Neuro‐oxysterols and neuro‐sterols as ligands to nuclear receptors, GPCRs, ligand‐gated ion channels and other protein receptors

The brain is the most cholesterol rich organ in the body containing about 25% of the body's free cholesterol. Cholesterol cannot pass the blood–brain barrier and be imported or exported; instead, it is synthesised in situ and metabolised to oxysterols, oxidised forms of cholesterol, which can pass the blood–brain barrier. 24S‐Hydroxycholesterol is the dominant oxysterol in the brain after parturition, but during development, a myriad of other oxysterols are produced, which persist as minor oxysterols after birth. During both development and in later life, sterols and oxysterols interact with a variety of different receptors, including nuclear receptors, membrane bound GPCRs, the oxysterol/sterol sensing proteins INSIG and SCAP, and the ligand‐gated ion channel NMDA receptors found in nerve cells. In this review, we summarise the different oxysterols and sterols found in the CNS whose biological activity is transmitted via these different classes of protein receptors.


| INTRODUCTION
It has been known for decades that C 18 -C 21 steroids are ligands to nuclear receptors and since the 1990s that oxysterols are ligands to "orphan," now "adopted," nuclear receptors (Evans & Mangelsdorf, 2014). Importantly, nuclear receptors are expressed in the brain where oxysterols are also abundant (Warner & Gustafsson, 2015).
These receptors work via regulating gene and hence protein expression. N-methyl-D-aspartate receptors (NMDARs) are expressed in nerve cells and work on a much shorter time scale. They are ligand-gated ion channels activated by the neurotransmitter glutamate, critical to the regulation of excitatory synaptic function.
Much of this article was written during the Coronavirus  outbreak. The article is dedicated to all healthcare workers throughout the world.
For simplicity, in this article, we use the term oxysterol to cover oxidised forms of cholesterol and its precursors (Schroepfer, 2000) and use the term sterol to specifically embrace cholesterol and its cyclic precursors. By extension, neuro-oxysterols and neuro-sterols are the respective terms to define oxysterols and sterols found in the CNS. The sterol definition is at variance with the formal definition of sterols by lipid maps to include all molecules based on the cyclopentanoperhydrophenanthrene skeleton and ring-opened versions thereof (Fahy et al., 2005).
G protein-coupled receptors (GPCRs) are cell membrane receptors that are activated by molecules outside the cell and activate signal transduction pathways within the cell. Cell membranes are rich F I G U R E 1 Biosynthesis of simple oxysterols. Enzymes are marked in blue. Non-enzymatic oxidation marked in red. Oxysterol names are linked to their abbreviations and biological activities in Table S1 in cholesterol and membrane cholesterol plays an important role in GPCR structure and function (Sengupta & Chattopadhyay, 2015).
Several cholesterol interaction sites have been identified in GPCRs whose occupancy may modulate GPCR activity. Smoothened (SMO), a member of the Frizzled-class of GPCRs and a critical component of the hedgehog (Hh) signalling pathway, has a cholesterol binding site within its extracellular cysteine rich domain, mutational modification of which impairs the ability of SMO to transmit hedgehog signals (Byrne et al., 2016). The hedgehog pathway is critical for tissue patterning during development and abnormal function is associated with birth defects and cancer (Kong, Siebold, & Rohatgi, 2019). Defective hedgehog signalling is implicated in Smith-Lemli-Opitz syndrome, which presents not only with dysmorphology but also learning and behavioural problems, highlighting the importance of hedgehog signalling in the brain (Cooper et al., 2003;Kelley et al., 1996).
Smith-Lemli-Opitz syndrome results from deficiency in 7-dehydrocholesterol reductase (DHCR7), an enzyme near the end of the cholesterol biosynthesis pathway. Numerous oxysterols have also been found to activate the hedgehog signalling pathway by binding to the cysteine rich domain of SMO (Nedelcu, Liu, Xu, Jao, & Salic, 2013;Qi et al., 2019;Raleigh et al., 2018), many of these are found in the brain and are hence neuro-oxysterols (Yutuc et al., 2020). GPR183, also known as Epstein Barr Virus-Induced G protein-coupled receptor 2 or EBI2, is involved in the trafficking of immune cells towards their EBI2 ligands, that is, oxysterols with a dihydroxycholesterol (diHC) structure, where one additional hydroxy group is at C-7α and the second on the side-chain (Hannedouche et al., 2011;Liu et al., 2011).
