FGF2 and FGFR1 signaling regulate functional recovery following cuprizone demyelination

In demyelinating diseases, such as multiple sclerosis, remyelination offers the potential to recover function of viable denuded axons by restoring saltatory conduction and/or protecting from further damage. Mice with genetic reduction of fibroblast growth factor 2 (Fgf2) or Fgf receptor 1 (Fgfr1) exhibit dramatically improved remyelination following experimental demyelination with cuprizone. The current studies are the first to test neurobehavioral outcomes with these gene deletions that improved remyelination. The cuprizone protocols used did not produce overt abnormalities but did reduce bilateral sensorimotor coordination (complex wheel task) and increase sociability (two chamber apparatus with novel mouse). A significant effect of genotype was observed on the complex wheel task but not in the sociability apparatus. Specifically, complex wheel velocities for Fgf2 nulls improved significantly after removal of cuprizone from the diet. This improvement in Fgf2 null mice occurred following either acute (6 weeks) or chronic (12 weeks) demyelination. Plp/CreERT:Fgfr1(fl/fl) mice administered tamoxifen at 10 weeks of cuprizone treatment to induce Fgfr1 knockdown also showed improved recovery of running velocities on the complex wheels. Therefore, constitutive deletion of Fgf2 or Fgfr1 knockdown in oligodendrocyte lineage cells is sufficient to overcome impairment of sensorimotor coordination after cuprizone demyelination.


Introduction
Remyelination, which may restore nerve conduction and protect axons, is significantly greater in early stage multiple sclerosis (MS) lesions than in chronic disease [5,6]. Failed differentiation of oligodendrocyte lineage cells (OLCs) may contribute to poor remyelination in chronic MS lesions and prolonged neurological deficits [9]. Multiple molecular signaling pathways inhibit differentiation of oligodendrocyte progenitor (OP) cells and limit remyelination in experimental models [1]. Modifying inhibitory signals in lesion areas could potentially enhance functional recovery in MS patients by improving the remyelination capacity of immature OLCs that persist in MS lesions [4,9].
The current studies are the first to evaluate recovery of function in Fgf2 null mice as well as in mice with Fgfr1 knockdown in OLCs after cuprizone demyelination. A complex wheel task differentiates cuprizone treated mice due to impaired bilateral sensorimotor coordination that reflects corpus callosum dysfunction [7,16]. This complex wheel assessment detects deficits during acute and chronic cuprizone demyelination and with latent axon loss in the corpus callosum [7,10,11]. In addition, social interaction is evaluated as a distinct behavioral domain that is altered with cuprizone demyelination [7].

Mice and cuprizone demyelination
All procedures were approved by the USUHS IACUC. Fgf2 and Plp/CreERT:Fgfr1 fl/fl mice were bred, fed cuprizone and treated with tamoxifen as previously described [2,21].

Wheel running
Cohorts of up to 12 mice were tested on the wheels simultaneously as previously detailed [7].

Sociability
A sociability apparatus (Ugo Basile, Italy) along with AnyMaze software (Stoelting Co., Wood Dale, IL) was used to quantify sociability [13]. During a 15-min habituation period, mice explore between two chambers (top and bottom) -each has one side empty and the other contains a wire cage mouse carrier. An unfamiliar mouse is placed within the carrier in the top chamber during a subsequent 15-min social period.

Myelination status
Immunohistochemistry for myelin oligodendrocyte glycoprotein (MOG) was performed and quantified as described previously [2].

Data analysis
Prism (GraphPad Software, La Jolla, CA) was used to perform ANOVA followed by Tukey's multiple comparison test for statistical analyses. Cohorts were run simultaneously for gene deletion and control comparisons by investigators blinded to condition. Behavioral cohorts included 3-6 mice per condition.

