FXTAS neuropathology includes widespread reactive astrogliosis and white matter specific astrocyte degeneration

Abstract

Objective:

Fragile X- associated Tremor/Ataxia Syndrome (FXTAS) is a late-onset progressive genetic neurodegenerative disorder that occurs in FMR1 premutation carriers. The temporal, spatial, and cell-type specific patterns of neurodegeneration in the FXTAS brain remain incompletely characterized. Intranuclear inclusion bodies are the neuropathological hallmark of FXTAS, which are largest and occur most frequently in astrocytes, glial cells that maintain brain homeostasis. Here, we characterized neuropathological alterations in astrocytes in multiple regions of the FXTAS brain.

Methods:

Striatal and cerebellar sections from FXTAS cases (n=12) and controls (n=12) were stained for the astrocyte markers Glial Fibrillary Acidic Protein (GFAP) and aldehyde dehydrogenase 1L1 (ALDH1L1) using immunohistochemistry. Reactive astrogliosis severity, the prevalence of GFAP+ fragments, and astrocyte density were scored. Double label immunofluorescence was utilized to detect co-localization of GFAP and Cleaved Caspase 3.

Results:

FXTAS cases showed widespread reactive gliosis in both grey and white matter. GFAP staining also revealed remarkably severe astrocyte pathology in FXTAS white matter - characterized by a significant and visible reduction in astrocyte density (−38.7% in striatum and −32.2% in cerebellum) and the widespread presence of GFAP+ fragments reminiscent of apoptotic bodies. White matter specific reductions in astrocyte density were confirmed with ALDH1L1 staining. GFAP+ astrocytes and fragments in white matter were positive for cleaved caspase-3, suggesting that apoptosis-mediated degeneration is responsible for reduced astrocyte counts.

Interpretation:

We have established that FXTAS neuropathology includes robust degeneration of astrocytes, which is specific to white matter. Since astrocytes are essential for maintaining homeostasis within the CNS, a loss of astrocytes likely further exacerbates neuropathological progression of other cell types in the FXTAS brain.

INTRODUCTION

Fragile X-associated Tremor/Ataxia Syndrome (FXTAS) is a progressive genetic neurodegenerative disorder that occurs in a subset of FMR1 premutation (PM) carriers later in life^{1, 2}. In addition to the core symptoms of intention tremor and cerebellar ataxia¹, patients often experience cognitive impairment³, autonomic dysfunction⁴, and peripheral neuropathy⁵. FMR1 premutation occurs when there is an expansion in the CGG triplet repeat in the non-coding 5’ region of the X-linked FMR1 gene, in the range of 55–200 repeats, which confers a toxic gain of function on the encoded FMR1 messenger RNA. FMR1 PM is relatively prevalent, estimated to occur in 1/150–300 females and in 1/400–850 males^{2, 6, 7}. Although not fully penetrant, FXTAS occurs at high rates with ~50% of male and ~20% of female PM carriers developing the disorder by age 70^{8, 9}. Individuals with FXTAS show widespread and progressive reductions in regional brain volume, with the cerebellum, putamen, brainstem, and corpus callosum most impacted^{10, 11}. White matter disease is particularly severe, characterized by extensive volumetric loss and the presence of focal T2 FLAIR hyperintensities¹². While there is presumably widespread cell loss that drives these volumetric changes – the temporal, spatial, and cell-type specific patterns of neurodegeneration remain largely uncharacterized to date.

The neuropathological hallmark of FXTAS is the formation of intranuclear inclusion bodies¹³. FXTAS inclusion bodies are most prevalent and largest in astrocytes¹³, glial cells present throughout the brain. Astrocytes have numerous functions, all largely designed to maintain brain homeostasis in various ways¹⁴. They regulate extracellular ion and neurotransmitter concentrations^{14, 15}, maintain brain pH¹⁶, and play a crucial role in supporting neuronal metabolism¹⁵. They modulate neuronal signaling through direct contacts at neuronal synapses¹⁷ and axonal nodes of Ranvier¹⁸. Astrocyte processes ensheath brain capillaries - modulating blood flow and forming an essential component of the blood-brain barrier¹⁹. Astrocytes also play an essential role in neuroinflammation. In response to brain tissue trauma or damage, they undergo reactive astrogliosis - becoming hypertrophic, upregulating expression of glial fibrillary acidic protein (GFAP), secreting cytokines and neurotrophic factors, and in extreme cases proliferating to form a ‘glial scar’ that isolates aberrant damaged brain tissue^{14, 20}.

Until now, all that was known about astrocytes in FXTAS is that they show a uniquely high inclusion burden and that there are patches of reactive astrogliosis in the brain. Here, we used GFAP staining to better characterize regional differences in reactive astrogliosis in the FXTAS brain, with a specific focus on the cerebellum and striatum. We confirmed and further characterized regional differences in patchy astrogliosis in the FXTAS brain, which is more severe in white matter. More importantly, we discovered that severe astrocyte degeneration occurs in FXTAS white matter, in both cerebellum and striatum, as indicated by a 30–50% reduction in astrocyte density, the widespread presence of GFAP+ fragments reminiscent of apoptotic bodies, and the co-localization of GFAP+ astrocytes and fragments with the apoptosis marker cleaved caspase 3.

METHODS

Cases and Tissue

Formalin fixed human postmortem brain tissue from 12 FMR1 premutation carriers with clinically diagnosed FXTAS and 12 age/sex matched controls were utilized (Table 1). Data on the presence of tremor and ataxia was available for 11 cases – with ataxia present in 11/11 and tremor in 9/11 FXTAS donors. All controls underwent neuropathological analysis at autopsy and were confirmed to be free of neurodegenerative disease. FXTAS diagnosis was confirmed neuropathologically – ubiquitinated intranuclear inclusion bodies were present in all FXTAS subjects, detected using immunohistochemistry. Tissue was obtained from the UC Davis Fragile X Brain Repository. All donations were carried out with approval from the UC Davis Institutional Review Board and with informed consent from donors and/or next of kin.

Table 1.