As T-cell trafficking plays a major role in the neurodegenerative autoimmune disease multiple sclerosis and in its animal model, experimental autoimmune encephalitis (EAE), GPR183 and neuro-oxysterols have been studied in this regard (Duc, Vigne, & Pot, 2019). 7α,25-Dihydroxycholesterol (7α,25-diHC) and 7α,(25R) 26-dihydroxycholesterol (7α,(25R)26-diHC, also known as 7α,27dihdroxycholesterol, note, if stereochemistry at C-25 is not indicated it is assumed to be 25R) are both GPR183 agonists which are present in the brain  and 7α,25-diHC has been F I G U R E 2 Cartoon representations of the regulation of SPEBP-2 processing and of hedgehog (Hh) signalling by cholesterol and oxysterols. (a) Upper panel:-SCAP transports SREBP-2 from the endoplasmic reticulum to the Golgi where it is cleaved to its active form which translocates to the nucleus and activates transcription of genes coding for enzymes of the cholesterol biosynthesis pathway and of the LDL receptor. org/pathways/pathway_lipids_list.php for exact details), is regulated by the master transcription factor SREBP-2 (sterol regulatory element-binding protein 2) which also regulates the expression of the LDL receptor (Horton, Goldstein, & Brown, 2002). When cholesterol levels are low, SREBP-2 is transported by SCAP (SREBP cleavage activating protein) from the endoplasmic reticulum to the Golgi apparatus where it is processed to its active form, which translocates to the nucleus and activates target gene transcription, hence up-regulating cholesterol biosynthesis and import via LDL receptors and restoring cholesterol levels in the cell ( Figure 2) (Goldstein, DeBose-Boyd, & Brown, 2006). When cholesterol levels are elevated, cholesterol in the endoplasmic reticulum binds to SCAP, which becomes tethered to the resident endoplasmic reticulum protein INSIG (insulin induced gene) and prevents transport of SREBP-2 to the Golgi apparatus for activation (Goldstein et al., 2006;Sun, Seemann, Goldstein, & Brown, 2007).
In this way, cholesterol regulates its own biosynthesis and import via the LDL receptor. Oxysterols also inhibit cholesterol biosynthesis, but in this case by binding to INSIG and tethering the SCAP-SREBP-2 complex in the endoplasmic reticulum ( Figure 2) (Radhakrishnan, Ikeda, Kwon, Brown, & Goldstein, 2007). Although the turnover of cholesterol in the adult brain is slow (0.4% per day in mouse, 0.03% in human), the rate of synthesis is much higher during development (Dietschy & Turley, 2004). In the fetal mouse, the blood-brain barrier (BBB) is formed at E10-E11 and after E11-E12, the brain is the source of essentially all new sterols (Tint, Yu, Shang, Xu, & Patel, 2006). Importantly, cholesterol 24-hydroxylase (cytochrome P450 46A1, CYP46A1) expression is low until E18, conserving cholesterol and restricting the biosynthesis of the neuro-oxysterol 24S-HC (Tint et al., 2006). CYP46A1 is the dominant cholesterol hydroxylase in the brain and hydroxylation of cholesterol to 24S-HC is responsible for 50-60% of all cholesterol metabolism in adult brain (Russell, Halford, Ramirez, Shah, & Kotti, 2009).
The expression of the liver X receptor target gene Apoe was upregulated and the transport of cholesterol from astrocytes to neurons partially explained the improved phenotype (Kacher et al., 2019). It should be noted that the anti-HIV drug efavirenz activates CYP46A1 and could provide an alternative route to enhancing 24S-HC and 24S,25-EC formation in the brain (Mast et al., 2017).
Studies on the Lxrα −/− Lxrβ −/− adult mouse reveal progressive accumulation of lipids in the brain and loss of spinal cord motor neurons and ventral midbrain dopaminergic neurons, indicating further that these receptors are important in the brain in both the developing and adult mouse (Wang et al., 2002). Surprisingly, studies of the CYP46A1tg adult mouse found little change in the mRNA of liver X receptor target genes in the brain (Shafaati et al., 2011), despite elevation in 24S-HC and 24S,25-EC levels (Shafaati et al., 2011;Theofilopoulos et al., 2019). This may be a consequence of making global measurements, where changes in specific brain regions are lost by averaging signal levels in bulk tissue.
Remarkably, 3β-HCA is neurotoxic towards motor neurons, while 3β,7α-diHCA is protective and 3βH,7O-CA promotes maturation of precursor cells into motor neurons, with each activity mediated by liver X receptors (Theofilopoulos et al., 2014). Theofilopoulos et al. suggested that the loss of motor function in two diseases resulting from inborn errors of metabolism, cerebrotendinous xanthomatosis (deficiency in CYP27A1) and hereditary spastic paresis type F I G U R E 4 Biosynthesis of cholestenoic acids. Enzymes are marked in blue. Non-enzymatic oxidation marked in red. Oxysterol names are linked to their abbreviations and biological activities in Table S1 5 (deficiency in CYP7B1) is a consequence of a reduced production of 3β,7α-diHCA in cerebrotendinous xanthomatosis, with the additional over production of neurotoxic 3β-HCA in spastic paresis type 5 providing a double-hit mechanism in the latter disease (see Figure 4 for CYP catalysed reactions) (Theofilopoulos et al., 2014).
