Sterols and oxysterols in plasma from Smith-Lemli-Opitz syndrome patients

Smith-Lemli-Opitz syndrome (SLOS) is a severe autosomal recessive disorder resulting from defects in the cholesterol synthesising enzyme 7-dehydrocholesterol reductase (Δ7-sterol reductase, DHCR7, EC 1.3.1.21) leading to a build-up of the cholesterol precursor 7-dehydrocholesterol (7-DHC) in tissues and blood plasma. Although the underling enzyme deficiency associated with SLOS is clear there are likely to be multiple mechanisms responsible for SLOS pathology. In an effort to learn more of the aetiology of SLOS we have analysed plasma from SLOS patients to search for metabolites derived from 7-DHC which may be responsible for some of the pathology. We have identified a novel hydroxy-8-dehydrocholesterol, which is either 24- or 25-hydroxy-8-dehydrocholesterol and also the known metabolites 26-hydroxy-8-dehydrocholesterol, 4-hydroxy-7-dehydrocholesterol, 3β,5α-dihydroxycholest-7-en-6-one and 7α,8α-epoxycholesterol. None of these metabolites are detected in control plasma at quantifiable levels (0.5ng/mL).

The DHCR7 gene is encoded by nine exons, and over 100 mutations have been identified in SLOS patients [7]. Genotype-phenotype correlations are poor, although many missense mutations result in residual enzyme activity which is associated with a less severe phenotype [7]. SLOS has a high carrier frequency in Caucasians. In European populations the combined carrier frequency of two of the most common mutations c.964-1G > C (IVS8-1G > C) and p.W151X ranges from 1 to 2.3% [8]. Considering these numbers, the clinical incidence of SLOS (1:10,000-1:70,000 in Northern and Central European populations, 1:50,000 in the USA) is much lower than that predicted [5]. This is most likely due to several factors, including under-diagnosis of mild cases, and early prenatal pregnancy loss in severe cases. It is tempting to speculate that the high carrier frequency, particularly in populations from Northern and Central Europe, conveys a heterogeneous advantage [5]. 7-DHC is a precursor of vitamin D 3 (Fig. 1B), and increased vitamin D 3 levels in the skin could protect against vitamin D deficiency.
SLOS can be diagnosed biochemically based on increased 7-DHC in serum and tissues [9]. 7-DHC levels are typically more than 50fold elevated in SLOS cases, although there are equivocal cases of SLOS with serum 7-DHC levels just above normal levels [5]. Gene sequencing of DHCR7 is an alternative to biochemical analysis, but is limited by known pathogenic mutations.
Dietary supplementation with cholesterol to reduce de novo synthesis of 7-DHC and increase cholesterol levels is a standard treatment for SLOS. Dietary cholesterol supplementation is reported to improve behaviour [10], but as cholesterol does not pass the blood brain barrier (BBB), this improvement may be mediated by cholesterol metabolites, e.g. oxysterols, which can cross the BBB. Theoretically, statin therapy should also reduce 7-DHC biosynthesis and also tissue levels [11].
Although the underlying enzymatic defect in SLOS is well established there are likely to be multiple mechanisms responsible for SLOS pathology. For instance, cholesterol has numerous biological functions and substitution of 7-DHC for cholesterol, and 7-DHD for desmosterol, may alter physiochemical properties and function of membranes. Also 7-DHC, its isomer 8-DHC, their metabolites and 7-DHD analogues may have a direct toxic effect on cells [12]. Cholesterol is the precursor of steroid hormones and bile acids and dehydrocholesterol analogues of pregnenolone, pregnanetriol, dehydroepiandrosterone and androstenediol have been reported [13]. 7-DHC derived bile acid precursors have been reported to be formed in liver mitochondrial incubations from a rat model of SLOS, including 26-hydroxy-7-dehydrocholesterol (26-OH-7-DHC, cholesta-5,7-diene-3b,26-diol) and 26-hydroxy-8-dehydrocholesterol (26-OH-8-DHC, cholesta-5,8(9)-dien-3b,26diol) (Fig. 1C) [14]. Note, we use here the systematic nomenclature where a hydroxy group introduced to the terminal carbon of the sterol side-chain is at carbon-26 [15]. Unless stated otherwise, this is assumed to introduce R stereochemistry at carbon-25. Further metabolism remains to be fully elucidated, although Natowicz and Evans reported unusual bile acids in the urine of SLOS patients [16]. These results have not been confirmed by others. 26-OH-7-DHC and 26-OH-8-DHC have been reported to be present in plasma from SLOS patients at levels of 0.04-0.51 mM (16-204 ng/mL), the D 7 isomer being an inhibitor of sterol synthesis and a ligand to the liver X receptor a [17]. The mitochondrial enzyme, cytochrome P450 (CYP) 27A1, oxidises cholesterol to 26-hydroxycholesterol (26-OHC, cholest-5-en-3b,(25R) 26-diol) and it is likely that this is the mitochondrial enzyme which also oxidises 7-and 8-DHC to D 7 and D 8 analogues of 26-OHC (Fig. 1C) [18]. In a study of infants with SLOS, Björkhem et al. found reduced plasma levels of 24Shydroxycholesterol (24S-OHC, cholest-5-ene-3b,24S-diol), but increased levels of 26-OHC [19]. The reduced level of brain derived 24S-OHC was not surprising in light of the reduced abundance of its precursor, cholesterol, but the elevated level of 26-OHC was less easy to explain [19].
Historically, biochemical diagnosis of SLOS has been by gas chromatography (GC) À mass spectrometry (MS) based on 7-DHC levels in blood [9], although in recent years atmospheric pressure ionisation (API)-and liquid chromatography (LC)-MS methods have been adopted [20,[29][30][31]. In addition to 7-DHC, its metabolites, formed either enzymatically or non-enzymatically, have a potential as biochemical markers [20]. An advantage of profiling for 7-and/or 8-DHC metabolites is that their identity may reveal more details of the aetiology of the SLOS phenotype. In the current study we have exploited an LC-electrospray ionisation (ESI)-MS with multistage fragmentation (MS n ) approach for the profile-analysis of cholesterol, 7-DHC, 8-DHC and their oxysterol metabolites [32,33]. By using a charge-tagging approach analytical sensitivity is maximised (Supplementary Fig. S1A).

