Development and validation of a liquid chromatography-tandem mass spectrometry assay to quantify plasma 24(S)-hydroxycholesterol and 27-hydroxycholesterol: A new approach integrating the concept of ion ratio

Accurate quantification of 24(S)-hydroxycholesterol and 27-hydroxycholesterol holds substantial biological significance due to their involvement in pivotal cellular processes, encompassing cholesterol homeostasis, inflammatory responses, neuronal signaling

The central nervous system (CNS) contains approximately 25% of the human body cholesterol.Cholesterol is not able to cross the blood-brain barrier (BBB) and the brain relies on its own biosynthesis [6].Brain cholesterol is synthesized by astrocytes and then transported to neurons where it is oxidized into 24(S)-OHC by CYP46A1.Subsequently, the latter crosses the BBB to reach the peripheral circulation.Almost all plasma 24(S)-OHC is derived from the CNS making it an appealing biomarker of brain cholesterol homeostasis [6].Furthermore, 27-OHC, being the most abundant oxysterol in human plasma, is primarily synthesized by hepatic cells and can traverse the BBB, thereby entering the brain from the systemic circulation [7,8].Thus, there is a crosstalk of these oxysterols with fluxes through the brain in opposite directions [7].

Method development 2.2.1. Sample preparation
Human EDTA-K2 plasma was obtained from four healthy volunteers (at Centre intégré universitaire de santé et de services sociaux de l′Estrie -Centre hospitalier universitaire de Sherbrooke) after centrifugation for 5 min at 2100 g and immediately stored frozen at − 80 • C (the lag time between blood draw and plasma storage at − 80 • C was up to 90 min The sample preparation was performed following the methodology previously described by McDonald et al., 2012 [29].A volume of 200 µl plasma was used a result of a thoughtful process informed by various factors.This volume aligns with the established laboratory practices where it is routinely employed for quantifying diverse analytes in clinical applications.Additionally, the suitability of our Agilent 6460 machine for this volume parameter enhances its appropriateness.Moreover, our selection finds endorsement in earlier studies focused on plasma oxysterols [29,43], where the 200 µl plasma volume was similarly chosen.20 µl of internal standards (IS) mixture (27-OHC-d6 and 24 (R/S)-OHC-d7 at 24.5 µM (10 ng/µl) each dissolved in isopropanol) were added to of plasma.Initially, we performed two consecutive lipid extractions with 3 ml of DCM: MeOH (1:1) containing 50 µg/ml of BHT.A nitrogen stream was applied for several seconds to remove oxygen and the tube was capped.Then, samples were vortexed for 30 s and incubated in an ultrasonic bath at 30 • C for 10 min, and centrifuged at 2200 g for 5 min.The organic layer obtained from both extractions were combined.Then, we performed the hydrolysis with 300 µl of 10 N KOH followed by 90 min of incubation at 37 • C.After hydrolysis, 3 ml of DPBS was added to the extraction tube, vortexed for 30 s, and centrifuged at 2200 g for 5 min.The organic layer was transferred to a clean glass tube and set aside.Afterward, 3 ml of DCM was added and mixed to the aqueous layer and centrifuged at 2200 g for 5 min.The organic phase was transferred to the previously collected organic phase.The latter was dried under nitrogen flow.The oxysterols were suspended in a solution containing 50 µl of HPLC-water with 5 mM ammonium formate and µl of MeOH with 5 mM ammonium formate, placed in an ultrasonic bath at 30 • C for 5 min, then, transferred into another tube and centrifuged at 16,200g for 5 min.Finally, samples were transferred in autosampler vials and stored at 4 • C until instrumental analysis.

Liquid chromatography (LC)
For chromatographic separation, a UPLC instrument (1290 Infinity, Agilent, Santa Clara, CA) with a Luna Omega Polar C18 column (2.1 × 100 mm, particle size 1.6 µm, Phenomenex, Torrance, CA) was used.The injection volume was 20 µl.Mobile phases were (A) HPLC-water with mM ammonium formate and (B) MeOH with 5 mM ammonium formate.Isocratic separation of oxysterol at 80% mobile phase B was achieved at a flow rate of 0.3 ml/min with column temperature at 30 • C. The total run time was 17.5 min, and the complete mobile phase program can be found in Supplementary Table 1.

