Analysis of bioactive oxysterols in newborn mouse brain by LC/MS.

Unesterified cholesterol is a major component of plasma membranes. In the brain of the adult, it is mostly found in myelin sheaths, where it plays a major architectural role. In the newborn mouse, little myelination of neurons has occurred, and much of this sterol comprises a metabolically active pool. In the current study, we have accessed this metabolically active pool and, using LC/MS, have identified cholesterol precursors and metabolites. Although desmosterol and 24S-hydroxycholesterol represent the major precursor and metabolite, respectively, other steroids, including the oxysterols 22-oxocholesterol, 22R-hydroxycholesterol, 20R,22R-dihydroxycholesterol, and the C(21)-neurosteroid progesterone, were identified. 24S,25-epoxycholesterol formed in parallel to cholesterol was also found to be a major sterol in newborn brain. Like 24S- and 22R-hydroxycholesterols, and also desmosterol, 24S,25-epoxycholesterol is a ligand to the liver X receptors, which are expressed in brain. The desmosterol metabolites (24Z),26-, (24E),26-, and 7α-hydroxydesmosterol were identified in brain for the first time.


Animals
Brain extracts from newborn mice were from Children's Hospital, Oakland Research Institute (USA) and shipped to Swansea (UK) for sterol analysis. All animal work conformed to National Institutes of Health guidelines and was approved by the Institutional Animal Care and Use Committee.

Isolation of sterols/oxysterols from newborn mouse brain
Newborn mouse brain from phenotypically normal animals with a predominant C57BL/6 background ( 31 ) was homogenized and extracted in methanol-chloroform (1:1, v/v), and reextracted in methanol. The combined extracts were taken to dryness. The lipid extracts, equivalent of 100 mg of wet tissue, were solubilized in 1.05 ml of 99.9% ethanol containing 50 ng of 24(R/S)-[ 2 H 6 ] hydroxycholesterol and 50 g of [ 2 H 7 ]cholesterol and ultrasonicated for 15 min at room temperature. Then 0.45 ml of water was added, and the mixture was sonicated for another 15 min. The mixture was centrifuged at 14,000 g at 4°C for 60 min, and the supernatant was collected. The extraction procedure was repeated on the lipid residue (with a second 1.05 ml ethanol containing reference standards followed by addition of 0.45 ml water), and the supernatants were combined (fi nal volume 3 ml).
To minimize the possibility of autoxidation of cholesterol during sample preparation, oxysterols were separated from cholesterol in an initial reverse-phase (RP)-SPE step. A Certifi ed Sep-Pak C 18 (200 mg; Waters) cartridge was washed with 4 ml of 99.9% ethanol followed by an equilibration rinse with 6 ml of 70% ethanol. removed from brain by two mechanisms ( 6 ), one of which is metabolism to 24S-hydroxycholesterol, which is able to cross the BBB and is transported via the circulation to the liver for further metabolism ( 7 ), and the other has yet to be fully described. In the fetal mouse, the expression of cholesterol 24-hydroxylase (Cyp461a1) is low, possibly to conserve cholesterol ( 4 ). However, during the fi rst 2 weeks of life, there is a signifi cant increase in Cyp46a1 mRNA levels in parallel with an increase in 24S-hydroxycholesterol levels ( 8,9 ).
24S-hydroxycholesterol, like 24S,25-epoxycholesterol, is a ligand in vitro to the liver X receptors (LXRs) ( 10,11 ); this is also true of 22R-hydroxycholesterol. 22R-hydroxycholesterol is an intermediate in the conversion of cholesterol to pregnenolone by the cytochrome P450 (Cyp) side-chain cleavage enzyme (P450 SCC , Cyp11a1). This enzyme converts cholesterol fi rst to 22R-hydroxycholesterol then to 20R,22R-dihydroxycholesterol, and ultimately to pregnenolone, thus offering a route for neurosteroid biosynthesis ( 12 ). 22R-hydroxycholesterol has been reported to be present in human brain and to also have neuroprotective effects in vitro against amyloid ␤ toxicity ( 13,14 ).
Side-chain oxidized sterols can have a regulatory role in cellular cholesterol homeostasis in that they are ligands to Insigs (insulin-induced genes), which, when bound to oxysterols bind to SCAP (SREBP cleavage-activating protein), anchoring the complex in the ER and preventing the transport by SCAP of SREBPs (sterol-regulatory element binding proteins) to the Golgi for processing to their active forms as transcription factors ( 15 ). SREBP-2 activates genes encoding the enzymes of the mevalonate pathway required for cholesterol biosynthesis and also for cholesterol uptake ( 16 ). Cholesterol can also regulate the processing of SREBPs, but in this case, by binding to SCAP, which triggers binding to Insigs, anchoring SCAP in the ER and preventing transport of SREBPs to the Golgi for processing ( 17 ). Current concepts suggest that although cholesterol is the major regulator of its own synthesis, oxysterols have a role in fi ne-tuning acute cholesterol homeostasis and protecting the cell against over-accumulation of newly synthesized cholesterol ( 18 ).
