Reduced axonal caliber and white matter changes in a rat model of Fragile X syndrome with a deletion of a K Homology domain of Fmr1

Fragile X syndrome (FXS) is a neurodevelopmental disorder that is caused by mutations in the FMR1 gene that are known to cause neuroanatomical alterations. The morphological underpinnings of these alterations have not been elucidated. Furthermore, while alterations have been identified in both male and female individuals, neuroanatomy in female rodent models has not been assessed. We identified structural differences in regions that are also altered in FXS in male and female rat models, including the splenium of the corpus callosum. Interestingly, different sets of regions were disrupted in male and female rat models and, remarkably, male rats had higher brain-wide diffusion than female rats overall. We found reduced axonal caliber in the splenium, offering a mechanism for its structural changes. Our results provide insight into which brain regions are vulnerable to a loss of Fmr1 expression and suggest a potential mechanism for how its loss causes white matter dysfunction in FXS.


Introduction
Fragile X syndrome (FXS) is a monogenic disorder caused by mutations in the FMR1 gene, which encodes fragile X mental retardation protein (FMRP). It is the leading monogenic cause of autism spectrum disorder (ASD), the most frequent known form of inherited intellectual disability (ID), is often comorbid with attention-deficit/hyperactivity disorder (ADHD), and can cause sensory hyperarousal [1,2]. With the use of magnetic resonance imaging (MRI), alterations in brain structure have been identified in both gray and white matter regions in 3 individuals with FXS and associated with aberrant cognitive phenotypes. However, the integrity of brain regions and the white matter that connects them has not been directly assessed in a rat model of FXS. The regions that should be given priority in studies of FXS rat models have therefore not been elucidated. Furthermore, the mechanism by which a lack of FMRP causes structural alterations is poorly understood and can only be studied in rodent models of FXS.
In humans with FXS, impairments that change across development have been identified in frontal and striatal regions and the white matter that connects them [3]. It is hypothesized that these alterations underlie the reported deficits in attention in FXS. However, other regions and white matter tracts that are also implicated in attention show anatomical alterations in FXS, such as the thalamus, internal capsule, and the splenium of the corpus callosum [4]. Structural deficits in white matter have also been identified in Fmr1 knockout (KO) mice and resemble those seen in individuals with FXS [5][6][7]. Importantly, this includes regions known to be recruited during visuospatial attention, including the superior colliculus, and the tracts that connect them, including the splenium of the corpus callosum and the white matter of the medial prefrontal cortex. There are also deficits in the degree to which networks containing spatially disparate regions fire together while the individual is at rest, called functional connectivity, in individuals with FXS and Fmr1 KO mice [8,9]. One example of this is hypoconnectivity in the somatosensory network of Fmr1 KO mice, again suggesting an anatomical link between decreased Fmrp expression and deficits in sensory perception [9]. We predict that similar regions will be impaired in a rat model of FXS. 4 FMRP could play a role in structural integrity through its repression of protein translation of mRNA targets [10]. In FXS, without FMRP, the protein expression of these mRNA targets is elevated because their translation is no longer repressed. Many of the mRNA targets of Fmrp encode key presynaptic and postsynaptic proteins that affect synapse development, including cytoskeleton scaffolding and remodeling, as well as transcripts involved in the development of myelin [1]. Dysregulation of their expression in the absence of FMRP is thought to contribute to the heightened levels of long, immature dendritic spines and increased spine density that is commonly seen in the pyramidal neurons of individuals with FXS and animal models [1].
Therefore, these alterations in neurons and oligodendrocytes could influence the brain's functional anatomy, however, a mechanism by which this might occur has not yet been elucidated. Thus, it is pertinent to perform an unbiased study of structural integrity in rats that screens across regions.
We recently identified a deficit in sustained attention in both sexes of a rat model of FXS that has a deletion of an mRNA-binding domain, K-homology 1 (KH1), the Fmr1-Δ exon 8 rat [11].
Auditory deficits have also been modeled in this rat model with impaired cortical representation of speech sounds [12]. We chose to assess neuroanatomical integrity in this Fmr1-Δ exon 8 rat model of FXS. We used structural MRI to measure regional volumes and tissue densities, diffusion tensor imaging (DTI) to assess white matter integrity, and electron microscopy to uncover deficits in ultrastructure that could reflect functional perturbations. Both male and female Fmr1-Δ exon 8 rats were included in these studies to test for an effect of sex and for a 5 dose-dependent effect of Fmrp-Δ KH1 in a rat that is heterozygous in expression for Fmr1-Δ exon 8, a phenomenon only possible in females.

