Mining for Oxysterols in Cyp7b1−/− Mouse Brain and Plasma: Relevance to Spastic Paraplegia Type 5

Deficiency in cytochrome P450 (CYP) 7B1, also known as oxysterol 7α-hydroxylase, in humans leads to hereditary spastic paraplegia type 5 (SPG5) and in some cases in infants to liver disease. SPG5 is medically characterized by loss of motor neurons in the corticospinal tract. In an effort to gain a better understanding of the fundamental biochemistry of this disorder, we have extended our previous profiling of the oxysterol content of brain and plasma of Cyp7b1 knockout (-/-) mice to include, amongst other sterols, 25-hydroxylated cholesterol metabolites. Although brain cholesterol levels do not differ between wild-type (wt) and knockout mice, we find, using a charge-tagging methodology in combination with liquid chromatography–mass spectrometry (LC–MS) and multistage fragmentation (MSn), that there is a build-up of the CYP7B1 substrate 25-hydroxycholesterol (25-HC) in Cyp7b1-/- mouse brain and plasma. As reported earlier, levels of (25R)26-hydroxycholesterol (26-HC), 3β-hydroxycholest-5-en-(25R)26-oic acid and 24S,25-epoxycholesterol (24S,25-EC) are similarly elevated in brain and plasma. Side-chain oxysterols including 25-HC, 26-HC and 24S,25-EC are known to bind to INSIG (insulin-induced gene) and inhibit the processing of SREBP-2 (sterol regulatory element-binding protein-2) to its active form as a master regulator of cholesterol biosynthesis. We suggest the concentration of cholesterol in brain of the Cyp7b1-/- mouse is maintained by balancing reduced metabolism, as a consequence of a loss in CYP7B1, with reduced biosynthesis. The Cyp7b1-/- mouse does not show a motor defect; whether the defect in humans is a consequence of less efficient homeostasis of cholesterol in brain has yet to be uncovered.


Introduction
Cytochrome P450 (CYP) 7B1 (cytochrome P450 family 7 subfamily B member 1) was first identified in 1995 and found to be primarily expressed in brain in rodents [1]. CYP7B1 is an oxysterol-and steroid-7α-hydroxylase, accepting many oxysterols and cholestenoic acids as substrates as well as steroids including dehydroepiandrosterone (DHEA) [2][3][4]. In humans, deficiency in the enzyme was

Analysis
The oxidised/derivatised oxysterol-rich fractions were analysed by LC-MS (MS n ) using an Ultimate 3000 LC system (Thermo Fisher Scientific, Loughborough, UK) and LTQ-Orbitrap mass spectrometer (Thermo Fisher Scientific, Loughborough, UK) as described in Meljon et al. [20] and Crick et al. [23,24]. In brief, GP-derivatised oxysterols were separated on a reversed phase Hypersil Gold C 18 column (Thermo Fisher Scientific) using a methanol/acetonitrile/0.1% formic acid gradient. The eluent was directed to an electrospray ionisation source (ESI) and analysed by high-resolution (60,000 at m/z 400) MS and MS 3 ([M] + →[M-Py] + →, where "-Py" corresponds to the loss of the pyridine group from the molecular ion M + ) scans performed in parallel in the Orbitrap and LTQ linear ion-trap, respectively. Quantification was performed using the isotope dilution method.
In the Cyp7b1+/+ mice, the concentration of 25-HC is below the limit of detection of the LC-MS method (<0.01 ng/mg); however, in the Cyp7b1-/mice, the level of 25-HC is significantly elevated (p < 0.001) in both 13 (0.89 ± 0.14 ng/mg, mean ± SD) and 23 (2.09 ± 0.45 ng/mg) month-old mice. This is particularly evident when using an extended chromatographic gradient (33 min) which resolves 25-HC from other side-chain hydroxycholesterols ( Figure 2). Unlike 24S-HC, 24R-HC is only a minor oxysterol in mouse brain [21], but like 24S-HC its concentration in the Cyp7b1-/mouse brain was not found to differ from that measured in wild-type (wt) animals (0.20 ± 0.09 ng/mg and 0.30 ± 0.03 ng/mg at 13 and 23 months, respectively, in wt).
As reported earlier [12], the concentration of 26-HC is elevated in the Cyp7b1-/mice (Figure 2a,b). This oxysterol is formed by oxidation of cholesterol by CYP27A1 (cytochrome P450 family 27 subfamily A member 1) [25]. CYP27A1 can also oxidize desmosterol (24-DHC, 24-dehydrocholesterol) to 26-HD [26] and its cis (Z) and trans (E) geometric isomers have been reported to be present in newborn mouse brain [20]. Although only minor oxysterols, both isomers are found to be elevated in concentration in Cyp7b1-/mouse brain, most prominently in 23-month-old animals, where the concentration of the (E) isomer was raised from 0.01± 0.00 ng/mg in the wt to 0.04 ± 0.00 ng/mg in the Cyp7b1-/animals (p < 0.001, Figures 1 and 3).   As reported earlier [12], the concentration of 26-HC is elevated in the Cyp7b1-/-mice ( Figure  2a,b). This oxysterol is formed by oxidation of cholesterol by CYP27A1 (cytochrome P450 family 27 subfamily A member 1) [25]. CYP27A1 can also oxidize desmosterol (24-DHC, 24dehydrocholesterol) to 26-HD [26] and its cis (Z) and trans (E) geometric isomers have been reported    to be present in newborn mouse brain [20]. Although only minor oxysterols, both isomers are found to be elevated in concentration in Cyp7b1-/-mouse brain, most prominently in 23-month-old animals, where the concentration of the (E) isomer was raised from 0.01± 0.00 ng/mg in the wt to 0.04 ± 0.00 ng/mg in the Cyp7b1-/-animals (p < 0.001, Figures 1 and 3).      7α-HC is also a minor oxysterol in wt mouse brain (0.01 ng/mg), but in brain from the Cyp7b1-/-23-month-old mice its concentration is raised to 0.02 ± 0.00 ng/mg (p < 0.05). Likewise, 7β-HC, another minor oxysterol, is more abundant in brain from the 23-month-old Cyp7b1-/-mice (0.03 ± 0.01 ng/mg) than the wt animals of the same age (0.02 ± 0.00 ng/mg, p < 0.05).