Mass spectrometry (MS/MS)
The quantification of oxysterols was carried out on a triple quadrupole mass spectrometer instrument equipped with an Electrospray Ionization (ESI) source (6460 MS/MS with ESI jet stream, Agilent, Santa Clara, CA).The ESI parameters were optimized for both oxysterols as follows: gas and sheath gas temperatures were respectively 100 • C and 200 • C, gas flow and sheath gas flow were 12 L/min and 4 L/min; nebulizer pressure was 15 psi; capillary and nozzle voltages were 5500 V and 2000 V, respectively.
The Multiple Reaction Monitoring (MRM) acquisition mode was employed in this study.For each quantified oxysterol, specific transitions were designated as quantifier (Q), qualifier (q), and internal standard (IS).These transitions were carefully selected and optimized to ensure both high sensitivity and specificity, as shown in Table 1.The MassHunter Workstation Quantitative Analysis software (Version B.10.0.707.0,Agilent Technologies) was used for data acquisition and processing.

Validation studies
The choice of guidelines was based on established practices and standards in Canada where the CLSI [44,45] and CSA Z316.8-18 [46], are commonly followed for clinical validation purposes.

Linearity
An eight-point calibration curve, ranging from 20 to 300 nM (20 nM, 30 nM, 40 nM, 50 nM, 75 nM, 100 nM,150 nM, and 300 nM) was prepared for 24(S)-OHC and 27-OHC in 1% BSA in 0.9% NaCl.The linearity was assessed by analyzing one calibration curve per day, for five consecutive days.The solutions went through the same process as human samples.The data were fit to a weighted least squares linear regression model that used to calculate the equation relating the peakarea ratio of each oxysterol versus IS and the concentration of each analyte in the calibrators.Furthermore, the inverse of the concentration (1/x) was used as weighting factor.
A coefficient of Determination R 2 ≥ 0.995 and a standard error of the estimate for each calibration point ≤ 15% were considered as acceptance criteria [44].
D. Rojas et al.

Analytical measurement range, precision, and accuracy
Five concentrations of 24(S)-OHC (20 nM, 30 nM, 40 nM, 75 nM et 100 nM) and 27-OHC (20 nM, 30 nM, 50 nM, 75 nM et 150 nM) prepared in BSA 1% in 0.9% NaCl, were analyzed in quintuplicates for 5 consecutive days.Precision and reproducibility were verified by calculating the intra and inter-day coefficient of variation (CV).Accuracy was assessed by calculating the intra and inter-day bias.The lower limit of quantification (LLOQ) was defined as the lowest concentration obtained with a signal-to-noise ratio (S/N) > 10, a CV < 20%, and a bias < 15%.The acceptance criteria for all other concentrations were a CV < 15% and a bias < 15% [44].The S/N ratio was determined following Agilent MassHunter Workstation Software Quantitative Analysis.

Extraction efficiency and matrix effects
The spiking concentrations used for the following validation parameters correspond to highest concentrations reported, respectively, for 24(S)-OHC and 27-OHC in healthy individuals [47].

Carryover
To determine carry-over, three samples were prepared using BSA 1% in 0.9% NaCl: a blank (B), a mixture of 24(S) and 27-OHC at low concentration of each 20 nM (L), and a mixture of 24(S) and 27-OHC at high concentration of 300 nM each (H)\.The first experiment aimed to determine the carry-over of the low-concentration sample (L) when injected immediately after the high-concentration sample (H).Specimens were injected in the following order: (LLLHHLHHLLLLHHLHHLHHL). Carryover was non-significant if the result of the L obtained after the injection of the H was less than 3 standard deviation of the low results [48].The second experiment aimed to determine the contamination of the blank when injected immediately after the high-concentration sample.Specimens were injected in the following order:(BBLLLLHHBHHBHHBHHBBB.Carryover was non-significant if the result of the B obtained after the injection of the H was less than 25% of the mean of LLLL results [48].The data from the first experiment were analyzed using EP Evaluator (version 12.2.0.7,Data Innovations, Inc.South Burlington, VT, USA).

The chromatographic resolution (Rs).
The chromatographic resolution (Rs) was calculated as follows: the difference in retention times between two peaks * 2)/ (half widths sum of the two peaks * 1.7.An Rs ≥ 1.25 was considered as an acceptable separation [44].