Experimental Design
The objective of this controlled laboratory experiment was to use an unbiased approach to identify brain regions that are affected by a loss of Fmr1 expression and then determine the mechanism that could underlie this change. We hypothesized that brain regions that are affected in individuals with FXS would also be altered in the Fmr1-Δ exon 8 rat model of FXS. Once we found that diffusion in the splenium of the corpus callosum was altered, we hypothesized that this could be due to alterations in axon integrity. Sample sizes used in the MRI experiments were determined based on a previous study of a rodent model of autism [13] and sample sizes for electron microscopy mirrored similar studies. Sex was considered a variable and its effect was reported where meaningful. Analysis of the MRI data was replicated with two different analytical pipelines and electron microscopy was replicated in two separate cohorts, each constituting a sample size of three. The degree to which the MRI results were validated in the second analysis are detailed in the Results section and the findings from the electron microscopy experiments were substantiated in the second cohort. During image acquisition, manual quality control of the MRI segmentations, and tracing of the electron microscopy images, the experimenters were blind to genotype. Furthermore, each manually performed operation was validated across two experimenters to limit subjectivity. Each experimental animal was scanned in the MRI the same day as a same-sex control animal. Images with obvious artifacts or masks 6 that did not align to the image were excluded before the statistical analyses commenced.
Individual data points that were outside 1.5 times the interquartile range were omitted from the analysis. No outliers were included in the results reported.

Generation of the Fmr1-
Δ exon 8 rat model The Fmr1-Δ exon 8 rat model was generated using zinc finger nucleases (ZFNs) in the outbred Sprague-Dawley background. The design and cloning of the ZFN, as well as the embryonic microinjection and screening for positive founder rats were performed by Horizon Labs (Boyertown, PA USA) as previously described [14]. The best performing ZFN pair targeting the CATGAACAGTTTATCgtacgaGAAGATCTGATGGGT sequence, located between 18686bp-18721bp in the Fmr1 gene (NCBI reference sequence NC_005120.4), was used for embryo microinjection. Positive Sprague-Dawley founder animals with a deletion in the Fmr1 gene were mated to produce F1 breeding pairs. Polymerase chain reaction amplification at the target sites followed by sequencing analysis revealed the exact deletion of 122 bp at the junction of intron 7 and exon 8 (between 18588bp-18709bp).

Animal breeding, care, and husbandry
This study used age-matched male and female littermate rats. To produce both male genotypes 22±2°C. Animals were pair-caged with food and water available ad libitum. All animal procedures were approved by the Institutional Animal Care and Use Committee at the Icahn School of Medicine at Mount Sinai.

