Identification of unusual oxysterols biosynthesised in human pregnancy by charge-tagging and liquid chromatography - mass spectrometry

The aim of this study was to identify oxysterols and any down-stream metabolites in placenta, umbilical cord blood plasma, maternal plasma and amniotic fluid to enhance our knowledge of the involvement of these molecules in pregnancy. We confirm the identification of 20S-hydroxycholesterol in human placenta, previously reported in a single publication, and propose a pathway from 22R-hydroxycholesterol to a C27 bile acid of probable structure 3β,20R,22R-trihydroxycholest-5-en-(25R)26-oic acid. The pathway is evident not only in placenta, but pathway intermediates are also found in umbilical cord plasma, maternal plasma and amniotic fluid but not non-pregnant women.


Introduction
Amongst other functions, the placenta plays a key role in the transport of cholesterol from the mother to the fetus (1). The placenta is rich in cholesterol metabolising enzymes, particularly those involved in progesterone and estrogen synthesis (2). Hence, it should also be a site for oxysterol synthesis and further metabolism. Cytochrome P450 (CYP) 11A1 (also known as P450 SCC ) is the enzyme that generates pregnenolone from cholesterol via consecutive 22R-and 20R-hydroxylations followed by side-chain cleavage (3). Progesterone is then formed from pregnenolone by oxidation at C-3 and D 5 -D 4 isomerisation by hydroxysteroid dehydrogenase (HSD) 3B1 (See Figure 1). Although 22R-hydroxycholesterol (22R-HC) and 20R,22R-dihydroxycholesterol (20R,  are known intermediates in CYP11A1-mediated side-chain cleavage of cholesterol to give pregnenolone (4), few studies have explored the oxysterol profile of placenta (5,6). One important, but until now not replicated, finding made in the early 2000's was the presence of 20S-hydroxycholesterol (20S-HC) in human placenta (7). 20S-HC is an enigmatic oxysterol with many biological properties, but seldom reported in mammalian systems (8). Note, the actual stereochemical location of the 20S-hydroxy group in 20S-HC is the same as that of the 20R-hydroxy group in 20R,22R-diHC (see structures in Figure 1), but they are named 20S and 20R, respectively, according to rules of chemical priority. Similar to the situation with placenta, there are few reports of the oxysterol profiles of umbilical cord blood (5), i.e. blood of fetal origin that remains in the placenta and in the attached umbilical cord after childbirth, or of amniotic fluid, the fluid that acts as a cushion for the growing fetus and serves to facilitate the exchange of biochemicals between mother and fetus. Interestingly, however, oxysterols have been found as their sulphate esters in meconium, the earliest stool of a mammalian infant, including FIGURE 1 Proposed metabolism of cholesterol to the C 27 bile acid 3b,20R,22R-triH-D 24 -CA (green background). For comparison the acidic pathway to the C 27 bile acid 3b,7a-diH-D 24 -CA is shown (blue background) as is the pathway to progesterone (yellow background). Enzymes are indicated in blue. When shown in bold the enzymatic reactions can be found in the literature, when shown in normal typeface and with a broken arrow they are postulated. Cholesterol is shown in red with full stereochemistry and numbering system on a salmon background. For simplicity the 3b,20R,22R-triH-D 24 -CA and 3b,7a-diH-D 24 -CA are shown as the acids rather than the CoA-thioester products of SLC27A5 (bile acid CoA ligase). 20S-HC is shown on an orange background, although the enzyme responsible for its formation is unproven, it can be converted to 20R,22R-diHC by CYP11A1. Note the stereochemistry of metabolites down-stream of 20R,22R-diHC towards and including 3b,20R,22R-triH-D 24 -CA are assumed based on biosynthetic considerations. 20,22-diHC, 22-HC, 23-hydroxycholesterol (23-HC) and 24hydroxycholesterol (24-HC) (9,10).
Here we report liquid chromatography (LC)mass spectrometry (MS)-based methods for the identification of oxysterols in full term placenta, plasma derived from cord blood and pregnant female blood (maternal blood) and in mid-gestation amniotic fluid. The methods are based on high mass resolution MS with multistage fragmentation (MS n ) exploiting charge-tagging to enhance analyte signal (Supplemental Figure S1A).