Analysis of sterols, oxysterols, and other steroids in brain can be challenging. This is because of the high level of cholesterol in comparison to that of other steroids and its propensity to undergo autoxidation ( 19,20 ). The major auto xidation products of cholesterol are the oxysterols 7-oxo-, 7 ␤ -hydroxy-, 7 ␣ -hydroxy-, 25-hydroxy-, 5 ␤ , 6 ␤ -epoxy-, 5 ␣ ,6 ␣ -epoxy-, and 5 ␣ ,6 ␤ -dihydroxycholesterols, whereas other steroids, e.g., pregnenolone, dehydroepiandrosterone, can also be formed ( 19,20 ). To minimize this problem, many workers utilize butylhydroxytoluene (BHT) as an antioxidant during extraction and/or separation of oxysterols (and neurosteroids) from cholesterol prior to analysis ( 5,(21)(22)(23)(24). MS, coupled with either GC or LC, offers the most-sensitive method of oxysterol and steroid analysis in brain ( 25 ). However, GC-MS requires derivatization of oxysterols to avoid thermal decomposition ( 26 ), whereas LC-MS, in the absence of derivatization, is usually operated in the multiple reaction monitoring mode to achieve suffi cient sensitivity. We and others have above ( ‫ف‬ 3 ml, ‫ف‬ 70% methanol) was applied to the column, and effl uent was retained. The sample tube was then washed with 1 ml of 70% methanol and applied to the column, followed by a column wash of 1 ml 35% methanol. The effl uents were pooled (now ‫ف‬ 5 ml) and diluted with 4 ml of water to give 9 ml of 35% methanol. This solution was then applied to the column, and the fl ow-through was collected, followed by a wash with 1 ml of 17% methanol. The fl ow-through and wash were diluted with 9 ml of water to give 19 ml of ‫ف‬ 17% methanol. This solution was again loaded onto the column, followed by a wash with 6 ml of 10% methanol. At this point, all the derivatized sterols/oxysterols were retained on the column whereas excess derivatization reagent was in the 10% methanol wash. Derivatized sterols/oxysterols were then eluted in three separate 1 ml portions of 100% methanol (SPE2-FR1, -FR2, -FR3) followed by 1 ml of 99.9% ethanol (SPE2-FR4). ESI-MS analysis revealed that the derivatized oxysterols were present in the fi rst and second ml of methanol (see supplementary information, Results). Thus, for subsequent analysis of oxysterols, the fi rst two milliliters were combined, whereas for sterol analysis, all three milliliters were combined.

LC-MS and MS n analysis
The LC-MS(MS n ) system consists of an LTQ-Orbitrap XL (Thermo Fisher Scientifi c, UK) equipped with an ESI probe, and a Dionex Ultimate 3000 LC system (Dionex, UK). Chromatographic separation was performed using a Hypersil Gold RP column (50 × 2.1 mm, 1.9 µm; Thermo Fisher Scientifi c) at room temperature. Mobile phase A consisted of 0.1% formic acid in 33.3% methanol, 16.7% acetonitrile. Mobile phase B consisted of 0.1% formic acid in 63.3% methanol, 31.7% acetonitrile. Two different gradients were employed. Gradient 1: After 1 min at 20% B, the proportion of B was raised to 80% B over the next 7 min and maintained at 80% B for further 5 min, before returning to 20% B in 0.1 min. The column was reequilibrated for a further 3.9 min, giving a total run of 17 min. Gradient 2; After 1 min at 20% B, the proportion of B was raised to 55% B over the next 19 min using concave curve gradient (curve 8 according to Chromeleon software). From minute 20, the proportion of B was increased linearly to 80% over 10 min, before returning to 20% B in 0.1 min. The column was reequilibrated for 3.9 min, giving a total run time of 34 min. The longer gradient, Gradient 2, was utilized to enhance the resolution of closely eluting oxysterols but offered reduced throughput and lower sensitivity because peaks became broader. The fl ow rate for each gradient was 200 µl/ min, and the eluent was directed to the ESI source of the LTQ-Orbitrap XL mass spectrometer.
The LTQ-Orbitrap XL mass spectrometer was calibrated using "Cal mix" (Thermo Fisher Scientifi c) and tuned using GP charge-tagged 19-hydroxycholesterol prior to each analytical session. For subsequent LC-MS and LC-MS n system tests, GP chargetagged 19-hydroxycholesterol (1 pg/ l in 60% methanol, 0.1% formic acid) was injected (20 l) onto the RP column and eluted into the ESI source at a fl ow-rate of 200 µl/min. The analysis of this reference compound was repeated periodically during the analytical batch (every 10 injections) to ensure mass accuracy and to monitor chromatographic performance. For the analysis of GP charge-tagged sterols/oxysterols from brain, the methanol eluents from the SPE2 fractions were diluted to give solutions of 60% methanol, 0.1% formic acid, 3 pg/ l 24(R/S)-[ 2 H 6 ] hydroxycholesterol. Twenty microliters of solution was injected onto the LC column (equivalent to 0.06 mg brain). Note that while stable in 100% methanol, the stability of GP-tagged steroids is variable in 60% methanol/0.1% formic acid, and analytes should not be stored in this solution for a prolonged period (>48 h). The brain extract in 70% ethanol (3 ml) was applied to the cartridge and allowed to fl ow at a rate of 0.25 ml/min. To aid the fl ow through the cartridge, a negative pressure was applied to the outlet of the cartridge via a vacuum manifold (Vacuum Processing Station; Agilent, Waghaeusel-Wiesental, Germany). The fl owthrough was collected. This fraction contains oxysterols ( 32 ). A wash with 4 ml 70% ethanol was similarly collected and combined with the fl ow-through to give fraction SPE1-FR1 (see supplementary Fig. I). The column was washed further with 4 ml of 70% ethanol (SPE1-FR2), and cholesterol was eluted from the cartridge with 2 ml of 99.9% ethanol (SPE1-FR3). A fi nal 2 ml of 99.9% ethanol was applied to the column to elute more hydrophobic sterols from the cartridge (SPE1-FR4). Each fraction was divided into two equal volumes, A and B, and all fractions were dried in a Scanspeed vacuum concentrator (Coolsafe; West Sussex, UK) and reconstituted in 100 µl of propan-2-ol.