Complex wheel assessment of bilateral sensorimotor coordination in Fgf2 null mice
Cohorts of Fgf2 null and control mice were assessed on the complex wheel task (Figs. 1 and 2). Fgf2 wild type mice show reduced maximum velocities during acute cuprizone treatment that do not improve significantly after return to normal chow ( Fig. 2A), consistent with our results in C57BL/6 mice [7]. In contrast, Fgf2 null mice improve significantly during the recovery period on normal chow (Fig. 2B). With chronic cuprizone demyelination, both Fgf2 The running wheel was configured either as a "training wheel" with all rungs in place (left) or as a "complex" wheel with a non-uniform rung distribution (right) to increase the task difficulty. (B) Fgf2 mice behavioral assessment timelines showing weeks with training (T) or complex (C) wheels and sociability (S) testing. (C) Fgfr1 mice behavioral assessments shown weekly along body weight record. Cuprizone ingestion significantly reduced body weight in Plp/CreERT:Fgfr1 fl/fl mice (p < 0.001 "cup" vs. "no cup") and Fgfr1 fl/fl mice (p < 0.001 compared to Plp/CreERT:Fgfr1 fl/fl "no cup"). Tamoxifen (tam) administration did not adversely affect body weight.
wild type and null mice show impairment during acute and chronic cuprizone treatment periods ( Fig. 2C and D). Notably, only Fgf2 null mice exhibit significant improvement following this chronic demyelination (Fig. 2D).
An additional cohort of Fgf2 null mice was fed 0.3% cuprizone attempting to further differentiate the demyelination deficits from the recovery phase (data not shown). As with 0.2% cuprizone, this 0.3% dose did not produce overt changes in behavior or appearance. Similar to the data presented for 0.2% cuprizone (Fig. 2), significant improvement was again observed with return to normal chow after chronic demyelination with 0.3% cuprizone in Fgf2 null mice (p < 0.05 for recovery vs. chronic). A similar extent of remyelination was also observed. After chronic demyelination and a 6-week recovery period on normal chow, 87.45% ± 1.18 of the area of the corpus callosum immunostained for MOG in Fgf2 null mice treated with 0.3% cuprizone as compared to 87.84% ± 4.72 with 0.2% cuprizone (p = 0.9308).    both show a significant increase in sociability with acute cuprizone treatment (*p < 0.05, with vs. without cuprizone) prior to tamoxifen, which is therefore due to cuprizone and not due to Fgfr1 knockdown. Correspondingly, values before and after tamoxifen are similar between Plp/CreERT:Fgfr1 fl/fl (black bars) and Fgfr1 fl/fl (gray bars).

Sociability assessment in Fgf2 null mice
Using a sociability apparatus, Fgf2 mice were tested after 12 weeks of chronic cuprizone followed by a 6-week recovery period (Fig. 3). During the habituation period, without a second mouse present, both Fgf2 null and control mice spent equivalent time exploring each quadrant (Fig. 3A). During the social phase, the mice spent more time exploring the quadrant with the novel mouse ( Fig. 3B; top "mouse" vs. bottom "carrier"; p < 0.05). However, there was no difference between genotypes (Fig. 3B).