Interestingly, 3β,7α-diHCA is found to be most abundant in mouse cerebellum (Yutuc et al., 2020) but is absent in cerebellum from cerebrotendinous xanthomatosis patients. Cerebellar ataxia, impaired co-ordination of voluntary movements due to underdevelopment of the cerebellum, is a common characteristic of cerebrotendinous xanthomatosis and may occur in spastic paresis type 5 also (Bjorkhem, 2013;Clayton, 2011), linking cholestenoic acids and liver X receptors to brain development. Besides the cholestenoic acids, (25R)26-hydroxycholesterol ((25R)26-HC, also called by the non-systematic name 27-hydroxycholesterol, 27-HC, note if stereochemistry at C-25 is not defined it is assumed to be 25R), a weak liver X receptor agonist (Fu et al., 2001), also accumulates in CSF of spastic paresis type 5 patients (Schols et al., 2017;Theofilopoulos et al., 2014) and Hauser et al. have suggested that neurotoxic effects of (25R)26-HC are major contributors to the spastic paresis type 5 phenotype (Hauser et al., 2019). It is noteworthy that neither Cyp27a1 −/− or Cyp7b1 −/− mice shows a motor neuron phenotype. A possible explanation for these differences between human and mouse is the prevalence of alternative pathways to produce neuroprotective 3β,7α-diHCA in mouse Meljon et al., 2019) or alternatively the presence of lower levels of neurotoxic neuro-oxysterols in mouse than in human (Hauser et al., 2019). Gene therapy has been suggested as a potential treatment for spastic paresis type 5 (Hauser et al., 2019) and also cerebrotendinous xanthomatosis. Interestingly, CYP27A1 the deficient enzyme in cerebrotendinous xanthomatosis, is a mitochondrial enzyme and mitochondrial dysfunction has been linked to several neurodegenerative diseases (Fivenson et al., 2017).
24S-HC through its activation of liver X receptors has been suggested to be protective against glioblastoma, the most common primary malignant brain tumour in adults (Han et al., 2020). Efavirenz, an antiretroviral medication that crosses the BBB and activates CYP46A1 (Mast et al., 2017), was shown to inhibit glioblastoma growth, the effect being explained by enhanced synthesis of 24S-HC, activation of liver X receptor and inhibition of the cholesterol synthesis pathway by inhibition of SREBP processing (Han et al., 2020). The results reported by Han et al. are interesting in a number of regards (Han et al., 2020):-(a) Despite the name, the cellular origin of glioblastoma is unknown, but if glia are the origin of the cancer the action of 24S-HC must be via a paracrine mechanism as CYP46A1 is expressed in neurons not glia (Lund, Guileyardo, & Russell, 1999); (b) the SREBP studied by Han et al. was SREBP-1, not SREBP-2 the dominating transcription factor regulating cholesterol synthesis (Horton et al., 2002).
24S-HC was found to reduce nuclear SREBP-1, which is expected (Wang, Muneton, Sjovall, Jovanovic, & Griffiths, 2008), but SREBP-1 primarily activates fatty acid not cholesterol synthesis (Horton et al., 2002), hence the link to reduced cholesterol synthesis and uptake via LDL receptor is less clear. Nevertheless, the beneficial effects of 24S-HC suggest CYP46A1 as a potential therapeutic target.

| Oestrogen receptors
The oestrogen receptors α (ERα, NR3A1) and β (ERβ, NR3A2) are both present in rodent brain, although in different locations (Warner & Gustafsson, 2015) as are their classical steroid hormone agonists (Li & Gibbs, 2019) and oxysterol modulators Meljon et al., 2019). ERβ has a role in the migration of cortical neurons in the developing brain, where it is detectable at E12.5 (Wang, Andersson, Warner, & Gustafsson, 2003) and in adult in the maintenance of 5-hydroxytryptamine containing neurons in the dorsal raphé nucleus, involved in fear, anxiety and depression (Suzuki et al., 2013).
ERα is expressed in spinal cord motor neurons where it has a role in protecting neurons against cytokine toxicity (Das et al., 2011). Both ERα and ERβ are protective against experimental autoimmune encephalitis, a mouse model of multiple sclerosis (Spence et al., 2013).