Human material
Maternal blood was taken 24 -48 hr prior to elective caesarean section at 37+ weeks of gestation for reasons that did not include maternal or fetal anomaly. Umbilical cord blood was collected at delivery of the baby. Control plasma was from non-pregnant females. Amniotic fluid was obtained at 16 -18 weeks of pregnancy during diagnostic amniocentesis; only samples with no fetal chromosomal abnormality were used. All samples were collected with approval from an appropriate Health Research Authority Research Ethics Committee. All participants provided informed consent and the study adhered to the principles of the Declaration of Helsinki.

Sterol and oxysterol extraction 2.3.1 Placenta
Sterols and oxysterols were extracted from placental tissue using a modified protocol previously used to extract oxysterols from brain and liver tissue (12,13). Approximately 400 mg of tissue was cut from the maternal side of fresh placenta, weighed and washed three times in PBS to remove blood. The tissue was then transferred to a gentleMACS ™ C tube (Miltenyi Biotec, Woking, UK) followed by 4.2 mL of absolute ethanol containing 50 ng of [ 2 H 7 ]24R/S-HC and 50 ng of [ 2 H 7 ]22R-HCO. The tissue was homogenised for 2 min. The homogenate was transferred to a 15 mL corning tube and sonicated for 15 min. Whilst sonicating, 1.8 mL of HPLC grade water was added dropwise to give 6 mL of homogenate at 70% ethanol. The homogenate was then centrifuged for 1 hr at 4,000 x g. The supernatant was transferred to a fresh 15 mL corning tube and the remaining pellet re-suspended in a further 4.2 mL ethanol containing 50 ng of [ 2 H 7 ]24R/S-HC and 50 ng of [ 2 H 7 ]22R-HCO. The suspension was vortex mixed and transferred back into the original gentleMACS ™ C tube where it was then homogenised for a further 2 min. The homogenate was removed and sonicated for 15 min, then 1.8 mL of water added to give 6 mL of 70% ethanol. The supernatants from the two extractions were combined to yield 12 mL in 70% ethanol. This was mixed by vortex and sonicated for a further 10 min followed by centrifugation for 1 hr at 4,000 x g. Ten % (1.2 mL) of the total supernatant was added to 300 µL of 70% ethanol under sonication. The 1.5 mL of sample was subjected to solid phase extraction (SPE) by a procedure modified from an earlier protocol (12,13) to allow the collection of C 21 steroids besides sterols and oxysterols including sterol acids.
The sample from above was loaded onto a Certified Sep-Pak C 18 , 200 mg (3 cm 3 , Waters Inc. Elstree, UK) reversed-phase SPE column previously conditioned with ethanol (4 mL) followed by 70% ethanol (6 mL). The sample flow-through (1.5 mL) was combined with a column wash of 70% ethanol (5.5 mL) resulting in SPE1-Fr1 (7 mL) which contained oxysterols, sterol acids and C 21 steroids. A second fraction was obtained by further washing with 70% ethanol (4 mL) and collected as SPE1-Fr2. Cholesterol and other sterols of similar hydrophobicity were eluted from the SPE column with absolute ethanol (2 mL) to give SPE1-Fr3. A final fourth fraction was eluted with a further 2 mL of absolute ethanol (SPE1-Fr4). Each of the four fractions was divided equally into A and B sub-fractions and dried overnight under vacuum by centrifugal evaporation (ScanLaf ScanSpeed vacuum concentrator, Lynge, Denmark).
Each lyophilised sample was reconstituted in propan-2-ol (100 µL) and mixed thoroughly by vortex. To fractions (A), 50 mM K 2 HPO 4 buffer, pH 7 (1 mL) containing cholesterol oxidase solution (3.0 µL, 2 mg/mL in water, 44 units/mg of protein) was added. The samples were mixed by vortex and incubated at 37°C for 1 hr in a water bath. The reaction was then quenched by the addition of methanol (2 mL). Fractions (B) were treated in parallel in an identical fashion to fractions (A) but in the absence of cholesterol oxidase. Glacial acetic acid (150 µL) was added to fractions (A) and (B) and mixed by vortex. [ 2 H 5 ]Girard P (GP) reagent (11) (190 mg, bromide salt) was added to fractions (A) and [ 2 H 0 ]GP reagent (150 mg, chloride salt, TCI Europe, Oxford UK) was added to fractions (B). The samples were mixed by vortex until the derivatising reagent had dissolved. The reaction was left to proceed overnight at room temperature protected from light.
An OASIS HLB 60 mg (3 cm 3 ) SPE cartridge was washed with methanol (6 mL), 5% methanol (6 mL) and conditioned with 70% methanol (4 mL). Sample from above (3.25 mL, 69% organic) was loaded onto the column and the flow-through collected. The sample tube was rinsed with 70% methanol (1 mL) which was then loaded onto the SPE column, and the eluent combined with the earlier flowthrough. The column was re-conditioned with 35% methanol (1 mL) and the eluent combined with the earlier collection. The total eluent (~5 mL) was diluted with 4 mL of water to give~9 mL of 35% methanol. The 9 mL sample solution was loaded onto the column and the flow-through collected. The sorbent was re-conditioned with 17.5% methanol (1 mL) and the eluent combined with earlier flowthrough. To the combined 10 mL, water (9 mL) was added to give a 19 mL of 17.5% methanol. This solution was loaded onto the column and the flow-through collected once more. The sorbent was reconditioned with 8.75% methanol (1 mL) and the flow-through combined with the earlier collection. The total combined eluent of 20 mL was diluted with 19 mL of water to give 39 mL 8.75% methanol. The solution was loaded onto the column and the flow-through discarded. A 5% methanol solution (6 mL) was used to wash the column before the analytes were eluted. The samples were eluted into four separate 1.5 mL microcentrifuge tubes using 3 x 1 mL methanol followed by 1 mL ethanol to give SPE2-FR1, -Fr2, -Fr3, -Fr4. Oxysterols originating from SPE1-Fr1 elute across SPE2-Fr1 and SPE2-Fr2, cholesterol originating from SPE1-Fr3 elutes across SPE2-Fr1,-Fr2, Fr-3. Here we report data only for oxysterols and more polar metabolites.
Immediately prior to LC-MS analysis of oxysterols, equal volumes of SPE2-Fr1A and SPE2-Fr2A were combined with equal aliquots of SPE2-Fr1B and SPE2-Fr2B and diluted with water to form a solvent composition of 60% methanol.