FXTAS and Control Cases

								Striatum			Cerebellum
ID	Group	Sex	Age	PMI	CGG	Ataxia	Tremor	PUT	CD	IC	GCL	MS
F1	FXTAS	F	52	NA	(36) 75	Y	Y	Y	N	N	Y	Y
F2	FXTAS	M	58	14	97	Y	Y	Y	Y	Y	Y	Y
F3	FXTAS	F	65	240	NA	Y	Y	Y	Y	Y	Y	Y
F4	FXTAS	M	66	NA	100	Y	Y	Y	Y	Y	Y	Y
F5	FXTAS	M	70	2.5	75	Y	Y	Y	Y	Y	Y	Y
F6	FXTAS	M	77	24	74	Y	Y	Y	N	N	Y	Y
F7	FXTAS	M	78	NA	106	NA	NA	Y	N	N	Y	Y
F8	FXTAS	F	80	5	(30) 63	Y	Y	Y	Y	Y	Y	Y
F9	FXTAS	M	82	NA	67	Y	N	Y	Y	Y	Y	Y
F10	FXTAS	F	84	12	(27) 59	Y	Y	Y	Y	Y	Y	Y
F11	FXTAS	M	85	127	86	Y	Y	Y	Y	Y	Y	Y
F12	FXTAS	M	87	NA	65	Y	N	Y	Y	Y	Y	Y
C1	Control	M	50	94.5	NA	-	-	Y	Y	Y	Y	Y
C2	Control	M	54	64.3	NA	-	-	Y	Y	Y	Y	Y
C3	Control	M	65	170	NA	-	-	Y	N	Y	Y	Y
C4	Control	M	68	52.7	NA	-	-	Y	Y	Y	Y	Y
C5	Control	M	68	108	NA	-	-	N	Y	N	Y	Y
C6	Control	F	68	44.3	NA	-	-	Y	Y	Y	Y	Y
C7	Control	M	72	35.6	NA	-	-	Y	Y	Y	Y	Y
C8	Control	M	74	17.9	NA	-	-	Y	N	N	Y	Y
C9	Control	M	76	55.8	NA	-	-	Y	N	N	Y	Y
C10	Control	F	80	43	NA	-	-	Y	Y	Y	Y	Y
C11	Control	F	81	17.3	NA	-	-	Y	Y	Y	Y	Y
C12	Control	F	81	NA	NA	-	-	Y	Y	Y	Y	Y

PMI – Post Mortem Interval

CGG – FMR1 CGG triplet repeat length

PUT – Putamen (grey matter)

CD – Caudate (grey matter)

IC – Internal Capsule (white matter)

GCL – Granule Cell Layer (grey matter)

MS – Medullary Substance (white matter)

Y/N – Yes/No

NA – Not available

Tissue blocks from striatum and cerebellum were dissected (Fig. 1a-c), immersed in 30% sucrose until sunk, embedded in O.C.T. Compound (Tissue-Tek), and sectioned at 12μm using a cryostat (Leica). All sections were mounted on superfrost plus slides (Fisher), and stored at −20C. All cerebellar sections had cerebellar folia and medullary substance present - subject N = 24 (n=12 FXTAS and n=12 controls). Some striatal blocks/sections used here were generated in a previous study, and a few did not contain all 3 striatal subregions of interest (outlined in Table 1), Accordingly, the final subject N for each subregion were as follows: N= 23 for putamen (n=12 FXTAS and n=11 controls), N=18 for caudate (n=9 FXTAS and n=9 controls), and N=18 for internal capsule (n=9 FXTAS and n=9 controls).

Figure 1. Tissue and Reactive Astrogliosis Scoring.

(A) Postmortem human brains from FXTAS and control subjects were utilized, with tissue from (B) striatum and (C) cerebellum dissected, sectioned, and stained for GFAP. Astrocytes were quantified within striatal grey (caudate and putamen) and white matter (internal capsule), and within cerebellar grey (granule cell layer) and white matter (medullary substance) (4x montage). (D) Reactive astrogliosis severity was scored in all four subregions using an established I-IV scale based on GFAP stain intensity, soma size and shape, and branch thickness, as outlined here (40x images, scale bar =20μm).

Immunohistochemistry

Striatal and cerebellar sections from all subjects were stained for Glial Fibrillary Acidic Protein (GFAP) using enzymatic immunohistochemistry for reactive gliosis, astrocyte density, and fragment analyses. Sections were also stained for Aldehyde Dehydrogenase 1 Family Member L1 (ALDH1L1) - to confirm alterations in astrocyte density detected with GFAP staining. Tissue underwent heat mediated antigen retrieval in a DIVA antigen decloaker (Biocare Medical). Endogenous peroxidases were blocked using 3% hydrogen peroxide (Fisher), and non-specific antibody binding was blocked using 10% donkey serum. Sections were incubated in GFAP antibody (1:400, Agilent Dako Z0334) or ALDH1L1 antibody (1:1000, Abcam AB190298) overnight, washed, then incubated in biotinylated secondary antibody for 1.5 hours. Signal was amplified using an ABC kit (vector labs) and developed with DAB (ImmPACT DAB Substrate Kit, Vector SK-4105). Staining was enhanced for ALDH1L1 using nickel intensification. Sections were dehydrated and coverslipped with permount (Sigma).

A separate set of FXTAS and control sections were assessed for the presence of Cleaved-Caspase 3+ astrocytes using double label immunofluorescence. Here, sections underwent antigen retrieval (DIVA, Biocare Medical), were blocked in 10% donkey serum, incubated in rabbit anti Cleaved-Caspase 3 (1:25, Asp-175, Cell Signaling Technology, #9661) and mouse anti GFAP (1:25, Agilent Dako M0761) overnight, then incubated in the appropriate secondary antibodies – AlexaFluor-488 anti-rabbit (1:200, Thermofisher A-21206) and AlexaFluor-594 anti-mouse (1:400, Thermofisher A-21203). Sections were counterstained with DAPI, and autofluorescence was quenched using TrueBlack (Biotium #23007). Sections were cover slipped with Mowiol (Sigma-Aldrich #81381).

Imaging and Quantification

GFAP stained sections were imaged on an Olympus BX61 Microscope. For each subject, three images were taken from each region of interest (Fig. 1B-C: Putamen, Caudate, Internal Capsule, Cerebellar Granule Cell Layer, and Cerebellar Medullary Substance). Each image was a montage with consistent dimensions (0.69mm²) acquired as a z-stack (3 layers spaced 1.5um apart) with a 40x objective. Images were acquired by one investigator (TB) blind to diagnostic group. Double label immunofluorescence sections (GFAP/Cleaved-Caspase 3) from control and FXTAS striatum and cerebellum were imaged on a Nikon A1 Confocal microscope. All images included are maximum projection z-stacks and were acquired at either 40x or 60x.