MRI
All imaging was performed by the BioMedical Molecular Imaging Institute using a Bruker Biospec 70/30 7 T scanner with a B-GA12S gradient insert (gradient strength 440 mT/m and slew rate 3444 T/m/s). A Bruker 4 Channel rat brain phased array was used for all data acquisition in conjunction with a Bruker volume transmit 86-cm coil. All rats (N = 15/group) were imaged on a heated bed and respiration was monitored continuously until the end of the scan. The animal was anesthetized using isoflurane anesthesia (3% induction and 1.5% maintenance). After a three-plane localizer, a field map was acquired and the rat brain was MRI region-based analytical pipeline with manual editing 8 A magnetic resonance imaging processing pipeline was used to perform semi-automated nonbiased brain segmentation, while blinded to genotype (N = 15/group) [15]. The pipeline is composed of six major steps: rigid registration of images to each other, generation of a wholebrain mask for each image, averaging of all images, creation of a whole-brain mask for this averaged image, segmentation of the average mask by regions of interest (ROIs), parcellation propagation of the segmented mask to individual subjects, and ROI-based statistics for the individual images. The deformation necessary to warp each subject's image to the average was used to calculate the volume of the ROIs. After each mask was generated, it was improved manually in ITK-SNAP (www.itksnap.org). The segmentation into ROIs was determined by a template that was previously hand-segmented into 32 brain regions, listed in Supplementary   Table 1.
Segmented masks for the individual images that did not closely match the segmented average mask (one Fmr1-Δ exon 8 -/-T2 mask, two female WT T2 masks, and one male WT DTI mask) and individual data point outliers, defined as being outside 1.5 times the interquartile range, were excluded from the analysis. Males and females were analyzed in two separate pipelines because the difference in their brain size would skew the averaging step. The whole-brain masks were used to determine whole-brain measures. Mean voxel intensity for each ROI and across the wholebrain was measured in both the T2 and DTI images and the volume of each ROI and whole-brain was calculated from the T2 images. The olfactory bulb, cerebellum, and brainstem were not included in the DTI analysis because these images did not capture the entirety of these regions. Furthermore, regions without any known white matter were excluded from the DTI analysis, 9 including the aqueduct, periaqueductal gray (PAG), and third, fourth, and lateral ventricles. In the analysis of the male data, for each independent variable, if the distribution was nonparametric according to the Shapiro-Wilk's test, a Mann-Whitney U test was administered and if the distribution was normally distributed, a two-way ANOVA was applied. In the analysis of the female data, if the distribution was nonparametric, a Kruskal-Wallis test was administered, which was followed by a Dunn test to compare the individual means and an adjustment of the pvalues with the False Discovery Rate method, and if the data was parametric, an ANOVA was applied and followed by a post-hoc Tukey HSD test that compared the pairs of means and adjusted the p-values to account for the additional comparisons. Due to the fact that many comparisons were made across ROIs, the output was then assessed for the ability to survive a correction for multiple comparisons with a Bonferroni Correction, the most conservative method of this nature. Genotype was the only between groups factor. Custom scripts written in the R statistical programming environment were used for the statistical analysis (R Development Core Team, 2006).

Automated MRI region-based and voxel-based analytical pipeline
T2 images were aligned together in an iterative registration procedure using the PydPiper framework that included a series of linear and nonlinear alignment steps, as previously described [16]. Briefly, images were first linearly aligned (6 degrees of freedom: rotations and translations) to a model template, so that all images were roughly in the same space and roughly in the same orientation. These images were then linearly aligned (12 degrees of freedom: rotations, translations, scaling and shearing) to each other in a pairwise manner; each image was resampled with its average transformation to the other images. The linearly transformed resampled images were then averaged to create a linear (12-parameter) template. After creation of a linear template, images were then nonlinearly aligned with Advanced Normalization Tools (ANTs) [17] (www.picsl.upenn.edu/software/ants) to this template and averaged to create a new updated template. Nonlinear alignment and averaging were repeated for a total of three iterations. The output of this pipeline included a final study-specific nonlinear average template, transformations mapping each image to the template, and voxel-wise Jacobian determinants corresponding to each transformation that represent the extent to which each image locally deformed to match the template. To identify voxels that were altered in volume, a p-value significance was set to an false discovery rate (FDR)-corrected p-value < 0.1. We controlled for the FDR instead of applying the Bonferroni method in the analysis of the voxel-wise data because the dataset was much larger. The Bonferroni method could therefore yield falsenegatives.
DTI metrics were extracted for each subject from the raw imaging data with DTIFit (Functional Magnetic Resonance Imaging of the Brain Software Library). The raw T2 (no diffusion weighting, "S0") images were registered together using the same pipeline described above. This DTI S0 template was further aligned to the T2 template using ANTs, and concatenated transformations mapping each raw image to the T2 template were computed. The remaining images (fractional anisotropy (FA), mean diffusion (MD), axial diffusion (AD), radial diffusion (RD)) were resampled to the T2 template with their respective concatenated transforms to allow both voxel-wise and ROI analyses. 11 We used DTI to evaluate white matter specifically, as it is affected in individuals with FXS, with AD, RD, MD, and total FA as measures of possible changes. FA is enhanced when parallel diffusivity is facilitated and/or perpendicular diffusivity is restricted, RD is increased following myelin damage, and AD is reduced after axonal damage [18]. Furthermore, MD may reflect pathology if it is increased in white matter.
The template that was used for the T2 semi-automated analysis was aligned to the T2* template using ANTs in order to segment it into ROIs (see Supplementary Table 1) and resampled to these DTI templates. For T2, volumes for each subject were computed by summing the voxel volumes under each ROI, weighted by the Jacobian determinants. This was done using the RMINC software library in R (https://github.com/Mouse-Imaging-Centre/RMINC). Mean volume for each ROI and each DTI metric were computed in T2* space using RMINC. Whole brain values for the DTI metrics were calculated by summing the average metrics per ROI, weighted by volume of each ROI. The regional volume of the PAG was computed using the Waxholm Space Atlas (NITRC), a better estimate of this region, once volumetric changes in voxels within it were identified.
For this analysis, Fmr1-Δ exon 8 -/and Fmr1-Δ exon 8 -/y rats were compared to their WT male and female littermates (N = 15/group) so that the effect of sex could be examined. Data points that were outside 1.5 times the interquartile range were determined to be outliers and removed from the analysis. Statistical analysis began with the assessment of normality. Non-parametric data were evaluated with a linear model (LM) and parametric data was assessed with a two-way ANOVA, both with genotype and sex as factors. A Tukey HSD post hoc test was administered if 12 there was a main effect of genotype or sex to correct for post hoc multiple comparisons. The output was corrected for multiple comparisons with a Bonferroni Correction, just as in the semiautomated ROI-based analysis.