SLOS samples
Historical residual clinical plasma samples from SLOS patients were analysed along with samples from newly diagnosed patients and controls provided with written informed consent and institutional review board ethical approval and were collected according to the principles of the Declaration of Helsinki [29,32,34]. Data from two patient samples was previously reported in Ref. [34]. Additional control samples were those reported earlier in Ref. [35].

Analytical methods
Sterols and oxysterols were analysed by LC-ESI-MS n using a charge-tagging approach (enzyme-assisted derivatisation for sterol analysis, EADSA) described fully in [32,33] and in supplementary information. In brief, non-polar sterols including cholesterol, 7-DHC and 8-DHC were separated from more-polar oxysterols by reversed-phase solid phase extraction (RP-SPE). The separated fractions were individually treated with cholesterol oxidase to convert 3b-hydroxy-5-ene and 3b-hydroxy-5,7(or 8)diene to their 3-oxo-4-ene and 3-oxo-4,7(or 8)-diene equivalents, then derivatised with Girard P (GP) reagent ( Supplementary  Fig. S1A) to add a charged quaternary nitrogen group to the analytes which greatly improve their LC-ESI-MS and MS n response. When fragmented by MS 2 GP-tagged analytes give an intense [M-Py] + ion, corresponding to the loss of the pyridine (Py) ring (Fig. S1B), which can be fragmented further by MS 3 to give a structurally informative pattern ( Fig. S1C-J). Some sterols and oxysterols naturally contain an oxo group and can be differentiated from those oxidised to contain one by omitting the cholesterol oxidase enzyme from the sample work-up procedure [32]. No  [35] or if not measured in that study from a pool of 8 adult plasma samples analysed in this work. The absence of a metabolite in the controls is indicated by a double-headed arrow. 26-OH-7-DHC was present in the pooled plasma but below the limit of quantification (0.5 ng/mL). All data is for free sterols as a hydrolysis step was not carried out. *, P < 0.05; **, P < 0.01.   3 (534 ! 455 ! ) spectrum of 7-OC from the SLOS patient. All data is for free oxysterols as a hydrolysis step was not carried out. GP-derivatives give syn and anti conformers, some of which e.g. 7b-OHC are resolved, while others e.g. 7-OC give a single peak. hydrolysis step was carried out so free sterols and oxysterols were exclusively analysed.

Statistical analysis
All values are shown as mean (AEstandard error of mean). The unpaired two grouped two tailed Student's t-test was performed to asses significant differences.
There is evidence in the literature for enzymatically formed 24-OH-7-DHC and 25-OH-7-DHC from 7-DHC by CYP46A1 [22], and Xu et al. have identified the former compound in plasma from a rat model of SLOS [23,24]. Additionally, 26-OH-7-DHC and 26-OH-8-DHC, 4a-OH-7-DHC and 4b-OH-7-DHC have been identified plasma from SLOS patients [17,18,20]. We therefore searched for the presence of OH-7-DHCs and OH-8-DHCs in plasma samples from SLOS patients (Fig. 5A and B). We identified two chromatographic peaks with retention times and giving MS 3 spectra compatible with (i) either 24-OH-8-DHC (cholesta-5,8(9)-diene-3b,24-diol) and/or 25-OH-8-DHC (cholesta-5,8(9)-diene-3b,25diol), and (ii) 26-OH-8-DHC ( Fig. 5C and D). These compounds are present in SLOS plasma but absent from control samples. The earlier eluting peak may well be a composite of 24-and 25-OH-8-DHC, while the latter peak is predominantly 26-OH-8-DHC, but could contain a small amount of unresolved 26-OH-7-DHC. Identification of these components in the absence of authentic standards is facilitated by the MS 3 spectra recorded and by reference to the equivalent spectra of the DHC precursor molecules and to their hydroxycholesterol analogues. Chromatographic elution time provides another dimension for identification with the presence of an additional double bond reducing elution time by about 0.5 min. With our chromatographic system we cannot resolve 4a-from 4b-OH-7-DHC. However, we do find a compound eluting with the appropriate retention time (Fig. 5B) and giving an MS 3 spectrum identical to that of the authentic standard ( Fig. 5G and H). This compound was at a level below our limit of quantification (0.5 ng/mL).