An identical procedure was carried out with brains from 15 week young adult female mice.

Derivatization of 3-oxo-4-ene sterols/oxysterols with Girard P reagent
Neutral sterols/oxysterols are by defi nition neither acidic nor basic, and are poorly ionized in the atmospheric pressure ionization process; thus, a derivatization method was exploited to enhance their LC-MS analysis ( 22,25 ). Sterols/oxysterols possessing an oxo group (either naturally or via treatment with cholesterol oxidase; see supplementary Fig. II) were derivatized with the GP reagent, which contains a charged quaternary nitrogen ( 22,34 ). To each of the fractions from above (SPE1-FR1A/B → SPE1-FR4A/B), 150 l of glacial acetic acid and 150 mg of GP hydrazine were added (see supplementary Fig. II). The addition of glacial acetic acid increases the rate of hydrazone formation. The mixture was incubated at room temperature overnight in the dark. We call the combined process of oxidation with cholesterol oxidase and derivatization with GP reagent "chargetagging."

SPE recycling of sterols/oxysterols
Derivatized sterols/oxysterols were separated from the excess of GP hydrazine by a recycling method ( 22 ). Sterol/oxysterol molecules, even when derivatized to the GP hydrazones, are still prone to precipitation in highly aqueous solutions. To tackle this problem, a recycling method was used, where initially, analytes dissolved in 70% methanol were recycled on a second 200 mg Certifi ed Sep-Pak C 18 cartridge (SPE2). Prior to application of the sample, the column was washed with 6 ml of 100% methanol and 6 ml of 10% methanol before conditioning with 4 ml of 70% methanol. The fl ow of solvent through the column was aided by a negative pressure applied to the column outlet by a vacuum manifold. The derivatization mixture from RESULTS AND DISCUSSION Oxysterols in brain are accompanied by a 10 3 -fold excess of cholesterol ( 2 ). Therefore, even a minor degree of autoxidation of cholesterol will signifi cantly alter the oxysterol profi le of a sample. Thus, steps should be taken to minimize this possibility. One solution to the problem is to add the antioxidant BHT during sterol/oxysterol extraction and/or perform sample preparation under anaerobic conditions ( 21,24 ). Alternatively, and as performed here, oxysterols are separated from cholesterol at the earliest possible stage of the sample preparation procedure.
Following extraction and GP charge-tagging, the oxysterol and cholesterol fractions from brain were further resolved by RP chromatography, and analytes were identifi ed on the LTQ-Orbitrap XL. The identifi cation of compounds was based on retention time, exact mass, and MS n fragmentation pattern and comparison with authentic standards. In the absence of authentic standards, presumptive identifications were made based on these data. The availability of authentic standards is noted in supplementary Table II The LTQ-Orbitap XL was operated utilizing three scan events. First, a Fourier transform (FT)MS scan in the Orbitrap over the m/z range 400-605 at 30,000 resolution (full width at half-maximum height) was performed, followed by data-dependent MS 2  → ) events in the linear ion trap (LIT). These MS n scans in the LIT were performed in parallel to acquisition of the high-resolution MS scan by the Orbitrap. For the MS 2 and MS 3 scans, the precursor-ion isolation width was set at 2 (to select the monoisotopic ion) and the normalized collision energy at 30 and 35 (instrument settings), respectively. A precursor-ion inclusion list was defi ned according to the m/z of the [M] + ions of expected sterols/oxysterols so that MS 2 was preferentially performed on these ions in the LIT if their intensity exceeded a preset minimum (500 counts) (see supplementary Table I). If a fragment-ion corresponding to a neutral loss of 79 Da from the precursor-ion was observed in the MS 2 event and was above a minimal signal setting (200 counts

Recovery experiments
Experiments to evaluate the recovery of different oxysterols were performed by adding known amounts of oxysterol standards (22R-, 24S-, 25-, and (25R),26-hydroxycholesterol) to a portion of homogenized brain. This sample and an equivalent portion of homogenate without added oxysterols were processed and analyzed in parallel. Recovery was calculated as the measured oxysterol level divided by the calculated level.