Electron Microscopy
Preparation for electron microscopy was performed by the Icahn School of Medicine's Microscopy CoRE using protocols optimized to study the ultrastructure of nervous tissue. Male rats (WT: N = 7, Fmr1-Δ exon 8 -/y : N = 6) were anesthetized and perfused using a peristaltic pump at a flow rate of 35 ml/min with 1% paraformaldehyde/phosphate buffered saline (PBS), pH 7.2, and immediately followed with 2% paraformaldehyde and 2% glutaraldehyde/PBS, pH 7.2, at the same flow rate for an additional 10-12 min. The brain was removed and placed in immersion fixation (same as above) to be postfixed for a minimum of one week at 4 ºC. Fixed brains were sectioned using a Leica VT1000S vibratome (Leica Biosystems, Buffalo Grove, IL) and coronal slices (325 µm) containing the splenium of the corpus callosum were removed and embedded in EPON resin (Electron microscopy Sciences (EMS), Hatfield, PA). Briefly, sections were rinsed in 0.1 M sodium cacodylate buffer (EMS), fixed with 1% osmium tetroxide followed with 2% uranyl acetate, dehydrated through ascending ethanol series (beginning with 25% up to 100%) and infiltrated with propylene oxide (EMS) and then EPON resin (EMS). Sections were transferred to beem capsules and heat-polymerized at 60 °C for 72 h. Semithin sections (0.5 and 1 µm) were obtained using a Leica UC7 ultramicrotome, counterstained with 1% Toluidine Blue, coverslipped and viewed under a light microscope to identify and secure the region of interest. Ultrathin sections (80 nm) were collected on copper 300 mesh grids (EMS) using a 13 Coat-Quick adhesive pen (EMS), and serial sections were collected on carbon-coated slot grids (EMS). Sections were counter-stained with 1% uranyl acetate followed with lead citrate.
Sections were then imaged on an Hitachi 7000 electron microscope (Hitachi High-Technologies,Tokyo, Japan) using an advantage CCD camera (Advanced Microscopy Techniques, Danvers, MA). Ten areas from within the region of the splenium that contained cross sections of axons were chosen randomly to be imaged per section at 15,000 x magnification. Images were adjusted for brightness and contrast using Adobe Photoshop  14 was nested in the rat ID which was nested in the cohort number in order to determine the effect of genotype was computed using custom scripts written in the R statistical programming environment (R Development Core Team, 2006).