Non-enzymatically derived oxysterols
There is always considerable debate whether non-enzymatically formed oxysterols are generated in vivo or ex vivo during sample handling or storage [37,42]. The analytical protocol used in this work essentially eliminates ex vivo autoxidation during sample work-up by separating polar oxysterols from non-polar sterols like cholesterol, 7-DHC and 8-DHC. However, the possibility of autoxidation during sample collection and storage is difficult to eliminate when dealing with clinical samples from patients, especially when historical samples are analysed, as in the present study.
DHCEO is the major metabolic product of 7-DHC in SLOS fibroblasts generated through free radical oxidation (Fig. 1E) [27] and it has been identified in brain from a Dhcr7 knock-out mouse model of SLOS [25,26]. We thus searched for the presence of DHCEO in plasma from SLOS patients. Shown in Fig. 6 is the appropriate chromatogram for dihydroxycholestenones. DHCEO elutes late in the chromatogram and was not found in plasma from all SLOS patients. However, it was not detected in any control plasma (Fig. 2D).
As is evident from Figs. 5A and 6A there numerous other peaks present in the SLOS plasma which are absent from controls. Although their MS 3 spectra were recorded, their identification was not obvious.   [37]. Both panels are plotted on the same scale. MS 3 (550 ! 471 ! ) spectra of (B) 7a,25-dihydroxycholest-4-en-3one, (C) 7a,26-dihydroxycholest-4-en-3-one, (D) DHCEO and (E) authentic DHCEO. Oxysterols were analysed as GP-derivatives. All data is for free oxysterols as a hydrolysis step was not carried out.

Discussion
In the current study we are able to identify 24-or 25-OH-8-DHC and 26-OH-8-DHC at elevated levels in plasma of SLOS patients (Fig. 2D). This was found for each of the SLOS samples analysed. These molecules were not detected at quantifiable levels in control plasma (<0.5 ng/mL). Wassif et al. have previously identified 26-OH-8-DHC in SLOS plasma in the range of 16-204 ng/mL [17], somewhat higher than the current range of 1.41-15.75 ng/mL, but neither 24-or 25-OH-8-DHC have previously been found in human plasma. We also confirmed the earlier findings of elevated levels of 7-OC and 7,8-EC in SLOS plasma [20,34]. 7,8-EC was not found in control plasma and was detected in seven of the ten SLOS patients studied. 7-OC can be converted to 7b-OHC by HSD11B1 (Fig. 1E) [40,41] and every SLOS patient showed elevated 7-OC and/or 7b-OHC. In the present study we do not have data on disease severity so we are not able to correlate metabolite levels with SLOS severity. We also identified 4-OH-7-DHC in some SLOS samples, but only at the limit of detection (0.1 ng/mL), below the limit of quantification (0.5 ng/mL). 4a-and 4b-OH-7-DHC have previously been identified in SLOS plasma [20].
In an earlier study Björkhem et al. found reduced plasma levels of 24S-OHC in SLOS patients, this was readily explained by reduced cholesterol content of brain, the source of this metabolite [19]. 24S-OHC levels are known to vary with age, and in the current study we were not able to age match SLOS patients with controls so differences between SLOS and controls are likely lost in the case of 24S-OHC. In contrast to the previous study by Björkhem et al. we found lower levels of 26-OHC in SLOS plasma than controls [19]. The difference is likely to be methodological as in the earlier study total 26-OHC was measured following alkaline hydrolysis of sterol esters while here only free sterols were measured. 7-DHC is known to readily undergo free radical oxidation [23], one of the products of which, DHCEO (Fig. 1E), has been found in rodent models of SLOS. Here we were able to identify DHCEO in four of our ten SLOS samples. DHCEO is not detected in control plasma.
One of the SLOS patient samples investigated in this study showed an almost normal DHC to cholesterol ratio. Unfortunately, there was limited clinical information available relating to this patient. However, the pattern of plasma oxysterols from this patient clearly identifies SLOS. Of particular note were the high levels of 7b-OHC (24.46 ng/mL cf. 0.92 AE 0.49 ng/mL) and 7-OC (130.71 ng/mL cf. 3.86 AE 1.91 ng/mL) in plasma compared to controls. Further confirmation of SLOS was provided by the presence of elevated 26-OH-8-DHC (1.41 ng/mL cf. <0.5 ng/mL) in the patient plasma.
In summary, we have identified a number of metabolites derived from 7-or 8-DHC in SLOS plasma. Further studies will be directed at investigating how their values vary with disease severity and their merit as markers for disease stratification.