Plasma
The extraction protocol for sterols and oxysterols was essentially that described previously (11,13,14), with minor modification to allow for extraction of C 21 steroids. 100 µL of plasma was added dropwise to a solution of acetonitrile (1.05 mL) containing 20 ng of [ 2 H 7 ]24R/S-HC and 20 ng of [ 2 H 7 ]22R-HCO in an ultrasonic bath with sonication. After a further 5 min of sonication, 350 µL of water was added. The sample (1.5 mL), now in 70% acetonitrile, was sonicated for a further 5 minutes and centrifuged at 17,000 x g at 4°C for 30 min. The sample was subjected to SPE and prepared for LC-MS analysis exactly as for the placental extract with the following modification: SPE1, Certified Sep-Pak C 18 , 200 mg, was conditioned with 70% acetonitrile rather than 70% ethanol.

Amniotic fluid
The protocol for extraction of sterols and oxysterols from amniotic fluid was exactly as that described for plasma except the internal standards were 7 ng of [ 2 H 7 ]24R/S-HC and 7 ng of [ 2 H 7 ]22R-HCO.

LC-MS(MS n )
Analysis was performed on a Dionex Ultimate 3000 UHPLC system (Dionex, now Thermo Fisher Scientific, Hemel Hempstead, UK) interfaced via an electrospray ionisation (ESI) probe to an Orbitrap Elite MS (Thermo Fisher Scientific). Chromatographic separation was carried out on a Hypersil Gold reversed phase C 18 column (1.9 µm particle size, 50 x 2.1 mm, Thermo Fisher Scientific, UK). Details of the mobile phase and gradients employed are given in Supplemental Materials and Methods. MS analysis on the Orbitrap Elite was performed in the positive-ion mode with five scan events, one high resolution (120,000 full width at half maximum height at m/z 400) scan over the m/z range 400 -610 in the Orbitrap and four MS 3 scans performed in parallel in the linear ion trap (LIT). Mass accuracy in the Orbitrap was typically < 5 ppm. More details of the scan events are provided in Supplemental Materials and Methods. Injection volumes were 35 µL for plasma extracts and at 90 µL for amniotic fluid and placental extracts.