Luciferase reporter assay
The ability of sterols and oxysterols to activate LXR ␣ and -␤ was tested in luciferase assays. Transient transfection studies were performed in the mouse substantia nigra-like cell line SN4741. This cell line, which is of neural origin, was selected because the sterols and oxysterols were identifi ed in brain. Cells were plated in 24-well plates (5 × 10 5 cells/well) 24 h before transfection and transfected with 1 µg of plasmid DNA/well complexed with 2 µl of lipofectamine 2000 (Invitrogen). Cells were transfected with 400 ng of an LXR-responsive luciferase reporter construct, and 200 ng of LXR ␣ or LXR ␤ . A reporter gene expressing the Renilla luciferase (pRL-TK; Promega) was cotransfected in all experiments as an internal control for normalization of transfection effi ciency. After a 12 h incubation, the lipid/DNA mix was replaced with fresh 2.5% serum medium containing vehicle or appropriate ligand (10 µM), as specifi ed in each experiment. Luciferase activities were assayed 24 h later using the Dual-Luciferase Reporter Assay System (Promega), following the manufacturer's protocol. of cholesterol ( 8 ). The Cyp46a1 enzyme is normally expressed in neurons. In newborn mouse brain, the level of 24S-hydroxycholesterol was determined to be 0.510 ± 0.082 ng/mg (mean ± SD, n = 6). This value is considerably lower than that determined here for the young female adult (27.14 ± 1.71 ng/mg, Fig. 1 ; supplementary Table II) or in the published literature for the adult mouse (30-60 ng/mg) ( 9,21,25,35 ). This is not surprising, inasmuch as Cyp46a1 is not expressed until late embryonic development in mouse ( 4,9 ), and in human, Lütjohann et al. ( 37 ) have shown that CYP46A1 activity is low in neonates. GP-tagged 24S-, 25-, and 26-hydroxycholesterols (cholest-(25R)-5-ene-3 ␤ , 26-diol) give closely eluting peaks but characteristically different MS 3 spectra ( Figs. 2D-F ). This allows the identification of all three oxysterols. However, in this study, the comparatively high level of 24S-hydroxycholesterol and low levels of the other two closely eluting oxysterols (<0.010 ng/mg) made accurate quantifi cation of the minor oxysterols diffi cult. Nevertheless, by utilizing the extended chromatographic gradient (Gradient 2), it was possible to

Oxysterols in newborn mouse brain
Monohydroxycholesterols ( m/z 534.4054). As in adult mouse brain ( 9,21,24,25,35 ), the most-abundant monohydroxycholesterol found in newborn mouse brain is 24Shydroxycholesterol ( Fig. 1 ; supplementary Table II, see  also supplementary Table II in ( 25 ) and Table III in ( 36 ) for published data on adult mouse). Once GP chargetagged, this oxysterol gives an [M] + ion of m/z 534.4054. The RIC of this m/z from newborn mouse brain using chromatographic Gradient 1 is shown in Fig. 2A .   Table II for numerical data and further information. Shown in C and D are profi les of sterols and oxysterol, respectively, in 15 week female mouse brain . Data are mean ± SD. Cyp7a1 is expressed only in liver ( 45 ). Both 7 ␣ -and 7 ␤hydroxycholesterol can be produced by autoxidation of cholesterol during sample preparation and storage ( 20 ), or they may be formed endogenously by reaction of cholesterol with reactive oxygen species ( 46 ). In some, but not all samples, a peak giving MS 2 and MS 3 spectra identical to those of the authentic standard of 6 ␤ -hydroxycholesterol was observed (see supplementary Table II). 6 ␤ -Hydroxycholesterol is a product of 5,6-epoxycholesterol following hydrolysis and dehydration. The remaining chromatographic peak to be characterized corresponds to a hydroxycholesterol eluting at 8.91 min ( Fig. 2A) and at 22.52 min ( Fig. 2B) (0.045 ± 0.015 ng/mg). The MS n spectra suggest hydroxylation at a primary or secondary carbon on the C 17 side-chain ( Fig. 2G ). One possibility is cholest-(25S)-5-ene-3 ␤ ,27-diol; however, the retention time does not correspond to that of the authentic standard, which elutes just before the 25R isomer. Note: in this study, we use the nomenclature recommended by lipid maps where 26-hydroxycholesterol corresponds to cholest-(25R)-5-ene-3 ␤ ,26-diol ( 47 ). In summary, the pattern in newborn brain is of one dominating monohydro xycholesterol, i.e., 24S-hydroxycholesterol ( Fig. 1 ). This is similar to that observed here in the young female adult and to published data for adult mouse brain ( 21,35 ). However, it is noteworthy that an array of monohydroxycholesterol isomers are observed in newborn mouse brain, and that a similar situation exists in embryonic brain ( 5 ).