Subregions were scored for reactive gliosis severity, the prevalence of GFAP+ fragmentation, and astrocyte density (separately with two markers: GFAP and ALHD1L1) by a single blind rater (BD). All slides and images were coded. For each measure - scores for each subject reflect the average across 3 images. Reactive astrogliosis severity was scored following an established I to IV scoring system²¹, based largely on astrocyte morphology and staining intensity using ImageJ cell counter. Criteria for reactive scores are outlined in detail in Figure 1D, ranging from Score I (resting – light staining with small soma) to Score IV (severely reactive – intense staining, enlarged round soma, processes absent or fragmented). Only GFAP+ astrocytes with a clear soma/centrality were scored. Astrocyte density was calculated as the total of all astrocytes (I-IV) per field, and divided by the dimensions of the image (0.69mm²). The prevalence of GFAP+ fragments were also holistically scored by a single blind rater (BD) on a 0–4 scale, representing the proportion of the image field containing GFAP fragments (0 = 0% of field, 1 = 1–25%, 2 = 26–50%, 3 = 51–75%, 4 = 76–100%).

Statistics

We performed statistical analyses in JMP 16.0.0 (SAS Institute, Cary, NC), using a repeated measures mixed linear regression model to assess differences in reactive astrocyte, astrocyte density, and fragment scores between FXTAS subjects and matched controls. Mean scores (averaged across three images for each ROI) from each region (Putamen, Caudate, Internal Capsule, Cerebellar Granule Cell Layer, and Medullary Substance) were included in the same analysis as a repeated measure. Our final model included the random effect of subject (nested within diagnosis) and the fixed effects of diagnosis, (brain) region, and diagnosis*region interaction. Restricted Maximum Likelihood (REML) approach was utilized for fit. We first fit three separate full models, in addition to the final reduced model mentioned above, which separately included Sex, Age, or PMI; Fixed effects of Sex, Age, and PMI had no effect on any astrocyte measures in any brain region (alone or interactions, p>.10 for all), and thus were omitted from the final model. Pre-planned contrasts, utilizing F-tests, were carried out to assess differences between FXTAS and control astrocyte measures for each ROI, with critical alpha Bonferroni adjusted to account for multiple pairwise comparisons.

RESULTS

To better understand regional variation in reactive astrogliosis, and other aspects of possible astrocyte pathology in FXTAS, we assessed GFAP stained sections from 12 FXTAS subjects and 12 controls (CTL) that were matched for age (CTL: 69.8 ± 2.9 yrs., FXTAS: 73.7 ± 3.3 yrs., p=.801), sex (CTL: 8M/4F, FXTAS: 8M/4F), and postmortem interval (CTL: 64.0 ± 13.0 hrs., FXTAS: 60.6 ± 26.1 hrs., p=.798). Subjects are outlined in Table 1. Astrocytes were quantified in multiple subregions, including striatal grey matter (GM) (caudate and putamen), striatal white matter (WM) (internal capsule), cerebellar GM (granule cell layer) and cerebellar WM (medullary substance). Subregions were scored for reactive gliosis severity, the prevalence of GFAP+ fragmentation, and astrocyte density. Each measure was analyzed using a repeated measures regression model and all pairwise comparisons between CTL and FXTAS subjects within each region were made using post-hoc contrasts. To account for multiple pairwise comparisons, critical alpha was Bonferroni adjusted for striatum (3 subregions, α=.017) and cerebellum (2 subregions, α =.025). We detected striking differences between GM and WM astrocyte pathology in most analyses, including differences in reactive astrogliosis and astrocyte density. This GM/WM dichotomy was largely consistent between our two structures of interest – striatum and cerebellum. All results - including means, standard errors, and p-values are listed for all measures in Table 2.

Table 2.

Astrocyte Data Summary

					Control			FXTAS
Measure	Stain	ROI	Region	Type	n	Mean	SE	n	Mean	SE	Difference	α	p-value
% Reactive	GFAP	CD	Striatum	GM	9	23.4%	3.8%	9	44.2%	8.0%	+20.8%	0.017	0.006
% Reactive	GFAP	PUT	Striatum	GM	11	22.5%	4.5%	12	47.8%	6.7%	+25.3%	0.017	0.010
% Reactive	GFAP	IC	Striatum	WM	9	34.9%	3.7%	9	56.8%	9.5%	+21.9%	0.017	0.010
% Reactive	GFAP	GCL	Cerebellum	GM	12	15.8%	3.4%	12	24.0%	7.6%	+8.2%	0.025	0.350
% Reactive	GFAP	MS	Cerebellum	WM	12	30.6%	5.2%	12	74.1%	7.2%	+43.5%	0.025	<.001
% Severe	GFAP	CD	Striatum	GM	9	2.0%	1.4%	9	5.2%	3.3%	+3.2%	0.017	0.361
% Severe	GFAP	PUT	Striatum	GM	11	0.4%	2.0%	12	13.1%	6.9%	+12.7%	0.017	0.196
% Severe	GFAP	IC	Striatum	WM	9	3.7%	2.4%	9	34.8%	12.5%	+31.1%	0.017	<.001
% Severe	GFAP	GCL	Cerebellum	GM	12	0.1%	0.1%	12	7.9%	6.5%	+7.8%	0.025	0.386
% Severe	GFAP	MS	Cerebellum	WM	12	7.7%	3.9%	12	51.7%	11.3%	+44.0%	0.025	<.001
Fragment	GFAP	CD	Striatum	GM	9	0.30	0.22	9	0.63	0.29	+0.33	0.017	0.242
Fragment	GFAP	PUT	Striatum	GM	11	0.06	0.04	12	0.94	0.31	+0.88	0.017	0.091
Fragment	GFAP	IC	Striatum	WM	9	0.30	0.22	9	1.74	0.52	+1.44	0.017	0.003
Fragment	GFAP	GCL	Cerebellum	GM	12	0.03	0.03	12	0.58	0.32	+0.55	0.025	0.221
Fragment	GFAP	MS	Cerebellum	WM	12	0.81	0.43	12	2.61	0.48	+1.80	0.025	<.001
Cell Density	GFAP	CD	Striatum	GM	9	122.0	7.7	9	139.3	17.2	+14.2%	0.017	0.613
Cell Density	GFAP	PUT	Striatum	GM	11	110.0	10.4	12	103.4	12.2	−6.1%	0.017	0.796
Cell Density	GFAP	IC	Striatum	WM	9	145.6	10.1	9	89.3	10.2	−38.7%	0.017	0.005
Cell Density	GFAP	GCL	Cerebellum	GM	12	245.6	20.2	12	220.6	20.0	−10.2%	0.025	0.180
Cell Density	GFAP	MS	Cerebellum	WM	12	157.8	10.7	12	106	13.2	−32.2%	0.025	0.007
Cell Density	ALDH	CD	Striatum	GM	9	162.3	6.6	9	170.8	12	+5.2%	0.017	0.613
Cell Density	ALDH	PUT	Striatum	GM	11	145.6	5.1	12	144.1	11	−1.0%	0.017	0.903
Cell Density	ALDH	IC	Striatum	WM	9	155.8	6.1	9	92.1	12.1	−40.2%	0.017	<.001
Cell Density	ALDH	GCL	Cerebellum	GM	12	204.3	5.9	12	199	9.8	−2.6%	0.025	0.682
Cell Density	ALDH	MS	Cerebellum	WM	12	204.8	11.7	12	97.1	11.7	−52.6%	0.025	<.001