Results
The aim of this study was to identify oxysterols and any downstream metabolites in placenta, cord plasma, maternal plasma and amniotic fluid to enhance our knowledge of the involvement of these molecules in pregnancy. For side-chain oxysterols expected to be present quantitative measurements were possible by reference to an isotope-labelled standard, for unexpected metabolites only semiquantitative data was obtained, however, this could be used for relative quantification between sample sets (Table 1).

Identification of oxysterols in placenta
The placenta is a blood-rich organ. The maternal side contains less vascular tissue than the fetal side and was selected for analysis. During sample preparation tissue was washed three times with PBS to remove blood.   D/NM, detected but not measured. ND, not detected. Authentic standards are available for each metabolite unless otherwise noted. An asterisk preceding a fragment describing letter indicates that the fragment-ion has lost the pyridine ring. A prime before the fragment describing letter indicates that the fragment-ion is deficient in a hydrogen atom compared to a similar fragment formed by a homolytic cleavage, A prime after the fragment describing letter indicates that the fragment-ion has gained a hydrogen atom compared to a similar fragment formed by homolytic cleavage. See Supplemental Figure S1C for examples. 1 Extraction of oxysterols was performed according to methods designed primarily for brain and liver without further validation.
2 Single outlier removed. 3 Ring-oxysterols were not the focus of this study, and their measurement is only semi-quantitative. 4 20S-HC and 24S-HC give chromatographic peaks that were not completely resolved so were not quantified. 5 Previous studies have shown [ 2 H 7 ]24R/S-HC to be a satisfactory internal standard for the quantification of side-chain oxysterols and steroid-acids [Yutuc E, et al., 2021, Anal Chim Acta 1154: 338259]. 6 7a,26-diHC and 7a,26-diHCO were not differentiated. 7 Presumptive identification, no authentic standard available. for the additional presence an intense pair of peaks corresponding to the syn and anti conformers of [ 2 H 5 ]GP-derivatised 22R-HC in the placental sample. Note syn and anti conformers are a consequence of GP-derivatisation at C-3 of the sterol A-ring (see Supplemental Figure S1B). 22R-HC is usually only a minor oxysterol in adult plasma/serum (14, 16) and is essentially absent in the NIST SRM 1950 plasma sample (representative of the adult population of the USA) illustrated here (15). The observation of 22R-HC in placenta is not surprising as CYP11A1, the enzyme which generates this oxysterol in the pathway from cholesterol to pregnenolone, is abundant in placenta (2, 17).  (18,19). In brief, a 3b-hydroxy-5-ene function in the parent structure, with no additional substitutions on the ring system, gives following cholesterol oxidase treatment, GP derivatisation and MS 3  A minor unknown pair of peaks were also observed eluting much later in the MRM chromatogram. Their MS 3 spectra suggests that these correspond to the 22S-epimer ( Figure 2D). This was confirmed by analysis of [ 2 H 7 ]22S-HC which gave an identical MS 3 fragmentation pattern and co-eluted with the endogenous molecule. In many LC-MS/MS studies oxysterols are identified by MRM where the transition is often non-specific (16,20). This provides high sensitivity but relies on chromatographic separation of isomers and co-elution with isotopic labelled standards (21). In an initial interpretation of the data presented in Figure 2A (upper panel), the peak at 6.90 min was assumed to be 24S-HC, presumably from contaminating blood. However, closer scrutiny of the chromatogram and relevant MS 3 spectra suggested that the earlier eluting peak 6.64 min was one of the syn or anti conformers of 20S-HC. 20S-HC has a characteristic MS 3 fragmentation spectrum with a major fragment ion at 327.2 (*e', Figure 2F and Supplemental Figure S2B Figure S3 for the MS 3 spectrum of 24S-HC). Fortuitously, as both 20S-HC and 24S-HC give syn and anti conformers following GP-derivatisation, the first peak of 20S-HC (12.73 min) is completely resolved from both peaks of 24S-HC and it is only the second peak of 20S-HC and the first peak of 24S-HC that partially co-elute, leaving the second peak of 24S-HC (16.04 min) completely resolved from 20S-HC.