Δ exon 8 rats
Structural T2-weighted MRI was used to assess both volume and white versus gray matter density in male and female Fmr1-Δ exon 8 rats and their WT littermates. The brains were divided into ROIs to determine whether these measures in the whole brain or in 32 ROIs (see Supplementary Table 1) were affected by genotype. In the region-based analysis, ROIs were first determined based on a template (see Methods) in an averaged image that was calculated from all subjects. This method of segmentation was then applied to each individual image.
Because male rodents have larger brains than female rodents [19] due to increased body size, we decided to analyze the data two ways: one with the males and females separated and one with them combined. We will first describe the results of the analysis where they were split by sex.
There were no significant differences in whole-brain volume between the Fmr1- Additionally, there were no differences in mean tissue density across the whole brain and in individual ROIs, suggesting an equal distribution of gray and white matter in Fmr1- Furthermore, the volume across most ROIs did not differ by genotype and no regions differed by genotype when we normalized the volume of each ROI to the volume of the whole brain. The 15 only metric that was altered was the absolute volume of the superior colliculus, which had a significant nominal p-value that did not survive a Bonferroni correction for multiple comparisons, but was increased in the Fmr1-Δ exon 8 -/y rats (Supplementary Figure 1B; ANOVA, p = 0.011). Pertinent to this study, the superior colliculus is implicated in visuospatial attention across species [20], most specifically in spatial orienting.
Similar to the analysis of the male rats, there was no significant effect of genotype on wholebrain volume or density in the analysis of the female WT, Fmr1- adj. = 0.02 and p adj.= 0.023, respectively). Next, we examined the relative volume of each ROI by calculating its percentage of each rat's total brain volume. We found that the genu was again increased in Fmr1- Kruskal-Wallis followed by Dunn's test, p adj. = 0.02).

Volumetric changes are validated when males and females are combined
We next validated our findings with a second, fully automated pipeline that used the same method of segmentation (see Supplementary Table 1 for ROIs) and combined males and females. All brains were first registered together and then ROIs were determined for each subject individually, thus allowing us to directly compare males and females. There was a significant effect of sex on absolute and relative volume of the superior colliculus (Figure 2A and B; LM followed by Tukey HSD, p = 4.4 x 10 -13 and p = 5.08 x 10 -8 , respectively). Additionally, the relative volume of the lateral ventricle increased and the relative volume of the fourth ventricle decreased in Fmr1-Δ exon 8 rats (Figure 2A and B; two-way ANOVA followed by Tukey HSD, p adj. = 2.01 x 10 -7 and p adj. = 2.11 x 10 -4 , respectively).
The superior colliculus and fourth ventricle were larger in the male rats (Supplementary Figure   2A). 17 Because this pipeline transforms each image to a study-based average, we were additionally able to test for local effects that would not otherwise be captured at the structural level with a voxelbased approach. The PAG contained many voxels that were affected by genotype ( Figure 2C; pFDR < 0.1). After applying region segmentation using the Waxholm Space Atlas, we found that there was a main effect of genotype that survived a Bonferroni correction (Figure 2C; two-way ANOVA followed by Tukey HSD, p adj. = 2 x 10 -7 ) on relative volume of the PAG where the Fmr1-Δ exon 8 rats had an increased volume. There was no effect of sex ( Figure 2C; p = 0.267).

Voxel-based analysis identifies further volumetric changes
The absolute volume of the PAG was also significantly increased in Fmr1-Δ exon 8 compared to WT rats and in male rats compared to female rats ( Figure 2C; LM followed by Tukey HSD, p adj. = 0.002 and p adj. = 7.46 x 10 -13 , respectively); however, the effect of genotype did not survive a Bonferroni correction and the effect of sex was likely due to the overall increase in brain size in males.