Sterols
We have previously reported that the concentration of cholesterol in brain does not vary between the Cyp7b1+/+ and Cyp7b1-/genotypes (10.60 ± 1.43 ng/mg at 13 months old and 15.81 ± 0.65 at 23 months old in wt animals) [14]. Measurements of cholesterol precursors is limited in the current LC-MS setting by the dynamic range of the chromatographic system, where injection of sufficient sample to achieve peak areas of the necessary size to allow accurate measurements of cholesterol precursors overloads the LC column with cholesterol. As the goal of the current study was primarily to measure oxysterols, we only measured cholesterol precursors in a single brain of each genotype from 13-month-old animals. Interestingly, the levels of both desmosterol and 8(9)-dehydrocholesterol (8-DHC), an enzymatically derived isomer of 7-dehydrocholesterol (7-DHC), are reduced in the Cyp7b1/-/mouse ( Figure 5). diHC/7α,24-diHCO, (e) 7α,25-diHC/7α,25-diHCO, (f) unresolved mixture of 7α,24-and 7α,25dihydroxysterols, and (g) 7α,26-diHC/7α,26-diHCO from the Cyp7b1+/+ mouse and (h) 24S,26-diHC from the Cyp7b1-/-mouse. 24,25-diHC is the hydrolysis product of 24S,25-EC. 24S,26-diHC was detected but not quantified. Many GP-derivatised sterols give syn and anti conformers which may or may not be chromatographically resolved.
The biochemical results presented above reveal an increased concentration of oxysterol ligands to the LXRs and to INSIG (insulin-induced gene) in brain of the Cyp7b1-/-mouse. The biological consequence of this is likely to be increased transport of cholesterol between cells through enhanced expression of ABC transporters and APOE [27], and of reduced cholesterol biosynthesis via inhibition of the SREBP-2 pathway [15].

Sterols
We have previously reported that the concentration of cholesterol in brain does not vary between the Cyp7b1+/+ and Cyp7b1-/-genotypes (10.60 ± 1.43 ng/mg at 13 months old and 15.81 ± 0.65 at 23 months old in wt animals) [14]. Measurements of cholesterol precursors is limited in the current LC-MS setting by the dynamic range of the chromatographic system, where injection of sufficient sample to achieve peak areas of the necessary size to allow accurate measurements of cholesterol precursors overloads the LC column with cholesterol. As the goal of the current study was primarily to measure oxysterols, we only measured cholesterol precursors in a single brain of each genotype from 13-month-old animals. Interestingly, the levels of both desmosterol and 8 (9)    Although cholesterol levels do not differ between the two genotypes, the reduction of cholesterol precursors in Cyp7b1-/mouse brain does suggest a reduced cholesterol synthesis as a result of oxysterol-induced inhibition of the SREBP-2 pathway.