% Reactive – Percent of astrocytes in reactive state (II, III, IV)

% Severe – Percent of astrocytes in severe reactive state (IV)

Fragment – Scored proportion of field with GFAP+ fragments (0 = 0%, 1 = 1–25%, 2 = 26–50%, 3 = 51–75%, 4 = 76%−100%)

Cell Density – Number of positive cells/mm²

GFAP – Glial Fibrillary Acidic Protein

ALDH – Aldehyde Dehydrogenase 1 Family Member L1

CD – Caudate

PUT – Putamen

IC – Internal Capsule

GCL – Granule Cell Layer

MS – Medullary Substance

GM – Grey Matter

WM – White Matter

Difference – Reflects difference in FXTAS counts/scores relative to controls (+ = FXTAS increase)

Reactive Astrogliosis is Increased in both FXTAS Grey and White Matter

We first tested the hypothesis that there is an increase in reactive astrogliosis in the FXTAS brain, predicting that severity would vary between subregions. A reactive astrogliosis score was calculated which represents the proportion of all astrocytes in any reactive state (II-IV). We found significant fixed effects across all factors, including diagnosis (F_1,21=12.07, p=.002), Region (F_4,75=21.62, p<.001), and Diagnosis*Region (F_4,75=5.40, p<.001). Within striatum, FXTAS subjects showed significant increases in reactive astrogliosis scores that was similar in magnitude across all subregions, including both GM (Fig. 2I, +20.8% in Caudate, +25.3% in Putamen, p=.010 for both) and WM (Fig. 2I, +21.9% in Internal capsule, p=.010). Within cerebellum, reactive astrogliosis scores were significantly elevated in FXTAS WM (Fig. 3I, +43.5% in Medullary substance, p<.001) but not GM (Fig. 3I, +8.2% in Granule Cell Layer, p=.350). Representative images illustrating qualitative differences in reactive astrogliosis are shown in Figure 2A-H for striatum and Figure 3A-H for cerebellum. As had been described previously, large visible patches of increased GFAP immunoreactivity were present in FXTAS subjects, which are most visible in both striatal GM (Fig. 2B) and WM (Fig. 2D). Patches were also present in cerebellar WM (Fig. 3D), but largely absent in cerebellar GM (Fig. 3B). Astrocytes processes wrap around brain vessels, providing structural integrity and regulate blood flow and solute entry into parenchyma ²². These perivascular astrocytes and their lumen-surrounding processes are often visible with GFAP staining, as illustrated in control cerebellar WM (Fig. 4A-C). FXTAS WM often contained reactive perivascular astrocytes and processes (Fig. 4D-K).

Figure 2. Striatum - Reactive Astrogliosis and Astrocyte Density.

Representative images from GFAP stained striatum (40x, montage top row) show visible increases in reactive astrogliosis in FXTAS grey (putamen: B,F / caudate: not shown) and white matter (internal capsule: D,H) relative to controls (A,E/C,G). FXTAS subjects showed a significant increase in (I) the proportion of astrocytes with a reactive phenotype of any severity (II-IV) in both striatal grey (putamen and caudate) and white matter (internal capsule), (J) the proportion of severely reactive (IV) astrocytes in striatal white matter, and (K) the severity of GFAP+ fragmentation within striatal white matter (fragment scores represent proportion of field containing dense GFAP+ puncta). (L) Astrocyte density is significantly reduced in FXTAS striatal white matter, but not grey matter, relative to controls. Striatal tissue was also stained with the astrocyte marker ALDH1L1 (M-P), confirming that there is (Q) a significant reduction in astrocyte density in FXTAS cerebellar white matter. Mean ± SE shown, Bonferroni adjusted α=.017. Each datapoint represents a single subject’s score, and all scores represent counts averaged across three images per subject per sub-region). N = 18 (n=9 FXTAS, n=9 control) for caudate and internal capsule. N= 23 for putamen (n=12 FXTAS, n=11 control).

Figure 3. Cerebellum - Reactive Astrogliosis and Astrocyte Density.

Representative images from GFAP stained cerebellum (40x, montage top row) shows visible increases in reactive gliosis in FXTAS cerebellar white matter (medullary substance, D,H) relative to controls (C,G), but not in grey matter (granule cell layer: A-B, E-F). FXTAS subjects showed a significant increase in (I) the proportion of astrocytes with a reactive phenotype of any severity (II-IV) in cerebellar white matter but not in grey matter, (J) the proportion of severely reactive (IV) astrocytes in cerebellar white matter only, and (K) the severity of GFAP+ fragmentation within cerebellar white matter only (fragment scores represent proportion of field containing dense GFAP+ puncta). (L) Astrocyte density is significantly reduced in FXTAS cerebellar white matter, but not grey matter, relative to controls. Cerebellar tissue was also stained with the astrocyte marker ALDH1L1 (M-P), confirming that there is (Q) a significant reduction in astrocyte density in FXTAS cerebellar white matter. Mean ± SE shown, Bonferroni adjusted α=.025. All scores represent counts averaged across three images per subject. Each datapoint represents a single subject’s score, and all scores represent counts averaged across three images per subject per sub-region. N = 24 (n=12 FXTAS, n=12 control) for granule cell layer and medullary substance.

Figure 4. Perivascular Astrogliosis.

Representative images from GFAP stained cerebellar and striatal white matter. (A) Perivascular astrocyte processes are visible in control cerebellar white matter with (C) a cross-sectional [arrows] and (B) transverse orientation [arrowheads]. Low magnification of FXTAS (D, G) cerebellar and (J) striatal white matter, illustrating (E,F) particularly severe perivascular astrocyte reactivity, as well as reactive perivascular fibers in (H,K) cross sectional and (I,L) transverse orientation.

