Affected astrocytes in the spinal cord of the leukodystrophy vanishing white matter

Abstract Leukodystrophies are often devastating diseases, presented with progressive clinical signs as spasticity, ataxia and cognitive decline, and lack proper treatment options. New therapy strategies for leukodystrophies mostly focus on oligodendrocyte replacement to rescue lack of myelin in the brain, even though disease pathology also often involves other glial cells and the spinal cord. In this study we investigated spinal cord pathology in a mouse model for Vanishing White Matter disease (VWM) and show that astrocytes in the white matter are severely affected. Astrocyte pathology starts postnatally in the sensory tracts, followed by changes in the astrocytic populations in the motor tracts. Studies in post‐mortem tissue of two VWM patients, a 13‐year‐old boy and a 6‐year‐old girl, confirmed astrocyte abnormalities in the spinal cord. For proper development of new treatment options for VWM and, possibly, other leukodystrophies, future studies should investigate spinal cord involvement.


| I N TR ODU C TI ON
White matter disorders (WMDs) have many causes and affect both the brain and spinal cord of the central nervous system (CNS). The peripheral nervous system is variably affected. The genetic WMDs, known as leukodystrophies, include disorders like Canavan disease, Vanishing White Matter (VWM), Pelizaeus-Merzbacher disease (PMD) and metachromatic leukodystrophy (MLD). Leukodystrophies are rare disorders that typically show progressive involvement of the white matter (WM; Ashrafi & Tavasoli, 2017). Clinical signs of leukodystrophies vary and are dependent on the age of onset; they often include loss of motor function due to spasticity and ataxia. Cognitive decline of variable severity occurs as well. Since many patients with leukodystrophies show progressive decline, there is urgent need for better treatment.
Cell replacement therapy, where populations of healthy (macro-) glial cells (oligodendrocytes and astrocytes) or their precursor cells are transplanted in the CNS, is a promising treatment strategy for leukodystrophies (Osorio & Goldman, 2016). Several glial transplantations have been performed in the brains of rodent models with WM defects, which in successful cases resulted in myelin formation and increased survival (Izrael et al., 2007;Pouya, Satarian 2011; Wang et al., 2013). However, proof-of-concept studies in animal models that accurately mimic human leukodystrophies are still lacking.
A first clinical trial, where human neural stem cells were transplanted in the frontal white matter of patients with PMD, showed no adverse effects and indications of myelination on MRI in the transplanted regions (Gupta et al., 2012). Remarkably, most transplantation studies have so far focused on the brain, while many leukodystrophies also show spinal cord involvement, such as Alexander disease (Liu et al., 2016;van der Knaap et al., 2006), PMD (Koeppen & Robitaille, 2002), MLD (Toldo, Carollo, Battistella, & Laverda, 2005) and Krabbe disease (Wang, Melberg, Weis, Mansson, & Raininko, 2007). In rodent models of spinal cord injury, glial transplantations in the spinal cord resulted in decreased pathology and functional recovery (Haidet-Phillips et al., 2014;Li et al., 2015;Nicaise, Mitrecic, Falnikar, & Lepore, 2015). Also a clinical trial, where patients with spinal cord injury were transplanted with human neural stem cells, showed moderate clinical improvements (Shin et al., 2015). These findings suggest that glial transplantations could also repair abnormalities in the spinal cord of leukodystrophies.
The spinal cord is a compact structure with a high density of ascending somatosensory tracts and descending motor tracts. Consequently, spinal cord damage, either by spinal cord injury or WMDs, is readily associated with significant neurological handicap. This is in contrast to WM abnormalities in the brain, which may remain without clinical consequences depending on location and extent. Treatment of both the spinal cord and the brain in leukodystrophy patients may be essential for restoration of proper CNS function. Intravenous injection of neural and mesenchymal stem cells in a rodent model of multiple sclerosis (MS) resulted in increased myelination and reduction of pathology in the spinal cord, together with improved locomotor function, highlighting the importance of spinal cord treatment in WMDs (Mitra, Bindal, Eng Hwa, Chua, & Tan, 2015;Zhang et al., 2016). This implies that in leukodystrophies with spinal cord involvement, prospective cellular replacement therapies should also target the spinal cord to achieve CNS regeneration.

| Human tissue
The spinal cords of two VWM patients were examined. A cross-section of cervical spinal cord was collected at autopsy from a 13-year-old male VWM patient (VWM343), with two compound heterozygous mutations in the EIF2B5 gene (c.271A > G/p.Thr91Ala and c.1208C > T/p.Ala403Val). The thoracic-level spinal cord was collected from a 6-year-old female VWM patient (VWM367) with two compound heterozygous mutations in the EIF2B5 gene (c.338G > A/p. Arg113His and c.1208C > T/p.Ala403Val). As control, the tissue of a 10-year-old patient deceased of metastasized osteosarcoma was used.
For cryo-sectioning, the spinal cord of VWM343 was fixed in 4% PFA for 2 days, after which the tissue was embedded in optimal cutting temperature (OCT) mounting solution and stored at 2808C until further processing.
The human post-mortem spinal cord tissue was collected at the VU University Medical Center in Amsterdam, the Netherlands, with approval by the Institutional Review Board and informed consent of the parents.