Dihydroxycholesterols, trihydroxycholesterols, pregnenolone and progesterone
The second step in the conversion of cholesterol to pregnenolone by CYP11A1 is the generation of 20R,22R-diHC from 22R-HC (Figure 1). The RIC (m/z 555.4317 ± 5 ppm) for dihydroxycholesterols reveals two major peaks which appear at retention times, and give identical MS 3 Figure  S2E). Unsurprisingly, when a search was made for pregnenolone by constructing the appropriate RIC (m/z 450.3115) this steroid was evident as was progesterone (m/z 448.2959), its HSD3B1 oxidised metabolite, in the fraction not treated with cholesterol oxidase i.e. fraction B (Figures 3D, E). Besides 20R,22R-diHC and the presumptively identified 22,23-diHC, low levels of 7a,(25R)26dihydroxycholesterol (7a,26-diHC) were also found in placenta ( Figures 3A, C).
20R,22R-diHC extracted from placenta gives intense signals in LC-MS and based on this and on the additional presumptive identification of 22,23-diHC, it is likely that other hydroxylase activities are present in placenta besides those normally associated with CYP11A1, potentially resulting in the formation of trihydroxycholesterols (triHC). The RIC for triHC (m/z 571.4266 ± 5 ppm) reveals three major peaks ( Figure 3F upper panel). To tighten the search for hydroxylated metabolites of 20R,22R-diHC, MRM-like chromatograms were generated for the major side-chain cleavage fragment ions associated with the 20R,22R-diHC structure i.e.  Figure 3B). Again, three major peaks were evident in these chromatograms. Interrogation of the respective MS 3 Figure S3D) and 3b-hydroxycholest-5-en-(25R)26-oic acid (3b-HCA, Supplemental Figure S4) in placenta, and there is the possibility that 20R,22R-diHC may be a substrate for CYP27A1 and be metabolised via 20R,22R,26-triHC to the C 27 bile acid 3b,20R,22R-trihydroxycholest-5-en-(25R)26-oic acid (3b,20R,22R-triHCA) in placenta (Figure 1). The RIC appropriate for 3b,20R,22R-triHCA (m/z 585.4059 ± 5ppm) reveals two major and a minor peak ( Figure 4A) of which only the first gives an MS 3 spectrum compatible with a 3b,20R,22R-triHCA structure ( Figure 4B & Supplemental Figure S2J & K, cf. Supplemental Figure S2H, I). The similarity between the MS 3 spectra of 20R,22R-diHC, and the presumptively identified 20R,22R,26-triHC and 3b,20R,22R-triHCA can be visualised in Figure 4C where the low-middle m/z range of the three spectra are shown on the same m/z scale. While the low-middle m/z range provides evidence for the 20,22-dihydroxy structural motif (presumably 20R,22R-) the high m/z range ( Figure 4B) is indicative of a C-26 acid. Sterol acids show characteristic neutral losses from the [M-Py] + ion corresponding to the net loss of H 2 CO 2 + n(H 2 O), where n is the number of OH groups on the sterol beyond that derivatised at C-3 (19). In Figure 4B Figure S2L). Note, the third hydroxy group is the site of derivatisation. The other two chromatographic peaks give almost identical MS 3 spectra ( Figure 4D) presumably syn and anti conformers of a second isomer whose structure is not obvious from the MS 3 spectra, although fragment-ions at m/z 385, 397 and 471 are probably structurally significant.