Oxo-and epoxycholesterols and hydroxydesmosterols ( m/z 532.3898). Other oxysterols found by Wang et al. ( 5 ) to be present in embryonic brain include 24S,25-epoxycholesterol and 24-oxocholesterol, both of which give an [M] + ion at m/z 532.3898. RICs for this m/z , shown in Fig. 3A, B , are generated with the short and long gradient, respectively. The two peaks eluting at RT 6.80 and 7.01 min ( Fig.  3A) correspond to the syn and anti conformers of GPtagged 24S,25-epoxycholesterol (0.067 ± 0.012 ng/mg). In Fig. 3B, the corresponding peaks are at 12.47 and 13.93 min, respectively. The MS 3 spectra are identical to those of the authentic standard, giving the characteristic low-mass ions at m/z 151, 163, and 177, and side-chain cleavage fragment ions at 325 ('*e), 353 ('*f), and 367 ('*g) ( Fig. 3D ). The chromatographic peak eluting at 7.79 min ( Fig. 3A) was identifi ed as GP-tagged 24-oxocholesterol (0.031 ± 0.008 ng/mg). The presence of a 24-oxo group rather than the 24,25-epoxide is indicated by the elevated abun-  Fig. 3D, G ). 24-Oxocholesterol is an isomerization product of 24S,25-epoxycholesterol formed during the derivatization procedure. Analysis of the newborn mouse brain samples also revealed a peak with a retention time of 8.05 min ( Fig. 3A), which gave an MS 3 spectrum characteristic of GP-tagged 22oxocholesterol (0.060 ± 0.025 ng/mg) ( Fig. 3H ). Similar to improve the resolution of these isomers and to make quantitative estimates of oxysterol concentrations ( Fig. 2B ). 25-Hydroxycholesterol is formed from cholesterol either by cholesterol 25-hydroxylase (Ch25h) ( 38 ) or in Cyp46a1catalyzed reactions ( 8,39 ), whereas 26-hydroxycholesterol is formed in Cyp27a1-catalyzed reactions ( 40,41 ). Further scrutiny of the RIC of m/z 534.4054 reveals ( Fig. 2A) a minor peak at 8.24 min with retention time and MS 3 spectrum identical to charge-tagged 24R-hydroxycholesterol. Reference to the authentic standard indicates that this compound elutes as syn and anti conformers, with the earlier eluting component eluting at 7.88 min. These conformers elute at 19.21 and 20.71 min with the longer gradient ( Fig. 2B ). The level of 24R-hydroxycholesterol was determined to be 0.061 ± 0.006 ng/mg. 24R-hydroxycholesterol has not previously been characterized in brain. In mouse, Cyp46a1 accounts for production of 80% of the 24-hydroxycholesterol content of plasma, and Lund et al. ( 42,43 ) suggested that the remainder may be formed in a reaction catalyzed by Cyp27a1, inasmuch as earlier studies in pig revealed that the hepatic mitochondrial sterol 27-hydroxylase has 24-, 25-, and (25R),26-hydroxylase activity. It is uncertain whether the 24R-isomer found here in newborn brain is formed as a minor product of Cyp46a1 activity in brain or is an import product hepatically derived from Cyp27a1 oxidation of cholesterol. 24R-hydroxycholesterol was not detected in young adult female mouse brain. Preceding the major peaks in the RIC of m/z 534.4054 is a minor peak eluting at 6.22 min ( Fig. 2A) , identifi ed by retention time and its MS 3 spectrum ( Fig. 2C ) to correspond to GP-tagged 22R-hydroxycholesterol (0.010 ± 0.002 ng/mg). The presence of a hydroxy group at the C-22 is characterized by an abundant ion at m/z 355 (*f') resulting from a cleavage of the C-20-C-22 bond ␣ to the 22-hydroxy group, and at m/z 273 corresponding to the charge-tagged ABC rings ( 22 ). 22R-hydroxycholesterol is formed from cholesterol in the fi rst step of a P450 SCC (Cyp11a1) catalyzed reaction and may be further transformed via 20R,22Rdihydroxycholesterol into pregnenolone by the same enzyme. The enzyme is expressed in adult rodent brain and also during embryogenesis ( 44 ). Other monohydroxycholesterols identifi ed in newborn mouse brain were 7 ␤ -, 7 ␣ -, and 6-hydroxycholesterols, at levels of 0.039 ± 0.015, 0.048 ± 0.022, and 0.019 ± 0.014 ng/mg ( Fig. 2H ). The presence of an additional hydroxy group in the B ring of the steroid nucleus results in a changed pattern of lowmass fragment ions in the MS 3 spectra (see supplementary Fig. IIIC). 7 ␣ -or 7 ␤ -Hydroxylation causes a shift in the *b 3 -C 2 H 4 ion from m/z 163 to 179. There is an elevated abundance of the ion at m/z 151 (*b 1 -12) in the spectrum of the 7 ␤ -isomer compared with the 7 ␣ -isomer, whereas the peaks at m/z 231 and 394 are more pronounced in the spectrum of the 7 ␣ isomer. 7 ␣ -Hydroxycholesterol can be formed enzymatically in a Cyp7a1-catalyzed reaction, but X-hydroxycholesterol (X-HC); and H: 7 ␣ -hydroxycholesterol (7 ␣ -HC). Assignment is based on a comparison of retention time, exact mass, and MS n spectra with those of authentic standards. When charge-tagged with the GP-hydrazone, syn and anti conformers are formed about the hydrazone group that may or may not be chromatographically resolved. In this study, we use the nomenclature recommended by lipid maps where 26-hydroxycholesterol corresponds to cholest-(25R)-5-ene-3 ␤ ,26-diol ( 47 ). and Javitt ( 30 ) have shown that desmosterol can be 26-hydroxylated by human recombinant CYP27A1, whereas Heverin et al. ( 52 ) have identifi ed this oxysterol in the cholesterol-free mouse. The retention time of an authentic charge-tagged (24Z),26-hydroxydesmosterol (cholesta-5, 24Z-diene-3 ␤ ,26-diol) run with the short gradient is 7.16 min, and a peak appearing in the RIC of m/z 532.3898 at this retention time (0.048 ± 0.009 ng/mg) gives a spectrum essentially identical to that of the authentic standard ( Fig. 3E ). By utilizing the extended chromatographic gradient (24Z), 26-hydroxydesmosterol was completely resolved from the second peak of 24S,25-epoxycholesterol (cf. Fig. 3A, B ). Hydroxylation of the terminal carbon of desmosterol introduces Z or E stereochemistry about the C-24-25 double bond, and these isomers are readily resolved by utilizing the extended gradient, allowing the identifi