The protective effect of ERα is through astrocytes but that of ERβ through microglia . Microglia are the macrophages of the brain and when activated can damage healthy neurons in the region of insult. ERβ selective agonists dampen the activation of microglia and reduce the proinflammatory potential of invading T-cells (Warner & Gustafsson, 2015;Wu, Tan, et al., 2013). SERMs are ER ligands whose relative agonist/antagonist activities vary in a cell-and promotor-dependent manner (Wardell, Nelson, & McDonnell, 2014). Selectivity is based on the ability of ligands to induce alterations in the ER structure leading to differential recruitment of coactivators and co-repressors. (25R)26-HC induces conformational changes in both ERα and ERβ. Interestingly, (25R)26-HC is a competitive antagonist of ER action in the vasculature (Umetani et al., 2007) but has ER agonist activity in breast cancer cells (DuSell et al., 2008). Although most studies of (25R)26-HC and ER have been made in the context of breast cancer (Nelson et al., 2013;, the possibility exists that (25R)26-HC may act as a SERM in the CNS.
In mouse and human brain levels of (25R)26-HC are low ($1 ngÁmg −1 ) Heverin et al., 2004;Meljon et al., 2019) but are elevated in the Cyp7b1 −/− mouse ($5 ngÁml −1 )  and presumably in sufferers of spastic paresis type 5 where CYP7B1 is deficient and one route of (25R)26-HC metabolism blocked (Figure 4). A raised content of (25R)26-HC in spastic paresis type 5 brain has not been considered in the context of (25R)26-HC as a SERM, but the role of ERα in the protection of spinal cord motor neurons may be relevant with (25R)26-HC acting as a potential ERα antagonist in spastic paresis type 5 where motor neurons are lost. 25-Hydroxycholesterol (25-HC) is also a SERM (Simigdala et al., 2016) and is also elevated in Cyp7b1 −/− mouse brain .

| Retinoic acid receptor-related orphan receptors
There are three retinoic acid receptor-related orphan receptors (RORs): RORα (NR1F1), RORβ (NR1F2) and RORγ (NR1F3). RORγ exists as two isoforms RORγ1, or simply RORγ and RORγ2 also called RORγt. RORγt has a shorter N-terminus compared to RORγ but with otherwise identical domains. RORγt is highly expressed in thymus and is an essential transcription factor for Th17 cell development. Th17 cells are pro-inflammatory T-helper cells (CD4 + ) that express IL-17.
Th17 cells are mediators of experimental autoimmune encephalitis, the mouse model of multiple sclerosis (Duc et al., 2019). RORγ is expressed in multiple organs but not normally in the CNS. However, RORγt-immunoreactive cells have been found in the meninges, the three membranes that envelop the brain and spinal cord, of multiple sclerosis patients, presumably as infiltrating immune cells (Serafini et al., 2016). RORα and RORβ are both expressed in the brain. RORα is abundant in cerebellum and thalamus and plays a key role in development, particularly in the regulation of the maturation and survival of Purkinje cells (Jetten & Joo, 2006). RORβ is highly expressed in the brain and also the retina (Jetten & Joo, 2006). Oxysterols have been found to bind to the ligand binding domains of both RORα and RORγ (including RORγt), but not as yet RORβ (Duc et al., 2019). 25-HC, 7-oxocholesterol (7-OC), 7α-hydroxycholesterol (7α-HC) and 7β-hydroxycholesterol (7β-HC), all of which have been identified in mouse or human brain (Figure 1) (Yutuc et al., 2020). It is noteworthy that the side-chain oxysterols 20S-HC, 22R-HC 24S-HC, 24S,25-EC and 25-HC are also ligands to the LXRs (Janowski et al., 1996;Lehmann et al., 1997) and with the exception of 20S-HC, which has not been tested, to INSIG (Radhakrishnan et al., 2007). Cholesterol sulphate, the dominating neuro-sterol sulphate in rat brain (Liu, Sjovall, & Griffiths, 2003), has also been shown to bind to the ligand binding domain (LBD) of RORα (Kallen, Schlaeppi, Bitsch, Delhon, & Fournier, 2004), while desmosterol, one of the most abundant neurosterols and also an LXR ligand (Yang et al., 2006), has been shown to be a potent RORγ agonist (Hu et al., 2015). RORα is an unusual nuclear receptor in that it is constitutively active in the absence of ligand and despite binding to the LBD of RORα cholesterol sulphate does not appear to affect the transcriptional activity of RORα or RORγ (Wang, Kumar, Solt, et al., 2010). In contrast, Wang et al. reported 7α-HC, 7β-HC and also 7-OC to be ligands to RORα and RORγ and to supress the transcriptional activity of these receptors acting as inverse agonists . Although CYP7A1, the enzyme required to synthesise 7α-HC and 7-OC, is not expressed in the brain, 7α-HC, 7-OC and 7β-HC can cross the BBB and be imported from the periphery  or alternatively be formed via non-enzymatic reactions initiated by ROS (Griffiths & Wang, 2020;Murphy & Johnson, 2008). Interestingly, HSD11B1 (11-βHSD1) is expressed in the brain and will convert 7-OC to 7β-HC Mitic et al., 2013) (Figure 1).