Ion ratio.
Using the data acquired from the precision experiment, the intra and inter-day CV of the ion ratio for both 24(S)-OHC and 27-OHC were calculated and CV < 15% was considered acceptable.Moreover, the ion ratio was specific for different oxysterols using the same transitions.A standard of 25-OHC at 75 nM (high concentration) was spiked into a mixture of 24(S)-OHC and 27-OHC (30 nM each, low concentrations) in BSA 1% in 0.9% NaCl to determine the peak resolutions [44].
According to CLSI guidelines (C50-A) [45], the ion ratio from patient samples should not alter by ±20% from that of the mean ion ratio of calibrators.An alteration of the ion ratio in patient samples is indicative of an interferent.To test the effect of cholesterol in generating (S)-OHC and 27-OHC, a pool of human plasma was spiked with cholesterol 5 mM and 10 mM, prior extraction process.The plasma concentrations and ion ratio of 24(S)-OHC and 27-OHC were determined in unspiked and spiked samples.

Application of the method
Our validated method was used to quantify 24(S)-OHC and 27-OHC in a cohort of 20 healthy controls 14 males and 6 females aged from to 42 years.Fasting plasma were prepared and analysed following the described protocol.All specimens were measured in the same day using the same calibration curve.The written informed consent was obtained from each individual.

Linearity
To evaluate linearity, one calibration curve per day was analyzed for five consecutive day, detailed results of the calibration curve are shown in Supplementary Table 2.The linear regression plots for 24(S)-OHC and 27-OHC are respectively shown in Fig. 1A and B. The method was linear for the range of 20-300 nM, with an R 2 > 0.995 for both oxysterols.Difference plots of the bias were up to 12.6% and 14.1% for each calibration point of 24(S)-OHC and 27-OHC, respectively (Supplementary Fig. 1A and B).

Analytical measurement range, precision, and accuracy
The lowest concentrations to be reliably quantified was 20 nM for both 24(S)-OHC and for 27-OHC.Intra-day and inter-day CVs and biases of 24(S)-OHC and 27-OHC are shown in Table 2 and Supplementary Table 3.The inter-day CVs were lower than 8.5% for each tested concentration of 24(S)-OHC and 27-OHC.The inter-day biases were up to 9.6% and -8.1% for 24(S)-OHC and 27-OHC, respectively.Analytical sensitivity data showed an S/N of at least > 35 for each tested concentration for both oxysterols.

Carryover
The results for carryover are shown in Supplementary Table 4.The carryover of the low-high results was less than three standard deviations from the low-low results.In the blank samples, the concentration of 24 (S)-OHC and 27-OHC, was found to be less than 25% of the mean value of the low-low results.The calculated values for 24(S)-OHC and 27-OHC in the blank samples were 20.6% and 20.4% respectively, compared to the mean values obtained from the low concentration samples that indicates a low background signal and minimal interference in the blank samples.

Interference
The reference ion ratio value was calculated by averaging the ion ratios of calibration solutions.The ion ratio obtained for 24(S)-OHC was 192.5 with a CV < 7.9% for all tested concentrations.The ion ratio obtained for 27-OHC was 4.6 with a CV < 14.2% for concentrations > 20 nM as shown in Supplementary Table 5.
To verify if ion ratios of 24(S)-OHC and 27-OHC are able to identify the presence of an interferent analyte, the respective ion ratios were recalculated using the MRM transitions of each other.The ion ratio of 27-OHC using MRM transitions of the 24(S)-OHC was 8085, while the ion ratio of 24(S)-OHC using MRM transitions of the 27-OHC was 5.2.The ion ratio of 25-OHC using MRM transitions of 24(S)-OHC and 27-OHC were 34.4 and 18.9, respectively.The 25-OHC was separated from 24(S)-OHC and 27-OHC with Rs values of 1.79 and 3.04, respectively (Fig. 2).
MRM transitions of 24(S)-OHC were able to identify more specifically a related oxysterol as compared to MRM transitions of 27-OHC.No chromatographic peak was identified for the evaluated oxysterols including 22(R)-hydroxycholestérol, 7α-hydroxycholestérol, 7βhydroxycholestérol and 5α,6α-epoxycholestanol at the respective transition of Q and q for both 24(S)-OHC and 27-OHC as shown in Supplementary Table 6.
Concentrations of 24(S)-OHC and 27-OHC in unspiked specimens remained comparable to spiked ones.Ion ratios of endogenous 24(S)-OHC and 27-OHC remained lower than 20% of the mean of ion ratios' calibrators (means were 184.3 for 24(S)-OHC and 4.3 for 27-OHC) as shown in Supplementary Table 7.