Identification of oxysterols in cord and maternal plasma
To investigate if the placental oxysterols are transported to the fetus, umbilical cord plasma derived from umbilical cord blood was analysed for oxysterols. The data was compared to the oxysterol profiles in plasma from maternal blood, taken 1 -2 day before elective caesarean section and plasma from "control"  Figure 3, the 20R,22R-stereochemistry is assumed in 3b,20R,22R-triHCA and 3b,20R,22R-triH-D 24 -CA. non-pregnant females. As might be expected, the oxysterol profile of cord plasma resembles that of non-pregnant females, but with the additional presence of CYP11A1-derived oxysterols. 22R-HC is present in both cord and maternal plasma but is absent from controls ( Figures 5A, B, 6A-C; Table 1). If present, 20S-HC is at levels in cord, maternal and control plasma samples b e l o w t h e l i m i t o f d e t e c t i o n ( 0 . 1 n g / m L ) . T h e dihydroxycholesterol 20R,22R-diHC is the dominant oxysterol in cord plasma, it is also a major oxysterol in maternal plasma but is absent from control plasma (Figures 5C, D, 6D-F). Presumptively identified 20R,22R,26-triHC was near the limit of quantification in both cord and maternal plasma but was absent from controls ( Figures 5E, F, 6G, H). The presumptively identified C 27 bile acid 3b,20R,22R-triHCA was evident in cord plasma and just detected in maternal plasma but was absent from control plasma ( Figures 5G, H, 7A-C). Presumptively identified 3b,20R,22R-triH-D 24 -CA was only detected in cord plasma (Supplemental Figures S5A, B).
Besides the oxysterols discussed above, the profile of cord and maternal plasma was investigated for other oxysterols and sterol acids routinely found in adult plasma, this data is included in Table 1. While it was possible to make quantitative measurements for mono-and di-hydroxycholesterols thanks to the availability of authentic standards, the absence of standards for 20R,22R,26-triHC and 3b,20R,22R-triHCA means that the values determined for these metabolites are semi-quantitative, but by using the same internal standard for quantification across samples should give reliable measurements for relative quantification across the sample groups.
In cord plasma, but not plasma from pregnant and nonpregnant females, 22R-HCO was found but at a much lower level than 22R-HC. 20R,22R-diHCO was found in both cord plasma and plasma from pregnant females, but not non-pregnant females. Further details are reported elsewhere (23).

Identification of oxysterols in amniotic fluid
Amniotic fluid is derived from maternal plasma but also contains progressively more fetal urine as pregnancy continues. One of its functions is to facilitate the exchange of biochemicals between mother and fetus. Based on the data from analysis of placenta and cord plasma, it is reasonable to expect to find 22R-HC and its down-stream metabolites in amniotic fluid. As in cord and maternal blood 22R-HC ( Figures 6A-C), 20R,22R-diHC ( Figures 6D-F), 20R,22R,26-triHC ( Figures 6G, H) and 3b,20R,22R-triHCA ( Figures 7A-C) were identified and quantified in amniotic fluid (Table 1). Presumptively identified 3b,20R,22R-triH-D 24 -CA was also detected but not quantified (Supplemental Figures S5C, D).

Relative quantification between samples groups
As mentioned above although the measurement of some of the metabolites of 22R-HC is only semi-quantitative, relative values are likely to be accurate (Table 1). Shown in Figure 8 are plots displaying the quantities determined in the different samples for 22R-HC and its dominant metabolites. With respect to these metabolites, cord plasma is very different to control plasma from non-pregnant females with statistical differences also observed between cord and maternal plasma for 20R,22R-diHC and 3b,20R,22R-triHCA.

Quantification of other oxysterols
Besides the oxysterols discussed in the previous section, other oxysterols typically measured by the charge-tagging approach were also measured and are presented in Table 1 (11,14,19). Note the values for the oxysterols and sterol acids for which there is no authentic standard were quantified against [ 2 H 7 ]24R/S-HC and are only semi-quantitative values.