24S-HC has also been found to be an inverse agonist of RORα and RORγ supressing the constitutive activity of these receptors . Interestingly, 24S,25-EC selectively supresses the activity of RORγ . Conversely, 20S-HC, 22R-HC and 25-HC have been shown to be agonists towards RORγ, promoting the recruitment of co-activators (Jin et al., 2010). The crystal structure of the LBD of RORγ with 20S-HC, 22R-HC, or 25-HC bound showed the AF-2 helix in a conformation that is permissive for interactions with co-activator proteins. Importantly, mutations that disrupt the binding of these hydroxycholesterols abolish RORγ transcriptional activity, suggesting a critical role for hydroxycholesterols in activating RORγ (Jin et al., 2010). With respect to brain, it is interesting to note that CYP46A1, the enzyme that generates 24S-HC, is expressed in Purkinje cells (Lund et al., 1999) and that RORα plays a role in the regulation of the maturation and survival of these cells (Jetten & Joo, 2006). The cerebellum of RORα-deficient mice contains significantly fewer Purkinje cells and exhibits a loss of cerebellar granule cells (Jetten & Joo, 2006). It is tempting to speculate that 24S-HC Cyp46a1 −/− mice do display defects in motor learning (Kotti et al., 2006). Interestingly, Cyp7b1 has been reported as a target gene of RORα (Wada et al., 2008). CYP7B1 was first reported in the brain and functions as an oxysterol 7α-hydroxylase to side-chain oxysterols including 25-HC and (25R)26-HC (Rose et al., 2001), but not 24S-HC, where CYP39A1 is the 7α-hydroxylase (Li-Hawkins, Lund, Bronson, & Russell, 2000). Recent evidence suggests CYP39A1 is also regulated by RORα (Matsuoka et al., 2020).
There is good evidence for the production of 7α,(25R)26-diHC in the brain from either imported or in situ synthesised (25R)   and Smith-Lemli-Opitz syndrome patients where 7-dehydrocholesterol (7-DHC) is abundant, so mechanisms are available for its formation (Shinkyo et al., 2011), at least extra-cerebrally (Figures 1 and 3).  (Wanke et al., 2017). These data lead to the hypothesis that GPR183 and 7α,25-diHC play a role in the migration of inflammatory CD4 + T cells into the CNS. 7α,(25R)26-diHC is also a GPR183 ligand and it is perhaps no coincidence that this molecule is also an agonist towards RORγt.

| Smoothened
SMO belongs to the Class F (Frizzled, FZD) of GPCRs and of the 11 members of this class 10 are FZD paralogues, the other is SMO (Kozielewicz et al., 2020). SMO is a seven-pass transmembrane (7TM) protein, essential to the hedgehog signalling pathway critical in normal animal development, for example, activation of the hedgehog pathway is required for the differentiation of neural progenitor cells into motor neuron progenitors, as well as pathological malignancies, including medulloblastoma, the most common paediatric brain tumour.
The hedgehog signalling pathway is regulated in primary cilia where the transporter protein Patched 1 (PTCH1) and SMO are co-located (Rohatgi, Milenkovic, Corcoran, & Scott, 2009). PTCH1 is the receptor for hedgehog ligands, for example, sonic hedgehog (SHH). and in the absence of ligand inhibits SMO and the transduction of hedgehog signals across the plasma membrane ( Figure 2)  and cilia in their model (Kinnebrew et al., 2019). They defined three pools of cholesterol:-a fixed pool essential to maintain membrane integrity, a sphingomyelin sequestered pool of low accessibility and an available pool to interact with proteins and be transported to the endoplasmic reticulum (Kinnebrew et al., 2019). The idea of multiple pools of cholesterol was discussed earlier by Radhakrishnan et al. and in respect to brain by Dietschy and Turley (2004) (Radhakrishnan, Anderson, & McConnell, 2000). According to Kinnebrew (Corcoran & Scott, 2006;Dwyer et al., 2007;Kim, Meliton, Amantea, Hahn, & Parhami, 2007;Myers et al., 2013;Nachtergaele et al., 2012;Nachtergaele et al., 2013;Nedelcu et al., 2013;Qi et al., 2019;Raleigh et al., 2018). Conversely, 3β,5α-dihydroxycholest-7-en-6-one (DHCEO, Figure 3), an oxysterol derived from 7-DHC and identified in the brain of a mouse model of Smith-Lemli-Opitz syndrome where 7-DHC is abundant (Xu et al., 2012), binds to SMO and blocks hedgehog signalling (Sever et al., 2016). Of the above oxysterols, 24S,25-EC is abundant in embryonic mouse brain, particularly the ventral midbrain, while 24S-HC and 25-HC are also present during development but at concentrations about one order of magnitude lower than 24S,25-EC (Theofilopoulos et al., 2013;Wang et al., 2009). In the new-born mouse, 24S,25-EC is still the most abundant oxysterol (Meljon et al., 2012), but in the adult mouse, 24S-HC is by far the most abundant oxysterol and 20S-HC is also present but at a low level (Meljon et al., 2012;Yutuc et al., 2020). Smith-Lemli-Opitz syndrome phenocopies dysregulated hedgehog signalling (Cooper et al., 2003), however, at least in the new-born mouse the pattern of SMOactivating oxysterols in the brain is similar to the WT (Meljon, Watson, Wang, Shackleton, & Griffiths, 2013). Of the other oxysterols suggested to activate the hedgehog pathway through binding to SMO, we have identified 24-OC in the brain, although we used a LC-MS method incorporating a derivatisation step in which, as a sidereaction, the 24,25-epoxy function isomerises to the 24-oxo function.