cation also of (24E),26-hydroxydesmosterol (cholesta-5, 24E-diene-3 ␤ ,26-diol) eluting at 16.85 min in the chro matogram of newborn brain (0.015 ± 0.004 ng/mg; supplementary Fig. IVB). A minor peak is seen ( Fig. 3A) at 5.97 min, and the MS 3 spectrum ( Fig. 3C ) is again compatible with a GP-tagged hydroxydesmosterol, with an added hydroxy group on the side-chain (0.002 ± 0.002 ng/mg). Further inspection of the RIC of m/z 532.3898 ( Fig. 3A ) reveals peaks at 8.67 and 9.29 that give MS 3 spectra compatible with GP-tagged 7 ␤ -and 7 ␣ -hydroxydesmosterols, respectively (0.020 ± 0.006 and 0.024 ± 0.012 ng/mg) (see supplementary Fig. IVC, D). The 7 ␣ isomer is probably formed, at least in part, endogenously in a Cyp-catalyzed hydroxylation of desmosterol. Cyp7a1 has activity toward desmosterol, although 7 ␣ -hydroxydesmosterol was not found in plasma of the cholesterol-free mouse ( 52 ). A peak at 9.66 min ( Fig. 3A) gives an MS 3 spectrum commensurate with a GP-tagged 6-hydroxydesmosterol (0.023 ± 0.011 ng/mg), probably derived from 5,6-epoxydesmosterol (see supplementary Fig. IVE). The hydroxydesmosterol molecules described here have not previously been found in brain, and of the isomers characterized, only (24Z),26hydroxydesmosterol is found in the young adult.  ( 22 ). The *f' fragment ion results from a cleavage ␣ to the 22-oxo group, with elimination of either a ketene or CO and an alkene (see supplementary Fig. IIID). Further CO and NH moieties are eliminated from the remnants of the derivatizing group to give ions at m/z 327 and 312, respectively. 22-Oxocholesterol has not previously been reported to be present in brain tissue, nor do we fi nd it in the young female adult brain. However, there is a considerable volume of older literature debating its involvement in the side-chain shortening of cholesterol to pregnenolone ( 48,49 ). The dominant peak in the RIC of m/z 532.3898 is at 7.42 min in Fig. 3A  Only one peak is evident, appearing at 7.42 min, and the corresponding MS 3 spectrum is shown in Fig. 3F . The spectrum is compatible with a GP-tagged sterol unsaturated and hydroxylated in the side-chain, possibly 23-hydroxydesmosterol. There is a precedent for C-23 hydroxylation of sterols as part of the 25-hydroxylase pathway to bile acids, but whether this pathway is applicable to desmosterol is unknown ( 50,51 ). Perhaps signifi cantly, an essentially normal bile acid pattern is seen in the cholesterol-free mouse, in which desmosterol is the dominant sterol ( 52 ). It should be noted, however, that C-23 is an allylic site in desmosterol and is thus susceptible to nonenzymatic oxidation. An equivalent peak was not observed in the corresponding chromatogram from young adult female brain. The partial identifi cation of a hydroxylated desmosterol encouraged us to search for further isomers of this molecule. Pikuleva  ion at m/z 453. 24,25-Dihydroxycholesterol can he formed by hydrolysis of chemically labile 24S,25-epoxycholesterol. The acidic methanolic environment of derivatization can also cause methanolysis of 24S,25-epoxycholesterol, leading to formation of hydroxymethoxycholesterol isomers (0.294 ± 0.177 ng/mg) ( Fig. 4D ). Analysis of the RIC of m/z 564.4160 ( Fig. 4D ) showed two peaks with retention times and MS 3 fragmentation profi les identical to those obtained during the analysis of the GP-tagged reference standard ( Fig. 4E ). The [M-79-32] + ion ( m/z 453) is characteristic of the presence of a methoxy group that eliminates as methanol. The combined concentration of 24S,25-epoxycholesterol (0.067 ng/mg), 24-oxocholesterol (0.031 ng/mg), 24,25-dihydroxycholesterol (0.731 ng/mg), and hydroxymethoxycholesterol (0.294 ng/mg) equals 1.123 ± 0.304 ng/mg. 24S,25-Epoxycholesterol is formed in a shunt of the mevalonate pathway, so the comparatively high level of this oxysterol is not unexpected for a developing tissue with high rate of cholesterol synthesis ( 1 ). For comparison, the level of 24S,25-epoxycholesterol is only 0.64 ng/mg in the young adult female mouse. Close inspection of the RIC for m/z 550.4003 ( Fig. 4A) reveals a shoulder (4.48 min) on the leading edge of the peak at 4.62 min. This is partially resolved with the longer gradient (see supplementary Fig. VA) The MS 3 spectrum recorded suggests the presence of 20R,22R-dihydroxycholesterol (0.020 ± 0.007 ng/mg) ( Fig. 4C ) providing the fi rst direct evidence for the presence of this cholesterol metabolite in brain. 20R,22R-Dihydroxycholesterol is an intermediate in the conversion of cholesterol to pregnenolone. Its presence in newborn brain lends weight to the argument that neurosteroids are formed in brain, rather than just passengers from the circulation ( 12 ).

Neurosteroids in newborn mouse brain
Because we are able to identify both 22R-hydroxycholesterol and 20R,22R-dihydroxycholesterol in brain, the possibility exists that neurosteroidogenesis is proceeding in brain ( Fig. 5 ). Although we failed to identify pregnenolone, progesterone was conclusively identifi ed (0.001 ± 0.001 ng/mg) (see supplementary Fig. VI). Liu, Sjövall, and Griffi ths ( 23 ) also found progesterone to be the most-abundant neurosteroid in adult rodent brain (0.001-0.020 ng/mg).