| Experimental animals
The VWM mice were homozygous for a mutation in the Eif2b5 gene (Eif2b5 Arg191His/Arg191His ), on a C57Bl/6J background strain (referred to as 2b5 ho ; Dooves, Bugiani, et al., 2016). Heterozygous littermates, referred to as 2b5 he , were used as controls. Mice of embryonic ages E13.5 and E18.5 were harvested, after which they were fixated in 2% paraformaldehyde for 48 hours. The tails were used for genotyping.
Adult mouse spinal cords were harvested at 4, 7.5, 8, 9, and 10 months of age, after transcardial perfusion with 4% paraformaldehyde, followed by 48 hr post-fixation. The spinal cords were embedded in OCT mounting solution (Sakura Finetek Europe, Alphen a/d Rijn, Netherlands) and stored at 2808C until further processing.
Experimental procedures involving mice were in strict compliance with animal welfare policies of the Dutch government and were approved by the IACUC of the VU University, Amsterdam. water, stained with hematoxylin, and rinsed with tap water. The sections were then differentiated in 1% acid alcohol, rinsed again with tap water, incubated in 0.1% sodium carbonate, and rinsed again with tap water. Then sections were stained with Eosin Y, followed by washing steps with tap water and milliQ water. Then the slides were dehydrated with 50%-100% alcohol, cleared in Xylene, and embedded in Depex.

| Immunofluorescence
Spinal cords of patient VWM343 and VWM mice were cryo-sectioned at 12 mm thickness and immunofluorescently labeled. To wash away the cryo-protectant, the slides were washed six times for 5 min in PBS at room temperature. To increase antibody retrieval, the slices received a microwave pre-treatment at 908C in 0.1M citrate buffer pH 6 for 10 min, after which the slides were allowed to cool to room temperature. The slides were washed one time with PBS, followed by a 1-hr blocking step with blocking buffer (PBS 1 5% NGS 1 0.1% BSA 1 0.3% Triton X-100) at room temperature. Primary antibodies against nestin (1:1000, Molecular Probes, Eugene, Oregon, U.S.) were diluted in blocking buffer, and incubated for 1.5 hr at room temperature. The secondary antibody was washed away six times for 5 min in PBS at room temperature, followed by application of nuclear marker 4',6-diamidino-2-phenylindole staining (DAPI, 1:1000) in PBS for 2 min. The tissue slides were embedded in FluormountG (Southern Biotech, Birmingham, Alabama, U. S.) and cover-slipped.

| Quantification and statistics
A Leica DM6000B microscope (Leica Microsystems) was used to take pictures. To image entire spinal cords (Figure 4), two separate photos were merged in Adobe Photoshop CS6 using the "photo merge" tool. The spinal cords were cut out from the image, using the magnetic lasso tool, and pasted on a black background. For quantification of nuclear markers Sox2, Olig2, Sox9, and Id3, photos were taken of the white matter regions of the thoracic spinal cords.
Per animal, 3-6 serial sections were analyzed. Image J software was used to outline the white matter, in which the positive cells were counted and expressed as the number of positive nuclei/mm 2 tissue.
A t test of equal means was performed to compare means between mutant and controls. All statistical tests were performed using IBM Staining for vimentin (9-month-old animals, n 5 4 for both VWM and control mice) showed co-localization with GFAP and was therefore considered astrocyte-specific (Supporting Information, Figure S1b,c).
However, vimentin did not specifically label the affected astrocyte population, as it was also expressed in astrocytes outside the affected areas and in the astrocytes of control mice (Supporting Information, Figure S1b,c). The morphology of the blood vessels in the spinal cord was assessed with the marker a smooth muscle actin (a-SMA; 9month-old animals, n 5 3 for both VWM and control mice). No abnormalities in the blood vessels were observed (Supporting Information, Figure S2). In the affected WM areas, an increased density of cells could be observed when using DAPI staining (Supporting Information, Figure S1 and 2). As these cells did not show immunoreactivity for any of the tested markers (Sox2, NFIA, Sox9, Olig2, NG2, MBP, SMI, Reelin, Neurofilament, Nestin, Id3, GFAP, S100B, CD68; data not shown) their identity is unknown. To investigate whether these cell populations consist of apoptotic cells, immunohistochemistry was performed for apoptosis marker Cleaved Caspase 3 (CC3) in combination with nestin (Supporting Information, Figure S3; 9-month-old animals, n 5 3 for both VWM and control mice). Even though some CC3-expressing cells could be observed in the spinal cord (Supporting Information, Figure   S3a,b), these were present in both the affected and unaffected areas in the VWM mice, as well as in control mice. The majority of the unidentified cells within the affected areas did not express CC3 (Supporting Information, Figure S3c).