Similar to the situation with cholesterol, there appear to be multiple binding sites for oxysterols on SMO. Oxysterols can bind to the same cysteine rich domain pocket as cholesterol (Byrne et al., 2016) and also within a 7TM pocket (Qi et al., 2019;Raleigh et al., 2018). both HSD11B1 and HSD11B2 are expressed in the brain . Interestingly, HSD11B2 is expressed during brain development  and will catalyse the oxidation of the 7β-hydroxy group to a 7-oxo, while HSD11B1 catalyses the reverse reduction (Beck, Inderbinen, et al., 2019;Schweizer et al., 2004). Importantly, Hsd11b2 is enriched in mouse models of medulloblastoma and HSD11B2 is enriched in hedgehog-pathway associated human medulloblastoma (Raleigh et al., 2018). Raleigh et al. proposed a mechanism involving HSD11B2 and CYP27A1 by which hedgehog agonists 7β,(25R)26-diHC and (25R)26H,7O-C could be formed from 7β-HC and, remarkably, pharmacological inhibition of HSD11B2 reduced hedgehog signalling and tumour growth in mouse medulloblastoma (Raleigh et al., 2018). The contribution of HSD11B2 to oncogenic hedgehog signalling suggests that oxysterols produced by this enzyme are required for high-level pathway activity. Our interpretation of these data is that while (25R) 26H,7O-C and 7β,(25R)26-diHC are both agonists to SMO, (25R) 26H,7O-C must be more potent. Whether this is a consequence of differences in SMO binding or simply accessibility of ligand to receptor is unclear. While 20S-HC, 24S-HC, 25-HC and 24S,25-EC have been identified in the brain and are bona fide neuro-oxysterols, neither 7β,(25R)26-diHC nor (25R)26H,7O-C have been identified in the brain or medulloblastoma; however, their precursors 7-OC and 7β-HC have Meljon et al., 2019). Importantly, 7-OC and 7β-HC can traverse the blood-brain barrier and provide an extracerebral source of precursors for (25R)26-hydroxylation by CYP27A1 (Iuliano et al., 2015).  (Paul et al., 2013). While the presence of pregnenolone sulphate in the brain is debatable (Liu et al., 2003), 24S-HC is present in rodent and human brain at concentrations of 10-20 ngÁmg −1 (25-50 μM).
Importantly, 24S-HC does not alter membrane currents on its own in the absence of NMDA, indicating that the effect is independent of nuclear receptors, for example, LXRs. The synthetic 24-HC analogue, 24,, is also a positive allosteric modulator, somewhat more potent than the parent molecule and it likewise potentiates LTP and, remarkably, reverses behavioural deficits (e.g. spatial working memory) induced by NMDA receptor channel blockers (Paul et al., 2013). Importantly, the Cyp46a1 −/− mouse shows a deficiency in hippocampal LTP and a deficit in spatial, associative and motor learning (Kotti et al., 2006). This was explained by Kotti et al. to be a consequence of reduced flux of metabolites through the cholesterol biosynthetic pathway and through the mevalonate pathway towards geranylgeranyl diphosphate (Kotti et al., 2006). Administration of geranylgeraniol to hippocampal slices restored LTP to wild type levels indicating that it is a deficit in this molecule, or its diphosphate, which is responsible for the impaired LTP in the Cyp46a1 −/− mouse (Kotti et al., 2006;Kotti et al., 2008). In light of the data of Paul et al. indicating that 24S-HC is a NMDA receptor positive allosteric modulator (Paul et al., 2013), it is not unreasonable to speculate that a loss of this molecule in the Cyp46a1 −/− mouse may also be responsible for some of the deficiency F I G U R E 5 Biosynthesis of SMO ligands from 7-DHC. Enzymes are marked in blue. Non-enzymatic oxidation marked in red. CYP27A1 is the likely catalyst for reactions with broken arrows. Oxysterol names are linked to their abbreviations and biological activities in Table S1 in hippocampal LTP. It is likely that 24S-HC modulates NMDA receptors in an autocrine or paracrine manner at a distinct site from other positive allosteric modulators (Paul et al., 2013). Interestingly, the CYP46A1tg mouse shows improved spatial memory and increased hippocampal NMDAR activity (Maioli et al., 2013). Surprisingly, the other side-chain oxysterols 22R-HC and 20S-HC are not modulators of NMDA receptors, although 25-HC is a weak positive modulator (Linsenbardt et al., 2014;Paul et al., 2013). In fact, 25-HC is an antagonist of the positive allosteric modulator activity of 24S-HC to NMDA receptors but remarkably through a different site (Linsenbardt et al., 2014).