Plasma
We have previously reported the concentrations of 26-HC and its downstream metabolites, 7α,26-diHC/7α,26-diHCO, 3β-HCA and 7αH,3O-CA in plasma [12]. Here, we extend that study by reporting concentrations of further oxysterols, including 24S,25-EC, 25-HC and other 7-oxidised metabolites ( Figure 6 and Supplementary Materials, Table S1). Although cholesterol levels do not differ between the two genotypes, the reduction of cholesterol precursors in Cyp7b1-/-mouse brain does suggest a reduced cholesterol synthesis as a result of oxysterol-induced inhibition of the SREBP-2 pathway.

Discussion
CYP7B1 is a known 7α-hydroxylase towards 25-HC, 26-HC and 3β-HCA [8,35]. CYP7B1 is expressed in brain [1][2][3] and data presented here demonstrates its activity in mouse brain towards 25-HC, as seen by elevated levels of 25-HC and reduction in levels of its metabolite 7α,25-diHCO in brain of the Cyp7b1-/mouse. 26-HD exists as (Z) and (E) isomers and both are present at very low levels in brain [20]; however, their elevated abundance in the Cyp7b1-/mouse indicates that, like 26-HC, 26-HD(Z) and 26-HD(E) are substrates for CYP7B1 in brain.
The major route for cholesterol removal from brain is via oxidation by CYP46A1 to 24S-HC [36], although other mechanisms also exist, most likely involving CYP27A1 and CYP7B1 [37,38]. CYP7B1 has little activity towards 24S-HC, the product of CYP46A1 oxidation of cholesterol, where 7α-hydroxylation is catalysed by CYP39A1 (cytochrome P450 family 39 subfamily A member 1) [39]. By blocking one of the pathways of cholesterol metabolism in brain by deletion of Cyp7b1, it may be predicted that either cholesterol or 24S-HC levels would increase; however, neither of these events occurs. This can be explained by down-regulation of cholesterol biosynthesis by inhibition of the SREBP-2 pathway. Besides 25-HC, 26-HC and 24S,25-EC are elevated in brain of the Cyp7b1-/mouse. Each will interact with the endoplasmic reticulum resident protein INSIG and block transport by SCAP (SREBP cleavage-activating protein) of SREBP-2 to the Golgi for processing to its active form as the master transcription factor for genes of the cholesterol biosynthesis pathway [15]. Although we do not prove this explanation here, there is no reason to expect that the INSIG-SCAP-SREBP regulatory mechanism differs in cells expressing or not expressing CYP7B1, other than with respect to the abundance of oxysterols and steroids. 25-HC, 26-HC, 24S,25-EC and 3β-HCA are also ligands to the LXRs [16][17][18], both of which are expressed in brain [40]. LXR target genes include Abca1, which codes for the cholesterol efflux pump ABCA1, and Apoe which codes for the apolipoprotein APOE, which mediates transport of cholesterol within the brain [19]. The lack of a major neurological phenotype in the Cyp7b1-/mouse, despite the absence of a key sterol metabolizing enzyme, is likely a consequence of the safety net provided by side-chain oxysterols, INSIG-SCAP-SREBP and LXR working in concert to avoid detrimental concentrations of cholesterol and of oxysterols building up in neuronal cells.
In humans, deficiency in CYP7B1 leads to SPG5, a form of spastic paraplegia, defined by progressive neurodegeneration of corticospinal tract motor neurons. One hypothesis to explain the different phenotype between CYP7B1-deficient mouse and human is that, in humans, oxysterol levels in the central nervous system are less well regulated and reach a level toxic to corticospinal tract neurons resulting in SPG5 [11]. In fact, Schöls et al showed that the concentration of total (the sum of esterified and non-esterified) 26-HC in serum is correlated with SPG5 disease severity and duration [11]. They found that at concentrations of 25-HC and 26-HC close to those found in serum of SPG5 patients, both oxysterols were toxic towards motor neuron-like cells as well as cortical neurons derived from human induced pluripotent stem cells (iPSCs) [11]. It should be noted that concentrations of these molecules in serum from SPG5 patients (total 25-HC 177.9 ± 77.0 ng/mL, mean ± SD, n = 19; 26-HC 878.2 ± 207.1 ng/mL, n = 19) were far higher than the value measured for 26-HC in cerebrospinal fluid (CSF, 12.9 ± 4.3 ng/mL, n = 19, concentration of 25-HC in CSF were not reported) and that the free molecules (i.e., non-esterified oxysterols), which were used in toxicity tests, are likely to be present at concentrations of about 10% of the sum of non-esterified and esterified molecules. 3β-HCA was also found to be toxic to the cells but at concentrations higher than those found in SPG5 serum [11].
The concentrations of 25-HC, 26-HC and 3β-HCA (363.1 ± 82.1 ng/mL, n = 19) reported by Schöls et al. in SPG5 serum [11] are considerably higher than those measured here in plasma for the Cyp7b1 -/mouse (25-HC, 36.52-42.19 ng/mL; 26-HC, 23.80-16.93 ng/mL; 3β-HCA, 7.32-6.69 ng/mL, numbers are means at the two ages). However, we measured the concentrations of the free non-esterified molecules, whereas Schöls et al. measured the total non-esterified plus esterified molecules [11]. A better comparison is provided by the data reported by Theofilopoulos et al., where the values for the non-esterified molecules were measured as 25-HC 49.40 ± 11.38 ng/mL, 26-HC 97.75 ± 7.28 ng/mL and 368.40 ± 65.27 ng/mL, n = 9, in SPG5 plasma/serum [12]. The concentrations of non-esterified 26-HC and 3β-HCA in human plasma/serum are considerably greater than those in the Cyp7b1-/mouse, and this may explain the presence of a motor neuron disease phenotype in SPG5 patients but not in the Cyp7b1-/mouse.
Theofilopoulos et al. also measured the concentration of 3β,7α-diHCA in human CSF and plasma [12]. They found that in SPG5 patients the levels of this acid were significantly reduced in both fluids compared to controls and, based on its neuroprotective effects towards oculomotor neurons, suggested that its reduced levels in SPG5 patients may contribute towards their motor neuron loss [12]. In the current study, we failed to detect 3β,7α-diHCA in either mouse brain or plasma. This is probably a consequence of its greater rate of metabolism to 7αH,3O-CA by HSD3B7 (3β-hydroxysteroid dehydrogenase type 7) in mice than in humans. CYP7B1 is an essential enzyme in the acidic pathway of bile acid biosynthesis [28,41]. Surprisingly, this is not reflected in the plasma concentration of 7αH,3O-CA in mice where we did not observe a statistical difference between the two genotypes. This can be explained by the existence of an alternative route to 7αH,3O-CA avoiding CYP7B1 and rather utilizing 7α-hydroxylation of cholesterol by CYP7A1 [33,41]. The resulting 7α-HC is then oxidized at C-3, before or after oxidation at C-26 by CYP27A1 to give 7αH,3O-CA.
As mentioned in Section 3.1.2, the level of cholesterol was measured in each of the mouse brains from both genotypes. Its precursor desmosterol and the 7-DHC isomer 8-DHC are also present in brain but at much lower levels (see Supplementary Materials, Table S1). The limited dynamic range of the LC-MS system makes the simultaneous analysis of cholesterol and of its precursors challenging. As the main focus in this study was on oxysterols and not cholesterol precursors, these sterols were not investigated thoroughly. Nevertheless, the data from single animals suggest reduced levels of cholesterol precursors in brain of the Cyp7b1-/animals. Further studies are required to confirm or refute this suggestion. Similarly, further studies are required to investigate the concentration in plasma of 12α-hydroxy metabolites, which in this work were detected but not measured in all animals. Their presence or absence in brain also requires further investigation.