Severe Reactive Astrogliosis (IV) occurs prominently in FXTAS white matter

Severely reactive astrocytes (score IV) show a set of distinctive characteristics that differentiates them from less severe reactivity (II or III) - including very intense immunoreactivity and associated dark staining, marked enlargement and rounding of the soma, and often a lack of processes (Fig. 1D). To assess differences in severe reactive astrogliosis, we calculated the proportion of score IV astrocytes relative to all astrocytes (I-IV) within each image.

Severe reactive astrogliosis was significantly increased in FXTAS subjects (Diagnosis: F_1,20=9.25, p=.006), and is particularly severe in WM (Fig. 2H and Fig. 3H, Diagnosis*Region: F_4,74=6.68, p<.001). In control subjects, the presence of severe reactive gliosis was exceedingly rare across all subjects and regions. There was a significant increase in severe reactive astrogliosis scores in FXTAS striatal WM (Fig. 2J, +31.1% in Internal Capsule, p<.001) and cerebellar WM (Fig. 3J, +44.0% in Medullary, p<.001). In GM, severe reactive gliosis severity was elevated in FXTAS subjects but not significantly different from controls, in both striatum (Fig. 2J, +12.7% Putamen [p=.196] and +3.2% in Caudate [p=.361]) and cerebellum (Fig. 3J, +7.8% in Granule cell layer, p=.386).

Increased presence of GFAP+ fragments in FXTAS white matter

Many sections present in the analyses here also showed the widespread presence of fragmented GFAP+ puncta, which often occurred in high density clusters, across the ROI (Examples in Fig. 5F-O, Fig. 6F-O). We next assessed whether these fragments were more common in FXTAS subjects. All images used for reactive astrogliosis quantification were given a holistic score (0–4) representing the prevalence of GFAP+ puncta within the quantification field in the following way: 0 (Absent - 0% of field contained fragments), 1 (1 – 25%), 2 (26 – 50%), 3 (51 – 75%), and 4 (76 – 100%).

Figure 5. Astrocyte Degeneration in Striatal White Matter.

(A) GFAP stained striatal white matter typically reveals numerous resting astrocytes (B,C) and abundant fibers (D), even in areas of low immunoreactivity (E), as illustrated in control subject. In contrast, FXTAS subjects (F,K) showed a consistent but markedly different appearance with fewer astrocytes, a high proportion of dysmorphic reactive (G,L) and fragmented (H,M) astrocytes, large areas filled with clusters of GFAP+ fragments reminiscent of apoptotic blebbing (I,N). Even in areas with low immunoreactivity, blebbing was typically present, although GFAP+ astrocytes were absent (J,O).

Figure 6. Astrocyte Degeneration in Cerebellar White Matter.

(A) GFAP stained cerebellar white matter typically reveals numerous resting astrocytes (B,C) and abundant fibers (D), even in areas of low immunoreactivity (E), as illustrated in a control subject. In contrast, FXTAS subjects (F,K) showed a consistent but markedly different appearance with fewer astrocytes, a high proportion of dysmorphic reactive (G,L) and fragmented (H,M) astrocytes, large areas filled with often clustered GFAP+ fragments reminiscent of apoptotic blebbing (I,N). Even in areas with low immunoreactivity, blebbing was typically present, although GFAP+ astrocytes were absent (J,O).

GFAP+ fragment scores were significantly elevated in FXTAS subjects (F_1,21=8.97, p=.007), again occurring in a region-specific manner (Region: F_4,75=11.75, p<.001 / Diagnosis*Region: F_4,75=2.82, p=.031). GFAP Fragment scores were significantly increased in FXTAS WM, in both striatum (Fig. 2K, +1.44 in Internal Capsule, p=.003) and cerebellum (Fig. 3K, +1.80 in Medullary substance, p<.001). In striatal GM, we found a statistically trending but smaller magnitude increase in putamen (Fig. 2K, +0.88, p=.091), and a non-significant increase in caudate (Fig. 2K, +0.33, p=.242). A small non-significant increase was also detected in cerebellar GM (Fig. 3K, +0.55 in Granule cell layer, p=.221).

Reduced Astrocyte Density in FXTAS white matter

We next assessed whether there are alterations in astrocyte number in FXTAS subjects relative to controls. To do so, we calculated the density of GFAP+ astrocytes in all regions. In contrast with rodents which show poor expression of GFAP in GM astrocytes²³, GFAP is a highly specific pan-astrocyte marker in human tissue expressed constitutively in both WM and GM²⁴, and is thus suitable for assessing alterations in astrocyte number. We discovered a large and significant reduction in astrocyte density in the FXTAS brain (Diagnosis: F_1,22=4.82, p=.039) which occurred in a region-specific manner (Region: F_4,77=38.51, p<.001 / Diagnosis*Region: F_4,77=2.82, p=.031). As with other measures, reduced astrocyte density occurred specifically within FXTAS WM - we found a significant 38.7% reduction in striatal WM (Fig. 2L, Internal Capsule, p=.005) and a 32.2% reduction in cerebellar WM (Fig. 3L, Medullary substance, p=.007). There was not a significant difference in astrocyte density in either FXTAS striatal GM (Fig. 2L, 6.1% decrease in Putamen [p=.796], 14.2% increase in Caudate [p=.613]) or in FXTAS cerebellar GM (Fig. 3L, 10.2% decrease in Granule cell layer, p=.180).

To confirm that astrocyte density is reduced in FXTAS WM, we stained all sections with another pan astrocyte marker (Aldehyde Dehydrogenase 1 Family Member L1, ALDH1L1)²⁴, and again quantified astrocyte density in all subregions. Unlike GFAP which is predominantly localized to astrocyte fibers, ALDH1L1 is predominantly localized to astrocyte soma, facilitating a more accurate count. Again, we found a large and significant reduction in astrocyte density in the FXTAS brain (Diagnosis: F_1,22=15.10, p<.001) which occurred in a region-specific manner (Region: F_4,77=25.11, p<.001 / Diagnosis*Region: F_4,77=19.28, p<.001). Representative images are shown for striatum in Figure 2M-P and for cerebellum in Figure 3M-P. Again, we found large and significant reductions in ALDH1L1+ astrocytes that were specific to WM, including a 40.2% reduction in striatum (Fig. 2Q - Internal Capsule, p<.001) and a 52.6% reduction in cerebellum (Fig. 3Q - Medullary Substance, p<.001). FXTAS GM astrocyte densities were not significantly different from controls in striatum (−1% in Putamen [p=.903], +5.2% in caudate [.613]) or cerebellum (−2.6% in Granule cell layer, p=.682)