Discussion
In the current study we have investigated the oxysterol profile of placenta, cord plasma, maternal plasma, nonpregnant female plasma (control plasma) and amniotic fluid. In each of the pregnancy samples we identify metabolites derived from CYP11A1 which are essentially absent from non-pregnant females ( Figure 8; Table 1). There are two significant findings from the current study. Firstly, the rediscovery of 20S-HC and the discovery of 22S-HC in human placenta (7), and secondly the uncovering of a shunt pathway for 22R-HC metabolism to C 27 bile acids.
20S-HC is a controversial oxysterol as it has been detected in very few analytical studies (7,8,29,30) despite being biologically active in vitro. 20S-HC, like 22R-HC, is a ligand to the liver X receptors a and b (LXRa, LXRb) (31) and to the retinoic acid receptor-related orphan receptor g (RORg) (32), but unlike 22R-HC, activates the G protein-coupled receptor (GPCR) Smoothened (SMO), a key protein in the hedgehog signalling pathway, required for proper cell differentiation in the embryo (33, 34). 20S-HC also inhibits the processing of SREBP-2 (sterol regulatory element-binding protein 2) to its active form as the master transcription factor regulating cholesterol biosynthesis (35, 36), presumably by binding to INSIG (insulin induced gene) in a manner similar to other side-chain hydroxycholesterols (37). Recently, 20S-HC has been identified as a ligand to the sigma 2 (s2) receptor (38), also known as transmembrane protein 97 (Tmem97), which is expressed in the central nervous system (39), and has been suggested to be a chaperone protein for NPC1 (Niemann Pick C1), the lysosomal cholesterol transport protein (38). The enzyme required to biosynthesise 20S-HC has not been identified, although CYP11A1 has been reported to generate both 20-hydroxyvitamin D 3 and 20,22dihydroxyvitamin D 3 or 20,23-dihydroxyvitamin D 3 from vitamin D 3 (40,41). The high level of CYP11A1 in placenta (17), makes this a good candidate enzyme for biosynthesis of 20S-HC. Like 20S-HC, there are few reports of the detection of 22S-HC in biological systems (30), however, 22S-HC has been identified as the sulphate ester in human meconium (10), the earliest stool of a mammalian infant, and in the human cell lines HCT-15 and HCT-116 (42). Unlike 20S-HC and most other side-chain oxysterols, 22S-HC is not an LXR agonist (43), behaving more like an antagonist (44), neither does it activate the Hh signalling pathway through SMO (33).
22R-HC and 20R,22R-diHC are abundant oxysterols in cord plasma and placenta. 20R,22R-diHC, like 22R-HC and 20S-HC, is an LXR ligand and all three appears to have similar activating capacity (45). Although the primary function of the LXRs is considered to be the regulation of cellular cholesterol (46), LXRs also appear to have developmental functions, being required for the development of dopaminergic neurons in midbrain (47). In fact, LXRb also appears to have a protective role towards dopaminergic neurons, as the synthetic agonist GW3965 protects against the loss of dopaminergic neurons in a Parkinson's disease mouse model (48).
CYP11A1 is an inner mitochondrial membrane protein and catalyses the side-chain cleavage of cholesterol to pregnenolone. The intermediates in this reaction scheme i.e. 22R-HC and 20R,22R-diHC, bind more tightly to CYP11A1 and are converted to pregnenolone at a greater rate than cholesterol (49). It is generally considered that 22R-HC and 20R,22R-diHC remain in the active site until all three oxidation steps are complete (3), however, the abundance of 22R-HC, and the observation of 20R,22R-diHC, in cord plasma, maternal plasma and placenta in this study would argue that this is not always the case.
Pregnenolone is converted to progesterone by HSD3B1 which is localised in both mitochondria and the endoplasmic reticulum (50,51), and is highly expressed in placenta (52). Progesterone has many roles associated with the establishment and maintenance of pregnancy, including ovulation, uterine and mammary gland development and the onset of labour (53). Progesterone Concentration of 22R-HC and downstream metabolites. For each sample type: Control non-pregnant female plasma (plasma, n = 5); cord plasma (n = 14); maternal (pregnant female) plasma (n = 10); and amniotic fluid (n = 5). Concentrations of (A) 22R-HC, (B) 20R,22R-diHC, (C) 20R,22R,26-triHC and (D) 3b,20R,22R-triHCA were determined by LC-MS exploiting charge-tagging utilising GP derivatisation. Vales in (A) and (B) are quantitative (authentic standards available), those in (C) and (D) are semi-quantitative (authentic standards not available). The band represents the median where the whiskers extend to the most extreme upper and lower data points which are no more than 1.5 times the range between the first and third quartile. Non-parametric Kruskal-Wallis multiple comparisons test was used for comparison of data. *P < 0.05; **P < 0.01, ***P < 0.001 ****P < 0.0001. suppresses spontaneous uterine contractility during pregnancy and, in most mammals, a fall in systemic progesterone is required for the initiation of labour at term. However, in humans, labour occurs in the presence of elevated circulating levels of progesterone. Despite this, disruption of progesterone signalling by the progesterone receptor (PR) antagonist RU486 at any stage of pregnancy results in myometrial contractions and labour, strongly suggesting that reduced