Conclusions
Despite the elimination of a metabolic pathway for cholesterol removal in brain of the Cyp7b1-/mouse, the concentration of cholesterol is not elevated. This can be explained by a reduction in its synthesis due to a down-regulation of the cholesterol biosynthesis pathway by inhibition of SREBP-2 processing by side-chain oxysterols elevated in concentration in brain of the Cyp7b1-/mouse. Unlike human CYP7B1-deficiency, the Cyp7b1-/mouse does not have a pronounced neurological phenotype. Our results support the view that in mice the production of neurotoxic sterol metabolites is less marked than in humans, with the consequence that mice do not present with spastic paraplegia, while humans do.

Patents
The derivatization method described in this manuscript is patented by Swansea University, Swansea UK (US9851368B2) and licensed by Swansea Innovations to Avanti Polar Lipids and to Cayman Chemical Company.
Author Contributions: All authors contributed to writing, reviewing and editing the manuscript.
Funding: This research was funded by the UK Biotechnology and Biological Sciences Research Council, grant numbers BB/C515771/2, BB/I001735/1 and BB/N015932/1 to W.J.G., BB/L001942/1 to Y.W. and a studentship to A.M. Work in Edinburgh was supported by the Wellcome Trust (WT73429) and a Carter Fellowship to J.L.Y. from Alzheimer's Research UK.