Qualitative Pathological Indicators of Astrocyte Degeneration in FXTAS

After completing all included quantitative analyses, and now realizing that there may be astrocyte degeneration occurring in FXTAS WM as indicated by reduced astrocyte density, we broke blinding to qualitatively characterize the GFAP+ signal. Representative images from striatal and cerebellar WM appear in Figure 5 and Figure 6, respectively. We found qualitative indicators of severe astrocyte degeneration in FXTAS – with large patches of tissue showing a visible reduction in astrocyte soma density (Figs. 5F,K and 6F,K), a predominantly dysmorphic and beaded appearance of astrocytes that are present (Figs. 5G,L and 6G,L), neuropil filled with dense clusters of GFAP+ fragments (Figs. 5I,N and 6I,N), and some patches that are mostly absent of signal (Figs. 5J,O and 6J,O). The widespread appearance of GFAP+ puncta, which sometimes emanate from hypertrophied astrocytic processes, are reminiscent of apoptotic bodies. Dense clusters also occur that are lacking a clearly defined astrocyte soma, but may represent degeneration of the cell body while only apoptotic bodies remain (Figs. 5H,M and Fig. 6H,M). In contrast, control WM shows a relatively uniform distribution of predominantly resting fibrous astrocytes (Figs. 5A-C and 6A-C), with neuropil typically filled with visible but light GFAP+ fibers (Figs. 5D-E and 6D-E).

Confirmation of apoptosis with Cleaved Caspase 3

As the data thus far was highly suggestive of severe astrocyte degeneration in FXTAS WM, we speculated that widespread GFAP+ fragments were possibly apoptotic bodies. We hypothesized that if astrocyte degeneration is occurring in FXTAS during disease progression, we should be able to detect at least a subset of astrocytes actively undergoing degeneration at the time of donor’s death using apoptosis or other cell death markers. Thus, to further confirm that astrocyte degeneration occurs in FXTAS WM, we performed double label immunofluorescence for the apoptosis marker cleaved caspase 3 (CC3) in combination with the astrocyte marker GFAP.

Numerous CC3+ astrocytes were present in FXTAS striatal (Fig. 7A) and cerebellar WM (Fig. 7B), but absent in controls (Fig. 7A-B, bottom rows). Apoptotic CC3+ astrocytes were typically mildly to moderately reactive (II-III), rarely severely reactive (IV), and were exceedingly rare or absent in FXTAS GM. When present, CC3 was almost always localized to the astrocyte soma and processes, although nuclear localization was sometimes present. We also found discrete clusters of GFAP+ fragments that were also CC3+ (Fig. 7C), highly suggestive that the large swaths of GFAP+ fragments in FXTAS WM represent residual astrocyte apoptotic bodies. CC3 staining was widespread in brain vasculature, some of which included CC3+/GFAP+ perivascular astrocytic processes in FXTAS WM (Fig. 7D), suggestive of possible blood-brain barrier breakdown. Other CC3+ cells were visible in FXTAS tissue that could be easily identified due to localization and morphology, including Purkinje cells and large neurons in the cerebellar dentate nucleus.

Figure 7. Cleaved Caspase 3 positive astrocytes.

Numerous GFAP+ astrocytes in FXTAS (A) striatal and (B) cerebellar white matter were also positive for the apoptosis marker cleaved caspase 3. Representative images from two FXTAS subjects in top rows, control on bottom row. In most astrocytes, cleaved caspase signal was present in both soma and processes. (C) GFAP+ puncta, reminiscent of apoptotic bodies, were widespread in FXTAS white matter and were often positive for cleaved caspase 3. (D) GFAP+ perivascular astrocyte processes were also positive for cleaved caspase 3.

DISCUSSION

While there is presumably widespread cell loss in the FXTAS brain, as indicated by volumetric loss and white matter disease (WMD) – the temporal, spatial, and cell type specific patterns of neurodegeneration remain largely uncharacterized. Here, we demonstrate for the first time, that FXTAS neuropathology includes robust degeneration of astrocytes, predominantly within white matter (WM). We provide three pieces of evidence to support this finding: 1. Reduced GFAP+ and ALDH1L1+ astrocyte density in striatal and cerebellar WM; 2. Widespread presence of GFAP+ puncta throughout WM that are reminiscent of apoptotic cell bodies; and 3. Detection of cleaved caspase 3 positive (CC3+) astrocytes and puncta in WM, indicative of astrocytes actively undergoing apoptosis in FXTAS brains at the time of death. We also confirmed that patchy astrogliosis occurs in multiple regions of the FXTAS brain, which represents distinct phenomena regionally – GM patches consist of reactive astrocytes surrounded by mostly resting astrocytes, while WM patches consist of fewer and often severely reactive astrocytes that are surrounded by GFAP+ apoptotic bodies.

Focal degeneration of astrocytes has been identified in a variety of neurodegenerative disorders, including frontotemporal dementia (FTD)²⁵ and Amyotrophic Lateral Sclerosis (ALS)²⁶. While degeneration is not severe enough to result in overall reduced astrocyte density in FTD and ALS, apoptotic astrocytes show enlarged and sometimes fragmented soma surrounded by apoptotic bodies^{25, 26}, similar to our findings here. Astrocyte pathology is also central to many Leukodystrophies (LDs), genetic neurodegenerative disorders characterized by white matter disease and cerebellar ataxia²⁷. Additionally, ischemia, epilepsy, and a variety of other brain insults can induce clasmatodendrosis, a process driven by autophagy and characterized by morphological alterations in astrocyte processes²⁸. While it is difficult to rule out clasmatodendrosis conclusively in FXTAS here, as specific immunohistochemical markers are lacking, we did not observe the characteristic vacuolation²⁹ and linear patterns of beaded varicosities²⁸. Instead, the morphological changes we detected are more similar to the focal/clustered GFAP+ apoptotic bodies and fragmented astrocyte soma associated with astrocyte degeneration in FTD²⁵ and ALS²⁶, but occurring on a larger scale. Our finding of astrocyte degeneration here is not only supported by the presence of apoptotic markers, which are not typically present in clasmatodendrosis^{28, 30}, it is also consistent with existing FXTAS literature. In a recent bioinformatics study, Dias et al.³¹ found that astrocytes were underrepresented in nuclei isolated from frozen frontal cortex in FMR1 premutation carriers. As nuclear collection was unbiased, this likely reflects reduced astrocyte abundance in the initial samples - consistent with our finding of FXTAS astrocyte degeneration. In another recent bioinformatics study³², the most severe proteomic alterations detected in FXTAS brain homogenates includes more than a two-fold reduction in proteins expressed predominantly or exclusively in astrocytes (CD38 and Tenascin C) – also likely reflecting bulk astrocyte loss.