Δ exon 8 rats
We used DTI to evaluate white matter specifically, as it is affected in individuals with FXS, with axial, radial, and mean diffusion (AD, RD, and MD, respectively) and total FA as measures of possible changes. FA is enhanced when parallel diffusivity is facilitated and/or perpendicular diffusivity is restricted, RD is increased following myelin damage, and AD is reduced after axonal damage [18]. Furthermore, MD may reflect pathology if it is increased in white matter.
Similar to the T2 region-based analysis, we first analyzed the male and female rats separately and assessed both the whole-brain and the 32 brain ROIs (see Supplementary Table 1). When we compared Fmr1-Δ exon 8 -/y rats to their male WT littermates, we found that the nominal p- 18 values for the DTI metrics of multiple ROIs were significant, though only a few survived a Bonferroni correction. Remarkably, RD was increased across the whole brain (Figure 3A and B; ANOVA, p = 0.043). This could be due to RD being increased in many forebrain regions, including the largest region of the brain in our analysis, the neocortex (Figure 3A and C; ANOVA, p = 0.015). RD also increased in the piriform cortex, hypothalamus, and thalamus

Combining sexes shows effect of genotype and uncovers major sex differences
We validated our findings with the Fmr1-Δ exon 8 -/+ rats excluded from the analysis using the segmented average from the semi-automated pipeline as a template (see Supplementary Table 1 for ROIs) and found multiple regions to be altered. FA was reduced in the anterior commissure  (Figure 5B; LM, p = 0.14), when we multiplied the mean caliber of the myelinated axons by the number represented in a given field, we found that the myelinated axons occupied significantly less of the total field area in Fmr1-Δ exon 8 -/y rats ( Figure 5C; LM, p = 0.0012). This suggests that there is more room for diffusion outside of the myelinated axons, which could lead to an increase in MD and RD. However, it is unclear whether MRI is sensitive enough to capture these types of changes. Notably, we did not see deficits in myelin integrity. The thickness of the myelin was roughly equal in male WT rats and Fmr1-Δ exon 8 -/y rats (Figure 5D; LM, p = 0.28). Interestingly, however, the mean g-ratio was reduced in the Fmr1-Δ exon 8 -/y rats (Figure 5E; LM, p = 0.02). To assess whether reduced axonal caliber could be driving this decrease, we examined the relationship between g-ratio and axon caliber. We found a significant interaction between genotype and axon caliber where smaller axons had decreased g-ratios in Fmr1-Δ exon 8 -/y rats (Figure 5F; LM, p = 0.0012). These lower g-ratios in the smaller axons could be due to thicker myelin or reduced internode length.
Additionally, thicker myelin surrounding the smaller axons could further restrict diffusion along these axons.
Unlike myelinated axons, the caliber of unmyelinated axons was unaffected in the Fmr1-Δ exon 8 -/y rats (Figure 5G; LM, p = 0.1). However, interestingly, when we evaluated the proportion of axons at each caliber, we found a significant difference comparing the cumulative frequencies of axonal calibers in the Fmr1-Δ exon 8 -/y rats compared to their WT littermates of both myelinated and unmyelinated axons (Figure 5H and I; Kolmogorov-Smirnov test, p < 2.2 x 10 -16 and p < 9.78 x 10 -11 , D = 0.19 and D = 0.14). These frequency plots show that the fields from the Fmr1-Δ exon 8 -/y rats contained smaller myelinated (< 750,000 nm 2 ) and unmyelinated (~50,000 to 23 250,000 nm 2 ) axons and did not contain larger (> 750,000 nm 2 ) myelinated and unmyelinated (> 250,000 nm 2 ) axons that were present in the WT images. Overall, the impairments in the splenium in Fmr1 KO mice that were discovered using DTI appear to be at least partially explained by a general reduction in axonal caliber.