progesterone signalling is responsible for labour in women (54). In the current study we have uncovered a shunt pathway that operates in parallel to pregnenolone/progesterone biosynthesis in the placenta (Figure 1). Beyond 20R,22R-diHC we identified three trihydroxycholesterol isomers, one of which gives an MS 3 fragmentation pattern consistent with 20R,22R,26-triHC, and two dihydroxycholestenoic acid isomers one of which gives a fragmentation pattern we assign to 3b,20R,22R-triHCA. A downstream metabolite 3b,20R,22R-triH-D 24 -CA was also presumptively identified. Here we assign stereochemistry based on the assumption of 20R,22R-diHC being the precursor, but we await the chemical synthesis of these metabolites to definitively confirm their identification, this will be required whether LC-MS/MS or gas chromatography -MS is the identification method. However, their presence during pregnancy would define a new pathway of C 27 bile acid biosynthesis (Figure 1). Most of the metabolites of this pathway are also observed in cord plasma, maternal plasma and amniotic fluid (Table 1). Notably, the amniotic fluid samples were from 16 -18 weeks of gestation and the other pregnancy samples 37 + weeks indicating a pathway operational throughout pregnancy. CYP27A1 is the likely sterol hydroxylase which will convert 20R,22R-diHC to 20R,22R,26-triHC and on to 3b,20R,22R-triHCA. Like CYP11A1, CYP27A1 is an inner mitochondrial membrane enzyme and is expressed in placenta (6,25). Although the C 27 bile acid 3b,20R,22R-triHCA has not previously been identified, a C 27 bile acid with 22R-hydroxylation has been identified in a patient with Zellweger's syndrome (55). It should be noted that 20S-HC will also act as a substrate for pregnenolone formation via a CYP11A1 catalysed reaction (30), presumably via 20R,22R-diHC (56). Thus, a potential route for 20S-HC metabolism is through 20R,22R-diHC and on to C 27 bile acids.
How important is the 20R,22R-diHC shunt pathway? At present we can only speculate, but the LXR-activating capacity of 22R-HC and 20R,22R-diHC and the expression of both LXRa and b during mammalian development makes it tempting to speculate that 20R-HC, 20R,22R-diHC and also 20S-HC by activating LXR, and in the case of 20S-HC by binding to Smo and Tmem97, are important for development of the embryo (Figure 9). Interestingly Tmem97 is a SREBP target gene (57), meaning that 20S-HC is both a Tmem97 ligand and a regulator of its synthesis. Tmem97 is associated with NPC1 in the endosomal-lysosomal compartment linking it to cholesterol transport. Little is known about the biological activities of trihydroxycholesterols and trihydroxycholestenoic acids and it is unknown whether 20R,22R,26-triHC, 3b,20R,22R-triHCA FIGURE 9 Schematic showing some cholesterol metabolites identified in placenta and their interactions with protein receptors. 22R-HC, 22S-HC and 20S-HC are formed from cholesterol, 22R-HC by CYP11A1 which may also be the enzyme that catalyses the formation of 20S-HC. Oxysterols are shown on a light green background, steroids on a light blue background, bile acids on a brown background and enzymes on a dark green background. Nuclear receptors are shown on a purple background, GPCR on an orange background, sigma-2 receptor on a black background and INSIG on a mustard background. Blue arrows indicate a "process", red arrows a chemical reaction, T signifies inhibition of a process, arrows with a diamond arrowhead indicate activation of a receptor, and green oval arrowheads indicate catalysis. Pink double headed arrows link processes. and 3b,20R,22R-triH-D 24 -CA are simply inactive intermediates on the road to bile acids or biologically active molecules themselves. A final point of note, during the course of this study we found evidence that HSD3B1 can oxidise sterols as well as steroids. Further details are reported elsewhere (23).

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author. Datasets generated during this study are available in the Open Science Framework repository.

Ethics statement
The studies involving human participants were reviewed and approved by Local ethics committee at Swansea University Medical School. The patients/participants provided their written informed consent to participate in this study.

Author contributions
AD wrote the first draft of the manuscript, acquired, analysed and interpreted the data. EY analysed data and contributed to writing the manuscript by revising it critically for important intellectual content. CT provided essential materials and contributed to writing the manuscript by revising it critically for important intellectual content. YW conceived and designed the work, interpreted the data and contributed to writing the manuscript by revising it critically for important intellectual content. WG conceived and designed the work, interpreted the data and wrote the manuscript. All authors contributed to writing of the manuscript and approved the final version.

Funding
This work was supported by the Biotechnology and Biological Sciences Research Council (BBSRC, grant numbers BB/I001735/1, BB/N015932/1 and BB/S019588/1 to WG, BB/ L001942/1 to YW), the European Union through European Structural Funds (ESF), as part of the Welsh Government funded Academic Expertise for Business project (to WG and YW). AD was supported via a KESS2 award in association with Markes International from the Welsh Government and the European Social Fund.