Complex wheel assessment of bilateral sensorimotor coordination in Plp/CreERT:Fgfr1 fl/fl mice
Plp/CreERT:Fgfr1 fl/fl mice administered tamoxifen at 10 weeks of cuprizone show enhanced remyelination after chronic cuprizone treatment [21]. Therefore, Plp/CreERT:Fgfr1 fl/fl mice were tested on the complex wheel task to evaluate behavioral effects (Fig. 4). To avoid confounding variables, tamoxifen was administered during cuprizone week 10 before placement of the training wheels for week 11 and complex wheels during week 12. Plp/CreERT:Fgfr1 fl/fl mice fed normal chow throughout served as age matched nondemyelination controls and showed consistent maximum wheel velocities across time points with the exception of a reduction following tamoxifen administration (Fig. 4A). In Plp/CreERT:Fgfr1 fl/fl mice fed cuprizone (Fig. 4B), a dose of 0.3% was used to ensure a deficit on the complex wheels while the training wheel period for stabilization of running behavior was shortened to accommodate tamoxifen administration. Plp/CreERT:Fgfr1 fl/fl mice show significantly reduced running velocities during the acute cuprizone treatment (Fig. 4B). After tamoxifen administration, running velocities improved significantly even though the mice were still ingesting cuprizone and velocities remained elevated after a return to normal chow (Fig. 4B). Fgfr1 fl/fl mice without Plp/CreERT were used as a cuprizone-treated comparison so that tamoxifen could be administered without deletion of Fgfr1 (Fig. 4C). These mice had significantly reduced running velocities during acute cuprizone treatment that continued during the chronic cuprizone period with progressive recovery after return to normal chow (Fig. 4C). 3.5. Corpus callosum remyelination after chronic demyelination with 0.3% cuprizone in Plp/CreERT:Fgfr1 fl/fl mice Our previous studies in Plp/CreERT:Fgfr1 fl/fl mice used 0.2% cuprizone [21]. Remyelination following the current 0.3% dose of cuprizone was evaluated after the completion of behavioral assessments (Fig. 6). Plp/CreERT:Fgfr1 fl/fl mice administered tamoxifen showed significantly greater MOG immunolabeling after chronic demyelination and recovery (Fig. 6D-F). Additional littermates (Fgfr1 fl/fl and Fgfr1 fl/wt mice) were fed 0.3% cuprizone in parallel with the wheel cohorts to confirm effective demyelination after 6 weeks (42.82% ± 2.61, p < 0.05) and 12 weeks (50.79% ± 1.39, p < 0.05) of 0.3% cuprizone as compared to Plp/CreERT:Fgfr1 fl/fl mice fed normal chow (90.53% ± 2.16) (Fig. 6B and C).

Discussion
The current studies show that genetic reduction of either Fgf2 or Fgfr is sufficient to improve sensorimotor coordination after cuprizone demyelination. Furthermore, Plp/CreERT conditional deletion of Fgfr1 targets this functional recovery to OLCs after tamoxifen administration at 10 weeks of chronic cuprizone demyelination.
Our findings in Fgf2 null and Fgfr1 transgenic mice demonstrate behavioral deficits on the complex wheel task with cuprizone demyelination, confirming the usefulness of this assay as identified in C57BL/6 mice [7,10,11]. Importantly, Fgf2 and Fgfr1 genotype affected running behavior; cuprizone deficits were reduced in mice with genetically reduced Fgf2 and Fgfr1 signaling. Cuprizone also alters social behavior [7]. The current studies use a sociability apparatus which is more objective than the video analysis of social interaction used in C57BL/6 mice [7] yet similarly shows an increase of social behavior in cuprizone treated mice. However, sociability did not show an effect of Fgf2 or Fgfr1 genotype effect. This sociability assessment may be less sensitive than the complex wheel assessment in our experiments for multiple reasons, including sampling time of minutes versus days, not assessed in a home environment, or less dependent on corpus callosum myelination.
Cuprizone is a systemically administered neurotoxicant so that specifically connecting a complex behavior to a given neuroanatomical change is challenging. Cuprizone treatment produces widespread demyelination of the corpus callosum with a reproducible rostrocaudal pattern [19]. Cuprizone can demyelinate the hippocampus and cerebral cortex [7,8,17] which was observed in C57BL/6 mice with complex wheel deficits indicating potential involvement of these sites, in contrast to cerebellum [7] or spinal cord (data not shown). Importantly, Fgf2 null mice and mice with Fgfr1 knockdown in OLCs show significantly improved remyelination in the corpus callosum based on immunohistochemical detection of myelin as well as electron microscopic analysis of myelin thickness and the percentage of myelinated axons [18,21]. Furthermore, this increased remyelination in the corpus callosum corresponded with reduced axon damage which may also be an important factor contributing to improved behavioral performance in the current study.

Conclusions
These studies of FGF2 and FGFR1 roles in experimental models provide evidence that removing inhibitory signals can enhance the repair capacity of endogenous cells sufficiently to enable recovery of neurological functions impaired by demyelinating disease.