Application of the method
Plasma levels of 24(S)-OHC and 27-OHC for 20 healthy individuals were respectively 88.87 ± 27.49 nM and 264.00 ± 64.20 nM.The distribution of results is shown in Supplementary Fig. 2. The ion ratio observed in patient samples remained within a ±20% range of the mean ion ratio of the calibrators as shown in Supplementary Table 8.

Discussion
Several LC-MS/MS methods have been previously described for the quantification of 24(S)-OHC and 27-OHC, however, a limited number of them performed a formal method development and validation based on CLSI Guidelines C62-A [44].Most of the reported methods used derivatization to enhance the sensitivity, despite the fact it can be a laborious, time-consuming procedure that might lead to interference problems related to the derivatization process.Here, we developed a sensitive LC/ESI-MS/MS method for the detection and quantification of total plasma 24(S)-OHC and 27-OHC, without derivatization to simplify the sample preparation.
Previous methods determined the ion ratio of the quantifier and qualifier transitions; however, they did not perform a validation assay [30,31,49].It's important to highlight that the ion ratio has never been utilized for the validation of oxysterols quantification.Monitoring of ion ratio is used as an additional confirmation criterion that provides confidence in reporting results, avoiding false positive [50] Considering the presence of other oxysterols with very similar structures, but different physiological roles, it is recommended to determine and validate the ion ratios for 24(S)-OHC and 27-OHC.This is the first LC/ESI-MS/MS method that determined consistent ion ratios with CVs < 15% for 24(S)-OHC and 27-OHC at concentrations of ≥ 20 nM and ≥ 30 nM, respectively.We first identified specific ion ratios for both 24-OHC and 27-OHC.Our method was assessed for its specificity and it was shown that the following potential interferences 25-OHC, 22(R)-hydroxycholesterol, 7α-hydroxycholesterol, 7β-hydroxycholesterol, and 5α,6α-epoxycholestanol did not affect the quantification nor the ion ratio of the 24-OHC and 27-OHC.Previously, 25-OHC had been considered an interfering analyte for the quantification of 24 (S)-OHC and 27-OHC in LC/MS/MS [51], and their chromatographic separation has proven challenging [28,29,36,40,43].
We should note that ion ratio of 27-OHC was significantly different at low concentration of 20 nM.Although we can precisely and accurately quantify 27-OHC at 20 nM, we can exclude the presence of interferants for concentrations ≥ 30 nM.It has been shown that ion ratio might be affected by the analyte concentration [52].The variation in ion ratio is more pronounced at lower concentrations compared to higher ones.At low concentrations, the S/N of the qualifier transition is too low, which significantly impacts the calculation of the ion ratio (quantifier transition / qualifier transition).Moreover, matrix effects did not exert an influence on the ion ratio.The ion ratios remained consistent whether the surrogate matrix or the plasma matrix was used to determine 24-OHC and 27-OHC.This stability is attributed to the intrinsic and reproducible pattern fragmentation of each individual molecule when