Sterols in newborn mouse brain
Cholesterol and its precursors desmosterol (Bloch pathway) and 7-dehydrocholesterol (Kandutsch-Russell pathway) elute from the fi rst SPE column in fraction 3 (SPE1-FR3). Following GP charge tagging, this fraction was also analyzed by LC-MS. Due to the high concentration of cholesterol, it was necessary to dilute the eluent from the second SPE column by a factor of 10,000-20,000 prior to injection on the LC determined a similar value for 1 day-old animals. These values compare with levels in the young female adult, determined here to be 16.026 ± 0.758 µg/mg, and with literature values for the adult of 10-20 µg/mg ( 9,21,31,35 ). Cholest-4-en-3-one is also present in newborn brain, but at a level of about 0.1% that of cholesterol. Lathosterol is an isomer of cholesterol and a precursor in the Kandutsch-Russell pathway. Like cholesterol, it gives an [M] + ion of m/z 518.4105. The dynamic range of our LC-MS system is insuffi cient to observe this isomer against the dominating background of cholesterol. Others have also experienced similar problems using LC-MS for the analysis of sterols in plasma ( 53 ). The RIC for m/z 516.3948, corresponding in m/z to GP-tagged desmosterol and 7-dehydrocholesterol, revealed one major peak, with a second appearing as a doublet (see supplementary Fig. VIIIA). The fi rst peak, eluting at 11.33 min, corresponds to GP-tagged desmosterol (493.986 ± 18.062 ng/mg) (see supplementary Fig.  VIIIB). This corresponds to about 20% of the total sterol content of newborn brain and agrees well with the value determined by Tint et al. ( 4 ) for a 22day fetus. Tint et al. also showed that in the fetus and newborn animal, the expression of 24-dehydrocholesterol reductase ( Dhcr24 ) in brain is low, explaining the high level of desmosterol. For comparison, in the young female adult, the desmosterol proportion of total sterols is less than 1%. The fi rst component of the second peak, eluting at 12.07, gives an MS 3 spectrum identical to that of GP-tagged 7-dehydrocholesterol (see supplementary Fig. VIIIC). The second component, at 12.28 min, gives an MS 3 spectrum that we identify as cholesta-4,6-dien-3 ␤ -ol. This may be an isomerization product of 7-dehydrocholesterol. Each analyte is present at high concentrations in newborn brain, i.e., 210.484 ± 8.319 ng/mg, and 46.692 ± 6.736 ng/mg, respectively. The prevalence of cholesterol precursors indicates a high rate of cholesterol synthesis during neonatal development and is in good agreement with earlier reports showing that in the developing fetus and suckling newborn, the rates of sterol accumulation in brain are greatest ( 54 ). Quan et al. ( 54 ) measured both desmosterol and cholesterol levels in the central nervous system of mice, and at their earliest time point of 2 days old, desmosterol made up 30% of total sterols, whereas at day 15, desmosterol had "essentially disappeared." The high levels of desmosterol in the newborn animal are particularly interesting in light of recent data indicating that desmosterol can suppress SREBP processing and activate LXR target genes ( 55,56 ). The precursor of desmosterol is 7-dehydrodesmosterol. A component eluting at 10.85 min in the RIC for m/z 514.3792 (see supplementary Fig. IXA) gives an MS 3 spectrum commensurate with that expected for GP-tagged 7-dehydrodesmosterol (41.159 ± 2.503 ng/mg), and a later-eluting peak at 11.59 min is assigned to cholesta-4,6,24-trien-3 ␤ -ol (113.33 ± 7.018) (see supplementary Fig. IX). This may be an isomerization product of 7-dehydrodesmosterol. Again, a high level of cholesterol precursors in newborn brain can be explained by a low expression of Dhcr24.
A disadvantage of the analytical method employed in the present study is the laborious requirement to work up In the absence of cholesterol oxidase, only sterols/oxysterols with a natural oxo group will be detected, inasmuch as cholesterol oxidase is required for the detection of sterols/oxysterols with a 3 ␤ -hydroxy-5-ene function. A further disadvantage is that cholesterol oxidase from Streptomyces sp. is inactive toward lanosterol and 14-des methyllanosterol ( 33 ), presumably because of the two methyl groups on C-4; so these two molecules are transparent to analysis. It is worth noting that Roberg-Larsen et al. ( 57 ) have now developed an automated SPE-LC-MS/MS method for analysis of oxysterols derivatized with the GP reagent.

Sterols and oxysterols as LXR ligands
Many of the sterols and oxysterols identifi ed above have previously been tested for their activation capacity toward nuclear receptors ( 10,11,56 ). Here we compliment earlier work by testing the activity of desmosterol, the chemically synthesized 24Z and 24E isomers of 26hydroxydesmosterol, and the structural isomers 24-and 22-oxocholesterol in a cell line of neural origin ( Fig. 6 ).

CONCLUSIONS
There are three major fi ndings of the current study. First, we detect the C 21 -neurosteroid progesterone and its biosynthetic precursors 22R-hydroxycholesterol and 20R,22R-dihydroxcholesterol in newborn brain ( Fig. 5 ). Second, we fi nd high levels of the LXR and Insig ligand 24S,25-epoxycholesterol in newborn brain ( Fig. 1B ), and third, we identify numerous oxysterols derived from desmosterol. We also fi nd that desmosterol, but not its 26-hydroxy metabolites, is a ligand to LXR in neural cells ( Fig. 6 ). Furthermore, this is the fi rst identifi cation of the (24Z),26-hydroxydesmosterol isomer in a biological system. These fi ndings have been made utilizing a "charge-tagging" approach for oxysterol/sterol LC-MS(MS n ) analysis, which offers high sensitivity and allows structural characterizations.