The specific cell and molecular mechanisms driving FXTAS WM astrocyte degeneration remain unclear. Across cell types, FMR1 mRNA overexpression is believed to be the molecular origin of cell autonomous FXTAS neurotoxicity³³. However, consensus is lacking on how this toxicity further contributes to disease progression at a molecular and cellular level², up to cell death. The leading proposed mechanisms predict that: 1. FXTAS intranuclear inclusions sequester important intracellular proteins, leading to impairments in cell function³⁴, 2) FMR1 overexpression leads to DNA r-loop formation and a sustained DNA damage response that becomes toxic³⁵, and 3) Repeat Associated Non-AUG (RAN) translation causes formation of toxic protein species, particularly FMRpolyG³⁶. FXTAS inclusions are largest and most prevalent in astrocytes¹³, rendering them particularly vulnerable to possible inclusion-associated toxicity. Astrocyte specific FMR1 premutation expression in mice causes both FMRpolyG production and inclusion formation³⁷, as well as motor symptomology, suggesting that RAN translation may also be an important intermediary. It’s unclear why WM astrocytes are more vulnerable than those in GM, given that astrocyte inclusion burden is similar or possibly even reduced in WM relative to GM^{13, 38}. There are no known differences in FMR1 expression or toxicity in WM relative to GM astrocytes. However, WM astrocytes do have metabolic differences, including reduced basal glycogen and ATP content and a more reduced NADH/NAD+ redox ratio than GM astrocytes³⁹. It’s possible that these differences render WM astrocytes less able to cope metabolically with FMR1 toxicity. In addition to astrocytes, WM is composed predominantly of oligodendrocytes and myelinated axons, as well as vasculature and microglia. There is qualitative evidence of axonal and myelin loss in FXTAS¹³, although it is not known how severe or widespread these pathologies are. Unlike astrocytes and neurons, oligodendrocytes do not form FXTAS inclusion bodies¹³, suggesting that they may escape cell autonomous FMR1 toxicity and neurodegeneration.

WMD is severe and widespread in FXTAS, consisting of regional atrophy and the presence of focal WM hyperintensities (WMH) detected by T2 FLAIR MRI^{2, 11, 40}. WMD is sometimes considered primum movens⁴¹, as it’s occurrence tracks with FXTAS onset and progression in PM carriers^{42, 43}, that neuropathological analyses fail to detect significant grey matter neuronal degeneration or cortical thinning⁴⁴, and that the severity of WMD alone correlates well with FXTAS motor and cognitive deficits^{3, 41}. Here, we have identified that astrocyte degeneration is an important component of WMD in FXTAS, possibly the key constituent in regional atrophy given the severity of loss. Although unclear for FXTAS, WMH are also strongly associated with cerebral microbleeds in other disorders⁴⁵. Microbleeds do frequently occur in FXTAS, particularly in WM⁴⁶, suggesting that FXTAS WMH also reflect vascular disease and cerebral bleeding. Cerebrovasculature is comprised primarily of vascular endothelial cells (VECs) that form vessel walls, in combination with pericytes (PCs) that regulate blood flow and the entry of peripheral immune cells⁴⁷. Surrounding the VECs and PCs is the glia limitans perivascularis, a basement membrane that contributes to vascular structural integrity which is created and regulated by astrocytes²². Brain vessels are also wrapped by astrocytic endfeet¹⁹ - which provides an added physical barrier that helps to regulate of solute flow into brain parenchyma (blood brain barrier)²², and also enables astrocytic regulation of blood flow through pericyte contacts^{19, 22}. Here we detected widespread cleaved caspase 3 (CC3) staining in cerebrovasculature, suggesting that microbleeds may occur secondary to vascular degeneration. VECs do form FXTAS inclusion bodies⁴⁶, suggesting that they are vulnerable to degeneration. We also found numerous perivascular astrocytic processes that were CC3+, illustrating that astrocyte pathology may contribute to deficits in vascular structural integrity. It will be important to clarify the relative contributions of possible VEC, PC, and astrocyte pathology on FXTAS microhemorrhage, and how each may contribute to the T2 FLAIR WMH signal.

Since astrocytes regulate brain homeostasis broadly, providing support and regulation for most cell types in the brain - astrocyte degeneration likely has severe negative consequences. Astrocytes are key regulators of neuronal metabolism¹⁴ - providing lactate as an energy source for neurons during periods of increased activity⁴⁸, and storing glycogen which is released as an energy source for adjacent cells in times of need⁴⁹. Systemic bioenergetic deficits have been detected in FXTAS⁵⁰, presumably reflecting widespread intracellular metabolic impairment due to FMR1 mRNA toxicity⁵¹. Intracellular deficits in metabolism, combined with a loss of astrocytic metabolic support, may synergistically result in severe pathology or even degeneration of other cell types. Additionally, astrocytes regulate alterations in regional blood flow, augmenting oxygen and glucose uptake, in response to brain activity¹⁹. Thus, astrocyte loss would negatively impact neuronal and glial metabolism in multiple ways, providing a secondary insult in addition to intracellular stress due to FMR1 toxicity. Brain pH dysregulation due to loss of astrocytic control¹⁶ would invariably lead to extracellular stress and possible generation of reactive oxygen species (ROS)⁵², both implicated in FXTAS disease progression⁵¹. Our perivascular astrocyte pathology findings here, including CC3+ perivascular astrocytic endfeet and perivascular astrogliosis, suggests that BBB integrity may be impaired in FXTAS which would likely cause dysregulation of the chemical composition of brain parenchyma. Astrocytes are also essential for oligodendrocyte survival and maturation during development, and for maintenance of myelination in the adult brain⁵³. Focal ablation of astrocytes in the adult spinal cord results in robust demyelination and impaired myelin compaction⁵³, thus providing a possible mechanism for FXTAS demyelination¹³.

