist.
An alternative pathway modulating myelination may involve BDNF/TrkB signaling. In EAE, deletion of FGFR1 in OL resulted in an upregulation of BDNF/TrkB and less myelin loss [16]. BDNF increases inflammation in the CNS, while acting as a protective factor for axons in EAE [50]. Its deletion accordingly leads to more severe clinical activity, reduced axonal density and increased demyelination in the acute phase of EAE [51]. Thus, a neuroprotective effect of BDNF/TrkB signaling in an immune-dependent demyelinated setting, such as in EAE, supports the concept of neuroprotective autoimmunity [52]. Furthermore, administration of a TrkB agonist resulted in neuroprotective effects in EAE, including decreased OL loss and inflammation, less demyelination and increased remyelination [53]. A promyelinating effect of BDNF/TrkB signaling also seemed to be present in an oligodendrocyte monoculture setting without immune cells. Findings from our in vitro experiments show that FGFR inhibition in OLN-93 cells inhibits canonical downstream signal transducers, while simultaneously activating the BDNF/TrkB pathway, that was previously considered to be independent. Taken together, reduced ERK/Akt expression associated with increased BDNF/TrkB levels suggest that OLs are capable of compensatory responses following FGFR inhibition.
In agreement with our hypothesis, inhibition of FGFR resulted in increased expression of myelin-specific proteins. Promotion of cell cycle switches in mature OLs with potential for myelination may take place. Even though FGFs are required for developmental myelination and may enhance proliferation of OPCs [49,54], their effects vary dependent on the microenvironment, presence of other cells, and the differentiation state of OLs. Concordantly with our results from cell-specific FGFR knockout studies in EAE [16,17], in vitro data support the view that inhibition of FGFR signaling in OLs results in enhanced myelination. Furthermore, the inhibitor of remyelination SEMA3A was reduced in OLspecific FGFR1 and FGFR2 knockout mice in EAE [16,17]. SEMA3A has not only been suggested to inhibit remyelination, it may also modulate immune responses by balancing T cell inflammation [32]. TGFβ, which has several cell-specific immunomodulatory roles, exerts anti-inflammatory effects on the microglia in the CNS [55,56]. Taken together, FGFR inhibition in OLN-93 cells does not only result in an increase of myelin proteins, it also decreases the expression of myelin inhibitors. Thus, inhibition of FGFR signaling in OLs induces myelination and intrinsically decreases inhibition of myelination.
The effects of the two inhibitors used in this study are similar but with different mechanisms. AZD4547 altered cell responses on the genetic level. It distinctly lowered the amount of FGFR1 mRNA, increased BDNF and TrkB mRNA and, more subtly, the CNP mRNA level in comparison to dovitinib. AZD4547 reduced myelin inhibitors TGFβ and SEMA3A. Surprisingly, alteration of expression patterns did not always result in higher protein expression. Here, dovitinib treatment resulted in distinct effects on downstream signaling (pERK, pAkt) compensatory pro-myelinating signals (BDNF, TrkB) and a slightly higher effect on myelin proteins. However, none of the differences between the two substances were statistically significant-except for a difference in mRNA levels of TrkB. The trend towards stronger action by dovitinib may point towards the relevance of other tyrosine kinases inhibited by dovitinib. These include VEGFRs, c-kit, colony stimulating factor 1 receptor (CSF1R), FMS-like tyrosine kinase 3 (FLT3) and PDGFRs [57]. CSFR1 is also inhibited by AZD4547 at a similar IC 50 and will thus not explain the differences. All of the above-mentioned receptors are presumably expressed by oligodendrocytes at least either after injury or in progenitor states [58][59][60][61][62]. It is conceivable, that the effects of dovitinib are enhanced by its lower selectivity. Especially PDGFRs in mature OLs may be important for the enhanced response in the presence of dovitinib, as this receptor acts analogously to FGFRs [63,64].
The in vitro results from this study may be difficult to transfer to the CNS, where various cell types interact. However, EAE studies demonstrate that FGF/FGFR signaling affects myelination. Furthermore, most of the data from MS and demyelinating models show that FGFs accumulate in demyelinated lesions [13,14,65,66]. In EAE, effects of FGFR signaling are different between disease phases, but FGFR signaling is detrimental in the pathogenesis of EAE [16,17]. FGF2 and FGF9 were downregulated in the chronic phase of EAE in OL-specific FGFR2 knockout mice [17]. MS research has focused on ligands rather than receptors. FGF2 levels in the CSF are elevated in patients [15] and FGF2 + microglia are present in active MS lesions [13]. FGF9 is increased primarily in OLs in active as well as in the boundary zone of chronic active lesions suggesting that it blocks myelination [14]. In contrast, it has been proposed that FGFs act as recruiting factors for OPCs thus promoting remyelination. Taken together, FGFs may initially be conducive to recruitment and proliferation, but their continued presence may inhibit remyelination.
Both AZD4547 and dovitinib led to a significant decrease of FGFR1 + cells upon inhibition of the receptor. Other than expected, OLN-93 cells showed no response to FGFR1 inhibition as explained by a proposed negative feedback loop that upregulates FGFR1 expression. Further, FGFR1 could be either post-transcriptionally degraded (dovitinib) or its mRNA expression decreased (AZD4547), suggesting a double negative feedback loop. In fact, this scenario has been described earlier, in which FGFR1 mRNA levels in OPCs decline after exposure to FGF2 [67]. However, our results showed a reverse effect to the absence of FGFR1-dependent signaling. This may underline the relevance of high abundance of FGF2 in a disease state such as EAE and MS, where FGF2 will continuously activate the signaling pathway which is otherwise down-regulated by auto-negative feedback. In this scenario, continuously increased levels of FGF2 from various sources activate the cellular pathway that is otherwise heavily confined by self-regulatory mechanisms, thus leading to detrimental effects for remyelination.
In summary, we showed that pharmacological inhibition of FGFRs successfully disrupted the FGFR signaling cascade with alterations in intrinsic cellular responses and myelin production. It is conceivable that decreased phosphorylation of ERK and Akt induces an upregulation of BDNF/TrkB signaling leading to enhanced production of myelin-specific proteins. Considering that inhibition of FGFRs is currently being used in various cancer trials, these tyrosine kinase inhibitors should be administrated in preclinical studies to evaluate their potential for remyelination and, if successful, applied in MS research trials.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/cells10061318/s1, Table S1. List of antibodies used in this study, Table S2. List of primers used in this study.