In recent years, there has been some controversy over whether and to what extent neurosteroidogenosis proceeds in brain ( 12,19,23,60 ). Neurosteroids are C 18 -C 21 steroids synthesized in brain. In general, they mediate their actions not through classical steroid hormone (nuclear) receptors, but via ion-gated neurotransmitter receptors, in which case they are called neuroactive steroids. Recent data have also implicated neurosteroids with the infl ammatory response in brain, but in this case, via selective modulation of the estrogen receptor ␤ ( 61 ). It is often unclear whether a steroid found in brain is synthesized in the brain itself or imported from the circulation. Data presented here, confi rming the presence of intermediates of a neurosteroid biosynthetic pathway in developing mouse brain, lends weight to the neurosteroid hypothesis. However, it should be noted that detection of a steroid in situ does not necessarily equate to its synthesis in situ. The present study was not designed to specifi cally analyze neurosteroids; however, this could be achieved by minor modifi cation to the sample isolation procedure ( 23 ), and to the charge-tagging procedure by using cholesterol oxidase from Brevibacterium , which also catalyzes the oxidation of 3 ␤ -hydroxy-5-ene steroids of the C 19 and C 21 series, rather than from Streptomyces , which prefers a longer steroid side-chain ( 33 ), and by loading greater quantities of sample on the LC column.
In earlier studies, we have shown that embryonic mouse brain contains appreciable amounts of 24S,25-epoxycholesterol (0.1-0.4 ng/mg at E11.5), confi rming an active mevalonate pathway during brain development ( 5 ). Interestingly, at E11.5, the levels of 24S,25-epoxycholesterol exceed those of the oxysterols enzymatically derived from cholesterol (<0.1 ng/mg) ( 5 ). A similar picture is seen here in newborn brain, in which the 24S,25-epoxycholesterol levels (1.123 ± 0.304 ng/mg) exceed other oxysterols, including 24S-hydroxycholesterol (0.510 ± 0.082 ng/mg) ( Fig. 1B ), the most-abundant oxysterol found in the adult. In the newborn animal, a wide range of oxysterols are identifi ed in brain ( Fig. 1B ; see supplementary Table II). This is also true in embryonic brain ( 5 ), but not in the adult, probably because of the predominance of 24Shydroxycholesterol, which tends to obscure the observation Luciferase assays were performed in the mouse substancia nigra-like cell line SN4741. Both LXR ␣ and -␤ have been shown to be expressed in developing brain ( 58,59 ), and we have previously shown 24S,25-epoxycholesterol, 22R-, 24S-, and 25-hydroxycholesterols to be LXR ligands in this cell line ( 32,58 ). 26-Hydroxycholesterol was not found to be an LXR ligand in this cell line. In the current study, we have extended our earlier work and show that desmosterol is an LXR ligand in SN4741 cells, but neither of the 26-hydroxydesmosterol isomers are ligands. Of the two oxo compounds tested, only 24oxocholesterol was found to be an LXR ligand. However, this is derived from 24S,25-epoxycholesterol during our derivatization reaction. Admittedly, in this cell line, the response to LXR activation by desmosterol is rather low, but it is found to be statistically signifi cant and reproducible. To confi rm the results of the luciferase assay, immunoblot analysis was performed against ABCA1, an LXR target protein (see supplementary information). Both 24-oxocholesterol and desmosterol increased the level of ABCA1 in SN4741 cells (see supplementary Fig. X). Although the 10 µM concentration of ligand used is high, the level of desmosterol in newborn brain of ‫ف‬ 500 ng/mg suggests that high concentrations could be reached at specifi c locations. of other oxysterols present at low levels. The explanation for the high 24S,25-epoxycholesterol-to-24S-hydroxycholesterol ratio in embryonic and newborn brain compared with the adult is likely to be multifactorial, but refl ects a high fl ow of metabolites through the mevalonate pathway, including the shunt pathway, in the developing brain, and a muchreduced fl ow rate in the adult ( 1,3 ). Furthermore, the low expression of Dhcr24 in the embryo and newborn animal ( 4 ) may divert sterol biosynthesis from the Bloch and Kandutsch-Russell pathways to the shunt pathway, which does not utilize Dhcr24 and generates 24S,25-epoxycholesterol. In addition, the expression of Cyp46a1, cholesterol 24hydroxylase, is low in fetal mouse. What then might be the consequence of high levels of 24S,25-epoxycholesterol in newborn brain? 24S,25-epoxycholesterol is an LXR ligand, and LXR ligands promote neurogenesis in developing brain ( 58 ). Thus, it seems possible that this oxysterol, which is particularly abundant in brain, is important for neurogenesis, and indicates yet another important function of an active mevalonate pathway. In addition, 24S, 25-epoxycholesterol is a ligand to Insig and may be involved in the fi ne-tuning of acute cholesterol homeostasis.
Desmosterol is particularly abundant in the fetus and newborn animal ( 4,54 ), in part because of the low expression of Dhcr24 , as discussed above. This leads to the possibility of oxysterol formation using this substrate. This has been shown to be possible both in vitro and in vivo ( 30,52 ). Here, we show that oxysterols derived from desmosterol are present in newborn mouse brain, with hydroxylation occurring in both the steroid nucleus and side-chain. Desmosterol, unlike its 26-hydroxy metabolites, is also an LXR ligand ( 52,56 ), and here, we show that it is a ligand to both the ␣ and ␤ isoforms of the receptor in a neural cell line. From the present study, it is impossible to determine the relative importance of the LXR ligands found in newborn brain. Although desmosterol is present in far greater abundance than 24S,25-epoxycholesterol, or any other LXR ligand, it is likely that the majority is trapped in cell membranes and unavailable to act as a signaling molecule. On the other hand, the greater cytoplasmic solubility makes 24S,25-epoxycholesterol a better candidate as a active LXR ligand in newborn brain.