| The embryonic VWM mouse spinal cord shows no defects in glial cell development
To study whether glial defects start during early developmental stages, we studied the spinal cord of VWM mice at E13.5 and E18.5. To identify the different neural progenitor populations, which arise during early patterning, we performed immunohistochemical stains for NFIA in combination with regional identity markers Nkx6.1, Pax6, Nkx2.  Figure S1a). However, vimentin also visualized the dysmorphic morphology of the affected astrocytes, indicating blunt and short processes (Supporting Information, Figure   S1a).

| D ISC USSION
While many leukodystrophies have spinal cord involvement, detailed investigation of the pathology is often lacking. Although spinal cord pathology may significantly contribute to neurological disease, the spinal cord is often overlooked as potential therapeutic target. In this showing that astrocytes are central in the brain pathophysiology of VWM (Bugiani et al., 2011;Dooves, Bugiani, et al., 2016). We report that pathology starts in the dorsal and lateral WM tracts, with most severe defects in the sensory tracts, followed later by the vestibulospinal motor tracts. The pyramidal motor tracts appear unaffected. We further show that the pathology starts at the thoracic level, followed by the cervical level and, later, also by the lumbar level. Recent studies indicate that astrocytes are regionally, morphologically and functionally heterogeneous (Bayraktar, Fuentealba, Alvarez-Buylla, & Rowitch, 2014;Chaboub & Deneen, 2012;Molofsky & Deneen, 2015;Schitine, Nogaroli, Costa, & Hedin-Pereira, 2015;Yoon, Walters, Paulsen, & Scarisbrick, 2017). Subpopulations might be selectively affected and might correlate with variation in WM pathology. Indeed, the brains of various leukodystrophies show variation in vulnerability, with for instance sparing of the U-fibres underneath the cortex (van der Knaap et al., 1997;van Rappard, Boelens, & Wolf, 2015). Microenvironmental factors may contribute to this heterogeneity in pathology, such as regions with neuronal subpopulations with distinct axonal signals, distribution of the vasculature bed, timing of myelination programs, and presence of other neural cells. Therefore, while many environmentderived factors that regulate myelination processes have been described , regulatory processes can be specific for particular brain and/or spinal cord regions and thus determine local pathology.
Furthermore, Nkx2.2 marks ventral regions, while Pax3-expressing cells are located in the dorsal domain. While our results show that embryonic neural tube patterning and glial specification are unaffected in VWM, we found that, in adulthood, the dorsal and lateral WM astrocytes are affected first, followed by ventral WM astrocytes. Although adult astrocytes no longer express the embryonic regional markers, earlier studies showed that the regional identity of astrocytes remains unchanged after transplantation, aging, or injury (Krencik, Weick, Liu, Zhang, & Zhang, 2011;Tsai et al., 2012). The location of the affected astrocyte subpopulation can therefore already indicate the embryonically pattered subtype. New RNA sequencing methods, such as single cell sequencing (Marques et al., 2016) and fluorescence in situ sequencing of RNA (FISSEQ; ), will soon provide more insight into the glial subtypes and hopefully insight into the basis of selective vulnerability in VWM patients.
To start autologous glial cell transplantation in the affected CNS regions, including the spinal cord, there are still issues to overcome.
Firstly, we need to clarify which type of cells requires replacement.
One strategy would be the transplantation of neural stem cells. While these cells have high proliferative capacities and migration potentials, the differentiation of multipotent stem cells upon transplantation is hard to control; they can potentially generate undesired neural cell types. As astrocytes appear most affected and the myelin sheaths look normal, the transplantation of immature astrocytes would be a promising option. However, the cellular density of Olig2-positive OPCs is increased, similar to in the brain (Bugiani et al., 2011). This indicates an increased number of immature premyelinating OPCs, and therefore primary oligodendrocyte pathology cannot be excluded (Bugiani et al., 2011). As the human induced pluripotent stem cell (iPSC) technology is increasingly advancing, disease-modelling studies could identify vulnerable glial cell types in VWM (Nevin et al., 2017). Secondly, the generation of transplantable cells that are safe and are not rejected by the patient is challenging. Autologous transplantation of patients' own genetically corrected iPSC-derived astrocytes could be a promising treatment strategy for leukodystrophies (Osorio & Goldman, 2016).
Thirdly, as the spinal cord is a highly compact and structured part of the CNS, cellular injection forms a challenge for transplantation. Administration of cell transplants via the cerebral spinal fluid (CSF) could be considered (Liu & Huang, 2008). As we demonstrated that the most affected WM of the spinal cord is located under the pial surface, directly adjacent to the CSF, limited cellular integration could already target the affected regions. Finally, the involvement of the (diseased) microenvironment needs attention . Transplanted astrocytes can integrate and become functional , but depend on the host microenvironment to give functional recovery in the injured spinal cord (Noble, Davies, Mayer-Proschel, Proschel, & Davies, 2011). Therefore, the transplantation of healthy glial cells might not be sufficient in VWM, and possibly in many other leukodystrophies. Instead, a combined treatment of astrocyte transplantation, together with modulating interventions to make the affected microenvironment more permissive, could increase the success of the therapy in leukodystrophies .

CON FL I CT OF I N TE RE S T S T AT EM E NT
The authors declare that they have no conflict of interest.