Discussion
In this study, we discovered structural perturbations in several brain regions of Fmr1- The decrease in volume of the fourth ventricle could be due to the increase in size of the PAG and the increase in volume of the lateral ventricle could be due to subtle changes in volume of the brain regions surrounding it.
While a region-based approach is sensitive to volumetric changes at the anatomical level, we also applied a voxel-based approach further to probe changes at a smaller scale. Using this voxelbased approach, we found the PAG to be increased in both Fmr1-Δ exon 8 -/y and Fmr1-Δ exon 8 -/rats. The PAG is also increased in volume in Fmr1 KO mice on the FVB background strain [5].
Notably, the PAG has not garnered much attention in studies of individuals with FXS and there are no reports of its volume in this population. Because it is enlarged in both a mouse and rat model of FXS, crucially in both sexes of a rat model, its volume deserves consideration in individuals with FXS.
In our primary semi-automated region-based analysis of DTI images, we discovered alterations 25 in diffusion in both gray and white matter regions that are also impaired in Fmr1 KO mice and individuals with FXS. In the forebrain, the caudate/putamen and internal capsule are similarly smaller and the fimbria and inferior colliculus are similarly enlarged in the Fmr1 KO mouse on the FVB background strain compared to WT controls [5]. This reduction in size of the caudate nucleus is opposite of what is often documented in individuals with FXS, but the caudate and putamen are separate nuclei in humans, making them difficult to compare to their rodent homolog. Unlike the caudate nucleus, the fimbria and inferior colliculus have been unexplored as regions that could underlie FXS pathology. While the internal capsule also changed in the opposite direction to what is documented in humans with FXS and in Fmr1 KO mice on the FVB background strain, it has increased MD and RD in its anterior limbs [4], which complements what we found in Fmr1-Δ exon 8 -/y rats. Further complementing our results, the splenium and thalamus show increased AD, MD, and RD in girls with FXS [4], the thalamus has increased gray matter volume [23], and FA is decreased in the splenium of Fmr1 KO mice [7].
Interestingly, however, no prior studies have examined the anatomy of the substantia nigra in rats, which could offer a potential mechanism for the reduction in axonal caliber we identified. Interestingly, we previously found that PGC-1α (also known as PPARGC1A), a master regulator of mitochondria biogenesis that is also an Fmrp target, is decreased in expression in Fmr1-Δ exon 8 -/y rats [11]. It would be worthwhile to determine whether PGC-1α plays a role in this deficit in axonal caliber. Weakening of the cytoskeleton could also cause a reduction in caliber. Fmrp regulates the translation of mRNAs that encode for proteins that regulate cytoskeleton remodeling. Because Fmrp localizes to axons and associates with translational machinery in rats [34], it is possible that the expression of proteins that regulate the integrity of the cytoskeleton is dysregulated in Fmr1-Δ exon 8 -/y rats, leading to malformed axonal cytoskeleton. The brain regions we found to be altered that are also impaired in FXS and involved in attention networks are worth further exploration as targets for treatment of attention deficits in FXS because they are also implicated in ADHD. For example, in ADHD, MD in the caudate, 28 putamen, and thalamus is correlated with increased reaction time on a flanker test [35].
Furthermore, the internal capsule has impaired white matter integrity [36]. Lastly, there is higher MD and RD, and lower FA [37] in the splenium. The lower FA phenotype is especially prominent in patients with the inattention phenotype [36], the specific deficit we discovered in the Fmr1-Δ exon 8 rats [14]. Therefore, the relationship between the altered structure of these regions and attentional functioning in FXS models should be studied more directly in the same animals.
Sensory deficits have also been replicated in FXS models [38]. Many of the brain regions that were perturbed in our study are implicated in sensory perception. To begin with, three regions that are involved in olfaction, the piriform cortex, olfactory bulb, and anterior commissure were altered in Fmr1-Δ exon 8 rats. Olfaction has not been well characterized in FXS, but it is impaired in rodent models of FXS [39]. Individuals with FXS can also be hypersensitive to auditory and visual stimuli. We found that the inferior and superior colliculus, which are in the auditory and visual pathways, respectively, are altered in Fmr1-Δ exon 8 rats. Additionally, we identified alterations in the PAG, which, among other functions, is involved in somatosensory perception and the thalamus, which relays sensory information to the cortex, to be increased in diffusion. Cfos expression in the PAG and thalamus is excessive after an auditory stimulus that causes audiogenic-induced seizures in Fmr1 KO mice [40,41]. The role that these regions play in hypersensitivity in FXS has not garnered much attention. The connection between the structural alterations we identified and the sensory deficits that are often observed in FXS models should be further explored. 29 In summary, we have shown here that a specific deletion of exon 8 of the Fmr1 gene is sufficient to cause FXS-like changes in neuroanatomy in rat, specifically in the axons of a major pathway. Therefore, white matter in FXS deserves further examination. The regions we found to be disrupted could serve as potential cross-species non-invasive biomarkers for FXS.