We used oligodendrocyte stage-specific markers, PDGFRα for OPCs, BCAS1 for premyelinating oligodendrocytes, and CAII for myelinating oligodendrocytes for analysis. The staining results showed that there was no significant change in PDGFRα-positive or CAII-positive oligodendrocytes in the ventral horn of the spinal cord of 11-month-old LXRβ-/- mice (Supplementary Fig. 10A-F). Interestingly, the number of BCAS1-positive oligodendrocytes was reduced, and notably, the number of BCAS1-positive cells with mature oligodendrocyte-like morphology or myelin morphology, with four or more visible processes was reduced in the spinal cords of 11-month-old LXRβ-/- mice (Supplementary Fig. 10G-I). Therefore, LXRβ inactivation may specifically affect a particular stage of development while the overall oligodendrocyte lineage pool appears to be unchanged. In fact, LXRβ plays an important regulatory role in the differentiation and function of glial cells in the early stages of neural development. We will analyze the role of LXRβ on oligodendrocytes in postnatal mice.

In the present study, we confirmed that there is loss of neuronal dendrites and motor neuron degeneration in the spinal cord of LXRβ mice. This was accompanied by decreased MCT1, HMGCR and GS expression in oligodendrocytes, but no significant change in the number of total oligodendrocytes. In vitro differentiation of ES cells into motor neurons/oligodendrocytes showed that compared to WT ES cells, there was reduced differentiation into mixed motor neurons and oligodendrocytes in LXRβ ES cells.

We reported the loss of motor neurons in the spinal cord of LXRβ mice in 2005 [23]. Motor neurons were surprising because these cells do not express LXRβ, and we were forced to look at LXRβ-expressing cells other than motor neurons to find the cause of motor neuron death. We have shown that oligodendrocytes, astrocytes, and microglia [24,25,26,27] do express LXRβ as do the epithelial cells of the choroid plexus [28]. Although many genes have been identified as causative or disease-modifying genes in ALS, it is remarkable that loss of a single gene, LXRβ, does lead to motor neuron degeneration.

In the early stages of several neurodegenerative disorders, neuronal dendrites and dendritic spines undergo pathological changes, including spine loss [29]. In two models of ALS, C9orf7 [30] and SOD1 [31, 32], significant changes in dendrite density, shape and branching have been described, suggesting that morphological changes of motor neuron dendrites and dendritic spines may contribute to ALS pathogenesis. Ferraiuolo et al. have shown that oligodendrocytes contribute to motor neuron death in ALS via a SOD1-dependent mechanism [33]. This study used co-culture of mouse oligodendrocytes from mSOD1 or WT mice and WT Hb9-GFP motor neurons. They found that compared to motor neurons co-cultured with WT oligodendrocytes, motor neurons co-cultured with mSOD1 oligodendrocytes had a reduced survival. In the present study, we confirmed that there is degeneration of motor neurons in the spinal cord of LXRβ mice and, not surprisingly, this was accompanied by a considerable loss of dendrites. In motor neurons cultured from ES cells, there were fewer axonal structures and dendrites in LXRβ cultures than in WT motor neurons. Furthermore, the expression of PLP1 was lower in LXRβ than in WT oligodendrocytes in these cultures. Given that this is a mixed culture system of motor neurons and oligodendrocytes, and that LXRβ is expressed in oligodendrocytes but not motor neurons, it appears that LXRβ affects the differentiation and function of oligodendrocytes, which in turn leads to changes in the morphology of motor neurons.

There are mainly three MCT isoforms in the CNS, MCT1, MCT2 and MCT4. MCT1 is expressed in both oligodendrocytes and astrocytes, but MCT4 is not expressed in oligodendrocytes [34]. Loss of MCT1 expression in the oligodendrocytes with age leads to marked axonal degeneration with concomitant hypomyelination [35]. Knockout studies have shown that MCT1 but not MCT4 is essential for motor neuron survival [36]. The idea that defective oligodendrocytes could be the cause of ALS is not new. What is new is the role of defective LXRβ in oligodendrocytes as an etiological factor in motor neuron disease. A proteomic analysis of sera from ALS patients and controls revealed that the LXR/RXR pathway is one of the most significantly regulated pathways, and that both LXRα and LXRβ were genetic modulators of the ALS phenotype in patients [37, 38]. To date, there has been no report indicating that LXRβ provides motor neuron protection by modulating oligodendrocyte metabolic support functions. In the present study, we found that the expression of MCT1 was reduced in spinal cord oligodendrocytes of LXRβ mice as early as 4-month-old. Given that abnormalities in oligodendrocyte metabolic support function precede the neuronal dendrites loss and motor neuron degeneration in LXRβ mice, the present study suggests that defective oligodendrocytes are, at least in part, responsible for motor neuron disease in LXRβ mice.