| INSIG, SCAP AND HMG-COA REDUCTASE
Sterols and oxysterols regulate cholesterol synthesis by binding to SCAP or INSIG and via SREBP-2 modulate the expression of the enzymes of the cholesterol biosynthesis pathway (Figure 2) (Radhakrishnan et al., 2007;Sun et al., 2007). At times of cholesterol excess, cholesterol in the endoplasmic reticulum binds to SCAP, which is already in complex with SREBPs and causes a conformational change in SCAP which results in binding of the SCAP-SREBP complex to the endoplasmic reticulum-anchor protein INSIG. The consequence of this event is retention of SREBP-2 in the endoplasmic reticulum and prevention of its transport by SCAP to the Golgi for processing to its active form as a transcription factor for the enzymes of the cholesterol biosynthesis pathway and for the LDL receptor (Goldstein et al., 2006;Sun et al., 2007). Oxysterols also have a role in regulating cholesterol biosynthesis (Kandutsch, Chen, & Heiniger, 1978) but in most cases by providing "fine tuning" of this regulation (Gill et al., 2008). The brain is an organ with high oxysterol concentrations (24S-HC 10-20 ngÁmg −1 cf. cholesterol 10-20 μgÁmg −1 ) and it may be in the CNS that the involvement of oxysterols in regulation cholesterol biosynthesis is most significant (Saeed et al., 2014). Another case when oxysterols may be especially important in regulating cholesterol biosynthesis is in response to infection, both bacterial and virus (Bauman et al., 2009;Blanc et al., 2013). Rather than bind to SCAP, side-chain oxysterols, that is, 22R-HC, 24S-HC, 25-HC, (25R)26-HC and 24S,25-EC, bind to INSIG and tether the SCAP-SREBP-2 complex in the endoplasmic reticulum (Radhakrishnan et al., 2007) (Crick et al., 2017).
Based on Reboldi et al.'s work (Reboldi et al., 2014), Crick et al. suggested that a reduced production of 25-HC in multiple sclerosis patients may result in reduced activation of LXRs and enhanced activity of the cholesterol biosynthesis pathway leading to overproduction of IL-1β, both of which lead to enhanced activity of macrophages recruited to the CNS during multiple sclerosis relapse (Crick et al., 2017). 4,and lanosterol did not bind to either INSIG or SCAP and did not inhibit the processing of SREBP-2 (Radhakrishnan et al., 2007). Conversely, lanosterol, the first sterol in the cholesterol biosynthesis pathway (Figure 3 were potent suppressors of HMGCR (Axelson, Larsson, Zhang, Shoda, & Sjovall, 1995). The first three of these compounds have been found in the brain Meljon et al., 2019). However, Axelson et al.'s findings were before the involvement of INSIG in HMGCR degradation was uncovered and a mechanism of HMGCR suppression by oxysterols was not described by these authors (Axelson et al., 1995 (Radhakrishnan et al., 2007). Lanosterol is a 4,4-dimethyl sterol and a very recent study has confirmed its ability to stimulate HMGCR degradation and its inability to inhibit SREBP-2 cleavage (Chen et al., 2019). Surprisingly, other 4,4-dimethyl sterols in the cholesterol biosynthesis pathway promoted both HMGCR degradation and inhibited SREBP-2 cleavage (Chen et al., 2019).

| ANALYSIS OF NEURO-OXYSTEROLS AND NEURO-STEROLS
Much of our understanding of the biochemistry of neuro-sterols and neuro-oxysterols depends on their reliable detection and quantification in brain tissue, CSF, plasma and in cells of the nervous systems.
Both GC-MS and LC-MS methods have been extensively applied.
Gold standard GC-MS methods are based on the classic paper by Dzeletovic et al. describing the analysis of oxysterols in plasma (Dzeletovic, Breuer, Lund, & Diczfalusy, 1995;Heverin et al., 2004).
There are many LC-MS method, with and without the use of derivatisation. A major study using LC-tandem MS (MS/MS) was performed by Russell and colleagues in Dallas in which they analysed over 3,000 serum samples (Stiles et al., 2014). Their analytical method relied on multiple reaction monitoring (MRM), exploiting a tandem quadrupole to achieve maximum sensitivity, with the use of multiple isotope labelled standards for quantification (McDonald, Smith, Stiles, & Russell, 2012). To achieve high sensitivity, others have exploited derivatisation in combination with LC-MS/MS (Honda et al., 2009;Roberg-Larsen et al., 2014;Sidhu et al., 2015;Xu et al., 2011).