We confirmed that reactive astrogliosis occurs in the FXTAS cerebellum and striatum, but that there are regional differences in its presentation. Astrogliosis was consistently present in WM, including increased abundance of severe reactive phenotype (type IV) which was lacking in GM. Type IV astrocytes may represent a morphological phenotype that precedes degeneration, considering that astrocyte loss was limited to WM. We detected inconsistencies in GM astrogliosis, which was present in caudate and putamen but absent from the cerebellar granule cell layer (GCL). Neuronal degeneration has only been detected in specific neuronal populations in FXTAS (e.g. cerebellar Purkinje neurons⁵⁴, possibly cortical interneurons³¹), with investigators failing to detect widespread degeneration of neurons or oligodendrocytes in the FXTAS brain^{13, 31, 38}. It is possible that regional differences in GM reactive astrogliosis reflects selective loss of specific neuronal subpopulations. There is evidence of axonal degeneration in FXTAS WM¹³, which could contribute to reactive astrogliosis in a cell non-autonomous manner. However, we cannot rule out that reactive astrogliosis represents FMR1 overexpression induced cell autonomous pathology in astrocytes. It will be important to further assess the relationship of FMR1 expression levels, RAN translation, and the absence/presence of inclusions on the reactive phenotype of astrocytes in future studies.

While reactive astrogliosis typically occurs as a response to brain injury or trauma with characteristic transcriptional and morphological alterations, recent reevaluations suggest that two reactive isoforms exist: A1 astrocytes that are pro-inflammatory and potent inducers of apoptosis in adjacent cells, while A2 astrocytes are neuroprotective and anti-apoptotic²⁰. It is unclear whether one form predominates in FXTAS. TNF, a potent inducer of the A1 phenotype, is significantly increased in the FXTAS brain⁵⁵, suggesting that the A1 phenotype may predominate. A1 astrocytes cease supportive functions of other cells, including metabolic support of neurons, synaptic modulation, and neurotrophic factor secretion²⁰. A1 and A2 phenotypes are not distinguishable by GFAP staining alone; future studies should utilize A1 and A2 specific markers to clarify which form predominates. Additionally, astroglial senescence (AS) occurs in healthy aging⁵⁶, at increased levels in neurodegeneration⁵⁷, and could plausibly contribute to FXTAS neuropathological progression. Although AS causes similar morphological alterations to typical reactive astrogliosis, there is a specific senescence-associated secretory phenotype (SASP) that includes release of pro-inflammatory cytokines (Interleukins 6 and 1β, TNFα) and tissue remodeling factors (Matrix Metalloproteinases) and reduced secretion of neurotrophic factors (nerve growth factor [NGF] and insulin-like growth factor 1 [IGF1])⁵⁸. SASP can be induced by DDR and ROS, both of which are induced by FMR1 mRNA toxicity⁵⁶. Loss of the nuclear lamina protein Lamin B1 is a marker for cellular senescence⁵⁹. Interestingly, FMRpolyG causes disruptions of nuclear laminae⁶⁰, and Lamins A/C become aggregated in FXTAS inclusion bodies ⁶¹. It’s an intriguing possibility that astroglial senescence could precede and facilitate nuclear breakdown and inclusion formation in FXTAS. It will be important to assess whether severe (type IV) reactive phenotype corresponds with A1 or A2 astrocytes or AS.

Our data here identifies for the first time that severe astrocyte pathology occurs in FXTAS, as characterized as increased levels of reactive astrogliosis and profound WM astrocyte degeneration. Understanding the time course of astrogliosis and degeneration, the relative loss of astrocytes to other cell types, and the intersection of astrocyte pathology with other aspects of known FXTAS neuropathology will be essential to characterize in future studies. The specific mechanisms that underlie both the reactive and degenerative phenotype of astrocytes in FXTAS remains undefined. However, the increased relative abundance of astrocyte inclusion formation suggests that cell autonomous astrocyte pathology due to inclusion formation³⁴, RAN translation³⁶, and/or alterations in DNA-damage repair³⁵ is a likely contributor, given that inclusion formation tracks with conversion to FXTAS in PM carriers¹³. As outlined above, it remains unclear whether reactive astrogliosis in FXTAS represents a pathological phenotype itself, or rather an adaptive response to broad FXTAS neurodegenerative processes such as neuron degeneration⁵⁴, iron accumulation⁶², or microbleeds⁴⁶. In contrast, astrocyte degeneration is clearly a pathological phenotype, compromising the critical function of astrocytes in maintaining brain homeostasis¹⁴, and thus their loss may exacerbate other aspects of known FXATS neuropathology. For example - FXTAS WM disease, which includes spongiosis¹³ and volumetric loss¹¹, may be driven directly by the profound loss of astrocytes as they are a predominant cellular component of this tissue. Tenascin C, an important extracellular matrix protein that is synthesized by astrocytes and present predominantly in WM⁶³, is reduced in FXTAS³² and in turn further contribute to the WM disease phenotype. Astrocyte degeneration may contribute to microbleeds⁴⁶ or other types of vascular pathology in FXTAS, due to a loss of astrocyte mediated vascular structural support and blood brain barrier (BBB) integrity¹⁹. A loss of BBB integrity would also likely facilitate iron accumulation in the brain⁶⁴, common in FXTAS⁶². A loss of astrocytic metabolic support for other surrounding cells¹⁴ likely causes cellular stress to other cell types that are already compromised due to FMR1 associated toxicity, possibly leading to degeneration. Together, our current findings together emphasize the importance of astrocyte pathology in FXTAS neurodegeneration, particularly as a contributor to white matter disease, and also point to numerous areas that will need to be clarified in future studies. Additionally, these findings suggest that astrocyte targeted interventions may be an important avenue for FXTAS therapeutics moving forward.

SUMMARY FOR SOCIAL MEDIA IF PUBLISHED.

1. If you and/or a co-author has a Twitter handle that you would like to be tagged, please enter it here.

   • No twitter handle for authors.
2. What is the current knowledge on the topic?

   • Astrocytes show uniquely high inclusion burden in FXTAS, and there is existing evidence of patchy reactive astrogliosis.
3. What question did this study address?

   • This study further characterized regional patterns in reactive astrogliosis. This study also assessed the possibility that astrocyte degeneration is a part of FXTAS neuropathology.
4. What does this study add to our knowledge?

   • This study is the first to identify that robust astrocyte degeneration occurs specifically in FXTAS white matter. Additionally, this study shows regional differences in reactive astrogliosis in the striatum and cerebellum, which is more severe in white matter.
5. How might this potentially impact on the practice of neurology?

   • While many neurological disorders involve astrocyte pathology beyond just reactive astrogliosis (ALS, FTD, leukodystrophies, hypoxia/ischemia), this is the first disorder to our knowledge that is now known to be characterized by white matter astrocyte degeneration that is sufficient to substantially reduce cell density/counts. Given that inclusion burden disproportionately affects astrocytes in FXTAS – it may serve as an example of how astrocyte degeneration may further contribute to symptomology or neurological disease progression more broadly.

Data Availability

Datasets are available upon request