LXRβ is also expressed in astrocytes. In this study, MCT1 and BDNF are still highly expressed in astrocytes of the spinal cord of LXRβ knockout mice. We conclude that astrocytes may not be the main players based on the markers currently measured, but other mechanisms or specifically expressed transporters cannot be ruled out.

Oligodendrocytes play a critical role in supporting neuronal function and survival through neuroprotective functions mediated by myelin proteins on axons, as well as providing direct trophic and metabolic support to neuronal cells [39, 40]. In the present study, we have shown that LXRβ inactivation may specifically affect a certain developmental stage, whereas the total oligodendrocyte lineage pool does not appear to be altered.

The brain does not take up cholesterol from the circulation but makes its own cholesterol. Cholesterol level in the brain is controlled by metabolism and excretion both of which are LXR regulated. In addition, the endogenous ligands for LXR are cholesterol metabolites [1]. Defective cholesterol homeostasis in the brain is implicated in motor neuron disease. Defects in cholesterol excretion are the cause of spastic paraplegia type 5 (SPG5) and frontal lobe dementia [41]. The reduced expression of HMGCR, the rate-limiting enzyme for cholesterol biosynthesis, in spinal oligodendrocytes of LXRβ mice reported herein, indicates that loss of LXRβ leads to abnormal cholesterol synthesis in oligodendrocytes. Thus, decreased cholesterol synthesis may be a contributing to motor neuron loss seen in LXRβ mice.

Treatment with the LXRβ agonist, GW3965, corrected lipid droplet defects in SPG3A astrocytes, promoted cholesterol efflux from astrocytes, and rescued of axonal degeneration in SPG3A cortical projection neurons [42]. In a Drosophila SPG61 model, LXR agonists (LXR-623 and GW3965) blocked lipid droplet accumulation, restored axonal ER organization, and improved locomotor function [43]. LXRs, particularly LXRα, play a role in regulating the expression of CYP7B1 [44], which functions in the alternative pathway of bile acid synthesis [45]. Patients with SPG5 display mutations in the CYP7B1 gene [46].

GS, a key enzyme that metabolizes glutamate to glutamine, is expressed in myelinating oligodendrocytes but not in oligodendrocyte progenitors of the mouse and human ventral spinal cord. It has been reported that selective removal of oligodendrocyte GS in mice led to reduced levels of brain glutamate and glutamine, impairing glutamatergic synaptic transmission without disrupting myelination [47]. Mice with a conditional knockout of GS in oligodendrocytes showed transient and specific reductions in peak force, with unaffected locomotion and motor coordination [48]. In our previous study, we demonstrated that in LXRβ mice, GS expression was decreased in oligodendrocytes of optic nerve and spiral ganglion neurons of cochlea [3, 4]. In the present study, the expression of GS in spinal cord oligodendrocytes of LXRβ mice was reduced, and the resulting glutamate neurotoxicity was one of the factors contributing to the loss of motor neurons.

In this study, global knockout mice were used. To strengthen the mechanistic conclusion, the use of cell type-specific knockout mice targeting oligodendrocytes will be a valuable future direction to confirm that the function of LXRβ in oligodendrocytes is necessary for neuronal survival. To this end, we have created mice with LoxP sites inserted around the LXRβ gene [49, 50], which will enable cell-specific knockout of LXRβ.

The present study presents evidence that it is abnormalities in oligodendrocyte maturation and function, including metabolic support (MCT1), abnormal cholesterol synthesis (HMGCR), and glutamate neurotoxicity (GS), that are responsible for motor neuron death in LXRβ mice. LXRs are essential for normal brain development, influencing neurogenesis, neuronal and glial cell differentiation and migration, synaptogenesis, and myelination during early fetal life. Due to its widespread effects throughout the body, detailed analysis is still needed to determine whether general dysfunction of either receptor leads to the selective degeneration of certain neurons without causing other significant defects in the body [51]. By crossing specific LXRβ knockouts, we will be able to determine whether the combination of defective oligodendrocytes with defective astrocytes and others leads to motor neuron death.

In summary, we have shown that loss of LXRβ in mice leads to the loss of neuronal dendrites and the degeneration of motor neurons in the spinal cord. Abnormalities in oligodendrocyte maturation and function precede the loss of motor neurons in LXRβ mice, and defective oligodendrocytes appear to be contributors of motor neuron disease in LXRβ mice. Since these abnormalities are found in human ALS, we suggest that LXRβ signaling should be examined in patients with motor neuron diseases.