Our preference is to use the Girard P (GP) derivatisation reagent in combination with enzymatic oxidation in a methodology called "enzyme-assisted derivatisation for sterol analysis" EADSA Griffiths et al., 2013). The method is applicable to any sterol/oxysterol with an oxo group or with a hydroxy group amenable to enzymatic oxidation to a carbonyl. The most commonly used enzyme is cholesterol oxidase which converts 3β-hydroxy-5-ene sterols to 3-oxo-4-enes and 3β-hydroxy-5α-hydrogen structures to 3-ones (Karu et al., 2007). The oxo group is then reacted with the GP hydrazine reagent to give a GP hydrazone which effectively tags the sterol/oxysterol with a positive charge (see Figure S1). The consequence of charge-tagging is a major improvement in LC-MS sensitiv- hybrid and tribrid mass spectrometers. We have extensively utilised this method to identify and quantify neuro-sterols and neuro-oxysterols in the brain and in CSF and to monitor the import and export of these molecules into and out of brain (Iuliano et al., 2015). We have now extended this method to on-surface analysis of tissue slices to localise neurooxysterolsand neuro-sterolsinmousebrain(Yutucetal.,2020).

| THERAPEUTIC MANAGEMENT OF CNS DISEASE
Throughout this review, we have attempted to draw attention to potential therapies to neuro-sterol and neuro-oxysterol related CNS disorders and here we proved a summary of these therapeutic possibilities.
Treatment is available for cerebrotendinous xanthomatosis in the form of chenodeoxycholic acid (Bjorkhem, 2013;Clayton, 2011); however, while proving effective in stabilising systemic manifestations of the disease, it fails to prevent worsening of the neurological outcome and ultimately death from the disease (Bjorkhem, 2013). A curative approach to cerebrotendinous xanthomatosis is more likely to require complementation of 26-hydroxylated metabolites deficient in this disease by, for example, gene transfer of CYP27A1. Similar to cerebrotendinous xanthomatosis, oxysterol 7α-hydroxylase (CYP7B1) deficiency has been treated with chenodeoxycholic acid and this has proved successful in respect to improving liver function (Dai et al., 2014). However, whether this treatment improves neurological function in the spastic paraplegia form of this disease, that is, spastic paresis type 5, is unclear (Marelli et al., 2018). Hauser et al. have suggested CYP7B1 gene therapy as a potential treatment for spastic paresis type 5 (Hauser et al., 2019). Gene therapy in the form of AAV-CYP46A1 shows promising results in mouse models for the treatment of Alzheimer's disease, Huntington's disease and spinocerebellar ataxias (Boussicault et al., 2016;Burlot et al., 2015;Kacher et al., 2019;Nobrega et al., 2019) and perhaps also Parkinson's disease . An alternative way of enhancing CYP46A1 activity is through the anti-HIV drug Efavirenz which activates CYP46A1 and is able to cross the BBB from the circulation (Mast et al., 2017). 24S,25-EC one of the products of CYP46A1 catalysis has been suggested to have value in cell replacement therapy to treat Parkinson's disease based on its role in promoting neurogenesis of dopaminergic neurons via activation of liver X receptors (Theofilopoulos et al., 2013). Another liver X receptor ligand, 3β,7α-diHC may have value in the treatment of motor neuron disease based on its protective action towards motor neurons (Theofilopoulos et al., 2014).

| CONCLUSIONS
The dominating neuro-oxysterol and neuro-sterol in the brain are 24S-HC and cholesterol, respectively. During embryonic development and in the new-born, 24S,25-EC is more abundant than 24S-HC in the brain. Besides the dominating neuro-oxysterols, there are numerous other oxysterols present in the brain but at lower concentrations. This is also true for the neuro-sterols where comparatively low levels of cholesterol precursors are evident. It is, however, the most abundant species that appear to be most biochemically active. For example, both 24S-HC and 24S,25-EC are ligands to LXRs, 24S-HC is an inverse agonist to RORα and RORγ, while 24S,25-EC is an inverse agonist to RORγ only (see Table S1). Both 24S-HC and 24S,25-EC are reported as agonists to SMO and to activate hedgehog signalling, while 24S-HC is a positive allosteric modulator of the NMDA receptors, but 24S,25-EC has not been tested. In addition, both 24S-HC and 24S,25-EC will bind to INSIG and modulate SREBP-2 regulated cholesterol biosynthesis. Why this redundancy? The answer may be provided by the Cyp46a1 −/− mouse, which does not synthesise 24S-HC via its usual pathway and 24S-HC is only present in trace amounts in its brain. Hhowever, other than some learning difficulties the