Table 2
Intra and inter-day CV, inter-day bias, and S/N data for 24(S)-OHC and 27-OHC (n = 25).submitted to a specific energy (collision-induced dissociation; collision cell voltage with the inert gas); a process almost never affected by the matrix.Briefly, the ion ratio for 24-OHC and 27-OHC are analyte specific, less consistent at low concentrations, and not affected by the matrix.Mass spectrometers calculate ion ratio for each sample run, and the ion ratio falling outside the acceptability criteria will be flagged.An alteration of the ion ratio in patient samples will suggest the presence of an interferent and will lead to further investigation.Borah et al. [35] developed an LC-ESI-MS/MS method using a nucleoside C18 column (2 mm, particle size 5 µm) and obtained a chromatographic separation of 24(S)-OHC, 25-OHC, and 27-OHC with retention times of 15.98 min, 16.34 min, and 17.05 min, respectively.However, they did not report the Rs, a reference parameter used to assess the separation of two peaks.Here we used a column with a smaller particle size (1.6 µm vs 5 µm) to increase the column efficiency.Indeed, the differences in column characteristics might explain the optimal peak resolution (Rs ≥ 1.25) of three oxysterols we obtained with our method.
Due to the substantial difference in abundance between cholesterol and oxysterols, a minor degree of cholesterol autoxidation can generate oxysterols of greater abundance than those present endogenously.Therefore, we tested the impact of high cholesterol concentrations in generating our oxysterols of interest.No significant effect was observed in the plasma concentration of both 24(S)-OHC and 27-OHC.Several factors might prevent the autooxidation of cholesterol.For instance, blood was collected in EDTA tubes.The EDTA is an agent that quenches metal ions that might be involved in the autoxidation process [36] In addition, BHT (added to calibration solutions, quality control, and plasma samples) is an antioxidant agent that prevents auto-oxidation [53].Moreover, nitrogen purging, considered as an effective oxygen removal method, was an important step of sample preparation.Finally, C24 and C27 seem to be minor sites of cholesterol autooxidation.The 24 (S)-OHC and 27-OHC are mainly biosynthesized by enzymes, as opposed to 5,6-epoxides and 7-hydroxycholesterols that are autoxidation products and might be formed during sample preparation, prior to analysis [54].
In this method, we established a quantification limit of 20 and 30 nM for 24(S)-OHC and 27-OHC.It should be noted that previous studies established an LLOQ for plasma 24(S)-OHC and 27-OHC.Borah and colleagues [35] developed an LC-ESI-MS/MS method to quantify free oxysterols in different types of cell lines as well as PBMC.Following six independent measurements (two replicates for each), they determined an LLOQ of 11.7 nM (0.005 ng/µl) for both, 24(S)OHC and 27-OHC, with a CV < 20%, still without reporting the S/N parameter.Burkard and colleagues [36] developed an HPLC-MS method to quantify total plasma oxysterols.They determined an LLOQ of 99 nM (40 μg/l) for total plasma 24(S)-OHC and 62 nM (25 μg/l) for total plasma 27-OHC, with S/N > 3.Here we determined the LLOQ according to CLSI guidelines C62-A [44] based on more replicates (N = 25; 5 concentrations; 5 replicates, 5 days), and considering bias < 15%, CV < 20% and S/N > 10.The determination of LLOQ might be important in certain conditions associated with low oxysterol levels such as Smith Lemli Opitz [10], Multiple sclerosis [13], etc.
Our method showed adequate precision with inter-day CVs ranging up to 7.3% and 6.4%, for 24(S)-OHC and 27-OHC, respectively.Previous methods reported an inter-day CV < 10% for both oxysterols based on the results of up to six replicates of two different concentrations [28,36].The precision of our method relied on the results of 25 replicates of 5 different concentrations (for each oxysterol), which is in accord with CLSI guidelines [44].Recent research is focused on the implication of both, 24(S)-OHC and 27-OHC in neurocognitive or cancer diseases [55][56][57].and the use of a precise analytical method provides reliable measurements that certainly are important to draw valid conclusions.We verified the previously reported reference interval by collecting and analysing plasma specimens of 20 healthy controls from the local population.100% and 80% of tested subjects' values of 27-OHC and 24 (S)-OHC, respectively, felt inside the range of the previously reported interval [29,47,[58][59][60].There is thus a need to develop our own reference intervals, particularly for 24(S)-OHC, that are specific to our quantification method and local population.
In conclusion, we developed and validated an LC/ESI-MS/MS method without derivatization to quantify the 24(S)-OHC, and the 27-OHC with optimal precision.We improved the separation of these analytes and determined and validated specific ion ratios as additional quality control measures in the analysis of these oxysterols.

Fig. 1 .
Fig. 1.Linear Regression Analysis for 24(S)-OHC and 27-OHC.The linear regression plots demonstrate the obtained results versus theoretical value for 24(S)-OHC (A) and 27-OHC (B).The linear regression equation and the coefficient of determination (R 2 ) are indicated for each analyte.

Table 1
Mass spectrometry parameters optimized for each MRM transition.
D.Rojas et al.